Developments in Geochemistry 9

High-Pressure Geochemistry
and Mineral Physics

Basics for Planetology

and Geo-material Science
Developments in Geochemistry
1. W.S. Fyfe, N.J. Price and A.B. Thompson
FLUIDS IN THE EARTH’S CRUST

2. P. Henderson (Editor)
RARE EARTH ELEMENT GEOCHEMISTRY

3. B.A. Mamyrin and I.N. Tolstikhin

HELIUM ISOTOPES IN NATURE

4. B.O. Mysen
STRUCTURE AND PROPERTIES OF SILICATE MELTS

5. H.A. Das, A. Faanhof and H.A. van der Sloot

RADIOANALYSIS IN GEOCHEMISTRY

6. J. Berthelin
DIVERSITY OF ENVIRONMENTAL BIOGEOCHEMISTRY

7. L.W. Lake, S.L. Bryant and A.N. Araque-Martinez

GEOCHEMISTRY AND FLUID FLOW

8. N. Shikazono
GEOCHEMICAL AND TECTONIC EVOLUTION OF
ARC-BACKARC HYDROTHERMAL SYSTEMS
 Developments in Geochemistry 9
High-Pressure Geochemistry
and Mineral Physics

Basics for Planetology

and Geo-material Science
By

Sachinath Mitra

2004

Amsterdam – Boston – London – New York – Oxford – Paris

San Diego – San Francisco – Singapore – Sydney – Tokyo
ELSEVIER B.V. ELSEVIER Inc. ELSEVIER Ltd ELSEVIER Ltd
Radarweg 29 525 B Street, Suite 1900 The Boulevard, Langford Lane 84 Theobalds Road
P.O. Box 211, 1000 AE San Diego, CA 92101-4495 Kidlington, Oxford OX5 1GB London WC1X 8RR
Amsterdam, The Netherlands USA UK UK

q 2004 Elsevier B.V. All rights reserved.

This work is protected under copyright by Elsevier B.V., and the following terms and conditions apply to its use:

Photocopying
Single photocopies of single chapters may be made for personal use as allowed by national copyright laws. Permission of the Publisher and
payment of a fee is required for all other photocopying, including multiple or systematic copying, copying for advertising or promotional
purposes, resale, and all forms of document delivery. Special rates are available for educational institutions that wish to make photocopies
for non-profit educational classroom use.

Permissions may be sought directly from Elsevier’s Rights Department in Oxford, UK: phone (+44) 1865 843830, fax (+44) 1865 853333,
e-mail: permissions@elsevier.com. Requests may also be completed on-line via the Elsevier homepage
(http://www.elsevier.com/locate/permissions).

In the USA, users may clear permissions and make payments through the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers,
MA 01923, USA; phone: (+1) (978) 7508400, fax: (+1) (978) 7504744, and in the UK through the Copyright Licensing Agency Rapid
Clearance Service (CLARCS), 90 Tottenham Court Road, London W1P 0LP, UK; phone: (+44) 20 7631 5555; fax: (+44) 20 7631 5500.
Other countries may have a local reprographic rights agency for payments.

Derivative Works
Tables of contents may be reproduced for internal circulation, but permission of the Publisher is required for external resale or distribution
of such material. Permission of the Publisher is required for all other derivative works, including compilations and translations.

Electronic Storage or Usage

Permission of the Publisher is required to store or use electronically any material contained in this work, including any chapter or part of a
chapter.

Except as outlined above, no part of this work may be reproduced, stored in a retrieval system or transmitted in any form or by any means,
electronic, mechanical, photocopying, recording or otherwise, without prior written permission of the Publisher.
Address permissions requests to: Elsevier’s Rights Department, at the fax and e-mail addresses noted above.

Notice
No responsibility is assumed by the Publisher for any injury and/or damage to persons or property as a matter of products liability,
negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein.
Because of rapid advances in the medical sciences, in particular, independent verification of diagnoses and drug dosages should be made.

First edition 2004

Library of Congress Cataloging in Publication Data

A catalog record is available from the Library of Congress.

British Library Cataloguing in Publication Data

A catalogue record is available from the British Library.

ISBN: 0 444 51266 7

1
W The paper used in this publication meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper). Printed in
The Netherlands.
 Dedicated to

Ashima

for her consistent cooperation and pampering her husband for over a decade
cherishing the desire that she sees him coherently reflected
in a likeable tome of its ilk
This page is intentionally left blank
 vii

Preface

In the second half of twentieth century, the discipline of planetary science has
witnessed three major episodes, which have revolutionized its approach and content: (i)
the plate-tectonic theory, (ii) human landing and discoveries in planetary astronomy, and
(iii) the extraordinary technical advancement in high P – T studies, which have largely
been abetted by a vast improvement in computational methods. Using these new
computational methods, such as first principles including ab initio models, calculations
have been made for the electronic structure, bonding, thermal EOS, elasticity, melting,
thermal conductivity and diffusivity. Indeed, significant achievements have been made at
the cross-roads of physics and planetary science.
In this monograph, the boundaries of the definitions of a petrologist, geochemist,
geophysicist or a mineralogist have been wilfully eliminated to bring them all under the
spectrum of ‘high-pressure geochemistry’ when they deal with any material (quintessen-
tially a chemical assemblage) — terrestrial or extraterrestrial — under the conditions of
high-pressure and temperature. Thus, a petrologist using a spectrometer or any instrument
for high-pressure studies of a rock or a mineral, or a geochemist using them for chemical
synthesis and characterization, is better categorized as a ‘high-pressure geochemist’ rather
than any other kind of disciplinarian.
The contents of this monograph will display the purpose for bringing under one
cover apparently disparate disciplines like solid-Earth geophysics and geochemistry as
well as material science and condensed-matter physics. Indeed, such interdisciplinary
activities led to the discovery of new phenomena such as high P –T behaviour in metal
oxides (e.g., Mott transition), novel transitions such as amorphization, changes in order –
disorder in crystals and the anomalous properties of oxide melts.
This monograph thus tries to focus more on the theme rather than on the discipline
(e.g., mineralogy, geochemistry or geophysics or whatever). Even a simple and innocuous
word like ‘layering’ bears different connotations for different disciplines of geoscience. To
geochemists, it means the extent to which the mantle experienced degassing of primordial
noble gasses (e.g., 40Ar) or was deprived of its large-ion lithophile elements (now residing
in the continents). To mineral physicists, it pertains to seismic changes with depth of the
mantle’s elemental (and mineralogical) composition. To geophysicists, it means the
degree to which convection is prevented from being whole mantle, preventing formation
of internal boundary layers. The mineral physicists, in general, may accept mantle
homogeneity (e.g., pyrolite composition) while, for geochemists, the evidence is too
strong to stand against it. Geophysicists are prone to believe that the mantle being stirred
from bottom to top mainly operates through the process of subduction and plume
generation. It is generally believed by geochemists that much of the Earth’s heat originates
deep in the mantle and that it must get out by mass transport if the mantle convects from
viii Preface

top to bottom. The nature of such disagreement reflects very well the orthogonality of the
two major groups: the geochemists and the geophysicists. Nevertheless, the models may
converge in the frame of time.
Mineral physics and its relationships to seismological data, particularly for the
mantle, have been dealt with in this monograph, which attempts to provide a snapshot of
the rapidly evolving field of geochemistry under high P, T environments. An attempt has
been made to cull and collate all diverse ideas, theoretical models and experimental results
into a certain degree of coherence with perhaps some constrained (delimited) success.
Discussions about chemical systems — that is, what the planets are — are allowed
to waft through a collection of ideas which often stray far afield. Self-evidently, a graduate
student is not expected to be an expert in all the fields but one is asked to call upon a degree
of courage to know how and why to read the disciplines that have been allied into the
broad spectrum discipline called ‘geochemistry’. In this process, it can possibly be more to
be gained by collaborating with oneself rather than with a host of others. This effort in
lateral thinking helps to secure a certain degree of comfort from converging disciplines
and from the use of their language for an integrated and holistic appreciation. Indeed, this
is expected to be the prevalent mood in the approach to research in the global village of
geoscience in the early twenty-first century.
How a planet should evolve depends on a number of physical properties besides P
and T. Structural distortions, defect chemistry and impurity (dopants) may affect elastic
constants, thermal and electrical conductivity, rheology, diffusion rates and other physical
properties. In a sense, the basic question of deep-Earth geochemistry is a peculiar sort of
inverse problem in materials science.
For a better appreciation of the major problems in planetary science and
geochemistry today, the synthesis and characterization of materials at high pressures and
high temperatures come almost as the quintessential procedures. The ever-increasing
experimental abilities to manipulate, simulate and synthesize materials hold the key to this
endeavour. The activity falls well within the premise of what is broadly known as
condensed-matter physics. Under the rubric of condensed-matter geophysics comes the
diverse studies and syntheses of materials, liquid glasses and crystalline solids to shed light
on planetary interiors, volcanology and the transport properties of planetary materials at
high pressure and temperature prevalent within planets.
Advances in HP studies have occurred chiefly as a result of breakthroughs in
experimental techniques, developed by cross-fertilization of conceptual and technical
ideas across disciplines and amongst laboratories. Development of quantum mechanics
and modern solid-state theory and accelerating developments in static and dynamic
compression techniques have helped investigations to range from planetary interiors to
terapascal (10 million atmosphere pressure), the domains of stellar interiors and ultra-
dense plasmas. But nuclear processes effected by pressure are left out in the present scope
of this monograph, except for a glide over Mossbauer and NMR spectroscopy under
pressure.
Recently, it has been shown that pair potentials that accurately describe the X-ray
and sound-velocity measurements in solids can predict well for lower-mantle temperatures
along the Hugoniot.
Preface ix

Surprisingly, many high-pressure techniques have their origins in geoscience

laboratories and, in many respects and more often than not, necessities of geoscience have
mothered the inventions and development of high-pressure techniques which have fed
back usefully to physics, chemistry and materials science. Micro-analytical techniques,
such as micro-spectroscopy and X-ray diffraction, developed for high-pressure research,
are finding uses in other fields. Highly collimated, coherent, monochromatic laser beams
can be focussed quite conveniently through diamond windows of an HP cell to provide a
versatile microprobe with mm spatial resolution to multi-megabar pressures. The use of
these laser microprobes has opened new areas for research employing HP Raman,
luminescence and Brillouin spectroscopies.
With the arrival of the third-generation synchrotron source and the development of
in situ micro-analytical techniques, HP mineralogy is reaching its maturity. Under high
P –T conditions (through experimental and/or theoretical simulations), a host of important
chemical and physical properties of minerals have been investigated. The pressure-
induced transformations include crystallographic transformations (both reconstructive and
displacive), electronic changes (including band structure, bonding and insulator – metal
transition), magnetic transition and so on.
Some most exciting observations have been made on simple molecular
compounds at megabar pressures. To cite just two: the discovery of a new class of
excitations in orientationally ordered low-temperature phases of solid hydrogen and
deuterium and the experimental evidence obtained by the scientists of Geophysical
Laboratory (Washington) for the non-molecular high-pressure modification of ice with
symmetric hydrogen bonds. The transition pressure of H2O-ice from its normal
molecular form to its high-density non-molecular state can be pinpointed. This transition
is associated with unusual quantum mechanical phenomena. Currently, 3D tomographic
maps of seismic velocity anomalies and regional radial profiles are being prepared by
geophysicists.

Acknowledgements

The author is indebted to a number of eminent scientists who generously allowed

permission to use their materials in the preparation of this monograph, which is essentially
a wide-spectrum review work. The author expresses his gratitude for the copyright
permission obtained from the following publishers for their numerous source journals
from which tables and diagrams have been redrawn and reproduced.
American Association for the Advancement of Science; American Institute of
Physics; American Physical Society; Geochemical Society; IOP Publishing Ltd.;
Mineralogical Society of America; Nature; Royal Society, London; Societa Italiana di
Fisica; Springer-Verlag, Heidelberg; University of Chicago; Wiley-VCH; American
Geophysical Union.
Generous cooperation by way of the supply of research outputs was received from
the Carnegie Institute, Washington, for over a decade in the preparation of the
manuscript.
x Preface

In the preparation of the typed copy, spontaneous cooperation was rendered by my

students, Dr. M. Bidyananda, Dr. Susanta K. Samanta and Mr. Dibyendu Mal (Research
student). In the artwork, considerable help was available from Mr. Tarapada Bhattacharya,
the recently retired draftsman of our department.

Sachinath Mitra
Department of Geological Sciences
Jadavpur University
 xi

Table of Contents

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix

Section A: Preamble and Preview

Section B: The Earth and Planetary System

Chapter 1. (A) Cosmochemistry and Properties of Light Element

Compounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.1.1. Range of pressure in the universe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
1.1.2. The proto-solar nebula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
1.2. Cosmochemistry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
1.2.1. Data source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
1.2.2. Chemical segregation in nebular condensation . . . . . . . . . . . . . . . . . . . . 24
1.2.3. The Solar System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
1.2.3.1. Meteorites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
1.2.3.2. Inner planets: major constituents and phases . . . . . . . . . . . . 30
1.3. Evolutionary history of the Solar System: terrestrial planetary formation . . . 31
1.3.1. Interplanetary flights of planetary materials . . . . . . . . . . . . . . . . . . . . . . . 31
1.3.2. Primary chemical elements for life . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
1.3.2.1. Microorganisms under pressure: clues to HP
genesis of life . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
1.3.2.2. Biogenesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
1.3.3. Primitive atmosphere . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
1.4. Charge density within planetary interiors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
1.4.1. Electrons under pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
1.5. Forces binding atoms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
1.5.1. Van der Waals forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
1.5.1.1. Van der Waals compounds: new materials . . . . . . . . . . . . . . 38
1.5.2. Ionic compounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
1.5.2.1. Simple ionic model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
1.5.2.2. Overlap- and self-energy: pair-potential . . . . . . . . . . . . . . . . 42
1.5.2.3. Ions in distorted lattice . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
1.5.2.4. Multipoles and polarization . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
1.5.3. Covalent and hydrogen bonding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
xii Table of Contents

1.5.3.1. Pressure rupturing of the binding forces . . . . . . . . . . . . . . . . 44

1.6. Helioseismology and Jovian structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
1.7. Planetary constituents under pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
1.7.1. Transition pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
1.8. Hydrogen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
1.8.1. Hydrogen molecular states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
1.8.2. Vibrational properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
1.8.2.1. Vibrational excitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
1.8.2.2. Experiments (P .300 GPa) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
1.8.2.3. Vibrons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
1.8.2.4. Phonons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
1.8.2.5. Rotons and librons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
1.8.2.6. Hydrogen bridges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
1.8.3. Quantum condensate, BEC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
1.8.3.1. Proton quantum tunnelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
1.8.4. Insulator –metal transition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
1.8.5. Solid hydrogen: frustrating metallic behaviour . . . . . . . . . . . . . . . . . . . . 60
1.8.5.1. Black hydrogen and metallization . . . . . . . . . . . . . . . . . . . . . . 61
1.8.5.2. .300 GPa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
1.8.5.3. Effective charge: phase III . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
1.8.5.4. Solid hydrogen: alkali metal(?) at 340 GPa . . . . . . . . . . . . . 63
1.8.6. Ortho- and para-hydrogen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
1.8.6.1. Ortho –para conversion: quantum solid state . . . . . . . . . . . 64
1.8.6.2. Conversion energy channels: EQQ . . . . . . . . . . . . . . . . . . . . . 65
1.8.7. Hydrogen in Jupiter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
1.8.8. H in terrestrial planets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
1.8.8.1. Hydrogen in the Earth’s minerals . . . . . . . . . . . . . . . . . . . . . . 67
1.8.8.2. Water in the Earth: D/H ratios . . . . . . . . . . . . . . . . . . . . . . . . . 68
1.8.8.3. H/H2O in mantle phases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
1.9. Water and ammonia in Uranus and Neptune . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
1.9.1. Electrical conductivity: “synthetic Uranus” . . . . . . . . . . . . . . . . . . . . . . . 71
1.9.2. Metallicity(?) of water and ammonia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
1.9.3. Water: structural order and anomalies . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
1.9.3.1. Proton (and oxygen) diffusion in water . . . . . . . . . . . . . . . . . 73
1.9.4. Superionic solid state . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
1.9.4.1. Ammonia: superionic state . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
1.10. H2 mixtures and clathrates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
1.10.1. H2 – O2 mixture: “Hard Spheres” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
1.10.2. CH4 – H2 and Ar –H2 systems: Laves phases . . . . . . . . . . . . . . . . . . . . . . 75
1.10.2.1. Laves phases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
1.10.3. N2 – CH4: Titan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
1.11. H2O . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
1.11.1. Bonding: covalency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
1.11.2. Hydrodynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
Table of Contents xiii

1.11.3. H2O –ice structure: “ice rules” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

1.11.3.1. Reflectance spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
1.11.4. Entropy of ice . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
1.11.4.1. Ferroelectric alignment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
1.11.4.2. Spin ice . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
1.11.5. Ice Ih, III, IV, V and VI: phase diagram . . . . . . . . . . . . . . . . . . . . . . . . . . 84
1.11.5.1. Ice VI in diamond . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
1.11.5.2. Ice Ih: stability boundary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
1.11.5.3. Proton ordering/disordering: new phase . . . . . . . . . . . . . . . . 85
1.11.5.4. Higher isomorphs: ices VII, VIII and X . . . . . . . . . . . . . . . . 86
1.11.5.5. Ice VII: as pressure medium . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
1.11.6. Supercooled water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
1.11.6.1. Amorphous ice polymorphism: high-density
and low-density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
1.11.6.2. Diffraction study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
1.11.6.3. LDA ice, ice Ih and quenched water:
vibrational spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
1.11.6.4. VHDA ice . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
1.12. Deuterium at high pressure: Saturn’s core . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
1.12.1. Deuterium in Mars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
1.12.2. D/H ratios in minerals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
1.12.2.1. D/H ratio in extraterrestrial and subsurface water . . . . . . . 95
1.13. Alkali metals: Li to Cs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
1.13.1. “Nearly-free electron” behaviour . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
1.13.2. Fermi pressure in lithium isotopes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
1.14. CO2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
1.14.1. Stability of CO2 polymorphs: CO2-V quartz-like . . . . . . . . . . . . . . . . . . 99
1.14.2. H2O–CO2 mixture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
1.15. Carbon in space and in the Earth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
1.15.1. Fullerites and nano-crystallites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
1.15.2. Carbon polymorphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
1.15.3. Carbon in the Earth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
1.15.4. Carbon in high P – T: stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
1.15.4.1. C, Si and Ge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
1.15.5. Carbon-bonding structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
1.15.6. Graphite and diamond phases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
1.15.6.1. Superhard graphite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
1.15.6.2. Resistivity and phase transition . . . . . . . . . . . . . . . . . . . . . . . . 110
1.15.6.3. Pre-solar nano-diamonds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
1.15.6.4. Terrestrial occurrence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
1.15.7. Carbon isotopes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
1.15.7.1. Oxygen and carbon-12 (C-12) evolution
on the Earth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
1.15.7.2. 12C/13C ratios: interstellar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
xiv Table of Contents

1.15.7.3. Raman line: P, T calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . 115

1.15.7.4. 14C diamond: elastic moduli . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
1.15.8. Optical behaviour of diamond: flow and
pressure-luminescence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
1.15.9. Carbon clusters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
1.15.9.1. Charged carbon clusters: low-P diamond . . . . . . . . . . . . . . . 118
1.15.9.2. C-nanotubes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
1.15.9.3. Fullerenes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
1.15.10. Organic minerals in meteorites: shock loading . . . . . . . . . . . . . . . . . . . . 124
1.15.10.1. Amino-acid racemization: chirality retention . . . . . . . . . . . 126
1.15.10.2. Vitrinite maturation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
1.16. Nitrogen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
1.17. Sulphur . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
(B) Terrestrial Planets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
1.18. Early geochemical evolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
1.18.1. Chondritic character of terrestrial bodies . . . . . . . . . . . . . . . . . . . . . . . . . 131
1.18.1.1. Chemical differentiation: siderophile elements . . . . . . . . . . 132
1.19. Accretionary evolution of the Earth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
1.20. Compositional characteristics of the Earth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
1.20.1. Magma ocean generation and crustal fractionation in
early Earth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
1.20.1.1. Early crust . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
1.21. Fluids within the Earth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
1.21.1. Water in the Earth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
1.21.2. Water in the magmatic processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
1.21.3. Fluids in the lower crust: granulites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
1.21.4. Mantle fluids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
1.21.5. Atmospheric noble gases in mantle melts . . . . . . . . . . . . . . . . . . . . . . . . . 143
1.21.5.1. Inert gases: solar-like? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
1.21.5.2. Ar, Kr and Xe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
1.21.5.3. Ar solubility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
1.21.5.4. The “missing xenon problem” . . . . . . . . . . . . . . . . . . . . . . . . . 145
1.22. Potassium budget in the Earth’s mantle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
1.22.1. K-feldspar/Hollandite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
1.22.2. Phlogopite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
1.22.3. Clinopyroxene . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
1.22.4. K2O in mantle solidus: seismic attenuation . . . . . . . . . . . . . . . . . . . . . . . 146
1.23. Mantle isotopes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
1.23.1. Sm – Nd ratios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
1.23.2. Eu anomaly . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
1.23.3. Sr, Nd and Hf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
1.23.4. Osmium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
1.23.5. 187Re and 187Os . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
1.23.6. U – Pb and Re – Os ratio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
Table of Contents xv

1.23.7. Isotopes in MORB and hotspots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150

1.23.8. Isotopes in UHP rocks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
1.23.8.1. 18O isotopes: non-equilibration . . . . . . . . . . . . . . . . . . . . . . . . . 152
1.23.8.2. d18O: an example for isotope separation . . . . . . . . . . . . . . . . 152
3 4
1.24. He/ He reservoirs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
(C) Heavy Element Compounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
1.25. Ferrous metals in rocky planets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
1.26. Element distribution in mineral system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
1.26.1. Partitioning of elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
1.26.1.1. Siderophile elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
1.26.1.2. Incompatible elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
1.27. Transition metals in magmas: CFSE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
1.27.1. Principles of metal distribution in magmatic differentiation . . . . . . . . 161
1.27.2. Transition-element partitioning in mineral systems . . . . . . . . . . . . . . . . 163
1.27.2.1. Ni2þ and Co2þ: pressure partitioning . . . . . . . . . . . . . . . . . . . 164
1.27.2.2. Cr3þ, Ni2þ, Fe3þ and Ti4þ ions . . . . . . . . . . . . . . . . . . . . . . . . 165
1.27.2.3. Cr2þ, Ni2þ and Co3þ ions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
1.27.2.4. Ni –Co partitioning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
1.27.2.5. Plutonic rocks and metal concentration . . . . . . . . . . . . . . . . . 167
1.27.2.6. Highly siderophile elements . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
1.28. Ca –Al and Mg – Si proportionation in the mantle . . . . . . . . . . . . . . . . . . . . . . . . . . 170
1.28.1. Critical ratios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
1.29. Core differentiation / heterogeneous accretion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
1.29.1. 182W fractionation and Hf/W ratio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
1.29.1.1. Hf/W in early history . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
1.29.2. Core:Re/Os . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
40
1.30. K in the core . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175

Chapter 2. Petro-Tectonic Features of Terrestrial Planets . . . . . . . . . . . . . . . . . . . . . 177

2.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
2.2. The Earth models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
2.2.1. The PREM model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
2.2.2. Seismological models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
2.2.2.1. Elastic constants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
2.2.3. Petrological models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
2.2.3.1. Pyrolite model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
2.2.3.2. Piclogite model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
2.2.4. Convection model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
2.2.4.1. Mantle convection at Archean –Proterozoic transition . . . 187
2.2.4.2. Mantle Raleigh number and flush instability
at late Archean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
2.3. Physical parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
2.3.1. Parameter changes with depths . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
2.3.1.1. Lithosphere . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
xvi Table of Contents

2.3.2. Parameterized PREM model: EOS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192

2.4. Seismic model: discontinuities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
2.4.1. Seismic discontinuities: Moho to the D00 zone . . . . . . . . . . . . . . . . . . . . . 194
2.5. Thermal structures of the Earth’s mantle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
2.5.1. Temperature –depth relation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
2.5.1.1. Heat sources and heat flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
2.5.2. Thermal anomalies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
2.5.2.1. Upper mantle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
2.5.2.2. Lower mantle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
2.5.2.3. Thermal structure of the core and CMB . . . . . . . . . . . . . . . . 201
2.5.3. Adiabatic gradient . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
2.5.3.1. Deviations from adiabaticity . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
2.5.3.2. Heat flow and plate tectonics . . . . . . . . . . . . . . . . . . . . . . . . . . 203
2.6. Elastic parameters of the Earth’s interior . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
2.6.1. Stress and strain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
2.6.1.1. Strain tensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
2.6.1.2. Zero-pressure bulk modulus, K0 : Eulerian strain . . . . . . . . 205
2.6.2. Seismic waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
2.6.2.1. P- and S-waves in seismic discontinuities
and in the core . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
2.6.2.2. Minor discontinuities (reflective) . . . . . . . . . . . . . . . . . . . . . . . 208
2.6.2.3. Wave velocities in the lower crust . . . . . . . . . . . . . . . . . . . . . 208
2.6.2.4. Wave velocities in lower mantle: T effects . . . . . . . . . . . . . 209
2.6.2.5. Crustal plates and earthquakes . . . . . . . . . . . . . . . . . . . . . . . . . 210
2.6.2.6. Strain transients and earthquakes, co- and
post-seismic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
2.6.2.7. Precursors to earthquakes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
2.6.3. Acoustic and ultrasonic wave-velocities . . . . . . . . . . . . . . . . . . . . . . . . . . 217
2.6.3.1. Ultrasonic velocities and Q in porous rocks . . . . . . . . . . . . 217
2.6.4. Subcrustal stress fields: ore localization . . . . . . . . . . . . . . . . . . . . . . . . . . 218
2.6.4.1. Gravitational field models: degree harmonics
and mantle flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
2.6.5. Tools for sub-surface studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
2.6.5.1. GPS in tectonic studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
2.6.5.2. Mars global surveyor (MGS) . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
2.7. The crust and cratons (“keels”) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
2.7.1. Continental lithosphere . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
2.7.2. Subcontinental mantle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
2.7.3. Plate tectonics, magmatism and hotspots . . . . . . . . . . . . . . . . . . . . . . . . . 222
2.8. The mantle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
2.8.1. Geochemistry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
2.8.1.1. Mantle end members . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
2.8.2. Petro-tectonics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
2.8.3. Xenoliths . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
Table of Contents xvii

2.8.4.
Deep-mantle flow and Wilson cycle: American Cordillera . . . . . . . . . 226
2.8.5.
Diversification of rock types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
2.8.5.1. Petrogeny’s residua system . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
2.8.5.2. Effusive rocks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
2.8.5.3. Calc-alkaline magmatism: LIL enrichment
and “Pb paradox” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
2.9. Earth’s rheology and dynamism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
2.9.1. Lithospheric rheology and dynamism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
2.9.2. Mantle rheology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
2.9.2.1. Decompression and magma fragmentation . . . . . . . . . . . . . . 231
2.9.3. Seismic tomography: Iceland hotspot and Nazca plate . . . . . . . . . . . . 231
2.9.3.1. Anomalous low-velocity zone . . . . . . . . . . . . . . . . . . . . . . . . . . 232
2.10. Convergent plate boundaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
2.10.1. Subducting slabs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
2.10.1.1. Slab tomography: volatiles and partial melting . . . . . . . . . 235
2.10.1.2. Deflections of seismic discontinuities:
NW Pacific subduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
2.10.1.3. Deep-focus earthquakes: fossil slab at
transition zone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236
2.10.2. Subducting mafic, ultramafic rocks and sediments . . . . . . . . . . . . . . . . 237
2.10.2.1. Subduction of oceanic lithosphere: upper to
lower mantle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
2.10.2.2. Mid-oceanic ridge basalt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
2.10.3. Mantle wedge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
2.10.3.1. Arc magmatism: alkali and H2O activity . . . . . . . . . . . . . . . 242
2.10.4. Hotspots and mantle plumes: OIB versus MORB . . . . . . . . . . . . . . . . . 243
2.10.4.1. Iceland mantle plume . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
2.10.4.2. Plumes and underplating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
2.10.4.3. Megaplumes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
2.11. Upper mantle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
2.11.1. Upper-mantle anisotropy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246
2.11.2. Mantle minerals versus discontinuities . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
2.11.2.1. Upper mantle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
2.11.2.2. 400, 520 and 670 km discontinuities . . . . . . . . . . . . . . . . . . . 249
2.11.2.3. Ca-phases in mantle discontinuities . . . . . . . . . . . . . . . . . . . . 250
2.11.3. Mantle melting and extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250
2.11.3.1. Deep-mantle melting: melt sinking . . . . . . . . . . . . . . . . . . . . . 251
2.11.3.2. Depletion and mixing: non-Newtonian
high-viscosity blobs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
2.11.4. Peridotite mineralogy at depths . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
2.11.4.1. Mantle silicate framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
2.12. Lower mantle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
2.12.1. Phases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
2.12.2. Solidus in the lower mantle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258
xviii Table of Contents

2.12.3. Fe, Si enrichment in the lower mantle . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258

2.12.3.1. Effects of Fe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
2.12.3.2. P- and S-velocities and shear modulus . . . . . . . . . . . . . . . . . 260
2.13. Core – mantle boundary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260
2.13.1. Minerals at CMB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
2.13.2. Hotspots and CMB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
2.13.3. Anisotropic structures at CMB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
2.13.3.1. Seismic anisotropy in D00 layer . . . . . . . . . . . . . . . . . . . . . . . . . 263
2.13.3.2. Anisotropy caused by paleo-slabs . . . . . . . . . . . . . . . . . . . . . . 266
2.13.3.3. Carribbean and Pacific evidence . . . . . . . . . . . . . . . . . . . . . . . 266
2.14. Reaction between mantle and liquid-core . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
2.14.1. Ultra-low-velocity zone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
2.15. The Earth’s magnetism and orbital obliquity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270
2.15.1. Mantle plume and geomagnetic reversals . . . . . . . . . . . . . . . . . . . . . . . . . 270
2.16. Mars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
2.16.1. Crust and mantle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
2.16.2. SNC and LHB: ALH 84001 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
2.16.3. Martian mantle composition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
2.16.3.1. Mantle geochemistry: SNC . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
2.16.3.2. Mantle-phase stability: MB versus KLB . . . . . . . . . . . . . . . . 276
2.16.3.3. Fe-rich Martian mantle: density increase . . . . . . . . . . . . . . . 277
2.16.3.4. Olivine –spinel phase transition . . . . . . . . . . . . . . . . . . . . . . . . 279
2.16.3.5. Mantle-flow: viscosity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
2.16.3.6. Magmatic water in Mars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
2.16.4. Magnetism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280
2.16.4.1. Core formation and magnetism . . . . . . . . . . . . . . . . . . . . . . . . 280
2.17. Venus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280
2.17.1. Gabbro ! eclogite transition in Venus . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
2.18. Mercury . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
2.19. Galilean satellites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
2.20. Interplanetary flights of planetary materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282

Chapter 3. Structural Types of Major Phases: AB, AB2, A2B3, ABX3,

ABX4, AB2X4 and A2B2X7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
3.1. AB structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
3.1.1. NaCl (B1): alkali halides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 286
3.1.1.1. Exciton in alkali halides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
3.1.1.2. NaCl structure at lower mantle . . . . . . . . . . . . . . . . . . . . . . . . . 289
3.1.2. CsCl (B2) structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
3.1.2.1. Cs-halide (B2), CsI: metallization . . . . . . . . . . . . . . . . . . . . . . 291
3.1.2.2. Convergence with rare gas: solid Xe . . . . . . . . . . . . . . . . . . . 291
3.1.3. NiAs (B8) structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
3.1.3.1. Chemical bonding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
3.1.3.2. Hexagonal close packing and c/a ratio . . . . . . . . . . . . . . . . . 294
Table of Contents xix

3.2. AB2 structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 294

3.2.1. SiO2 polymorphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 296
3.2.1.1. Si-coordination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297
3.2.1.2. Polarization and chirality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297
3.2.2. TiO2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 298
3.2.2.1. Cotunnite type: hardest polymorph . . . . . . . . . . . . . . . . . . . . . 300
3.2.2.2. Crystallographic shear (cs) planes . . . . . . . . . . . . . . . . . . . . . . 300
3.2.3. Post-stishovite phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300
3.2.3.1. Stishovite(TiO2) ! a-PbO2 structural transformation . . . 301
3.2.3.2. Baddeleyite-type structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
3.3. A2B3 structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302
3.3.1. Fe2O3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
3.3.1.1. Structural and spin transition . . . . . . . . . . . . . . . . . . . . . . . . . . . 304
3.3.1.2. X-ray emission spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
3.3.2. Al2O3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 306
3.3.2.1. Quadrupole polarizability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
3.4. ABX3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
3.4.1. Perovskite – ilmenite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
3.4.2. Ilmenite structure: stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308
3.4.2.1. Polymorphism of FeTiO3: LiNbO3 structure . . . . . . . . . . . . 309
3.4.3. Ilmenite solution in olivine: Alpe Arami massif . . . . . . . . . . . . . . . . . . . 309
3.5. ABX4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
3.5.1. Berlinite/scheelite structure: AWO4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310
3.5.2. Berlinite and crystobalite: AlPO4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310
3.5.2.1. GaPO4 and AlAsO4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313
3.6. A2BX4 structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313
3.6.1. Tetragonal structure: K2NiO4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313
3.6.2. Spinel structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315
3.7. A2B2X7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 316
3.7.1. Pyrochlore structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 316
3.7.1.1. Frustration and magnetic “spin ice” . . . . . . . . . . . . . . . . . . . . 316
3.7.1.2. Tl2Mn2O7: GMR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 317

Section C: Basics for Pressures Studies

Chapter 4. Principles of Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321

4.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321
4.1.1. Insulator –metal transition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 322
4.1.1.1. Mott insulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 326
4.1.2. High-pressure techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 326
4.1.2.1. Synchrotron source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 329
4.1.2.2. Synchrotron radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 329
xx Table of Contents

4.1.2.3. Multi-anvil and DAC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 330

4.1.2.4. Measurement techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332
4.2. Diamond-anvil cell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333
4.2.1. Properties of the gaskets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333
4.2.2. Pressure medium and calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 334
4.2.2.1. Quasi-hydrostatic stress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 336
4.2.2.2. Shear stress (s) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 336
4.2.3. Reference-phase transitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337
4.2.3.1. CaO – MgO – SiO2 system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337
4.2.4. Temperature control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 338
4.2.4.1. Cryogenic methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 338
4.2.4.2. Laser heating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 338
4.2.5. Ruby (Al2O3:Cr3þ) calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 338
4.2.6. Diamond window . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 340
4.3. Hydrothermal diamond-anvil cell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 340
4.3.1. Pressure calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 340
4.4. Diffraction and spectroscopic techniques in pressure studies . . . . . . . . . . . . . . . 341
4.4.1. Optical spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 341
4.4.1.1. Crystal-field under pressure: theory . . . . . . . . . . . . . . . . . . . . 342
4.4.1.2. Transition-metal compounds: ionic model . . . . . . . . . . . . . . 346
4.4.1.3. Pressure on crystal-field parameters . . . . . . . . . . . . . . . . . . . . 348
4.4.1.4. Racah parameters and band shifts . . . . . . . . . . . . . . . . . . . . . . 349
4.4.1.5. Examples: Cr3þ-bearing minerals . . . . . . . . . . . . . . . . . . . . . . 350
4.4.1.6. Intervalence charge transfer . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352
4.4.1.7. O ! M charge transfer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352
4.4.1.8. M– M bonding and Cr dimerization . . . . . . . . . . . . . . . . . . . . 353
4.4.1.9. Crystal-field effect on transition pressure . . . . . . . . . . . . . . . 353
4.4.1.10. Exchange-coupled pair bands . . . . . . . . . . . . . . . . . . . . . . . . . . 354
4.4.1.11. CFSE and elastic property change: Fe2þ . . . . . . . . . . . . . . . 355
4.4.2. Volume compressibility and crystal-field splitting . . . . . . . . . . . . . . . . . 355
4.4.3. Vibrational spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356
4.4.3.1. Soft modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 358
4.4.3.2. Pressure relationship . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 359
4.4.3.3. Infrared . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 360
4.4.3.4. Raman spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 362
4.4.3.5. Second-order Raman scattering: disordering . . . . . . . . . . . . 366
4.4.3.6. Non-linear optical methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . 367
4.4.3.7. Brillouin scattering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 367
4.4.4. Ultrasound spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369
4.4.4.1. Instrumentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 371
4.4.5. Fluorescence spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 371
4.4.5.1. Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 372
4.4.5.2. Side-band fluorescence ultrasonic technique . . . . . . . . . . . . 372
4.4.6. Photoluminescence spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 374
Table of Contents xxi

4.4.6.1. Photo-emission method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 375

4.4.7. X-ray diffraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 375
4.4.7.1. Radial X-ray diffraction (RXD): deviatoric stress . . . . . . . 377
4.4.7.2. Density determination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 380
4.4.7.3. High-pressure XRD (powder) study: An example
of MgSiO3 ilmenite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 380
4.4.8. Mössbauer spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 380
4.4.8.1. Pressure dependence of isomer shift (d0) . . . . . . . . . . . . . . . 382
4.4.8.2. Quadrupole splitting (DEQ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 383
4.4.9. NMR spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 384
4.4.9.1. Pressure effects on proton NMR spectra . . . . . . . . . . . . . . . . 385
4.4.10. Thermoluminescence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 386
4.4.10.1. ZnS phosphor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 387
4.4.10.2. Trap depths . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 387
4.5. Computational simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 387
4.5.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 387
4.5.1.1. Ab initio methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 388
4.5.1.2. First-principles approximations . . . . . . . . . . . . . . . . . . . . . . . . 388
4.5.1.3. Density functional theory: Kohn –Sham equations . . . . . . 390
4.5.2. LCAO model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391
4.5.2.1. Molecular dynamics simulation . . . . . . . . . . . . . . . . . . . . . . . . 392
4.5.2.2. Inter-atomic potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393
4.5.2.3. Tight-binding total-energy model . . . . . . . . . . . . . . . . . . . . . . 393
4.5.2.4. Potential-induced breathing model . . . . . . . . . . . . . . . . . . . . . 394
4.5.2.5. Variationally induced breathing model: MgO . . . . . . . . . . . 394
4.5.3. Electronic approximations: “Muffin tin”, KSS and Bloch’s
theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 395
4.5.3.1. Double exchange (DE) model . . . . . . . . . . . . . . . . . . . . . . . . . . 396
4.5.4. LMTO method vs. APW and KKR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 398
4.5.4.1. Average pair correlation function for NN geometry:
SiO2 glass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 398
4.6. Shock pressure studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 398

Chapter 5. (Crystalline) Materials Under High Pressure . . . . . . . . . . . . . . . . . . . . . . 401

5.1. Material properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 401
5.1.1. Thermodynamics, equilibrium and time interval . . . . . . . . . . . . . . . . . . 404
5.1.2. Many-body systems and broken symmetry . . . . . . . . . . . . . . . . . . . . . . . . 404
5.1.2.1. Crystalline symmetries: 5-fold symmetry,
icosahedra and quasi-crystals . . . . . . . . . . . . . . . . . . . . . . . . . . 404
5.1.2.2. Broken symmetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 405
5.1.2.3. Electron excitations and band gaps . . . . . . . . . . . . . . . . . . . . . 406
5.1.2.4. Dielectric properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 407
5.1.2.5. Electronic and magnetic behaviour . . . . . . . . . . . . . . . . . . . . . 407
5.1.2.6. Ionicity in bonding: Madelung forces . . . . . . . . . . . . . . . . . . 408
xxii Table of Contents

5.1.3. Covalent bonding and hardness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 409

5.1.3.1. Hardness and bulk moduli . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 409
5.1.3.2. Phonon-s and band states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 409
5.1.4. Elasticity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 410
5.1.4.1. Elastic anisotropy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 411
5.1.5. Elastic constants: crystal systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 412
5.1.5.1. Cauchy relation and its violations . . . . . . . . . . . . . . . . . . . . . . 415
5.1.6. Born’s stability criteria: B1 ; B2 and B3 . . . . . . . . . . . . . . . . . . . . . . . . . . . 416
5.1.7. Thermoelasticity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 418
5.2. Atomic vibrations in crystals: phonon-s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 419
5.2.1. Elastic waves in crystals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 420
5.2.1.1. Shock waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 420
5.2.1.2. Shock velocity and particle velocity . . . . . . . . . . . . . . . . . . . . 421
5.2.1.3. Shock-induced transitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 423
5.3. Inelastic and non-hydrostatic states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 424
5.3.1. Stress states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 424
5.3.1.1. Non-hydrostatic stress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 425
5.3.2. Crystallographic shear . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 426
5.3.2.1. Shear and deformational twinning . . . . . . . . . . . . . . . . . . . . . . 427
5.3.3. Strain anisotropy in crystalline mass: e.g., hcp iron . . . . . . . . . . . . . . . 427
5.4. Spontaneous strain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 428
5.4.1. Spontaneous strain and order parameter . . . . . . . . . . . . . . . . . . . . . . . . . . 429
5.5. Strain tensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 431
5.6. Bulk modulus of ionic compounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 432
5.6.1. Molar volume . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 432
5.6.2. Shear modulus: mantle perovskite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 432
5.7. Magnetic features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 433
5.7.1. Ferromagnetism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 433
5.7.1.1. Curie temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 433
5.7.2. Ferrimagnetism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434
5.7.3. Spin states of iron . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434
5.7.3.1. Electronic/magnetic ordering: examples . . . . . . . . . . . . . . . . 435
5.7.3.2. Magnetic collapse: oxides and perovskites . . . . . . . . . . . . . . 435
5.7.3.3. Magnetism in phase stability . . . . . . . . . . . . . . . . . . . . . . . . . . . 436
5.7.3.4. Magnetic frustration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 437
5.8. Polyhedral changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 437
5.8.1. Elasticity of MgO6 and SiO6 octahedra: MgSiO3 ilmenite . . . . . . . . . 440
5.8.2. Anisotropic deformation: decompression . . . . . . . . . . . . . . . . . . . . . . . . . 441
5.8.3. Radius ratio and coordination changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 442
5.8.4. Five-fold coordination: silicon and titanium . . . . . . . . . . . . . . . . . . . . . . 443
5.8.5. Thermal expansivity and deformation equivalence (a /b) . . . . . . . . . . 446
5.8.6. Volume compressibility: negative . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 448
5.8.6.1. Relative compressibilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 449
5.8.7. Thermodynamic parameters and EOS . . . . . . . . . . . . . . . . . . . . . . . . . . . . 449
Table of Contents xxiii

5.8.7.1. P – V – T data and EOS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 449

5.8.7.2. Birch EOS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 450
5.8.7.3. Equations of state: density ratio . . . . . . . . . . . . . . . . . . . . . . . . 451
5.8.8. Bulk moduli: isothermal and isentropic . . . . . . . . . . . . . . . . . . . . . . . . . . . 453
5.8.8.1. K of mineral mixture: Reuss bound and
Voigt bound . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 454
5.8.8.2. Crystal-field spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 454
5.8.9. Velocity– volume relationship . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 455
5.8.10. Velocity– density relationship: rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 455
5.8.11. Stretch densification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 456
5.8.12. Compressibility and Si –O – Si bending . . . . . . . . . . . . . . . . . . . . . . . . . . . 457
5.8.12.1. Ionic compressibilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 457
5.9. Free and thermal energies: phase boundaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 458
5.9.1. Free energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 459
5.9.1.1. Free energy change and phase boundary . . . . . . . . . . . . . . . . 460
5.9.1.2. Volume change and DH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 461
5.9.1.3. Activation volume and activation enthalpy . . . . . . . . . . . . . 461
5.9.1.4. Communal entropy: fluid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 462
5.9.1.5. Heat capacity, entropy and phase boundaries . . . . . . . . . . . 464
5.9.2. Thermal-expansion coefficient . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 466
5.9.2.1. a values: spectroscopic vs. volumetric . . . . . . . . . . . . . . . . . 467
5.9.3. Grüneisen parameter (g) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 469
5.9.3.1. Mode Grüneisen (M-G) parameter . . . . . . . . . . . . . . . . . . . . . 469
5.9.3.2. Thermal Grüneisen parameter (gth) . . . . . . . . . . . . . . . . . . . . . 469
5.9.3.3. Density and Grüneisen parameter . . . . . . . . . . . . . . . . . . . . . . 470
5.9.3.4. Debye model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 471
5.9.3.5. Anderson –Grüneisen parameter . . . . . . . . . . . . . . . . . . . . . . . . 471
5.9.3.6. Vinet equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 471
5.9.3.7. Holzapfel equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 472
5.9.3.8. Logarithmic equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 472
5.9.3.9. Microscopic and macroscopic . . . . . . . . . . . . . . . . . . . . . . . . . . 472
5.9.4. Thermal expansion and crystal-field changes . . . . . . . . . . . . . . . . . . . . . . 473
5.9.5. Radiative-heat transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473
5.9.6. Thermal pressure: Eularian strain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 474
5.9.6.1. Thermal pressure as a function of volume . . . . . . . . . . . . . . 475
5.10. Phase transitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 475
5.10.1. Mixed and quasi-stable phases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 476
5.10.2. Lattice disorder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 480
5.10.3. Silicon: b-tin ! hcp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 480
5.10.4. Cation distribution and order –disorder . . . . . . . . . . . . . . . . . . . . . . . . . . . 480
5.10.5. Incommensurate phases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 481
5.10.6. Order of transition: first order and second order . . . . . . . . . . . . . . . . . . . 481
5.10.7. Order parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 482
5.10.8. Superlattice ordering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 482
xxiv Table of Contents

5.10.9. Structural changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 482

5.10.10. Phase changes: principles and types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 483
5.10.10.1. Thermal transformations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 484
5.10.10.2. Soft modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 484
5.10.10.3. Order parameter (h). Free energy and transition
temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 484
5.10.10.4. Landau theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 486
5.10.10.5. Landau order parameter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 488
5.10.10.6. Origin of doubled-well potential, V(h) . . . . . . . . . . . . . . . . . 489
5.10.10.7. Rigid-unit mode: “split atoms” and energy spectra . . . . . . 489
5.10.11. Pressure-induced order –disorder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 490
5.10.11.1. Fe –Mg ordering in silicates . . . . . . . . . . . . . . . . . . . . . . . . . . . 491
5.10.11.2. Structural disordering and twinning . . . . . . . . . . . . . . . . . . . . 492
5.10.11.3. Free energy and order parameter (Q) . . . . . . . . . . . . . . . . . . . 493
5.10.11.4. Order parameter (Q) and strain (1) in
phase transition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 494
5.10.12. Isosymmetric transitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 494
5.10.12.1. Energetics of iso-symmetric transition . . . . . . . . . . . . . . . . . . 495
5.10.13. Growth rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 495
5.11. Charge distribution in ionic solids: valence and core states . . . . . . . . . . . . . . . . 496
5.11.1. Ionic solid under compression: MgO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 498
5.11.1.1. Band-gap change: implication in lower mantle . . . . . . . . . 498
5.11.2. High-spin –low-spin transition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 499
5.11.2.1. Energy change in spin transition . . . . . . . . . . . . . . . . . . . . . . . 499
5.11.2.2. Spin-pairing in the lower mantle . . . . . . . . . . . . . . . . . . . . . . . 503
5.11.3. Pressure dissolution and substitution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 504
5.12. Amorphization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 505
5.12.1. Pressure-induced amorphization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 505
5.12.1.1. Metastability and reversible amorphization . . . . . . . . . . . . . 507
5.12.1.2. Non-hydrostatic pressure and amorphization . . . . . . . . . . . . 507
5.12.2. Disordering and amorphization: Raman scattering . . . . . . . . . . . . . . . . 508
5.12.2.1. Non-bonded atoms and steric hindrances . . . . . . . . . . . . . . . 508
5.12.2.2. Memory glass: AlPO4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 509
5.12.3. Solid –liquid (melt) stability boundary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 509
5.12.3.1. Law of melting: Lindemann . . . . . . . . . . . . . . . . . . . . . . . . . . . 509

Section D: Mineral Systems

Chapter 6. MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes . . . . . . . . . 515
6.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 515
6.1.1. Stability of binary oxides and ternary phases . . . . . . . . . . . . . . . . . . . . . 518
6.1.2. MgO – FeO –SiO2: thermodynamic data and phase equilibria
in the mantle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 520
Table of Contents xxv

6.2. MFS system in Mars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 523

6.3. Mg-olivines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 524
6.3.1. Olivines in the mantle: pyrolite model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 525
6.3.1.1. Mg2SiO4 –Fe2SiO4 system: binary loop in
the mantle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 527
6.3.1.2. Fe –Mg in a – b phases: ordering . . . . . . . . . . . . . . . . . . . . . . . 530
6.3.1.3. Thermal properties of MgO –SiO2 system . . . . . . . . . . . . . . 531
6.3.2. Nucleation rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 532
6.3.3. Elasticity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 533
6.3.3.1. San Carlos olivine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 534
6.3.3.2. Mode-Grüneisen parameter . . . . . . . . . . . . . . . . . . . . . . . . . . . . 537
6.3.4. (Mn, Fe, Co) olivines: compressibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . 537
6.3.5. Post-spinel transitions: phase-boundary study . . . . . . . . . . . . . . . . . . . . . 538
6.3.6. OH2 ions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 539
6.3.6.1. OH-bearing planar defects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 540
6.3.7. Inter-diffusion and activation volume, V p . . . . . . . . . . . . . . . . . . . . . . . . . 542
6.3.8. Seismic and acoustic velocities: VP and VS . . . . . . . . . . . . . . . . . . . . . . . 542
6.3.8.1. a – b system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 543
6.3.8.2. Fe/Mg in velocity relation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 545
6.3.9. Minor element partitioning in a ! b transformation . . . . . . . . . . . . . . 546
6.3.9.1. Cr3þ and Al3þ in wadsleyite . . . . . . . . . . . . . . . . . . . . . . . . . . . 547
6.3.9.2. Ti4þ in olivine/wadsleyite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 548
6.3.9.3. Cr2þ in olivine structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 548
6.3.10. Partition coefficients: olivine– melt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 548
6.3.10.1. Al3þ partitioning: Onuma diagram . . . . . . . . . . . . . . . . . . . . . 549
6.3.11. Compressibility and amorphization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 551
6.4. b-Mg2SiO4 (wadsleyite) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 554
6.4.1. Single-crystal elasticity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 554
6.4.2. Hydrous wadsleyite, b-Mg22xSiH2xO4 (0.00 # x # 0.25) . . . . . . . . . 557
6.4.2.1. H2O in a – b transition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 560
6.4.2.2. Mg-vacant structural module . . . . . . . . . . . . . . . . . . . . . . . . . . . 562
6.4.2.3. Fe in wadsleyite II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 563
6.4.2.4. Fe3þ in protonation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 563
6.5. Olivine ! spinel transition: CFS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 564
6.5.1. Oxygen sublattice transformation (hcp ! fcc):
partial dislocations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 565
6.5.2. Olivine –spinel compressibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 566
6.5.3. g-Mg2SiO4 (ringwoodite and inverse ringwoodite) . . . . . . . . . . . . . . . . 567
6.5.3.1. Under pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 568
6.5.3.2. Thermodynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 570
6.5.3.3. Symmetry analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 571
6.5.4. Olivine-(enstatite) – spinel nucleation in subducting lithosphere . . . . 573
6.5.4.1. Hydrous ringwoodite (g-Mg2SiO4) . . . . . . . . . . . . . . . . . . . . . 573
6.6. Fe2SiO4 systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 574
xxvi Table of Contents

6.6.1. g-Fe2SiO4 spinel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 575

6.6.1.1. Fe2SiO4 –Fe3O4 system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 575
6.6.2. Cr2SiO4 : Cr2þ orthosilicates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 576
6.6.2.1. XRD and electronic spectroscopy . . . . . . . . . . . . . . . . . . . . . . 576
6.6.2.2. M– M bonding and Cr dimerization . . . . . . . . . . . . . . . . . . . . 577
6.6.2.3. Compressional anisotropy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 577
6.6.3. Ni2SiO4: deformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 577
6.6.4. Mg2GeO4 olivine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 577
6.7. Pyroxenes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 578
6.7.1. Structural chains and angles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 581
6.7.2. MgSiO3 –FeSiO3 system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 581
6.7.2.1. MgSiO3 orthopyroxene . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 582
6.7.2.2. Orthoenstitite –clinoenstatite: LCLEN ! HCLEN . . . . . . 585
6.7.2.3. Aluminous orthopyroxene: elasticity and
velocities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 588
6.7.2.4. MD simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 589
6.7.2.5. Ab initio simulation: Hartree – Fock . . . . . . . . . . . . . . . . . . . . 590
6.7.3. Clinopyroxene . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 590
6.7.3.1. C2/c clinoenstatite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 591
6.7.3.2. Diopside – hedenbergite join . . . . . . . . . . . . . . . . . . . . . . . . . . . . 592
6.7.3.3. Enstatite– diopside –jadeite join: garnet . . . . . . . . . . . . . . . . . 599
6.7.3.4. Clinopyroxene and anorthite . . . . . . . . . . . . . . . . . . . . . . . . . . . 602
6.7.3.5. Potassium in clinopyroxene . . . . . . . . . . . . . . . . . . . . . . . . . . . . 602
6.7.3.6. Pyroxene – garnet transition: Martian mantle . . . . . . . . . . . . 603
6.7.3.7. FeSiO3: clinoferrosilite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 606
6.7.3.8. Na-pyroxene . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 610
6.7.4. Akermanite, CaMgSi2O7: incommensurate to normal
phase transition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 612

Chapter 7. (K2O, Na2O, CaO) – Al2O3 – SiO2 System . . . . . . . . . . . . . . . . . . . . . . . . . 613

7.1. KAlSi3O8 – NaAlSi3O8 – CaAlSi3O8 felspars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 613
7.1.1. Bulk moduli . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 615
7.1.2. Compressibilities: M –O and kT –O –Tl . . . . . . . . . . . . . . . . . . . . . . . . . . . 617
7.1.2.1. Unit strains in felspars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 619
7.2. KAlSi3O8 system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 620
7.2.1. Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 621
7.2.2. Phase relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 622
7.2.3. Displacive-phase transition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 623
7.3. Hollandite-type compounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 623
7.3.1. Pb –hollandite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 624
7.4. Anorthoclase (KAlSi3O8 – NaAlSi3O8) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 625
7.4.1. Phase relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 625
7.5. Plagioclase felspars (NaAlSi3O8 –CaAl2Si2O8) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 626
7.5.1. Albite, NaAlSi3O8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 626
Table of Contents xxvii

7.5.1.1. Al –Si order –disorder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 627

7.5.1.2. Low albite: kAl– O – Sil change . . . . . . . . . . . . . . . . . . . . . . . . 628
7.5.2. Anorthite, CaAl2Si2O8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 629
7.5.2.1. Al –Si order –disorder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 630
7.5.2.2. Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 631
7.5.2.3. Phase diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 632
7.5.2.4. P1 $ I 1 transition: non-ferroic displacive . . . . . . . . . . . . . . 634
7.5.2.5. Amorphization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 635
7.5.2.6. Shock transition to glass: Raman results . . . . . . . . . . . . . . . . 635
7.5.3. P1 $ I 1 29Si MAS-NMR spectroscopic study . . . . . . . . . . . . . . . . . . . . . 636
7.6. Reedmergnerite, NaBSiO8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 637

Chapter 8. Al2O3 – SiO2 and (CaO – MgO) –Al2O3 – SiO2 Systems . . . . . . . . . . . . . 639
8.1. Al2O3 –SiO2 system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 639
8.1.1. Sillimanite and andalusite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 641
8.1.2. Kyanite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 642
8.1.2.1. Bulk modulus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 645
8.1.2.2. dP/dT slope and stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 646
8.2 CaO – MgO –Al2O3 –SiO2 (CMAS) system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 646
8.2.1. Thermodynamic equilibria parameters of CMAS system . . . . . . . . . . 648
8.2.2. Garnet structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 649
8.2.2.1. Andradite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 652
8.2.2.2. Pyrope Mg3 Al2(SiO4)3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 652
8.2.3. Pyrope ! ilmenite ! perovskite transformation: Al-content . . . . . . . 654
8.2.4. Almandine (Fe3Al2Si3O12) break-down . . . . . . . . . . . . . . . . . . . . . . . . . . . 654
8.2.5. Factors for garnet compression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 655
8.2.6. Bulk moduli . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 656
8.2.7. Thermal expansion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 657
8.2.8. YAG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 658
8.2.9. Mg –Cr –garnet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 658
8.2.9.1. Cr, Al fractionation in garnets . . . . . . . . . . . . . . . . . . . . . . . . . 660
8.2.10. Tetragonal garnets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 661
8.2.10.1. Majorite garnet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 662
8.2.10.2. Compressibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 668
8.2.10.3. Bulk modulus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 668
8.2.10.4. Vibrational modes: I41/a . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 669
8.2.11. Ca –garnets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 669
8.2.12. Andradite –skiagite solid solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 670
8.2.13. Calderite garnet, Mn3Fe3þ 2 Si3O12 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 671

Chapter 9. AB2X4 Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 673

(A) Oxide Spinels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 675
9.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 677
9.1.1. Normal/inverse spinels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 679
xxviii Table of Contents

9.1.2. CFSE in spinels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 680

9.1.3. JT effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 681
9.1.4. Crystal structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 681
9.1.4.1. Compressibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 682
9.1.5. High-pressure studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 683
9.1.5.1. Compressibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 683
9.1.5.2. Polyhedral bulk moduli, K . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 685
9.2. MgAl2O4 spinel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 685
9.2.1. Spectral models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 687
9.2.1.1. Cr3þ in MgAl2O4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 688
9.2.2. (MgAl2O4)x (Fe3O4)12x solid solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . 688
9.2.3. Order – disorder (OD): cation partitioning . . . . . . . . . . . . . . . . . . . . . . . . . 688
9.2.4. Magnetic behaviour: MS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 689
9.3. Magnetite, Fe3O4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 690
9.3.1. h-Fe3O4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 693
9.3.1.1. EOS and molar volume . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 694
9.3.1.2. Néel temperature, TN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 695
9.3.2. Pressure dependence of u and a . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 696
9.3.3. Polyhedral bulk modulus, K . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 697
9.3.4. Ca-ferrite structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 697
9.3.4.1. CaMn2O4 and Mn3O4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 698
9.3.5. Fe3O4, MgAl2O4 and g-Ni2SiO4 spinels . . . . . . . . . . . . . . . . . . . . . . . . . . 698
9.3.5.1. MgAl2O4 and MgO: elastic constants
and sound velocities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 699
9.3.6. Electrical resistivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 700
9.3.7. g-Fe2O3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 701
9.4. Cr-spinels, MCr2O4 (M ¼ Mg, Mn, Zn): decomposition . . . . . . . . . . . . . . . . . . . 701
9.4.1. Oxidation of Cr-spinel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 702
9.4.2. OD in Cr-spinels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 703
9.4.2.1. Thermopower (Q) and conductivity . . . . . . . . . . . . . . . . . . . . 703
9.4.3. Defects and electrical behaviour . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 704
9.4.3.1. Shocked chromite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 705
9.4.4. Chromite: post-spinel orthorhombic polymorph . . . . . . . . . . . . . . . . . . . 706
9.4.4.1. Chromite: Raman bands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 706
(B) Sulfide Spinels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 707
9.4.5. ZnCr2S4 spinel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 709

Chapter 10. ABX3, Perovskite – Ilmenite Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . 711

10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 711
10.1.1. Magnetic ordering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 713
10.1.2. Perovskite and mantle convection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 713
10.1.3. Layered mantle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 714
10.1.4. Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 714
Table of Contents xxix

10.1.5. Structure and types of perovskites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 714

10.1.5.1. Defects in oxide perovskites . . . . . . . . . . . . . . . . . . . . . . . . . . . 718
10.1.5.2. Fe2O3 perovskite: TM and “magnetic hardening” . . . . . . . . 720
10.1.6. MgSiO3 Perovskite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 720
10.1.6.1. Atomistic simulation: MEG . . . . . . . . . . . . . . . . . . . . . . . . . . . . 721
10.1.6.2. Phonon spectrum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 721
10.1.7. Perovskite melting and bouyancy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 726
10.2 MgO –(FeO)– SiO2 system: perovskites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 727
10.2.1. Tolerence factor, t . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 727
10.2.2. Silicate perovskites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 729
10.2.2.1. Orthorhombic – tetragonal – cubic transitions . . . . . . . . . . . . 729
10.2.2.2. Ferroelectricity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 730
10.2.2.3. MgSiO3 –FeSiO3 perovskites . . . . . . . . . . . . . . . . . . . . . . . . . . 731
10.2.2.4. Shear moduli: ultrasonic interferometry . . . . . . . . . . . . . . . . 731
10.2.2.5. Vibrational modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 732
10.2.3. Elasticity: modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 734
10.2.3.1. Wave velocities: anisotropy . . . . . . . . . . . . . . . . . . . . . . . . . . . . 736
10.2.4. Thermoelasticity and expansivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 737
10.2.4.1. Bulk modulus and EOS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 737
10.2.4.2. Lattice compressiblity and KT . . . . . . . . . . . . . . . . . . . . . . . . . . 739
10.2.5. XRD results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 740
10.2.6. (Mg,Fe)SiO3 – perovskite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 740
10.2.6.1. Iron in perovskite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 741
10.2.6.2. Fe2þ in perovskites: A-site occupancy . . . . . . . . . . . . . . . . . . 743
10.2.6.3. Temperature-dependent electron delocalization . . . . . . . . . 746
10.2.6.4. Defect equilibria and M3þ: physical properties . . . . . . . . . 748
10.2.7. Glassy phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 749
10.3. Transformations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 749
10.3.1. Activation energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 749
10.3.2. Perovskite breakdown: volume change . . . . . . . . . . . . . . . . . . . . . . . . . . . 750
10.3.3. Vibrational models: intrinsic anharmonic effects . . . . . . . . . . . . . . . . . . 750
10.3.4. Raman study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 750
10.3.4.1. Soft-mode transition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 751
10.3.5. Ilmenite structure (R3̄) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 752
10.3.5.1. MgSiO3 ilmenite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 752
10.4. CaO – SiO2 system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 755
10.4.1. CaSiO3 perovskite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 755
10.4.1.1. LAPW calculations: phonon spectrum
and transition temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 756
10.4.1.2. Density and acoustic velocity . . . . . . . . . . . . . . . . . . . . . . . . . . 757
10.4.2. Pseudo-wollastonite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 757
10.4.3. CaSiO3 – CaTiO3 join . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 758
10.5. MgO(CaO)– SiO2(GeO2) –Al2O3 system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 758
10.5.1. Ca –Al perovskite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 759
xxx Table of Contents

10.5.2. Ca –Ge perovskites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 760

10.5.2.1. IR modes: Ca translation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 760
10.6. Alkaline-earth perovskites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 761
10.6.1. Li(Nb,Ta)O3 ferroelectrics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 761
10.6.1.1. Ferroelectric and para-electric structures . . . . . . . . . . . . . . . 763
10.6.2. RE orthoferrites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 764
10.6.2.1. (Sr/Ca) FeO3 perovskite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 765
10.7. Titanate perovskites and ilmenites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 765
10.7.1. CaTiO3 – FeTiO3 join . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 766
10.7.2. MgTiO3 –FeTiO3 join . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 768
10.7.2.1. MgTiO3, geikielite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 771
10.7.2.2. High-temperature phase transition
(without order –disorder) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 774
10.7.2.3. FeTiO3 structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 774
10.7.2.4. Shocked FeTiO3 –ilmenite: Mössbauer study . . . . . . . . . . . 778
10.7.3. Xenoliths in Kimberlites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 780
10.7.4. BaTiO3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 781
10.7.4.1. Ferroelectricity and ferroelasticity . . . . . . . . . . . . . . . . . . . . . . 784
10.7.4.2. Linearized augmented plane wave (LAPW)
calculations: surface effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . 786
10.7.4.3. Multiple-site model for perovskite ferroelectrics . . . . . . . . 786
10.7.4.4. Ferroelectric instability: “rattling-ion” model . . . . . . . . . . . 787
10.7.5. PbTiO3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 788
10.7.6 Other titanates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 789
10.7.6.1. MgTi2O5 karrooite: order – disorder . . . . . . . . . . . . . . . . . . . . 789
10.7.7. Ti/Nb perovskites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 790
10.8. Mn – oxide perovskites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 791
10.8.1. “Ruddlestone-Popper” series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 792

Chapter 11. Silicate Melts and Rocks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 793

11.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 793
11.1.1. Magmatic melt under pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 795
11.2. Alumino-silicate melts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 796
11.2.1. CaO – Al2O3 – SiO2 melts: compressibilities . . . . . . . . . . . . . . . . . . . . . . . 796
11.2.2. Na2O –Al2O3 – SiO2 melts: Ab50NTS50 . . . . . . . . . . . . . . . . . . . . . . . . . . . 797
11.3. Viscosity: controlling factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 798
11.3.1. Diffusivity: Stokes– Einstein equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 798
11.3.2. Temperature dependence: Arrhenian approximation . . . . . . . . . . . . . . . 798
11.3.3. Alkali oxides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 799
11.3.4. Water effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 799
11.3.5. Pressure effects on viscosity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 799
11.3.6. Silicate polymerization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 800
11.3.7. Density and viscosity determination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 801
Table of Contents xxxi

11.3.8. Melt percolation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 801

11.3.9. Crystal – melt phase equilibria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 802
11.3.9.1. fO2, fH2O and aH2O . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 802
11.4. H2O in silicate melts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 803
11.4.1. K2O –SiO2 – H2O system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 803
11.5. REE patterns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 804
11.5.1. Fe3þ in glass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 804
11.5.2. Partition coefficient in melt/solid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 804
11.6. Rocks under pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 806
11.6.1. Transformation under shock: pseudotachylites . . . . . . . . . . . . . . . . . . . . 806
11.6.2. Terrigenous and pelagic sediments under subduction . . . . . . . . . . . . . . 806
11.6.2.1. Density change and buoyancy . . . . . . . . . . . . . . . . . . . . . . . . . . 807
11.6.2.2. Potassium mobility in subduction pressures . . . . . . . . . . . . . 809
11.6.2.3. Lead paradox . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 809
11.6.2.4. Subducting slabs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 810
11.6.3. Ultra-high-pressure metamorphism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 810
11.6.3.1. Coesite –diamond . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 810
11.6.3.2. Crustal metamorphic regimes . . . . . . . . . . . . . . . . . . . . . . . . . . 811
11.6.3.3. Hot and cold eclogites: collision/subduction zones . . . . . . 811
11.6.3.4. Dabie – Sulu collision zone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 812
11.6.3.5. Alpe Arami UHP lherzolite . . . . . . . . . . . . . . . . . . . . . . . . . . . . 813
11.6.3.6. Exsolutions in VHP minerals . . . . . . . . . . . . . . . . . . . . . . . . . . 815
11.6.4. Basalts and eclogites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 816
11.6.4.1. Dehydration melting of metabasalt
at 0.8 –3.2 GPa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 817
11.6.5. MORB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 818
11.6.6. Komatiite, picrite and lherzolite: CaO – MgO(FeO) –SiO2
systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 819
11.6.7. Garnet peridotites: “forbidden zone” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 820
11.6.7.1. Exsolutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 821
11.6.7.2. Emplacement of garnet peridotites . . . . . . . . . . . . . . . . . . . . . 821

Chapter 12. Simple Oxides and Carbonates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 823

12.1 Dioxides: SiO2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 823
12.1.1. a-Quartz: structural change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 826
12.1.1.1. Fracture strength . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 828
12.1.2. Stishovite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 828
12.1.2.1. Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 829
12.1.2.2. SiO6: densities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 830
12.1.2.3. Elastic moduli . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 830
12.1.2.4. Compressibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 831
12.1.3. Fluorite (CaCl2) structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 832
12.1.3.1. EOS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 834
12.1.3.2. Theoretical models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 837
xxxii Table of Contents

12.1.3.3. Columbite (a-PbO2) structure: MD simulation . . . . . . . . . . 839

12.1.3.4. Seismic velocities: discontinuities . . . . . . . . . . . . . . . . . . . . . . 840
12.1.4. Cristobalite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 840
12.1.4.1. Structure: phase transition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 840
12.1.4.2. Phase transition: symmetry change . . . . . . . . . . . . . . . . . . . . . 843
12.1.4.3. Cristobalite III . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 844
12.1.4.4. Raman study: I ! II transition . . . . . . . . . . . . . . . . . . . . . . . . . 844
12.1.5. a-quartz, coesite, stishovite and cristobalite . . . . . . . . . . . . . . . . . . . . . . . 846
12.1.5.1. Coesite to quartz transformation kinetics . . . . . . . . . . . . . . . 847
12.1.6. Instability and ferroelastic transition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 848
12.1.6.1. A new phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 849
12.1.7. Amorphization experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 849
12.1.7.1. a ! c growth rate: magma viscosity . . . . . . . . . . . . . . . . . . . 850
12.1.7.2. SiO2 glass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 850
12.2. ZrO2 –SiO2: shocked . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 853
12.3. TiO2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 854
12.4. Simple monoxides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 854
12.4.1. MgO, FeO, CoO, MnO and NiO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 854
12.4.1.1. MgO and CoO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 854
12.4.1.2. MnO and FeO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 855
12.4.1.3. Normal and inverse NiAs structures . . . . . . . . . . . . . . . . . . . . 857
12.4.1.4. FeO in D00 zone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 857
12.4.1.5. MgO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 857
12.4.1.6. Elasticity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 861
12.4.1.7. B1 – B2 phase transition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 862
12.4.1.8. Elastic constants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 862
12.4.2. FeO at high P –T . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 862
12.4.2.1. Magnetic-phase transition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 866
12.4.2.2. NiAs phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 867
12.4.3. FexO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 869
12.4.3.1. Wüstite (Fe12xO) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 871
12.4.3.2. Fe –FeO system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 872
12.4.3.3. Fe –FeO þ diluting elements: solid solution under P . . . . 873
12.5. Carbonates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 874
12.5.1. CaCO3 calcite ! aragonite polymorphism . . . . . . . . . . . . . . . . . . . . . . . . 874
12.5.1.1. Calcite, CaCO3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 876
12.5.1.2. Compressibility and bulk modulus . . . . . . . . . . . . . . . . . . . . . 877
12.5.1.3. Oxy-anion– cation packing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 878
12.5.2. Mg-carbonates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 880
12.5.2.1. Dolomite stability at depths . . . . . . . . . . . . . . . . . . . . . . . . . . . . 882
12.6. Other carbonates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 883
12.7. CaO – MgO –SiO2 – CO2 system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 883
12.7.1. CaO – MgO – SiO2 – CO2 –H2O system: XCO2 . . . . . . . . . . . . . . . . . . . . . . 883
Table of Contents xxxiii

Chapter 13. Hydrous Minerals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 885

13.1. Water in primary minerals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 885
13.1.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 885
13.1.2. Hydrous minerals under pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 888
13.1.2.1. Water in subducting slabs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 889
13.1.3. H(D) – O bonds in hydroxides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 893
13.1.3.1. OH bonds: dn/dP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 893
13.2. Geophysical effects of water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 894
13.2.1. Creep rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 895
13.2.2. Electrical conductivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 895
13.3. H2O in the mantle and magmatic melt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 895
13.3.1. H2O in plagioclase crystallization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 896
13.4. MgO –SiO2 – H2O ternary system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 897
13.4.1. DHMS phases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 898
13.4.1.1. MgO – SiO2 þ volatiles (H2O, F2, Cl2) system . . . . . . . . . . 900
13.4.1.2. Halogens in DHMS phases . . . . . . . . . . . . . . . . . . . . . . . . . . . . 902
13.4.2. NMR spectroscopic study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 902
13.4.3. Choke point . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 903
13.4.4. Serpentine and phase A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 904
13.4.4.1. Serpentine, Mg3SiO5(OH)4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 904
13.4.4.2. Phase A (Mg7Si2O8(OH)6) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 904
13.4.4.3. Chrysotile transformations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 907
13.4.4.4. Talc and phase A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 908
13.4.4.5. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 908
13.4.4.6. 10-Å phase, Mg3Si4O10(OH)2, n H2O . . . . . . . . . . . . . . . . . . . 909
13.4.4.7. 3.65-Å phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 909
13.4.5. Anhy-B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 909
13.4.5.1. Octahedral sites: M3 site . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 911
13.4.5.2. Phase B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 911
13.4.5.3. NMR study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 912
13.4.6. Phase D (MgSi2H2O6) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 917
13.4.6.1. Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 919
13.4.6.2. Density and bulk modulus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 921
13.4.6.3. Anisotropic compressibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . 922
13.4.7. Phase E (Mg2.08Si1.6H3.2O6) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 922
13.4.7.1. Related phases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 924
13.4.7.2. Phase F(?) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 925
13.4.8. Phase G and other MSH phases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 925
13.5. Humite group minerals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 925
13.5.1. Clinohumite and chondrodite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 925
13.5.2. Elastic properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 926
13.5.3. Clinohumite-OH and chondrodite-OH . . . . . . . . . . . . . . . . . . . . . . . . . . . . 927
13.6. MgO –Na2O – SiO2 – H2O system: hydrated aenigmatites . . . . . . . . . . . . . . . . . . . 927
13.6.1. Hydrated-Na – aenigmatite: crystal structure . . . . . . . . . . . . . . . . . . . . . . . 928
xxxiv Table of Contents

13.7. CaO – Al2O3 –SiO2 – H2O system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 928

13.7.1. Zoisite and clinozoisite, Ca2Al2 (Al12pFep) (O/OH/Si2O7/SiO4) . . . 928
13.7.1.1. Compressibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 930
13.7.1.2. Thermal expansivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 931
13.7.1.3. dKT /dT and Anderson – Grüneisen parameter . . . . . . . . . . . . 932
13.7.1.4. Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 932
13.7.1.5. Zoisite and lawsonite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 933
13.7.1.6. Subducting andesitic rocks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 935
13.8. CaO – Al2O3 –SiO2 – H2O system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 935
13.8.1. Amphiboles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 935
13.8.1.1. Kaersutitic amphibole: oxidation – hydrogenation
reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 936
13.8.1.2. Kaersutite in SNC meteorites: Martian H2O . . . . . . . . . . . . 938
13.8.2. High Fe3þ content and the aH2O . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 939
13.8.3. Lawsonite, CaAl2Si2O7·H2O . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 940
13.8.3.1. K0 and a values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 942
13.8.3.2. Compressibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 943
13.8.3.3. IR study: Grüneisen parameter and thermal
expansion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 944
13.9. MgO –Al2O3 –SiO2 – H2O system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 946
13.9.1. Muscovite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 947
13.9.1.1. Phlogopite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 947
13.9.1.2. Phase X . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 948
13.10. K2O – MgO – Al2O3 –SiO2 – H2O system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 948
13.10.1. Cold geotherms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 948
13.11. Al2O3 –SiO2 – H2O (ASH) system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 949
13.11.1. AlSiO3OH, “phase egg” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 950
13.11.1.1. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 953
13.11.1.2. Phase egg in subduction zone . . . . . . . . . . . . . . . . . . . . . . . . . . 954
13.12. Clay minerals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 954
13.12.1. Structural disorder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 955
13.12.2. 19- and 15-Å hydrate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 955
13.12.3. Interlayer cations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 956
13.12.4. Kaolinite: Raman study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 956
13.12.5. Chlorite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 957
13.13. Hydrous oxides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 957
13.13.1. Hydrous silica: Shergotty and LM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 957
13.13.2. AlO(OH), diaspore . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 958
13.13.3. Mg(OH)2, brucite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 959
13.13.3.1. XRD study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 961
13.13.3.2. IR study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 962
13.13.3.3. Raman study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 963
13.14. Portlandite (Ca(OH)2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 965
Table of Contents xxxv

Chapter 14. Iron and Siderophile Elements: The Earth’s Core . . . . . . . . . . . . . . . . . 967
14.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 967
14.1.1. Theories of iron under pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 968
14.1.1.1. Energy bands and electron transitions at core . . . . . . . . . . . 969
14.1.1.2. Phase predictions from theoretical calculations . . . . . . . . . 970
14.1.1.3. First-principles approximation: bcc and hcp . . . . . . . . . . . . 971
14.1.2. fcc and hcp phases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 972
14.1.2.1. P –r relationship . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 975
14.2. Iron core . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 977
14.2.1. Core iron . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 978
14.2.2. Anisotropy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 978
14.2.3. EOS and melting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 979
14.2.4. Density deficit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 980
14.2.5. Iron phases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 983
14.2.5.1. Stability of bcc and fcc phases . . . . . . . . . . . . . . . . . . . . . . . . . 985
14.2.5.2. b-Fe (dhcp) phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 986
14.2.5.3. 1 ! Pbcm iron transformation . . . . . . . . . . . . . . . . . . . . . . . . . 988
14.2.5.4. Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 990
14.2.5.5. Thermal Grüneisen parameter . . . . . . . . . . . . . . . . . . . . . . . . . . 990
14.2.5.6. Vibrational modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 991
14.2.6. Phase boundaries and the triple points . . . . . . . . . . . . . . . . . . . . . . . . . . . . 992
14.2.6.1. a –1 – g triple point and 1 – g transition . . . . . . . . . . . . . . . . . 992
14.2.6.2. 1 –g – l triple point . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 995
14.2.6.3. Liquid iron: structural change under P . . . . . . . . . . . . . . . . . 995
14.2.7. Elasticity and rheology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 996
14.2.7.1. Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 996
14.2.7.2. Shear viscosity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 998
14.2.8. Rigid core: slichter modes of translational motion . . . . . . . . . . . . . . . . 999
14.2.9. Outer core . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 999
14.2.10. Inner core . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1000
14.2.10.1. Heat sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1001
14.2.10.2. Rotation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1001
14.2.10.3. Crystalline structure: elastic/seismic behaviour . . . . . . . . 1002
14.2.10.4. Anisotropism: axial angle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1004
14.2.10.5. Discontinuities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1008
14.2.10.6. Geodynamo: convection and av-dynamo . . . . . . . . . . . . . . 1009
14.2.10.7. Geomagnetic-field propagation . . . . . . . . . . . . . . . . . . . . . . . . 1010
14.2.10.8. Magnetic field, heat flow and plate tectonics . . . . . . . . . . . 1011
14.3. 1-Fe and paramagnetism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1011
14.3.1. Hugoniot temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1012
14.4. Iron and tungsten: yield strengths . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1012
14.5. Fe– Ni alloy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1013
14.6. Fe– Si alloy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1015
14.7. Fe– H system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1016
xxxvi Table of Contents

14.8. Sulphur in the core . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1017

14.8.1. Oxygen and sulphur solution in iron . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1017
14.8.1.1. S, Se and Te . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1018
14.9. Iron sulphides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1018
14.9.1. FeS: five polymorphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1019
14.9.1.1. FeS III, monoclinic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1019
14.9.1.2. FeS IV, hexagonal (2a,c) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1020
14.9.1.3. FeS V: hexagonal (a,c) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1022
14.9.1.4. Spin state of ferrous iron . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1023
14.9.1.5. R (Fe – S) change and spin splitting . . . . . . . . . . . . . . . . . . . . 1026
14.9.2. FeS: Martian CMB and core . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1028
14.9.3. Fe –FeS system: eutectic points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1028
14.9.3.1. Fe3S2, Fe3S, Fe2S . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1028
14.9.4. FeS2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1029
14.9.4.1. FeS2, pyrite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1029
14.9.4.2. Pyrrhotite: magnetic transition . . . . . . . . . . . . . . . . . . . . . . . . . 1030
14.10. Fe3S2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1030
14.11. Mn – S system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1031
14.11.1. a-MnS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1031
14.11.2. MnS2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1031
14.11.3. (Fe,Mg)S and (Fe,Mn)S . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1032
14.12. Pressure behaviour of FeS vs. FeO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1033
14.13. Fe(Ni) –Cu – S compounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1033
14.13.1. Nickel sulphides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1033
14.13.2. Cubanite, CuFe2S3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1034
14.14. Phosphates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1035
14.14.1. Berlinite, AlPO4: memory glass (?) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1035
14.14.2. Farringtonite – Mg3 (PO4)2-II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1037
14.14.3. Apatite (Ca5(PO4)3, (F,Cl,OH)) – monazite topotaxy: REE . . . . . . . . . 1038
14.14.4. Bearthite (Ca2Al(PO4)2OH) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1038

Section E: Transport Properties at Deep Depths & Related Condensed

Matter Phenomena

Chapter 15. Transport Properties in Deep Depths and Related

Condensed-Matter Phenomena . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1041
15.1. Transport properties under pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1041
15.1.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1041
15.1.2. Electrical conductivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1042
15.1.2.1. Electrical conductivity and activation energy . . . . . . . . . . . 1044
15.1.2.2. Techniques employed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1045
15.1.2.3. Conductivity of minerals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1045
15.2. Electron/hole transfer and magnetic behaviour . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1047
Table of Contents xxxvii

15.2.1. Polarons: small and large . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1048

15.2.2. Ferroelectricity: regimes and local well potential . . . . . . . . . . . . . . . . . 1049
15.3. Insulator to superconductivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1050
15.3.1. Superconductivity and magnetism: “co-habitation” . . . . . . . . . . . . . . . . 1051
15.3.1.1. “Magnetic glue” and failed spin: ghost magnetism . . . . . . 1052
15.3.2. Novel physics: paramagnetic Meissner effect . . . . . . . . . . . . . . . . . . . . . 1053
15.3.2.1. Mesoscopic magnetism: frustration and
superconducting loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1053
15.3.3. Double exchange in magnetic transition . . . . . . . . . . . . . . . . . . . . . . . . . . 1054
15.3.4. “Skutterudites” and chalcogenides: “holey” and
“unholey” semiconductors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1055
15.4. RE-Mn perovskite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1056
15.4.1. La12xCaxMnO3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1056
15.4.1.1. Resistivity and magnetism under pressure . . . . . . . . . . . . . . 1057
15.4.1.2. CMR: RE12xAxMnO3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1059
15.4.2. Pressure on polarons, activation energy and charge
carrier mobility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1061
15.4.2.1. Electron– lattice coupling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1062
15.4.3. LaMnO3 perovskite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1063
15.4.3.1. M– I cohabitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1063
15.5. La12xSrxMnO3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1064
15.6. Pr-manganates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1066
15.6.1. Pr12xCaxMnO3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1067
15.6.2. (La12yPry)12xAxMnO3: short- and
long-range order . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1067
15.7. Fe3þ in perovskite and conductivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1067
15.8. Al2O3 content and conductivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1068
15.9. Conductive TiO2, SiO2, FeO, Fe2O3 and Fe3O4 . . . . . . . . . . . . . . . . . . . . . . . . . . . 1069
15.10. Thermal conductivity, k . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1071
15.10.1. k at mantle depths . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1072
15.10.1.1. Radiative and lattice contribution . . . . . . . . . . . . . . . . . . . . . . 1073
15.10.2. k at D00 zone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1074
15.10.3. k and convective power of core . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1077
15.10.4. k under shock pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1077
15.11. Ferroelectric transitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1077
15.11.1. Ferroelectric phenomena in large planets . . . . . . . . . . . . . . . . . . . . . . . . . 1078
15.12. Non-elastic transport properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1078
15.12.1. Power law: fractal distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1079
15.12.2. Diffusion: self and co-operative . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1080
15.13. Defects, dislocations and deformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1080
15.13.1. Defects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1080
15.13.2. Dislocations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1081
15.13.3. Dislocation recovery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1081
15.13.4. Deformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1082
xxxviii Table of Contents

15.13.4.1. Olivine deformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1083

15.13.4.2. Deformation: single crystal to polycrystalline
mass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1084
15.14. Diffusion, creep and viscoplastic deformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1084
15.14.1. Diffusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1084
15.14.1.1. Mg, Fe diffusion: olivine, pyroxene and garnet . . . . . . . . . 1085
15.14.1.2. Grain-boundary diffusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1085
15.14.2. Creep . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1086
15.14.2.1. Creep rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1086
15.14.2.2. Dislocation (power-law) creep ! diffusion creep . . . . . . . 1087
15.14.2.3. Dislocation creep and spinel deformation . . . . . . . . . . . . . . . 1089
15.14.2.4. Slip systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1090
15.14.2.5. Creep, diffusion rate and conduction . . . . . . . . . . . . . . . . . . . 1090
15.14.3. Diffusivity and viscosity: Stokes –Einstein diffusivity . . . . . . . . . . . . . 1091
15.14.3.1. Silicate melts: O, Si diffusion . . . . . . . . . . . . . . . . . . . . . . . . . . 1091
15.14.4. Intrinsic and extrinsic regimes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1092
15.15. Cation (Mg, Fe and Ni) diffusion in olivine: a ! b –g transition . . . . . . . . . . 1093
15.15.1. a , b , g transitions and upper-mantle rheology . . . . . . . . . . . . . . . . 1094
15.15.2. Chemical diffusion in the slab and transition zone . . . . . . . . . . . . . . . . 1095
15.15.2.1. Homogenization rate in mantle . . . . . . . . . . . . . . . . . . . . . . . . . 1096
15.15.2.2. Diffusion in lower mantle: MgO
(Ita and Cohen, 1998) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1096
15.15.3. Rheology of lower mantle: strain rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1097
15.16. Transformational plasticity: partial dislocation, martensitic
or synchro-shear mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1098
15.17. Oxygen fugacity in the Earth’s dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1099
15.17.1. Solid-state diffusion: f O2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1099
15.17.1.1. Olivine/iron buffer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1099
15.17.1.2. Ferric iron and redox zone: crustal recycling . . . . . . . . . . . 1100
15.17.1.3. Core –mantle partitioning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1100

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1103

Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1205

Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1223

 Section A

Preamble and Preview

This page is intentionally left blank
Preamble and Preview 3

It is believed that the Earth, the other eight planets and the satellites and asteroids were
formed along with the Sun about 500 million years ago. Earth accreted most of its mass
through catastrophic impacts of large planetesimals during its early phase. The Moon was
formed from the accumulation of dust cloud generated by the giant impact of a Mars-size
body on the proto-Earth, when both the impactor and the Earth already had a metal proto-
core. Numerical modelling indicates that much of the impactor core many have ploughed
its way through the mantle and merged with Earth’s core. Concurrently, the kinetic energy
buried deep within the Earth would have been sufficient to melt a large portion of the
silicate mantle. It could also have been that the metal from the proto-planets was
excavated, re-mixed and re-equilibrated with the molten mantle.
Very often on a clear night, we may see the shooting stars which are meteor falls.
These glowing objects constitute only the visible contribution of interplanetary and
interstellar material to the Earth but a huge amount of such extra-terrestrial material,
mostly non-glowing and cold, is being added to the mass of the Earth undetected every
minute, giving a total of about 10,000 tons per year. Scientists at the Physical Research
Laboratory, Ahmedabad have in recent years found evidence that some rock and mineral
fragments found in the oceanic bottom of the Pacific are cosmic in origin; that is, they
originated beyond our Solar System.
The Earth has segregated layers as shells within it and we live on a thin outer crust
composed of landmasses and oceanic bottoms. However, compared with the Earth, this
crustal thickness is no thicker than the skin to an apple, accounting only for less than 1% of
the volume. Below the continents, this is about 25 km thick while below the oceans,
particularly below the Pacific, it is much thinner at about 5 km.
Since the Earth’s and the planetary interiors are not directly available for sampling,
we know the interior of our planet from several sources such as direct sampling of rocks
brought to the surface by geologic process from the depths (#00 km), by seismology, from
the chemistry of meteorites and other geochemical arguments and from laboratory and
computational simulations of the conditions at depth.
Laboratory techniques are needed to characterize the materials satisfying the
observed large-scale geophysical (density distribution within the planetary interior,
seismic-waves velocity) and geochemical (abundance of chemical elements in the Solar
System, element distribution within planetary bodies) parameters governing the global
planetary processes such as thermal flow, the mantle convection regime and plate
tectonics.
Investigations into the behaviour of minerals under extreme conditions link the
scale of electrons and nuclei with the global processes of the Earth and other planetary
bodies. The material properties arise directly from the play of electrons and nuclei. The
materials are studied with respect to their fundamentals, starting from their chemical
properties — such as crystal chemistry, thermo-chemistry, element partitioning and
melting — to their physical properties — such as equation of state, elasticity rheology,
vibrational dynamics, bonding, electronic structure and magnetism. Under pressure, the
reactivity and chemical affinity are altered with changes in ionic radii, electronic structure
and bonding characters.
4 Section A

Experiments defining molar volume as a function of pressure clarify the nature and
character of phase transitions that may possibly occur in the crust and mantle. The molar
volume and bulk modulus show empirical inverse correlation and can be described in
terms of an electrostatic model of interatomic forces. Silicate minerals may be modelled as
ionic compounds with bond strengths determined to a first approximation by Coulombic
forces.
With increase in pressure, the difference of free energies between the reactants and
the products increases and the products become more and more stable relative to the
reactants. Also, as temperature increases, the transition pressure decreases. Structural
investigations using high-pressure – temperature (HPT ) spectroscopic methods have
helped our understanding of viscosity – density relationships, solubility and speciation of
various melt components, and often have offered insights into melt –melt and mineral –
melt relationships, such as element partitioning and the immiscibility of liquids.
In extending our knowledge how Earth’s materials combine, dissociate and
transform under high temperatures and pressures, pioneering contributions were made by
such stalwarts as Bernal, Goldschmidt, Bridgman, Birch, Ringwood and many others in
their likeness.
To understand what happens to the crustal and mantle minerals and also at cores
under P; T conditions inside the Earth, laboratory simulations are carried out which
involve the most versatile set-up with diamond cells, in which a tiny specimen is
compressed between the tips of two gem-quality diamonds. Since the surface area of the
tips are very tiny, the force per unit area can reach pressures beyond 500 GPa (which
occurs perhaps at the core of the giant planet, Jupiter). In pressure studies, the sample can
be observed right through the diamond, which may thus serve as a window as well.
To reach ultra-high-pressures (i.e., above 102 GPa), the volume of the sample
required is small. To receive a useful signal from such a micro-sample, synchrotron
radiation of high brilliance is employed. Newly developed high-pressure, high-precision
X-ray tools such as monochromatic radiation with modern detectors, with short-time
resolution and employing long duration times, now make it possible with third-generation
synchrotron sources to study the rheology of deep-Earth materials under pressure. In high-
pressure cells using synchrotron radiation deviatoric, stresses down to 0.1 GPa and to
pressure approaching 300 GPa can be quantified.
A combination of synchrotron X-ray diffraction and Brillouin spectroscopy yields
complete geophysical information comparable with seismic observations of the Earth’s
interior. Equation-of-state data play a central role in describing the convective dynamic of
the Earth’s deep interior. Unprecedented breakthroughs, deemed inconceivable in pressure
studies only a few years back, have been possible because of the synergistic coupling of
the precisional mights of the diamond-anvil cells with the powers of synchrotron radiation
and laser heating.
Considerable developments in state-of-the-art instrumentation have been made
possible. Megabar experimentations have become a reality on all kinds of materials from
several micron thick thin-films to single crystals. New prospective grounds seem to appear
in the seams between the traditional disciplines and new discoveries result from the
potentially high pay-off ventures.
Preamble and Preview 5

Molecular structures are effectively studied by X-ray spectroscopy — bright X-rays

are produced from large accelerators. The Geophysical Research Laboratory (GRL),
Washington uses synchrotron photon beam sources, available from second-generation
National Synchrotron Light Sources (NSLS) and third-generation Advanced Photon
Source (APS) at Argonne National Laboratory. GRL has devised and employed a fully
dedicated high-pressure synchrotron infrared spectroscopy from NSLS.
There are only a few centres around the world which can carry out research on
pressures up to 365 GPa (the pressure believed to be present at the centre of the Earth) and
beyond by simultaneously employing synchrotron radiation for studies on the diffraction
properties of the crystallites. The dimension of such endeavours can be gauged from the
description of one such as the Center for High-pressure Research (CHiPR), which is a
consortium of laboratories such as the Geophysical Laboratory, State University of
New York at Stony Brook, University of California at Davis and University of California
at San Diego. This consortium is well-linked with Lawrence Livermore Laboratory and
the University of Nevada at Las Vegas and also with the activities at Bayreuth Geoinstitute
in Germany and the Institute for Study of Earth’s Interior (ISEI) in Misasa, Japan. A
critical part of the consortium’s work is the use of the third-generation X-ray synchrotron
source at Brookhaven National Laboratory. The Advanced Light Source (APS)
synchrotron facility at Aragonne National Laboratory is also utilized in such high-
pressure studies on “geochemical materials”. The main objective of this linked centre is to
address the questions primarily relating the evolution, properties and dynamics of the
planetary bodies.
High shock pressure work using laser is used at several places, viz. the Nova laser
at Lawrence Livermore National Laboratory, California (Collins and his colleagues), the
Phebus laser in France, the Gekko laser in Japan, etc. The potentiality of this laser-shock
lies in studying iron and even water, which may form a good part of larger planetary cores.
The experiments at Livermore NL show how deuterium and hydrogen behave in hot,
pressurized interiors of giant planets.
As stated, the experimental simulation studies are often complemented by
theoretical computer simulations. In theoretical high-pressure mineral physics, funda-
mental principles of quantum physics are employed in the calculations (e.g., first-
principles calculation and so on) to predict and understand the experimental results. For
example, Ronald Cohen and his associates at Geophysical Laboratory use a 12-Processor
or Cray J916/8-1024 to meet the requirements of state-of-the-art first-principles
computations (which indeed are immense!) for geochemical problems. The areas that
are being unravelled include the equations of state, phase diagrams, elasticity and
anelasticity of important mantle and core phases and melts, and the energetics, nuclear
magnetic resonance effects and the dynamics of natural organic systems and organo-
metallics. Lattice dynamical calculations may be carried out using either ab initio (first-
principles) or empirically derived inter-atomic potentials. In addition to simulation
models, global positioning satellites and satellite altimetry have opened up new ranges of
observation of the Earth in space and time.
In the dynamic behaviour of the Earth’s mantle, the phase transformations under
high pressure play a significant role. An understanding of these transformations is
6 Section A

important for the evaluation of the transformation kinetics and the effects of phase
transitions on the mantle rheology and convection. The rheological changes associated
with phase transitions may involve triggering the deep-focus Earthquakes.
We know many details of the internal structure of the Earth from shock waves
generated artificially or from natural Earthquakes. These waves get reflected and refracted
from different layers within the Earth and we receive information about the structures
inside. Below the crust and separated by an elasticity break, known as Moho, is the mantle,
at the upper part of which melts accumulate and come out as lava flows during volcanic
eruptions. Volcanism, Earthquakes, continental drift and formation of mountain chains
like the Alpine-Himalayan, etc., can be attributed to continuous movement in the mantle.
Below the mantle is the core, which has a radius of about 1,800 miles and is mostly
composed of iron and nickel.
The atmosphere we live in has been formed mostly by the belching out of gas,
mainly oxygen and nitrogen, from inside the stomach of the Earth through volcanism and
fumeroles during the geological past of about 5 billion years. In modern hydrothermal
vents, transition-metal sulphides are known to exist. These can promote geochemical
reactions that mimic key metabolic reactions in living organisms, where enzymes are used
as catalysts. However, under hydrothermal conditions, several reactions in the citric acid
metabolic cycle proceed without enzymes. Again, to the citric acid cycle, a synthesis of
pyruvic acid serves as the key molecular entry point. In the presence of iron oxides and
sulphides, nitrogen gas can be reduced to ammonia, which could be a major constituent of
the planetary atmosphere and that of the early Earth.
Transition-metal sulphides can promote carbon addition reactions, which are
important in synthesizing larger organic molecules. Most of the sulphides promote
Fischer –Tropsch-type reactions. Co and Ni sulphides (and, less so, Fe sulphides) mediate
carbonylation reactions, which produce carboxylic acids from thiols. Sulphide minerals
can be reactants as well as catalysts in these processes. Results of this kind provide further
evidence for the complex pre-biotic chemistry of the Earth’s Archean oceans and
the plausible mechanisms for transition from the chemical world to the biological world
of today.
The major features of the Earth, currently understood through high-pressure and
high-temperature studies, are presented below as a preview of the mineral-related
experimental observations discussed in different chapters of this review volume.
The Earth is a layered spherical body. As a consequence of spinning, the Earth has a
larger diameter of 27 miles along the E – W dimension around the equator than along the
N – S polar axis, which has a diameter of 7900 miles. The outermost layer, the silicate
crust, is followed successively downwards by silicate mantle, lower metallic liquid outer
core, and metallic solid inner core. The mantle is demarcated by three velocity zones:
upper mantle (occupying 30– 400-km depth), transition zone (400 –670 km) and lower
mantle (670 –2,900 km). The lower mantle, in terms of both volume and mass, constitutes
the largest part of the planet. Each layer is characterized by differences in elasticity and
density arising out of structural (as well as electronic) changes. Seismological models are
based on velocity –depth profiles, determined from the travel-time – distance curves for
seismic waves and from periods of free oscillations. These models employ density,
Preamble and Preview 7

pressure and elastic moduli as functions of depth. The temperature in the mantle reaches
perhaps 3,000 K and in the core perhaps 6,000 – 7,000 K. Tomographic images using
arrays of seismometers have delineated vertical as well as lateral heterogeneities.
The crust — 5 – 10 km thick below the oceans and 20 – 70 km under the
continents — on average contain more silica and aluminium (hence acronymed “sial”)
than the underlying mantle, richer in ferrous metals and magnesium (acronymed as
“sima”). Crust is rich in felspars (NaAlSi3O8 – KAlSi3O8 – CaAl2Si2O8) and quartz.
Felspars in the solid solution series involving KAlSi3O8, NaAlSi3O8 and CaAl2Si2O8
make up the most abundant minerals in the Earth’s crust. Under pressure above 10 GPa,
these phases transform to hollandite (BaMn8O16) structure.
The crust is separated from the upper mantle by the Mohorovicic discontinuity or
Moho, where the change is marked by the disappearance with increasing pressure of
architecturally open structures such as quartz and felspars.
The crust of the Earth is fragmented into several plates, which move against one
another in the collisional zones where subduction of the crustal plates to mantle depths
occur.
The mantle just below the crust is rich in ferromagnesian minerals such as olivines
[(Mg,Fe)2SiO4], pyroxenes [(Mg,Fe,Ca)SiO3] and garnets [(Mg,Fe,Ca)3 Al2Si3O12].
These are succeeded downwards by silicate perovskites and Mg-wustite. The mantle
constitutes 70% of the Earth’s mass and offers us many unanswered questions: How much
of it was, and is, being stirred during geological time? How do the mantle convection
and plate tectonics work? Have the mantle’s composition and structures changed with
time? Clues to many of the answers are derived from geophysical data and
experimental results.
The upper mantle may be largely peridotitic (perhaps “pyrolitic”) in bulk
composition. Early partial melting of the mantle may have resulted in large-scale
differentiation between upper and lower mantle, although elemental partitioning data
appear not to support such a model. Chemical separation between upper and lower mantle
may have been generated by diffusive or convective processes but this requires an
inordinately long time.
The dominant phases in the mantle result from mixtures of the major elements such
as O, Si, Fe and Mg. The phases of Mg –Fe-silicates serve as model systems for studies of
high-pressure structures, phase transitions, vibrational dynamics and chemical bonding.
Also, the high-pressure behaviour of oxides, silicates and sulphides has drawn
considerable interest and the data are growing. Compatible elements such as Mg and
Fe2þ may dissociate, while incompatible elements like Fe and Si or H or P combine to
form alloys. Thus, under pressure, the paradigms of compatible lithophile, siderophile,
noble and volatile, etc. need to be paraphrased.
If the cosmochemistry of the solar nebula were predominantly chondritic then the
bulk Earth with pyrolitic mantle would be deficient in silicon. The calibration of
photospheric spectra brought solar abundances agreeing with chondritic chemistry.
However, recent studies have revealed that the lower mantle is enriched in silica relative to
a pyrolite upper mantle. Since direct sampling of the lower mantle has not been possible,
seismic tomography and electrical studies have offered valuable insights into studies of the
8 Section A

Earth’s interior. Rheologically, the mantle silicates are in a ductile more than in a brittle
regime.
Groundwater, rainwater, seawater, glacier ice and extraterrestrial water each have a
distinctive isotope ratio, which allows us to determine where the water originated from.
Subducted cold slabs with sediment cover carry significant amounts of H2O and CO2 to
depths greater than 700 km. In reverse, through upwelling in spreading centres and
“hotspots” (e.g., Hawaii or Iceland), the mantle materials come up to the surface. H2O is
present in deep mantle rocks in the form of OH-bearing high PT phases, such as dense
hydrous magnesium silicates (DHMS), mostly derived from subducted slab materials.
The exhumed subducted bodies studied at the Cambrian continental collision zone
in northern Kazakhstan show the presence of diamonds in a matrix of metamorphosed
sediments, which were subducted to depths of 125 km or more, and returned to the surface
without losing their isotopic characteristics. The Triassic collisional terrain in eastern
China manifests an interesting cold-climate signature inferred from depleted 18O and
deuterium isotopes. 18O/16O analyses and zircon studies reveal that a cold climate existed
750 million years ago, coincident with glacial deposits for the “Snowball Earth” era.
The upper mantle extends from the base of the crust (i.e., Mohorovicic
discontinuity) to a depth of 410 km. The mineralogy of the upper mantle is known from
the mantle-xenoliths (especially in kimberlite pipes). Petrologically, the common
xenoliths are: spinel – lherzolite, garnet – lherzolite and eclogites (pyrope garnet þ
omphacitic diopside).
At depths greater than 50 km, spinel – lherzolite yields to garnet – lherzolite because
of the reaction: spinel þ enstatite þ diopside ¼ pyrope (Ca-rich) þ olivine (Mg-rich). On
the basis of the equilibration temperature and pressure of this reaction, the depth from
which the natural garnet – lherzolites were derived is calculated to be in excess of 130 km.
Euhedral Cr-rich spinel or wormy symplectite intergrowths of spinel with
clinopyroxenes are seen in kimberlite xenoliths and Al-rich interstitial spinels are
restricted to xenoliths in alkali olivine basalt. In kimberlite xenoliths, decomposition of
garnet above 10 GPa may yield a diopside – ilmenite intergrowth. Diamond mostly occurs
in the kimberlite clan of rocks.
Below the depth of 130 km, diamonds are stable. The common inclusions in them
are olivine (Fo92 – 95), Cr-pyrope, chromite, enstatite and diopside. Some inclusions are
eclogitic (pyrope-almandine and omphacite bearing) and some minerals are anomalously
rich in chromium, such as pyrope-garnet with 63% knorringite, Mg3Cr2Si3O12 and 34%
nureyite, NaCrSi2O6.
Volatiles in the upper mantle were the ultimate source of the hydrosphere
and atmosphere. They go into the constitution of amphiboles (pargasitic), phlogopite,
Ti-clinohumite, carbonates and the fluid inclusions in mantle minerals. These volatiles
are responsible for partial melting. The incongruent melting of orthopyroxene extends
to pressure above 3.5 GPa when water is present. Thus, a partial fusion of the upper
mantle in wet conditions may lead to quartz-normative melts. In drier conditions, the
melts would be olivine-normative.
Discontinuous jumps in density and elastic properties at depths of 400 and 670 km
define the transition zone. Discontinuities in seismic velocity near 400 and 670 km are
Preamble and Preview 9

ascribed to the transformation of olivine (orthorhombic structure) to b-olivine and to b- to

g-spinel structures, respectively. Significantly, the silicon coordination changes from
quintessentially 4-fold (SiO4) in the crustal silicates to 6-fold (SiO6) in the silicates in the
transition zone and beyond to lower mantle. A mixture of both the species of silicon may
be seen in the transition zone. The effect of increasing coordination is to increase the
packing and hence the density. At this zone, the upper-mantle minerals, olivine, pyroxene
and garnet, transform first to a spinel-type structure and then to dense silicate perovskite.
The 670-km discontinuity is marked by the dominance of perovskite structure, with
octahedral silicon, formed from spinel as: (Mg,Fe)2SiO4 (spinel) ¼ (Mg,Fe)SiO3
(perovskite) þ (Mg,Fe)O (rock-salt). Below this depth, silicate perovskite makes up
80 –100% of the volume of the lower mantle and, therefore, is the most abundant single
phase in the planet. Such silicate perovskites may act as hosts for large ions, viz. REE
and K.
To match the pressures at 400- and 670-km depths, corresponding to the
geophysically observed transitions, temperatures around 1,400 and 1,7008C, respectively,
are required. In the transition zone, the seismic velocities increase much faster, suggesting
phase transitions. Long-period (low-frequency) seismic waveforms tend to be simple and
globally coherent. Long-period studies indicate that depths to the discontinuities vary by
up to 40 km between different regions.
In the transition zone, the rapid increase in seismic velocities is due to phase
transitions. Under the pressure at this zone, olivine transforms to spinel and pyroxene to
garnet structure. This zone is dominated by refractory phases such as b-(wadsleyite, or
“modified spinel”) and g-(ringwoodite, spinel) forms of (Mg,Fe)2SiO4, along with garnet
(Mg3(Mg,Si)Si3O12) (majoritic) and perhaps ilmenite phases of MgSiO3. Several high-
pressure hydrous phases, particularly of MgO – SiO2 – H2O and Al2O3 – SiO2 – H2O
systems, may be present in the transition zone.
The most widespread mineral inclusion in diamond from the transition zone is
majorite garnet. Majorite is high-P (.8 GPa) polymorph of pyroxene. It is recognized by
having .3 Si atoms per 12 O atoms and the excess Si (beyond that accommodated in the
low-P garnet structure) is in octahedral coordination. Unmixed majorite garnet (with
crystallograplically controlled lamellae of pyroxene) is present in mantle xenoliths and is
indicative of depth of origin greater than 300 km but with equilibrium at lower (2 – 3 GPa)
pressures. Green tetragonal garnet (a new mineral) is of lower-mantle origin. In the
mantle, the composition of a majorite garnet phase will likely range from pure garnet in
the shallow upper mantle to about 50 –70% majorite at the top of the lower mantle.
Ultimately, in the lower mantle (.24 GPa), all these assume predominantly the
silicate perovskite form accompanied by magnesiowüstite (Mg,Fe)O form. These two
stable forms continue all the way down to the bottom of the lower mantle where the
pressure reaches 135 GPa. It is noted that most of the lower mantle (,70% of it) is made
with perovskite (Mg,Fe)SiO3, i.e., ,40% of the entire Earth. Now we know that a large
number of technology-related compounds show perovskite structure, including those
known as high-temperature oxide superconductors. Perovskite structure can undergo
dozens of phase transitions based on octahedral tilting, cation ordering, cation
displacements and anion defects. While silicate perovskite may be the dominant phase
10 Section A

in the Earth’s lower mantle, a number of other dense silicate phases will compete for
elements such as K, Ba, Ca and Al.
Through long periods of time (tens of million years), the hot mantle under pressure
flows and results in a gradual overturn or convection. The 670-km discontinuity has long
been seen as a potential barrier to mantle convection, separating the mantle into two
distinct layers. This is opposite to the other model, known as whole-mantle convection, in
which the convection involves an overturn of material throughout the mantle with flow
across the 670-km discontinuity.
The change in silicate mineralogy is often associated with change in silicon
coordination. There are about 100 different crustal silicates with 4-fold Si (ivSi)
coordination but only a dozen high-pressure [vi]Si structures are known. From the
commonest quartz and felspars to the dozens of zeolites and other framework structures,
silicates manifest open, low-density topologies with correspondingly loose packing of
oxygen. The restriction in anion topology reduces the number of possible cation
configurations at depth. Despite the restriction, considerable structural diversity is possible
because of three factors — reversible phase transitions, cation positional ordering and
modularity, particularly based on different close-packed layer-stacking sequences.
Close packing of O atoms leads to modular structures with certain features
(e.g., edge-sharing octahedral chains of rutile, the double chains of hollandite, the corner-
sharing octahedral sheets of perovskite and the face-sharing topology of ilmenite) can link
together in many ways to form ordered superstructures of great complexity.
SiO2 in deep mantle (near the core – mantle boundary) is produced by pressure
breakdown of silicate perovskite/ilmenite. It may contribute to the lateral seismic
heterogeneity. Its high-pressure, high-density polymorph, stishovite (with VISi), trans-
forms to CaCl2-type form through an orthorhombic distortion of the tetragonal rutile
structure.
A significant partitioning of elements, such as S, O, Si, Ni and Co, between core-
forming iron-alloy and mantle silicate minerals and the partitioning of C and S between
liquid and solid iron (in the outer and inner core) has been observed. The abundances of Ni
and Co in the Earth’s upper mantle are best explained by equilibrium core – mantle
differentiation in a global deep-magma ocean on the primitive Earth.
Within the Earth’s mantle, the largest seismic discontinuity occurs near ,250 km
above the core – mantle boundary (CMB). The D00 region is the only dynamic portion of the
lowermost 2,100 km of the Earth’s mantle which manifests evidence for anisotropy,
possibly caused by the flowage texture of the material. Partial melting at this depth has
profound implications for the genesis of hotspots and thus possibly also for the continental
extension at the Earth’s surface and for the properties of the geomagnetic field.
Much too little is known about the ultra-low-velocity zone (ULVZ) at the CMB.
The questions that remain to be satisfactorily answered are: whether the ULVZ has been
gradually produced over time by a combination of core interactions and iron- and/or
volatile-enriched melts descending from above or whether it represents the last residue of a
terrestrial magma ocean (Tonks and Melosh, 1990) driven by a negative buoyancy at the
base of a thermal-boundary layer.
Preamble and Preview 11

The Earth’s core is an important boundary condition for the behaviour of the
entire globe. The pressure there reaches over 300 GPa and the temperature is
,6,000 K (about the surface temperature of the Sun). The seismologically observed
elastic anisotropy of the inner core relies heavily on the crystalline structure of iron
under a pressure of 330– 360 GPa and high temperature. The crystalline structure of
the inner core will largely determine its electromagnetic properties and, consequently,
the geometry and the energy source of the geomagnetic field. The continued
formation of the inner core through freezing of the overlying liquid outer core
releases latent heat.
Convection in the outer core perhaps acts as the ultimate source of power for plate
tectonics, which brings about mixing of crustal matter in the mantle.
The outer core’s density deficit is explained by the presence of lighter
elements, predominantly sulphur. Three new high-pressure Fe –S compounds have
been investigated (by GL) up to 21 GPa and at between 950 and 1,4008C. Although
fluid motion in the Earth’s liquid outer core generates a magnetic field, a finitely
conducting inner core (paramagnetic) stabilizes the geodynamo generating the
magnetic field.
The anisotropic elastic behaviour of the solid inner core is becoming better
understood with condensed-matter theory, advanced computational methods and
experimental techniques and measurements of the elasticity of iron and its possible
compounds at core pressures.
Iron occurs in several distinct polymorphs: bcc, d-bcc and fcc, all at RTP, while hcp
(1-phase) occurs at HP. A new metastable b-phase (dhcp) has been described by the
groups associated with Saxena, Andrault and others.
The available data indicate that the inner core has elastic anisotropy, which is
cylindrically symmetric and is aligned along the Earth’s spin axis. Iron at the inner core is
mostly 1-phase with predominantly NiAs structure. The inner core manifests evidence for
the possibility of super-rotation of the core.
However, our poor knowledge of the elastic properties of solid iron at inner-core
pressure (330 –360 GPa) and temperatures (4,000 – 6,000 K) makes the seismological
observations difficult to interpret.
Amongst the significant goals that are achieved through such studies are the
discovery of high-temperature superconductors such as YBa2Cu3O62x, high P – T
electronic and magnetic transition, finding new phases, and observation of contrasting
outputs from hydrostatic and shear stress.
Unfortunately, in this monograph, the details of the experimental procedures,
such as the experimental set-up, difficulties encountered in experimentation, constraints
involved in calibration, temperature and pressure control devices, etc., could not
be accommodated in the discussions of the outputs of the experimental results.
The cited references may be consulted for such details. However, the citing of a large
number of investigators on the same topics has often been avoided by giving only the
primary and some recent references, wherein the earlier works and forerunners are
generally cited.
12 Section A

The key publications in the field of such studies are:

Hemley, R.J. (1998) Ultra-High-Pressure Mineralogy: Physics and Chemistry of the Earth’ Deep Interior,
Reviews in Mineralogy. Mineralogical society of America, Vol. 37, 671 pp.
Manghnani, M.H., and Yagi, T. (1998) Properties of Earth and Planetary Materials at High-pressures and
Temperatures. Washington, DC: American Geophysical Union.
Nakahara, M. (1998) Review of High-Pressure Science and Technology. Kyoto: Japan Society for High-
pressure Science and Technology, Vol. 7.

In addition the most useful companion volumes that may be listed are
Mineral Physics and Crystallography, AGU Reference Self, Vol. 2, 1995, Washington, DC: American
Geophysical Union.
Anderson O.L. (1995) Equations of State of Solids for Geophysics and Ceramic Science. New York: Oxford
University Press.
Jackson, I. (1998) The Earth’s Mantle: Composition, Structure, and Evolution. Cambridge, UK: Cambridge
University Press.
Poole, C.P. Encyclopaedic Dictionary of Condensed Matter Physics. Amsterdam: Elsevier Physics.

It may be pointed out here that the reader who would insist on an exclusively
geochemical treatment of the subject may skip Chapter II with no risk of losing continuity
in the discussion of the high-pressure theme.

APPENDIX I
Generalized compositions of minerals appearing in the text

10-Å phase Mg3Si4O10(OH)2·2H2O

Albite NaAlSi3O8
Amphibole AX2Y5Z9O22(OH,F)2, where A ¼ Na, K; X ¼ Na,
Ca, Fe2þ, Mg; Y ¼ Mg, Fe2þ, Fe3þ, Al; Z ¼ Si, Al
Anthophyllite Mg7Si8O22(OH)2
Antigorite ðm ¼ 17Þ Mg48Si34O85(OH)62
Apatite Ca3(PO4)2(OH,F,Cl)
Brucite Mg(OH)2
Calcite CaCO3
Chlorite (clinochlore) Mg5Al2Si3O10(OH)8
Chondrodite Mg5Si2O8(OH)2
Clinohumite Mg9Si4O16(OH)2
Clinopyroxene XYZ2O6, where X ¼ Na, Ca; Y ¼ Mg, Fe2þ, Fe3þ, Al;
Z ¼ Si, Al
Cordierite Mg2Al4Si5O18
Corundum Al2O3
Diamond C
Diaspore AlO(OH)
Diopside CaMgSi2O6
Ellenbergerite [Mg,(Ti,Zr)]2Mg6(Al,Mg)6(Si,P)8O29(OH)10
Enstatite MgSiO3
Epidote X2Y3Si3(O,OH,F)13 where X ¼ Ca, REE3þ, Fe2þ;
Y ¼ Al, Fe3þ, Fe2þ,Ti
Forsterite Mg2SiO4
Garnet (Ca,Fe2þ,Mg)2(Al,Si)3O12
Preamble and Preview 13

APPENDIX I (continued)
Graphite C
Humite Mg7Si3O12(OH)2
Ilmenite (Fe,Mg)TiO3
Jadeite NaAlSi2O6
K-richterite K2Ca(Mg,Fe2þ)5Si8O22(OH)2
Kyanite Al2SiO5
Lawsonite CaAl2(Si2O7)(OH)2H2O
Lizardite (Mg,Al)3(Si,Al)2O5(OH)4
Magnesiowustite (Mg,Fe)O
Magnetite Fe3O4
MgMgAl-pumpellyite Mg4(Al,Mg)Al4Si6O21(OH)7
Orthopyroxene XYZ2O6, where X ¼ Na, Ca; Y ¼ Mg, Fe2þ, Fe3þ, Al;
Z ¼ Si, Al
Norbergite Mg3SiO4(OH)2
OH-topaz Al2SiO4(OH)2
Paragonite Na2Al4Al2Si6O20(OH)4
Phengite K2Al2(Mg,Fe)2Si8O20(OH)4
Phase A Mg7Si2O8(OH)6
Phase B Mg12Si4O19(OH)2
Phase C Mg10Si3O14(OH)4 (?)a
Phase D MgSi22xH2þ4xO6
Phase E Mg2.3Si1.25H2.4O6
Phase F MgSi22xH2þ4xO6b
Phase G MgSi22xH2þ4xO6b
Phlogopite K2(Mg,Fe2þ)6Si6Al2O20(H,F)4
Quartz SiO2
Pyrope Mg3Al2Si3O12
Rutile TiO2
Serpentine (chrysotile, lizardite) Mg3Si2O5(OH)4
Sphene CaTiSiO5
Spinel MgAl2O4
Staurolite (Mg,Fe2þ)2(Al,Fe3þ)9O6(SiO4)4(O,OH)2
Stishovite SiO2
Talc Mg3Si4O10(OH)2
Ti-chondrodite Ti0.5Mg4.5Si2O9(OH)
Ti-clinohumite (x ¼ 0.5) Ti0.5Mg8.5Si4O17(OH)
Tremolite Ca2Mg5Si8O22(OH)2
a
Phase C is often considered the same phase at superhydrous B.
b
Although phases F and G in the literature have been suggested to be individual distinct phases, it is now
considered that these may well be the same phase.
14 Section A

APPENDIX II
Densities of some HP mineral phases

Minerals Densities (g cm23)

Coesite 2.9
Wadeite (K2Si4O9) 3.1
Hollandite (KAlSi3O8) 3.9
Clinopyroxene 3.3
Kyanite 3.6
Garnet 3.7
CAS phase 3.8
NaAlSiO4 (calcium-ferrite structure) 3.9
CaSiO3 perovskite 4.2
Stishovite 4.3
 Section B

The Earth and Planetary System

This page is intentionally left blank
 17

Chapter 1
(A) Cosmochemistry and Properties of Light Element Compounds
This page is intentionally left blank
Cosmochemistry and Properties of Light Element Compounds 19

1.1. Introduction

Since the “Big Bang” (#14 Ga, determined by the Wilkinson Microwave
Anisotropy Probe, NASA), the elements in the universe have been formed by the
process called nucleosynthesis. The starting-point is assumed to consist of primordial
material of hydrogen mixed with tenth as many atoms of helium. Taking into account
Big Bang nucleosynthesis and stellar and galactic evolution, an age of the universe has
been obtained as 14 ^ 4 £ 109 yr, while that of the Solar System is found to be
about 4.6 £ 106 yr.
Hydrogen and helium in galaxies could collapse under gravity and the collision of
atoms would increase the temperature to such an extent as to start nuclear fusion reactions.
These would lead to more conversion of hydrogen to helium with emission of light and
heat. This continues in medium-sized stars such as our Sun. Apart from hydrogen and most
of helium, lithium and boron, which were produced at the high-temperature, high-density
stage occurring during the Big Bang, the remaining elements were synthesized by
thermonuclear reactions within stars. Elements which are “light” ðz , 20Þ were
nucleosynthesized by nuclear equilibrium processes (till iron peak elements), while the
“heavy” elements ðz . 31Þ were formed by slow (s) and rapid (r) neutron capture and
proton capture processes. The stellar abundances of elements in main sequence stars are
presented in Fig. 1.1 (from Fowler, 1983).
In massive stars, the nuclear fusion reactions proceed much faster and hydrogen
fuel is consumed in less than a hundred million years. At higher temperature, the
conversion of helium takes place to form carbon or oxygen. This conversion would not,
however, produce enough radiative energy to stop the gravitational collapse, which would
lead to a very dense state forming a neutron star or black hole. The outer region may
blow off in a tremendous explosion called a supernova, which for a short while outshines
all the stars in the galaxy. The heavier elements produced may be flung into the space to
produce the next-generation stars.
Our Sun contains ,2% of heavier elements from the debris of earlier supernovae.
The Sun itself may be a second- or third-generation star formed from debris of earlier
supernovae. A small amount of heavier elements collected together to orbit round the Sun
to form planets. This process occurred 5,000 m.y. ago.
The Solar-System elements were formed primarily from earlier supernova
explosions and also from the burning of such nuclei as nitrogen, etc., on the surface of
dwarf stars in binary systems. Thus, the materials composing the Solar System came from
various stellar sources. The formation of a new Solar System would require contributions
from several exploded stars at least. In the universes like ours, there are galaxies whose
separate stars are continually refashioning themselves through supernovae. About 1.9% of
the material from which the Galaxy was formed about 14,000 m.y. ago (or half as long ago,
depending on the true value of Hubble’s constant) has been converted by supernova
explosions and other means into atoms heavier than helium.
Olber’s paradox, causing dark nights, suggests an expanding Universe.
The Universe might be infinite and hierarchical with receding galaxies of galaxies.
20 Chapter 1

Figure 1.1. A schematic curve of atomic abundance (relative to Si ¼ 106) versus mass number A for the Sun and
similar main sequence stars. The symbols s, r and p stand for the slow and rapid neutron capture processes and the
proton capture process, respectively (adapted from W. Fowler, Nobel Lecture, August 8, 1983).

In microcosmos also clusters of clusters of hierarchical atomic arrangements can be the

alternative to strict crystalline arrangements and form a new type of condensed matter.
All the elements present in terrestrial matter must have originated at a later stage by
nucleosynthesis in novae, supernovae or other such cosmic phenomena. Such a process
had been responsible for the generation of such elements as fluorine, which is formed
predominantly on the surfaces of white dwarf stars, into which the larger companion star
lost its material.
There must have been an average of one supernova explosion every 3 years in the
early history of the Galaxy; the present rate is much slower. About 5 b.y. ago every part of
the Galaxy contributed to the formation of the Solar System (Davies and Koch, 1991). The
terrestrial samples serve as the homogenized Solar System material. We have many means
to a fair understanding of the interiors of the distant stars, while the nature and evolution of
the Earth’s interior is poorly understood, and it evades confident prediction with our
present experimental and theoretical understanding. We can calculate with a fair degree of
certainty the structure and evolutionary history of stars for the problem becomes simpler in
stars as, under extreme pressure, the electrons are severed from atoms and form uniform
electron gas, the calculation of which is more straightforward than the many-body problem
Cosmochemistry and Properties of Light Element Compounds 21

encountered in solid silicates under pressure. In stellar physics, the kinetic energy of
electrons increases with the charge density ðrÞ as r2=3 ; while the potential energy binding
the electrons to the nuclei increases only as r1=3 : Therefore, the kinetic energy dominates at
high pressure, while the electrons become unbound in stellar interiors.

1.1.1. Range of pressure in the universe

In the Universe, pressure varies by more than 60 orders of magnitude — ranging

from less than one atom per cubic centimetre in intergalactic space to an unimaginable
pressure at the centre of the neutron stars (wherein a mass of the Sun is compressed to a
size smaller than that of the Earth!). Under pressure, molecules strongly interact to exhibit
properties which can be explained only by a combination of classical and quantum
mechanics.
The pressure of 29.4 tera-Pascal (TPa), at which atoms ionize and form
degenerate electron gas, may be designated as one atomic unit. (Note: A pascal (Pa)
is a force of 1 N (newton) on an area of 1 m2. A newton accelerates a kilogram at
1 cm sec22.) In planetary bodies, the pressure is mostly less than one atomic unit, and
hence is far from being plasma-like. For this reason, the pressure – temperature domains
of planetary materials are more complex than simple electron gas and can only be
described as condensed matter. In these materials, the properties are dominantly
controlled by complex mixing of atomic orbitals, and strong perturbations of these cause
new and unsuspecting properties.
In planetary materials, the pressure scale involves the spacing between electronic
bands (1 eV); and the volume typically occupied by a valence electron in a mineral is
,20 Bohr3 (< 3 Å3). The charge density has been calculated from the observed mass
density by assuming that the number of electrons is one-half the number of nucleons.
Planetary materials would show not only phase transitions but also electronic transitions,
such as insulator to metallic or even metal to insulator. The pressure in the interiors of the
Earth and Jupiter are presented in Fig. 1.2 (read the caption).

1.1.2. The proto-solar nebula

In the earliest evolutionary stages of the Sun-like star, the following are seen to
occur. Initially, a quiescent contraction of a core takes ,106 –107 years, although swifter
collapse (,105 years) may be triggered by interstellar shock waves. Once a central
protostar has developed, the evolution of young stellar objects by accretion of the
surrounding envelope of molecular cloud material is relatively rapid. When a protostar
reaches its final mass, it reaches the classical T-Tauri phase, which typically lasts for a few
million years (but it may be as little as ,105 years).
The temperatures of the mid-plane, where high dust densities promote planetary
formation, the temperature, assume particular importance. On the basis of the measured
surface temperatures for Tauri disks, models suggest that mid-plane temperatures are in the
range of 200– 800 K at 1 AU (astronomical unit ¼ Sun –Earth distance) and 100– 400 K at
2.5 AU. These are the regions where the Earth and most meteorites were formed
22 Chapter 1

Figure 1.2. Pressure in the interior of Jupiter (Chabrier and Baraffe, 1997) and Earth (Dziewonski and Anderson,
1981) as a function of mass density (top) and charge density (bottom). Planetary structures are compared with
limiting high density equations of state for three values (1, 10, 26) of the atomic number, Z (Stixrude et al., 1998,
q 1998 Mineralogical Society of America).

with the fall of material onto the disk. The dissipation of the associated gravitational
energy caused heating of the terrestrial planets, and asteroids were formed by collection of
dust particles and aggregation to larger objects. Three operating stages in the nebula can be
visualized: (i) aggregation of a few mm-sized dust to 1- to 10-km diameter planetesimals,
(ii) runaway growth of the largest planetesimals to form planetary embryos, and
(iii) aggregation of embryos to form the terrestrial planets (e.g., Chambers and Wetherill,
1998; Kokubo and Ida, 2000).
Brownian motion, turbulence, coagulation of dust particles and differential setting
towards the mid-plane contribute to the formation of planetesimals. Weak turbulence
concentrated particles in eddies (Cuzzi et al., 2001). From over cm-sized particles, objects
of km sizes can be formed, which, however, would be lost to the Sun if the growth is not
rapid. When the sizes exceed 1 km in diameter, the interactions between planetesimals
cause a runaway growth of the object in a region, ultimately producing Moon- to Mercury-
sized planetary embryos (,1026 g).
At 1 AU from the Sun, this stage lasts ,105 years and, at 2– 3 AU from the Sun,
lasts ,106 years. At the final stage of planet formation, growth is more oligarchic and
embryos accrete with violent impacts. When the planets formed, there was a radial
temperature gradient leading to volatile-depleted terrestrial planets and asteroids on the
one hand, and the Jovian planets and their icy satellites on the other. The terrestrial planets
formed in the region of steeper thermal gradient.
Cosmochemistry and Properties of Light Element Compounds 23

The short-lived radionuclides were synthesized by energetic particles after the

collapse in the protosolar cloud. Typically, shock waves with velocities of 10– 50 km s21
can trigger core collapse and the formation of a star. The most likely sources of these shock
waves are supernovae, Wolf-Ray stars or asymptotic giant branch stars. The driving gas
producing shock waves from any of these stars will synthesize short-lived nuclides, which
could be injected into the collapsing Solar System. Most radionuclides were derived from
shock-wave injection, whereas 10Be and possibly 53Mn were produced in the Solar System
by particle irradiation rather than in stellar nucleosynthesis. However, in the solar nebula,
heterogeneity is observed with respect to 16O isotope as also in 53Mn.
The interstellar molecular cloud had an angular momentum lost through viscous
dissipation leading to the growth of a central star, the Sun. The rest of the nebular cloud
dissipated out to space leaving a tiny mass (,0.1% of the solar mass) in the disc to
eventually form the planetary bodies. Because of turbulence in the nebula, grains collided
and coalesced to form solid grains, which went on growing in size to form planetesimals.
Fast runaway accretion is simulated to produce planetary embryos of 1023 kg (,2% of
Earth’s mass) bodies within ,105 years (e.g., Wetherill, 1994). The early Sun in its
T-Tauri stage became violently active to remove gas and fine dust from the inner Solar
System. The accumulation of the embryos led to the formation of planets at different
heliocentric distances.

1.2. Cosmochemistry

1.2.1. Data source

As the space probe Giotto penetrated the atmosphere to within 1,300 km of the
nucleus of the comet, the main ion spectra obtained were for Hþ, Hþ þ þ
2 , C , OH , H2O ,
þ
þ þ þ
H3O , CO and S . Isotopic measurements show that O, S, C, Mg, Si and Fe in cometary
and stratospheric dust particles are typical of those commonly found in other Solar-System
materials. Type I carbonaceous chondrites, made up of low-temperature condensates,
closely approximate the condensable fraction of primordial Solar-System material.
Lunar soil (regoliths) and some gas-rich meteorites are repositories for implanted
solar wind elements. From the lunar surface, a number of light isotopes had preferentially
been lost to space through volatilization by impacts of micrometeorites or solar wind
bombardment. Neon gas studies in meteorites indicate a large solar flare irradiation during
the T-Tauri phase of the early Sun.
In the Kuiper belt (beyond Neptune’s orbit), objects such as comets are believed to
contribute to the interplanetary dust particles (IDPs) collected from the Earth’s
stratosphere. IDPs have been collected on adhesive surfaces of U-2 aircraft at high
altitudes. Viking Landers 1 and 2 measured the composition of the atmosphere and regolith
of Mars. CO2 is seen to be the major constituent of the Martian atmosphere, and 15N
isotope is enriched by a factor of 1.6 compared with the Earth’s atmosphere. The SNC
meteorites obtained from Antarctica and supposedly derived from Mars (particularly ALH
84001) bear characteristic chemical features of Mars and evidences for primitive life form
and fluid flow on its surface.
24 Chapter 1

1.2.2. Chemical segregation in nebular condensation

At the birth of the Solar System, a number of isotopes were formed, some of which
are now extinct, e.g., 129I, 53Mn, 107Pd, 129I and 182Hf. Most of these have half-lives two
orders of magnitude shorter than that of 235U. Hf isotopes can be used to investigate the
earliest history of the Earth (Amelin et al., 1999).
From the studies on parent/daughter element ratios, we know that the accretion of
small bodies in the solar nebula occurred within ,10 m.y. of the birth of the Solar System
(Lugmair and Shukolyukov, 1998). By 4.51– 4.45 b.y. ago, the Earth had reached its
present mass, with a metal core and primitive atmosphere (e.g., Halliday, 2000). The early
Earth probably developed a magma ocean, sustained by heat from impacts and the
blanketing effects of a dense atmosphere. The latter was largely lost through the process of
dispersion of solar nebula and collisions (e.g., Benz and Cameron, 1990).
The solar nebula, in equilibrium progressive cooling at 1024 bar with C/O , 0.6,
yielded the theoretical condensates. The early condensates may be classed as: silicates,
metals, high T volatiles (condensing at 1,300 –600 K) and low T volatiles (condensing at
,600 K). The element ratios in the nebula are taken as equivalent to those in C1
carbonaceous chondrite.

1.2.3. The Solar System

The Earth and the other eight planets along with the satellites and asteroids were
formed along with the Sun about 4,600 m.y. ago. Since then, the Earth has segregated
layers as shells within it and we live on its thin outer crust composed of its land masses and
oceanic bottoms. But compared with the Earth, this crustal thickness is no thicker than the
skin to an apple, accounting for only ,1% of the volume. Below the continents, this is up
to about 25 km thick, while below the oceans, particularly below the Pacific, it is as thin
as ,5 km.
Asteroids, representing planetesimals, concentrated in the gap between Mars and
Jupiter at 2– 3.2 AU, which is in accord with the Titius – Bode law. Early growth of Jupiter
caused the gravitational perturbation which hindered the growth of planetesimals much
beyond the planetesimal stage (,1021 kg). The spectral properties of an asteroid, Vesta,
indicate that it could be the parent body for the “HED” suite (howardite, eucrites and
diogenites) of meteorites.
That the solar nebula was homogeneous in isotopic composition is amply
demonstrated by the data of Solar-System materials. But some rare, exotic grains of pre-
solar origin have been detected in the most primitive of the chondritic meteorites, which
show isotopic anomalies. Such isotopic anomalies are seen to exist in Ca- and Al-rich
inclusions in carbonaceous chondrites. Such grains are presumed to have formed in the
interior of a star and were ejected into the interstellar medium, which eventually developed
into meteorites. Apart from these exotic materials the inner Solar System was derived from
a homogeneous nebula that was efficiently mixed. In general, the Solar-System material
(including the Earth) is presumed to have formed from a nebular material that had a
Cosmochemistry and Properties of Light Element Compounds 25

uniform characteristic chemical and isotopic composition equivalent to the solar chemistry
(also detected from the solar photosphere).
The asteroid materials formed the precursors to larger planets, hence asteroidal
meteorites should document the fractionation processes that occurred during the early
stages of planet-building.
The two main isotopes of uranium 235U (half life 700 m.y.) and 238U (half-life
4 b.y.) undergo spontaneous chain decay to form stable 207Pb and 206Pb, respectively. The
ratio of 207Pb/206Pb is used to determine the age. Using this method, the origin of the Solar
System has been set at 4.566 ^ 0.002 b.y. ago (Allegre et al., 1995). U – Pb dating is
the most powerful technique for determining the absolute age of material from the early
Solar System.
Unlike the Moon and Mars, no rocks much older than 4 b.y. are preserved on the
Earth’s crust because of destruction by meteorite bombardment and tectonic processes. An
intense bombardment of the Moon occurred up until ,3.9 b.y. ago, when also the earliest
crust of the Earth was destroyed. It may be presumed that the early crust was like lunar
Highlands, composed of blocks of crustals that floated on the magma ocean. The Vanished
crustal rocks contained highly resistant zircon grains, which survived destruction by
erosion and other processes and became incorporated in sedimentary rocks. In one such in
Australia, the U –Pb ages of most of the grains turned out to lie between 2.1 and 4.2 b.y.,
and some are even slightly older.
Zircons grow in granitic magmas, which are formed in the crust at depths greater
than 20 km by the melting of pre-existing continental crust above the subduction zones
where the oceanic crust is destroyed. The buoyant granitic mass floats up to create
mountain ranges such as the Andes. Zircon dating suggests the existence of the continental
crust as far back as 4.4 b.y. ago. U – Pb ages and oxygen isotopic study on one grain of
zircon appears to have formed 4.4 b.y. ago (Mojzsis et al., 2001; Wilde et al., 2001).
Granitic magmas accrete to create a proto-continent and then a continent.

1.2.3.1. Meteorites
Chondrites. Chondrites contain chondrules formed from molten droplets (from flash
heating at 1,700 – 2,100 K) in the solar nebula. The chondrules offer evidences for
recycling and are formed by localized heating by lightning discharges and shock waves.
These are the most common and most primitive type of meteorites. They were later
modified by aqueous alteration, thermal metamorphism and shock metamorphism. The
three principal types are: ordinary, carbonaceous and enstatite chondrite, each of which are
divided into subgroups, one such being C1 carbonaceous chondrite showing a similar
non-motile composition to that of the solar photosphere.
Some chondrites (e.g., Murchison and Allende: C1 types) showed mm to
cm-diameter Ca – Al-rich inclusions (CaAlIs), formed at high temperatures
(1,700 – 2,400 K). The rim/matrix of chondules are relatively rich in volatiles, viz. Na,
Cl, H2O organic matter, and contain various types of pre-solar grains that formed in
circum-stellar environments.
Some of the C1 chondrites aggregated in the T-Tauri phase of the Sun, when the
solar nebular temperatures were ,400 K. The evidences of aqueous alteration in these
26 Chapter 1

suggest an ambient temperature low enough to cause water – ice condensation. However,
cooling from 1,300 to 600 K takes several tens of thousands of years. Fractionation of
chondules from dust and gas, as well as of silicates from metal, is likely to be a natural
process in the solar nebula. The process of evaporation is accompanied by large isotopic
fractionations.
Carbonaceous chondrites represent the undifferentiated nebular chemistry.
Achondritic meteorites (the irons and stony meteorites) appear to have originated through
the break-up of accreted planetesimals having cores and mantles.
The chondritic meteorites are classed into three groups: ordinary (O), enstatite (E)
and carbonaceous (C) chondrites. Ordinary meteorites are divided into three groups: H for
high Fe, L for low Fe and LL for low Fe, low metal. The degree of recrystallization
increases from C through O to E types, while the proportion of volatile components
decreases. The most primitive chondrites belong to the C1 group, which show chemical
composition comparable with the solar photosphere (except the ice-forming elements,
e.g., C, H, C, O, N and rare gases). The light elements (e.g., Li, Be and B) are actively
consumed by nuclear reactions (a burning process) in the interior of the Sun. Also,
meteorite-abundance measurements may offer better figures for the composition of the
Solar System.
Chondritic meteorites are classed in up to six varieties of petrologic types.
Type 1 shows the most aqueous alteration. The least altered primitive meteorites are
assigned to petrologic type 3; primitive breccias fall in this category. Increasing thermal
metamorphism is successively shown by types 4 and 5. Type 6 chondrites are well-
equilibrated metamorphic rocks, with the peak metamorphic temperatures estimated at
,9508C. Ordinary chondrites showing incipient melting are often assigned to petrologic
type 7.
The most abundant type of ordinary chondrites are represented, though poorly, in
the compositional mapping of the asteroids by reflectance spectroscopy.
On the basis of partitioning, the elements are classed as: lithophile (which
preferentially partitions into silicates or oxides), siderophile (i.e., iron loving) (which
partitions into Fe-rich metal) and chalcophile (copper loving) (which partitions into
sulphides). However, above the temperature of the Fe – FeS eutectic, the distinction
between chalcophile and siderophile elements becomes redundant, as the sulphide and
metal melt into one S-containing metallic liquid (when C is poor). In nebular condensation,
major elements condense as minerals (e.g., Mg and Si as Mg2SiO4, olivine), whereas trace
elements condense in solid solution with major phases such as, Mn in olivine and Au in
FeNi metal. Solar gas composition must be reducing such that Fe condenses as metal and
early Mg-silicates are free of FeO. At lower temperatures, FeO gets into silicates.
Substantial separation of Fe and silicate occurred in chondrites, especially in
enstatite chondrites. In fact, the Fe content of C1 meteorites is poorly defined and it is not
clear as to what was the Fe/(Si þ Mg) ratio in the nebula. Minor element concentrations
of meteoritic troilite (FeS) vary considerably with redox state. A substantial amount of
chromium (1 wt%) and other chalcophile elements are present in the FeS component of
the Earth.
Cosmochemistry and Properties of Light Element Compounds 27

Pre-solar refractory grains and interstellar mixing. The most refractory elements, Hf,
Zr, Sc and Y, condensed very early in the history of nebular condensation. Refractory
metals, such as Re, Os, Ir, Ru, Pt, Rh, Mo and W, would condense as alloys.
Above the condensation temperatures of the major phases of meteorites such as
Mg-silicates and FeNi metal, other refractory elements condensed. Compounds of these
are Al, Ca and Ti oxides and silicates, such as perovskite (CaTiO3) and gehlenite
(Ca2Al2SiO7). Refractory inclusions of these have been reported from carbonaceous
chondrites. Mg-silicates and FeNi metals constitute about 90% of the chondritic
meteorites. Forsterite and FeNi metals have similar condensation temperatures. Their
simultaneous condensation considerably changed the opacity and therefore the heat
balance of the nebular disk. As more radiation was absorbed and converted to heat,
evaporation occurred. This thermostat caused a delay in cooling and a fractionation of the
refractory component delimited the olivine –metal separation.
Chromium should mostly condense into olivine and pyroxene as Cr2þ 2 SiO4 and
2þ
Cr SiO3 components. Chromium was indeed a mainly lithophile elements in nebular
processes.
The carbonaceous chondrite (e.g., the “fall” at Allende in 1969) shows refractory
inclusions of spinel, perovskite, pyroxene, melilite, etc. The carbonaceous chondrites, such
as Murchison and Murray, also contain refractory inclusions of corundum and hibonite. It
has been determined that hibonite represents a higher condensation temperature than the
assemblage of Allende inclusions.
These refractory inclusions which are pre-solar in age are seen to be associated with
carbon compounds such as diamond, SiC and graphite, which show distinctive d13C
pattern and anomalous noble gas component. The diamonds (average size ,2 mm) exhibit
characteristic enrichment in the heavy and light isotopes of Xe (Xe-HL). The diamonds
may have been produced by supernova shock waves passing through molecular solids or
by chemical vapour deposition (CVD). Progressive enrichment of isotopically exotic Xe
components with respect to solar composition led to the discovery of interstellar diamond
and SiC (Fig. 1.3). The light and heavy components of Xe-HL cannot be produced in the
same nucleosynthetic event and are probably the result of mixing. The Xe-S component
from SiC reflects a mixture between the composition produced in s-process nucleosynth-
esis and a near-normal component of Xe (Laeter, 1999).
Most of the heavy elements were products of s-process nucleosynthesis (Fig. 1.1),
whilst 12C and 15N enrichments show evidence of a mixture with CNO-cycle material.
Some of the SiC grains show decay products of short-lived radionuclides, indicating that
these may represent C-rich zone material from the outer layers of a supernova. The
interstellar grains are supposed to be mostly oxides but, other than Al2O3 (corundum) all
other oxides could hardly be isolated from the pre-solar carbonaceous material. In the acid
digestive procedures adopted in extracting the refractory inclusions, most of the silicates
and oxides are destroyed, but the small grains of corundum (Al2O3) remain, which show
isotopic anomalies in 26Mg, 17O and s-process Ti.

Early Sun, x-wind and CAIs. The question that intrigues many is: How were the cosmic
dust, ice and gas in the solar nebula processed chemically for the final formation of the
28 Chapter 1

Figure 1.3. Progressive enrichment of isotopically exotic Xe components in interstellar diamond and SiC
(see text) (from Laeter, 1999, q 1999 Academic Press).

planets? Astrophysicists working with Shu proposed that the young Sun blasted the nebular
material nearest to it with enough heat and radiation to form blobs, which were blown out
to the nebular disk in a magnetically driven wind (“x-wind”). Falling on the disk, these
droplets/blobs formed chondrules, which became the building blocks of both chondrules
and terrestrial planets. The astrophysical x-wind model offers a reasonable explanation for
the formation of calcium –aluminium inclusions (CAIs) in chondrites. CAIs show
signatures of short-lived isotopes of beryllium, which requires irradiation from a
young Sun.
In August 2000, the finding of traces of now extinct Be-10, with a half-life of
1.5 m.y. and produced by irradiation (not within stars), strongly points to x-wind having
been at work in the solar nebula. The evidence of the presence of shorter-lived isotope
Be-7 in CAI suggests the x-wind model still more strongly.
Cosmochemistry and Properties of Light Element Compounds 29

Aluminium-26 isotope clock studies indicate that CAIs formed 1 –4 m.y. before
chondrules (which generated planets) were formed. Evidently, this prompts the question:
How CAIs could wait millions of years for chondrules to form?

Shock pressure in chondrule formation. Mineralogical and geochemical evidences

suggest that chondrules have been heated to 1,800 – 2,100 K for several minutes and then
cooled over several hours; this can only be possible by heating through a powerful shock.
The shocked and heated gas in the solar nebula could have kept the chondrules hot for a
few hours before they radiated away the heat. Thus shock is regarded as the major
mechanism for chondrule formation.

Early isotopes. When the Solar System formed, short- and long-lived radionuclides
evolved. The short-lived radionuclides are regarded as being present in the early Solar
System. Their-half-lives, abundances and possible sources are shown in Table 1.1. These
are extinct, but their early presence can be inferred from the excess in their daughter
products, linked to the magnitude of the parent/daughter ratio.
The now-extinct short-lived nuclides were the potential heat sources for melting
and metamorphism of planetesimals. The short-lived radionuclides were synthesized by
energetic particles after the collapse in the protosolar cloud began. The long-lived nuclides
are now the principal heat sources in the terrestrial planets.
The oldest measured chondrite, the ordinary chondrite Marguerite (H4), has an
absolute age of 4,563 ^ 1 Ma (Gopel et al., 1994). The oldest absolute crystallization age
of any achondrite is 4,558 ^ 1 Ma (Lugmarin and Galer, 1992). However, 53Mn data of
the HED (howardite – eucrite – diogenite) parent body suggest that these and the asteroid 4
Vesta (,550 km across) probably underwent differentiation and core formation at
4,564.8 ^ 0.9 Ma within 0 –4 Ma of CAI formation (Lugmair and Shokolynkov, 1998).
Hf –W dating of iron meteorites also points to rapid differentiation of their parent bodies

TABLE 1.1
Short-lived radionuclides (from Laeter, 1999)

Parent isotope Daughter isotope Half-life (m.y.) Early Solar-System abundance

26
Al 26
Mg 0.74 26
Al : 27Al ¼ 5 £ 1026
53
Mn 53
Cr 3.7 53
Mn : 55Mn < 0.1–6.7 £ 1025
60
Fe 56
Fe 1.5 60
Fe : 56Fe < 4 £ 1029
107
Pd 107
Ag 6.5 107
Pd : 108Pd < 2 £ 1025
129
I 129
Xe 16 129
I : 127I < 1 £ 1024
146 142 146
Sm Nd 103 Sm : 144Sm < 0.005–0.015
244 244
Pu 238U 81 Pu : 238U < 0.004– 0.007
41
Ca 41
K 0.10 41
Ca : 40Ca < 1.5 £ 1028
182
Hf 182
W 9 182
Hf : 180Hf < 2 £ 1024
36
Cl 36
Ar 0.30 36
Cl : 35Cl < 1.4 £ 1026
92
Nb 92
Zr 35 92
Nb : 93Nb < 2 £ 1025
99
Tc 99
Ru 0.21 99
Tc : 99Ru < 1 £ 1024
205
Pb 205
Tl 15 205
Pb : 204Pb < 3 £ 1024
30 Chapter 1

(,5– 15 Ma after CAIs) (Halliday and Lee, 1999). It is determined that 1 km-sized embryo
could grow in a period of ,105 – 106 years.
Hf –W data suggest that the segregation of iron core from silicate mantle in the
Earth and also the formation of the Moon by a giant impact occurred simultaneously at
,50 ^ 10 Ma after the Solar-System formation (Halliday, 1999). The time scale required
for the formation of terrestrial planets is presumed to be ,108 years. Because core
formation on planetesimals occurred within 5– 15 Ma of Solar-System formation, the
terrestrial planets would have formed from already differentiated bodies (for a good
review, read Alexander et al., 2001).

Isotopic anomalies. When the measured isotopic ratios cannot be related to the terrestrial
isotopic composition of elements through a mass fractionation, the isotopic abundances are
said to be “anomalous”.
129
Xe anomaly (“special”) may have been caused by the presence of the extinct
radionuclide 129I. Such anomalies, e.g., for Xe and Ne, were produced by the spontaneous
fission of superheavy elements. The extinct 129I radionuclide had a half-life of 16 £ 106 yr.
In the 1970s, the refractory Ca –Al-inclusions (CaAIs) from Allende gave an excess
of 26Mg. The excess corresponds to the abundance of Al. This may indicate the early
presence of 26Al, with a half-life of 0.74 £ 106 yr. Most unaltered CaAIs have the highest
initial abundances of 10Be, 26Al, 41Ca and 53Mn of any measured Solar-System objects.
They also have the oldest measured absolute ages of 4,566 ^ 2 m.y. ago (Ma) (Alligre
et al., 1995). Chondrule ages (based on 26Al) range from ,0.7 to .5 Ma after the
formation of CaAIs (Kita et al., 2000).

1.2.3.2. Inner planets: major constituents and phases

The inner planets are assumed to form by exactly the same process as the ordinary
chondrites, which are regarded as meteorites representing some planetsimals. Based on the
data of a large number of workers, the model projects the Earth as (i) enriched in “early
condensate” and “metal” by about 1.5-fold with respect to Mg-silicate, (ii) depleted by
about 4-fold in “volatiles 1,300 – 600 K”, and (iii) enriched by about 50-fold in “volatiles
,600 K”. Depletions of H, C and N went in the scale as, H: 5 £ 1027, C: 5 £ 1024 and
N: 3 £ 1025.
When one scrutinizes the relation between the Earth composition and C1
composition and plots against the ordinate, which is the ratio of concentrations in Earth
to 1.5 £ C1 meteorite, and the abscissa as the column number of the Periodic Table, one
observes the following features.
The “lithophile volatiles 1,300 –600 K” scatter widely except for Rb, K and Mn.
The “siderophile volatiles 1,300 –600 K” match quite well within the large uncertainties.
The “volatiles ,600 K” scatter widely, but Th, Cl, Br and I, which concentrate in the crust,
match very well. Most of C and N can be assigned to the core by analogy with iron
meteorites, whereas H can be doubled assuming atmospheric escape. Present-day emission
of noble gases from the mantle argues for incomplete differentiation, and this may cause a
substantial retention of potassium in deep-seated mantle minerals.
Cosmochemistry and Properties of Light Element Compounds 31

Uranium played a key role as the control element for the early condensate. If U is
taken as the sole radioactive heat source, then its content in the bulk Earth would be near
0.030 ppm. But convection process could retard heat flow by about 2-fold. For C1
proportions of “early condensate”, “silicate” and “metal”, the U content should be
,0.014 ppm.
From the primitive solar nebula the accretion of the Earth is presumed to have taken
place either homogeneously or heterogeneously. In the former postulate, the accreting
Earth differentiated metallic core from the silicate outer layer. In the latter postulate the
refractory elements accreted first and the more volatile elements were added later. In the
Earth, high-grade metamorphism in the lower crust has caused substantial redistribution of
mobile elements.

1.3. Evolutionary history of the Solar System: terrestrial planetary formation

From the above discussion, the evolutionary history of the Solar System can be
chronologically arranged as below (cf. Halliday, 2001).

B.y. ago Processes

4.57 Sun and accretionary disk formed
4.56 Some asteroids differentiated
4.54 Mars accretion completed
4.51 The Moon formed during mid to late stages of Earth’s accretion
4.50 Loss of Earth’s early atmosphere
4.47 Earth’s accretion, core formation and degassing essentially complete
4.4 Earliest known zircon fragment
4.3 Upper age limit of most other zircon grains
4.0 Earliest surviving continental crust
3.9 End of intense bombardment

The evolutionary history of the Earth can be summarized as: After the accretionary
growth, the Earth experienced degassing for the first 100 m.y., when a hot, dense
atmosphere formed and a magma ocean was generated. Later, the surface cooled to form
the crust and the dense atmosphere dissipated out. With cooling, liquid water and the
earliest granitic crust formed with possibly a primitive form of life. Meteorite
bombardment destroyed the primitive crust, induced widespread melting and vaporized
the earliest hydrosphere. Later (,4 b.y. ago), the stable continental crust appeared with
oceans developed from degassing of the mantle and life appeared to evolve through the
Geological period. The dark ages before 4 b.y. are called Hadean, a period from which no
crustal rock seems to have survived (see also Section 1.19).

1.3.1. Interplanetary flights of planetary materials

The isotopic ratios of gas trapped in the unusual SNC meteorites indicate their
provenance; they are actual pieces from Mars (Bogard and Johnson, 1983). Meteorites
32 Chapter 1

that originated on the Moon were identified and one might be even from Mercury
(Palme, 2002).
Such interplanetary transport, lifted by giant cratering impacts, might have
“seeded” life on one planet from the other, may be from the Earth to Mars or Mars to
the Earth.
The Moon may be a repository for ancient meteorites from other planets, including
Venus, Mars and the Earth itself. However, its surface has been tranquil for the past
3.8 b.y. compared with the pervasive geological and geochemical evolution of planets such
as the Earth, Venus and Mars.
Armstrong et al. (2002) propose that ,20,000 kg of ancient terrestrial rocks may
have been implanted in a 10 £ 10 km2 region on the Moon; ,200 kg are from Mars and
1 –30 kg from Venus. However, most of these lie under the “mega-regolith”, hundreds of
meters deep, formed by ejecta blasted around impact basins several hundred kilometres
in diameter.
During the short 50 m.y. period between 3.9 and 3.85 b.y. ago, several dozen
impact basins were formed on the Moon. This period, known as Late Heavy Bombardment
(LHB), witnessed bombardment by asteroids of the Earth and other terrestrial planets;
possibly Jupiter’s moons were also affected. This was the period when on the Earth
100-plus basins were formed by bombardment just when life was struggling to gain a
foothold on our planet, and possibly on others (Levison et al., 2001).
Armstrong et al. (2002) estimate that ,7 parts per million of lunar surface materials
are of terrestrial origin. Hence, it is possible that the samples brought back by lunar
missions might contain a few grams of terrestrial material. But how could such needles
from Earth’s or Venus be searched in the lunar samples? The transplanetary flights of
rocks from the Earth or Venus were the result of blasting off in catastrophic explosions.

1.3.2. Primary chemical elements for life

Life on the Earth, the third planet of the Sun, where H2O condensed to be in liquid
form, evolved through 3,000 m.y. of natural selection under a reasonably constant ambient
temperature. The earliest microbial life forms are recognized in the Proterozoic rock
formations. Some, however, believe in panspermia — that is, life came to the Earth from
space. All life forms consist not only of carbon, oxygen and hydrogen, but also of elements
such as phosphorus (as in DNA), iron (as in haemoglobin) and even cobalt (as in vitamin
B12). Even humble Escherichia coli depends on 17 elements (mainly H, O and C),
compared with 26 for human beings. More strangely, some eukaryotes concentrate
elements such as Hg and Cd relative to the ambient environment, and have their DNA to
evolve detoxifying systems so as to avoid ill effects of these “toxic” elements.
Transition metal (Fe, Co, Ni, Cu, Zn, Sn, W and Pb) sulphides convert nitrogen
oxide to ammonia very rapidly (,90% conversion in 90 min). Even oxide minerals (Fe, Ni,
Cu and Mn) also reduce nitrate and nitrite to ammonia. Ammonia is not easily destroyed in
hydrothermal systems. Thus, deep-ocean hydrothermal systems might have served as the
major source of ammonia for the Archean Earth, and consequently were the cause for the
production of amino acids and other nitrogen-containing biomolecules.
Cosmochemistry and Properties of Light Element Compounds 33

Transition metal sulphides are known to exist in modern hydrothermal vents. These
can promote geochemical reactions that mimic key metabolic reactions in living organisms,
where enzymes are used as catalysts. However, under hydrothermal conditions, several
reactions in the citric acid metabolic cycle proceed without enzymes. Again, a synthesis of
pyruvic acid serves as the key molecular entry point. In the presence of iron oxides and
sulphides, nitrogen gas can be reduced to ammonia to the citric acid cycle.
Transition metal sulphides can promote carbon addition reactions, which are
important in synthesizing larger organic molecules. Most of the sulphides promote
Fischer –Tropsch-type reactions. Co and Ni sulphides (and less so Fe sulphides) mediate
carbonylation reactions, which produce carboxylic acids from thiols. Sulphide minerals
can be reactants as well as catalysts in these processes.
Results of this kind provide further evidence for the complex prebiotic chemistry of
the Earth’s Archean oceans, and the plausible mechanisms for transition from the chemical
world to the biological world of today.
The Earth’s Archean oceans were rich in organic molecules, and in the primitive
Earth life appears to have arisen as a natural geochemical process. Transition metal
sulfides are seen to act as catalysts promoting carbon fixation reaction in prebiotic age.
The chiral crystal surfaces of quartz and calcite are seen to absorb chiral molecule
selectively. These mineral surfaces differentiate and selectively absorb left- and right-
handed molecules. The chiral selection mechanism requires three non-collinear points of
bonding between the chiral mineral surface and the chiral molecule. Therefore, carboxylic
acids or amino acids, containing three charge groups, are more chirally selected than amino
acids or others containing no such charge groups.
Life’s origin on Earth is commonly believed to be traced to the processes operating
in the phobic zone at the ocean –atmospheric interface, where the energy for prebiotic
organic synthesis came from the ionizing radiation. The life-forming processes can also
occur in submarine hydrothermal environments at water – mineral interfaces by oxidation –
reduction environments. The life-killing impacts occurred many times in the geological
history, but the deep hydrothermal environments deep in the sea insulated life from such
surface-sterilizing events. The microbes surviving the high-temperature environments may
be the early ancestors of the living world. Life needs to be traced from the evolution of
organic compounds from prebiotic molecular synthesis and organization to cellular
evolution and diversification.
Life on Earth is based on carbons. On other worlds, however, different chemical
basis can be postulated. But within stellar clouds carbon compounds are identified
spectroscopically. Carbon can evolve from interstellar medium to terrestrial organisms.
Besides carbon and hydrogen, biomolecules need oxygen, nitrogen and sulphur.
The interplanetary dust particles (IDPs) collected from Earth’s stratosphere are
considered to be among the most primitive extraterrestrial materials. These are , 20 m in
size and are aggregates of silicates, metals, sulfides, and carbonaceous materials. IDPs
may also be important sources of organic matter delivered to the early Earth. The
IDPs display relative enrichment in the ratio of D/H and 15N/14N compared to
terrestrial values.
34 Chapter 1

Hydrothermal activity on chondritic meteorites leads to the formation of mineral

species, such as hydrated silicates and sulphides, which facilitated prebiotic organic
synthesis and promoting organic chemical reactions in the early Earth.

1.3.2.1. Microorganisms under pressure: clues to HP genesis of life

During the 1870s, HMS Challenger documented for the first time the abundance of
live biota at abyssal depths (Thompson, 1878). Corliss et al. (1981) reported a diverse
biosystem in the proximity of hydrothermal vents at the sea-floor spreading centre. These
biota at dark reaches of the ocean derive their energy from geochemical sources rather than
from the Sun. The evidence for their survival and proliferation, therefore, challenges the
paradigm that the Sun is the only source for life’s energy.
Microbes have been isolated from numerous deep crustal environments, e.g., oil
reservoirs, deep aquifers, deep oceanic sediments and igneous rocks — wherever water can
exist in liquid form.
Holger W. Jannasch and his co-workers at Woods Hole Oceanographic Institute
have documented microbial growth at pressures #100 MPa, and identified barotolerant
microbes, displaying reduced metabolic rates at high pressure, and barophilic organisms,
which manifest enhanced metabolism and growth rates at high pressure. Investigations
revealed that some microbes thrive at .1008C, and microbial metabolism may persist at
P . 10 kbar (Sharma et al., 2001). Indeed, there is a variety of pressure-dependent
transitions in lipid biolayers and lyotropic phases. There are systematic effects of pressure
on protein folding (Michels and Clark, 1997).
H2 and N2 in the presence of catalytic ferrous iron minerals react to form NH3,
which is central for the synthesis of amines and other N-bearing organics. This may offer
the reason for concluding that hydrothermal vents could have served as the major source of
NH3 in the Archaean period. Thus, hydrothermal organic synthesis may have been
responsible for evolving organic inventory of the pre-biotic Earth. Pressure, which
stabilizes aqueous phases to T . 1008C; may have played a key role in life’s origin. Thus,
life-forming processes might also have occurred in the wet, pressurized interiors of Mars,
Europa and other Solar-System bodies.
Microorganisms are capable of existing at extreme pressures equivalent to that of
50 km below the Earth’s surface and in the subsurface ocean of Europa. Thus, pressure
may not be a significant impediment to life in deep waters of Jupiter’s moons, under
the Martian polar caps or in the subduction zones on the Earth. Experiments at GL (see Yb
01-02) show that Shewanella oneidensis MR-1 and E. coli, both gram-negative bacteria,
are capable of surviving pressure greater than 1.5 GPa.

1.3.2.2. Biogenesis
The first act of life’s origin started with the development of carbon-based
molecules, which could copy themselves. The earliest carbon-based compounds were
carbon monoxides and dioxides and methane.
But for living organisms the essential building blocks are amino acids, sugars, and
membrane-forming lipids. Some of these molecules bond together to form chain-like
polymers.
Cosmochemistry and Properties of Light Element Compounds 35

In life-forming chemical reactions minerals could have played significantly. The

mineral surfaces can offer scaffolding on which the life-forming molecules assemble and
grow. Minerals could have acted as containers, scaffolds and templates that helped to select
and organize the molecular menagerie of the primitive Earth (Hazen, 2001). Again
elements released from the solution of minerals can go into the contribution of life-forming
molecules.
Recent knowledge of diverse ecosystems thriving at the superheated mouths of
volcanic vents on the seafloor reflects that the need of energy for the organisms is met not
from the solar light but from the Earth’s internal heat. Usually, amino acids decompose
when heated. But amino acids in the presence of iron sulfide minerals (as formed around
ocean bottom hydrothermal vents) under high pressure and temperature stay intact for
several days, enough time to react with critical molecules.
Bioscientists with a penchant for geology long postulated that mineral surfaces may
provide sites for biologically important molecules to assemble. Of the stipulated minerals
clay group of minerals are considered to play a very important role. Clay minerals are
ubiquitous, their surfaces carry electric charge that will attract organic molecules and hold
them in place. Amino acids can collect on clay surfaces and link up into short chains,
appearing like protein molecules. Indeed, clays can act as templates rather than scaffolds
for the building blocks of RNA molecules, which translates genetic instructions to
proteins.
Layered minerals such as clays can trap organic molecules between the sheets, and
the molecules under compression can react to produce more complex forms. Amino acid,
like many other organic molecules, show enantiomorphism (“chirality”) to left-handed
(“L ”) and right-handed (“D ”) forms. But amino acids in living organisms show remarkably
left-handedness.
Some astrophysicists weirdly suggested that the cloud of dust and gas in which the
Earth was formed had excess left-handedness than the other. Nevertheless, some minerals
manifest selective absorption between L - and D -molecules. Calcite, e.g., is seen to
preferentially select L -amino acids on its left-handed faces, and vice versa, with excesses
approaching 40% in some cases. Protein molecules may behave this way. Perhaps in the
early Earth self-replicating proteins were formed on the face of a calcite crystal, and by
chance the crystal face preferentially selected left-handed ones.
In 1988 Wachtershauser opined that iron- and nickel-sulphides abounding deep-sea
hydrothermal vents served as template, catalyst and energy source that drove the reaction
for biological molecule formation, specially on the positively charged surfaces of pyrite
(FeS2). Indeed, metabolic enzymes are known to have a cluster of metal and sulfur atoms at
the core. Many common minerals, such as oxides and sulfides of iron, copper and zinc
promote carbon addition by a routine industrial process known as Fischer – Tropsch (F –T)
synthesis. Chain-like organic molecules are formed from carbon monoxides and hydrogen.
First, CO and H2 react to form CH4 (methane), which on reaction with more CO and
H2 form ethane (C2H4) and the reaction repeats itself with the products having more and
more carbons in their structure. Such F –T reactions can build molecules with 30 or more
carbon atoms in less than 24 h in the sea floor hydrothermal condition. Thus, from simple
inorganic chemicals large organic molecules might have formed in the early Earth.
36 Chapter 1

Near the sulfide-rich hydrothermal vents in the sea carbon monoxides, thiol and of
course, water are available for reaction. The products such as hydroxide and carbonyl
groups can be reassembled to generate a variety of complex organic molecules.
At deep-sea vents, as the mineral-laden water rises from below and contacts the
frigid water layers, minerals are deposited. This leads to the formation of spectacular
columns of about a dozen feet tall.
In summary, the scenario stands as the molecules of life were manufactured in (i)
the nebula that formed the solar system, (ii) ocean surface at water – air interfacial region,
(iii) near deep-sea hydrothermal vents. But the molecules were of far greater diversity than
what was needed for life generation. From this chaos a selective order was brought in by
minerals (see Holland, 1998). By selective absorption of molecules the minerals may have
jump-started the first self-replicating molecular systems. The self-replicating molecules
began to evolve to more complexity and to metabolic cycles of the living cells. Thus,
minerals played a much more complex and vital part in the origin of life on the planet
Earth.

1.3.3. Primitive atmosphere

The atmosphere we live in has been formed mostly by the belching out of gas,
mainly oxygen and nitrogen, from inside the stomach of the Earth through volcanism and
fumeroles during the geological past of 4.5 b.y. How the structure of the Earth we live
upon and the environment we live in undergo changes is not yet completely known to us.
The primitive atmosphere on Earth began ,4,460 ^ 20 Ma on the basis of the
abundance of radio-activity-produced 40Ar and 129Xe (Allegre et al., 1995) proton carriers
(Cavazzoni et al., 1999; Chau et al., 1999). Again, this probably marks the end of major
impact events that would have removed much of any pre-existing atmosphere. But Earth
already had a continental crust and oceans by 4,300 –4,400 Ma (e.g., Wilde et al., 2001).
Earth’s inventory of radiogenic 40Ar suggests that about half of the Earth’s mantle
has been stripped of its incompatible and volatile elements to escape to the atmosphere –
hydrosphere. Noble gas isotope studies of oceanic island basalt (OIB) reveal a continuous
outgassing of primordial (solar) noble-gas components.

1.4. Charge density within planetary interiors

The structure of planetary interiors manifests some fundamentally important

aspects of matter at planetary depths. At planetary densities, the net Coulombic attraction
between nuclei and electrons plays a primary role. The difference in charge (and mass)
density between the Earth and Jupiter (Fig. 1.3) can be accounted for by the difference in
mean nuclear charges of these two planets — the former being much smaller than the
latter. The charge density can be calculated from the observed mass density by assuming
that the number of electrons is one-half the number of nucleons.
The core states in an atom remain sharp, delta-function-like states. These states are
raised or lowered relative to their positions in isolated atoms. The shifts are mainly a result
Cosmochemistry and Properties of Light Element Compounds 37

of the screened Coulomb potential from the rest of the atoms in the crystal. The valence
and conduction states broaden into energy bands under higher pressure.
Assuming that the nuclei are in close-packed arrangement, the equation of state
(EOS) becomes:

P ¼ 0:176rs5 ½1 2 ð0:407Z 2=3 þ 0:207Þrs  ð1-1Þ

where P is the static (athermal) pressure, Z the nuclear charge and rs the Wigner – Seitz
radius.

 1=3
3
rs ¼ ð1-2Þ
4pr

This radius is a measure of the average spacing between electrons. The first term in
equation (1-1) is the kinetic contribution, the second is due to the Coulombic attraction of
the nuclei for the electrons and mutual repulsion of the electrons and the third is related to
exchange. At high density, Coulombic interaction, which is smaller than exchange, is
neglected, as is the mutual repulsion of the nuclei.
Screening has a first-order effect on the EOS, accounting for the much lower
densities of planets at a given pressure than predicted by equation (1-1). In the vicinity
of the nucleus the charge density is much enhanced; thereby, the ability of the point
charges to attract the remaining valence electrons is reduced. The major part of the
screening is due to the tightly bound rigid core electrons. In the case of terrestrial
planets, the charge density near the nuclei is much higher than in the interstitial region.
But screening in Jupiter is much reduced because there lighter elements dominate and
the pressure is very high.

1.4.1. Electrons under pressure

Under pressure, electrons obey an almost trivial limiting behaviour of an uniform

electron gas. With increase in charge density ðrÞ; the kinetic energy of electrons increase
as r2=3 ; while the potential energy binding the electrons to the nuclei increases only as r1=3 :
At high pressure, the kinetic energy dominates and the electrons become unbound
(see Bukowinski, 1994).
One atomic unit pressure is 29.4 TPa, which corresponds to the pressure required
for complete ionization and the formation of a degenerate electron gas. The pressures
within planetary bodies are much less than one unit and are best described as condensed
matter, and hence much more complex than plasma-like matter.

1.5. Forces binding atoms

Forces binding atoms in a matter are primarily electrostatic in nature. An atom with
more electrons to contribute to bonding will produce a stronger bond than another atom
38 Chapter 1

with fewer electrons at the same bond length. As the bond lengthens, the electron density
becomes more diffuse and, therefore, the contribution to the bonding energy is lower. At
the same time, the forces that keep atoms from collapsing into each other arise from
(i) increased kinetic energy as the atoms are brought close together, (ii) the Pauli exclusion
principle that keeps electrons apart, (iii) the electrostatic repulsion between electrons, and
(iv) the electrostatic repulsion of the nuclei, as they are brought closer. The electrostatic
energies are of the order: ,2.3 £ 10211 erg, 14 eV or 160,000 K.
Interparticle forces are essentially determined by Coulombic attraction between
opposite charges and the lowering of the kinetic energy by delocalization of quantum wave
functions to reduce kinetic energy.
In minerals deep within the Earth, the types of bonding that exist are ionic, covalent,
metallic, Van der Waals and hydrogen bonding. Some of these are discussed in
the following sections, the others in relevant sections/contexts. In response to compression,
the relative energies of the valence electron change. Therefore, the bonding
property changes. For example, solid Xe (inert gas) with Van der Waals bonding becomes
similar to CsI with ionic bonding. At pressure above 200 GPa, both show metallic bonding
and transform to hcp-like structure (Mao et al., 1989).

1.5.1. Van der Waals forces

Van der Waals forces are constituted of fluctuating dipoles on separated atoms or
molecules. Non-overlapping charge densities do show an attractive force between them
and it varies as C6 =r 6 and higher-order terms at large distances. The dispersion coefficients
are invoked to explain weak interlayer bonding in graphite and binding in closed-shell
molecules in minerals. From Van der Waals interaction potentials for closed shell systems,
the EOS of fluids and solids are evaluated. Some new Van der Waals compounds
synthesized in recent years are discussed below.

1.5.1.1. Van der Waals compounds: new materials

Inert-gas molecules interact weakly through Van der Waals force and, therefore, no
stoichiometric compounds can form. Under pressure, new stoichiometric compounds, the
so-called Van der Waals compounds are formed from simple molecular mixtures,
including binary mixtures of H2, He, Ne, Ar and CH4. The high-pressure solid compounds
that have recently been discovered include He (N2)11, NeHe2 and Ar(H2)2 by Vos et al.
(1992) and Loubeyre et al. (1993, 1994), respectively. Under pressure, the molecular
Van der Waals compounds that have been seen to be stabilized are: He(N2)11 (Vos et al.,
1992), Xe(He)2Ne(He)2 (Loubeyre et al., 1993), H2 – H2O (Vos et al., 1993) and molecular
compounds such as (O2)3(H2)4 (Loubeyre and LeToullec, 1995) and H2 –CH4 system
(Somayazulu et al., 1996). In the H2 – CH4 binary system at pressures below 8 GPa, four
new compounds are reported. Beginning at ,1 GPa, simple molecular mixtures exhibit the
formation of stoichiometric compounds (or “order alloys”). He and N2 form the compound
He(N2)11 at 8 GPa (Vos et al., 1992).
Cosmochemistry and Properties of Light Element Compounds 39

Clathrate hydrate: H2/H2O. H2 and H2O are by far the most abundant gas and ice
components in the Solar System. H2 molecules are seen to fill small cavities in ice II and
ice 1c at high pressures (Vos et al., 1993). Clathrate hydrates are molecular “cages”
of frozen water, which can host frozen molecules inside (Fig. 1.4). Clathrate hydrates have
hydrogen-bonded networks of cages in which large gas molecules are held by
Van der Waals forces. Under pressure, the open network breaks down and high-density
clathrates are formed. In the H2 –H2O binary system, H2 and H2O form interlocking
networks, both with a diamond structure (Fig. 1.4). This hydrate clathrate is stable to
,30 GPa. Since H2 and H2O constitute the most abundant gas and ice molecules in the
universe, the large stability range of their clathrates suggests that these can be widely
pervasive.
The H2 and H2O mixture crystallizes into the sII clathrate structure with an
approximate H2/H2O molar ratio of 1/2. Water molecules form clathrate hydrates, which
consist of networks of cages containing guest molecules. The occupancy of a hydrogen-
bonded H2O cage of clathrates by guests is linked by the guest/H2O molecular ratio, 1/6
(Villard’s rule).
At the formation of clathrate, broad liquid water OH peaks at 3,000 – 3,600 cm21
transformed to sharp peaks typical of sII clathrates. Meanwhile, hydrogen roton peaks
appear at 300 –850 cm21 and vibron peaks at 4,100 – 4,200 cm21. Hydrogen rotons,
S0(0), S0(1) and S0(2), in the clathrate are similar in frequency to those of pure
hydrogen. Thus, H2 molecules in clathrate cages are still in free rotational states. H –H
vibrons of new clathrates are distinct from those other known phases in the H2 – H2O

Figure 1.4. Crystal structure of the high-density H2 –H2O clathrate. The dumbbells are idealized representations
of the rotationally disordered hydrogen molecules; the large spheres denotes the oxygen of the H2O molecules
(from Vos et al., 1993, q 1993, American Physical Society).
40 Chapter 1

system. H2O molecules are squeezed into small cavities in clathrates. Thus, hydrate
clathrates are stable under confining pressure. Similar to Raman vibron frequencies, IR
spectra of HH-s II also show a hydrogen vibron peak at 4,145 cm21 and a weak peak at
4,120 cm21. While the vibron intensity of Raman spectra is intrinsic to the H2O
molecules, that in IR is highly sensitive to the environment. Free H2 molecule
(homonuclear) lack a permanent electric dipole moment and cannot absorb IR radiation
(Mao et al., 2002).
The first direct evidence of molecular hydrogen frozen in inter-stellar ices was
discovered in the IR spectrum of WL5, a deeply embedded protostar in the Ophiuchus
cloud complex. The IR peaks of HH-sII spectra match the WL5 spectra. The HH-sH can be
synthesized at pressures .180 GPa, which is prevalent in the interiors of icy satellites.
Hydrogen could potentially be held at high temperatures in the structures of bodies, which
are considered as not holding hydrogen.
Confining H2 molecular clusters offer scope for studying novel interactions and
quantum effects, such as superfluidity and Bose – Einstein condensation of hydrogen
molecular clusters (see, for example, Knuth et al., 1995).
Interestingly, clathrates, such as He(N2)11, NeHe2, Ar(H2)2, (H2)4(O2)3, Ar(O2)3
and the CH4 – H2 system, are isostructural with a very different class of materials across
the great divide, namely, metallic alloys. Studies of these and H2O have led to the
discovery of novel, dense cage (clathrate) structures. Such high-pressure Van der Waals
compounds could form and condense in clouds of the dense atmospheres deep within the
large planets or as ices within their moons. Enclathration helps store hydrogen in larger
bodies from which it could be released gradually to form the H-rich atmosphere (in
possible life-sustaining planets in interstellar space; Stevenson, 1997).
In characterizing and understanding these compounds, Raman scattering has
proved to be the most useful tool. Hemley et al. (1998) reported the Raman measurements
up to 40 GPa of molecular compounds Ar(H2)2 and CH4(H2)2 with similar Laves phase.

Xenon clathrate, Xe – H2O. Xenon is one of the gases (others are CH4 and CO2) which
stabilize the clathrate hydrate structure through Van der Waals interactions. Such a
clathrate structure contains two types of cages: (i) two pentagonal dodecahedra and six
tetrakaidecahedra (14-sided polyhedra). In Xe-hydrates, all cavities are filled, so that
the formula contains 46 H2O molecules and eight guest molecules (Sanloup et al.,
2002). Xenon clathrate is observed to be stable up to 2.5 GPa before breaking down to
ice VII plus solid xenon. The bulk modulus and structure of two phases of Xe
clathrates are:
Xe clathrate A (cubic) (at 1.1 GPa): a ¼ 11.595 Å, V ¼ 1,558.9 Å3
Xe clathrate B (tetragonal) (at 2.2 GPa): a ¼ 8.320 Å, c ¼ 10.287 Å, V ¼ 712.1 Å3
The solubility of Xe-clathrate B (.2 GPa) corresponds to a depth of about 65 km
within the upper mantle of the Earth. This depth corresponds to ,6508C. For the Martian
xenon, the storage depth is translated to ,200 km because the gravity field is 1/3 that of the
Earth. Phase equilibria studies of rare gas – water systems under pressure show that hydrate
stability decreases from xenon to neon. Argon can enter or leave the cavity relatively
Cosmochemistry and Properties of Light Element Compounds 41

easily. It is likely that Xe-clathrates are thermodynamically more stable than Ar-clathrates
at high P – T conditions.

1.5.2. Ionic compounds

Before embarking on a discussion on ionic compounds under pressure, let us first

try to understand the concepts delineating the models of ionically and covalently bonded
compounds.
A system is called ionic when the properties can be described by models with
discrete close-shell ions having integer charges. The closed shells may not overlap
because of the Pauli exclusion principle. The ions are regarded as simple charged hard
spheres. The simplest ionic model involving charged hard spheres is embodied in
Born –Mayer-type pair potentials. In extended ionic model, the changes in an ion’s
properties, caused by changes in its environment, are considered by incorporating many-
body interactions. Effectively, this incorporates the effects of polarization, compression
and deformation.
An ion’s size and shape change with the coordination environment. The concept of
an ion’s shape as spherical rests on the remoteness of electronic states of high angular
momentum. The oxide ion is particularly compressible compared with halides. Cation
polarization becomes especially important when low-energy, dipole-allowed transitions
are possible.
To visualize the environmental effects, one can consider the potential experienced
with the charges of the other ions in the crystal. Since the ground electronic state of a
closed-shell ion is an S state, the angular potential will only be significant to the extent that
it can mix in excited states of G symmetry.
Ions are basically spherical in high-symmetry crystal structures. Distortions of low
multipolar order need to be considered to account for the changes in their properties in an
environment in a condensed phase. This is discussed in the following sections.

1.5.2.1. Simple ionic model

In the simple ionic model, a crystal structure consists of hard spheres as embodied
in pair potentials of Born –Mayer form. The stability of the structure depends on how
spheres of appropriate charge and radius pack together to maximize the Coulombic
interactions of unlike ions and minimize those of like interactions.
For example, the B1 (NaCl) structure may be viewed as a close-packed cubic lattice
of one species, with the other occupying the octahedral holes. This arrangement equalizes
the nearest-neighbour cation – cation and anion –anion separations (rþ þ ¼ r2 2 ), and
hence minimizes the charge – charge interactions (see Section 3.1). If cation occupancy
does not minimize the cation –cation interactions, “non-ionic” (layered) structures
develop. The ionic structures maximize the distance between the highly charged cations
and interpose an anion in between a pair of cations. The only parameter which varies from
one system to another is the cation radius.
42 Chapter 1

The simple ionic model favours highly symmetric structures, whereas polarization
favours pushing highly polarizable ions into unsymmetrical sites. Thus, in ZnO, the ions
are tetrahedrally coordinated, whereas the radius ratio would suggest a rocksalt structure.

1.5.2.2. Overlap- and self-energy: pair-potential

Under compression, the overlap between the charge densities of nearest-
neighbour anions and cations will enhance, causing an increase in the overlap energy of
the system (Uov). The electron densities may be considered as decaying exponentially
and the total overlap energy the sum of the overlap energies associated with each
cation – anion pair. Under compression, the walls of the spherical confining potential
will move in and each ion will adjust by shrinking its charge density. This shrinkage of
ions under the influence of confining potential will cost energy, which might be called a
self-energy, Uself, or “rearrangement energy”. The total energy associated with
compression is

U ¼ Uov þ Uself

The Uov and Uself show different dependencies on the coordination number. The
self-energy, (Uself), and Uov successfully account for the lowest energy for crystal
structures and energies for lattice parameters, and accurately predict the transition
pressures to high(er) density structures.
The spherical potential, Uo, tends to compress the electron density relative to that
of the free ion. This leads to a marked reduction in the polarizability, which is
responsible for the stability of the oxide ion in condensed matters. Nevertheless, for
cations, the effect is much smaller. The oxide ion, being unstable in the gas phase, is
particularly susceptible to environmental effects. Ab initio data (Pyper, 1995) indicate
that Uself is a smaller component of the total repulsive energy. The ab initio pair
potential (phase polarization) gives a good account of numerous properties of CaF2,
including its superionicity.
The pair potential calculated with the B1 data of MgO differs from that required to
fit the B2 and B3 results. Hence, these pair potentials are not applicable for different phases
of MgO. Pair potentials appropriate to each phase are obtained from the corresponding
values of Uov þ Uself.
Since oxide ions show greater sensitivity to the coordination environment than
halides, the Uself is found to show a greater contribution to the total U in oxides than in
halides. Pair potentials will thus have a wider domain of applicability in halides than in
oxides. The crystal-parameterized Born– Mayer pair potentials give, inter alia, a better
representation of the interactions in alkali halides (Woodcock and Singer, 1971).

1.5.2.3. Ions in distorted lattice

In a crystal, when ions are shifted off from their lattice sites, the spherical harmonic
expansion of the environmental potential will have a modified spherical term and also
contain angular momentum l ¼ 1; 2 components. The l ¼ 1 and 2 terms cause
deformations of the ionic electron density of dipolar and quadrupolar symmetry,
Cosmochemistry and Properties of Light Element Compounds 43

respectively. Thus, the central ion will acquire a non-zero electric dipole and quadrupole
moment, which will alter its energy through Coulombic interactions with the charges
and multipoles of other ions constituting the polarization energy, and the ion becomes
non-spherical (deformed).
The overlap energy between a particular pair of ions depends not just on the
distance between them, but also on the angle between the inter-ion vector and the internal
vector of each of the ions, and hence on the configuration of the other ions around it.
For large cations, the simple ionic structures with an anion interposed between the
cations emerge as most stable. At intermediate size, this bridge is bent, resulting in corner-
linked polyhedra. For very small cations and polarizable anions, the bending is such as to
develop edge-linked polyhedra, in which the induced dipoles on two anions screen the
cation – cation repulsion.

1.5.2.4. Multipoles and polarization

Polarization is effected by induced multipoles on ions in distorted lattice. If an ion
in a crystal is displaced off its lattice site, the effect on the potential felt by the electrons
in the central ion is simply the electric field ðl ¼ 1Þ plus the field gradient ðl ¼ 2Þ at
that site.
For cation polarization, the short-range effect enhances the dipoles and
quadrupoles above the values which would be expected from the Coulombic-induced
moments. When an anion is displaced outwards, the electric field generated will tend to
displace the cation-electrons towards the displaced anion. This is also the direction
favoured by the displacement of the cation-charge cloud. When the charge cloud is
distorted from a spherical shape, multipoles develop. The polarization energy results from
the classical Coulombic interactions of the multipoles with the charges and multipoles
of ions.

1.5.3. Covalent and hydrogen bonding

In covalent solids, the dominant bonding interaction is caused by hybridization

among the states on different atoms. Such bonds may be strongly directional since they are
formed from the linear combination of directional orbitals on the two atoms. These bonds
can be very strong, as between C’s in diamond, but less so in Si.
In strong covalent solids, orbitals are not fully occupied on the constituent atoms.
Tight-binding models offer an insight into such bonding interactions.
In molecular hydrogen (H2), each H atom separately has one 1s electron and, when
brought close, one molecular bond forms out of two 1s atomic orbitals, which is then doubly
occupied. But in He, the 1s states are filled and, as the two He atoms approach, electrons must
be boosted into higher energy states. That is how H forms strong bonds, while He does not.
In hydrous minerals, a hydrogen atom is covalently bonded to an oxygen
atom and weakly attracted to a neighbouring oxygen. Thus, a hydrogen atom is
asymmetrically disposed between two oxygen atoms. Both covalent and ionic interactions
are present in each bond type. At mantle and core pressures, conventional hydrogen
bonding is lost.
44 Chapter 1

Silicates show bonding which is half-covalent and half-ionic, while transition

elements like Fe show both metallic and covalent bonding. A strongly covalently bonded
molecular group, such as C– O bonded to a CO223 carbonate group, is ionically bonded to
Ca2þ ions.

1.5.3.1. Pressure rupturing of the binding forces

By identifying the characteristic energies and volume changes brought about by
pressure, one can crudely estimate the order of magnitude of the pressure required for
rupturing forces to bring about some phenomena of general interest to us (Daniels, 1993).
(a) Pressure required for significant changes in chemical binding would be

energies of chemical binding kilojoules=mole

¼ , 1 GPa
volume change in reaction cm3 =mole

(b) The pressure required for disruption of electronic “shell” structure of atoms to lead to
white dwarf stars can be determined as

Characteristic electronic energy ð;1 RyÞ

Volume of an atom ð,4p=3 £ ðBohr radiusÞ3 Þ

Hence P < 1 Ry=ð4p=3Þa30 < 3; 511 GPa

(c) To break down nuclei into neutron star the pressure required can be calculated as

nuclear binding energies millions of eV

; , 1022 GPa
nuclear volume ð3 fermiÞ3
The energy scale is set by the binding energy per atom of the solid. This is typically
of the order of a few electron volts (1.6 £ 10216 erg). The length scale is set by the inter-
particle spacing, which is typically of the order of ,2 Å.

1.6. Helioseismology and Jovian structures

The Sun can be regarded as the Rosetta stone of astronomy. Its internal structure
can be modelled using mathematical equations for mechanical and thermal equilibrium,
matched to the observable boundary conditions. Zones of convection and of radiation
transfer of energy can be delineated. Signals from deep inside the Sun come in the form of
neutrinos. The recent Sudbury Neutrino Observatory results offer clues to a solution for the
long-standing solar neutrino puzzle. Other signals are the more classical waves travelling
through the solar body, involving numerous modes and leading to the field of study called
helioseismology.
The interior structures of giant planets are modelled from helioseismology, based
on impact sources and ring perturbations. The observation based on the impact process was
carried out on Jupiter during the impacts of Shoemaker-Levy in July 1994. To understand
Cosmochemistry and Properties of Light Element Compounds 45

better the seismic response of Jupiter and other gaseous planets, some direct measurements
of sound velocities in planetary materials at elevated pressures are needed.
High-pressure experiments on hydrogen and related low-Z material from ,1 to
300 GPa have helped the acquisition of an indirect knowledge of the interiors of the outer
planets and their satellites. In the lower pressure range (,50 GPa), observations include
new compounds (clathrates and Van der Waals compounds) and accurate determination of
equations of state (EOS). Sound velocities have also been measured in this pressure range.
At higher pressures (100 GPa), spectroscopic methods have been employed for
characterization of the state of bonding of hydrogen, identifying phase transitions,
determining the subsolidus phase diagram, characterizing optical properties (visible and
infrared) and for elucidation of the electronic properties of the material. At pressures
.200 GPa, the conditions correspond to densities approaching 1 g cm23. By comparison,
the mean densities of Jupiter and Saturn are 1.2 and 0.6 g cm23, respectively. It is because
of the very high temperatures prevailing in the interiors of the giant planets that they are in
a fluid state.
After the cometary (the Levy-Schumacher comet) impacts, the global free
oscillations of Jupiter were noted for the first time. The frequency of oscillations is
determined from the velocity of sound and thus it depends on the EOS. The velocity of
sound is given by C 0 ¼ ðdP=drÞS ; where S is entropy. Guillot et al. (1994) proposed that
radiative layers may exist near the surface of Jupiter, Saturn and Uranus at temperatures of
,2,000 K. In such a situation, their interiors should be fully convective.
A slow cooling of Jupiter causes its radius ðrJ ¼ 7 £ 1010 mmÞ to decrease by
,1 mm yr21. Jupiter’s molecular layer above the metallic core is fully convective. As the
dominant heat-conduction mechanism, convection causes the temperature distribution to
approach the isentropic distribution. Convection by dynamo action also generates the
planetary magnetic field. In Fig. 1.2, pressure as a function of mass density and charge
density in Jupiter and Earth is shown (also see the figure caption).

1.7. Planetary constituents under pressure

Under pressure, volatiles can be bound in dense, high-pressure phases,

e.g., hydrogen in ice, mantle silicates and ferrous alloys. Pressure at depth may dissociate
a Fe2þ –Mg2þ combination from oxide or silicate phases and incompatible elements such
as Fe and K may form alloys.
The phenomena manifested by materials at pressures prevalent at the depths of the
Earth are not only first-rank problems of geosciences but stand at the forefront of modern
condensed matter physics (Hemley and Ashcrost, 1998). The pressure range within the
Earth can compress the rock-forming silicate oxides by factors of 2– 3 and the molecular
species and rare gases by well over an order of magnitude. Discrete magnetic and
electronic transformations such as metallization and magnetic collapse may occur
(e.g., Cohen et al., 1997). When an ion possesses a shell filled with electrons, it attains
greater stability. The interaction between ions through Madelung or strong electrostatic
forces enhances the crystal stability.
46 Chapter 1

Pressure studies using a diamond-anvil cell (DAC, see Section 4.2), multi-anvil
press (MAP) and temperature and shock compression have attained the P – T conditions,
spanning the range available in terrestrial planet interiors, and reach deep into the interiors
of giant planets. Also, computational capability and quantum statistical techniques are
leading to the modelling of crystal structures and bonding properties under extreme P – T
conditions. Static high-pressure experiments have offered an understanding of the
composition, global processes and evolution of the outer planets and satellites. The EOS of
many materials under high pressure has been determined with diamond-anvils or explosive
gas guns. [Note: Very high pressure implies pressure greater than 1 GPa (¼ 10 kbar).] But
neither anvils nor gas guns seem capable of simulating the high temperatures and pressures
at the core of large planets. Laser beams with a total power of 100 W have been used in
DAC with CO2 (Boehler and Chopelas, 1992), YAG (Shen et al., 1996) and YLF lasers
(Shen et al., 1998).
The P – V – T EOS is one of the critical parameters for understanding the properties
of minerals under conditions prevalent in planetary bodies including satellites. In EOS
studies, pressure becomes important because it changes the interatomic spacing, electronic
configurations, structure and bonding. The EOS links the observable properties of a planet,
such as seismic velocity structure, moment of inertia, etc., with the P – T conditions within
the planet and the atomic structure of the constituent minerals. Besides mineral science,
P – V – T EOS finds important applications in studies of condensed matter physics,
astrophysics and materials science.
Typical shock velocities are of the order of 10 km s21, with the corresponding
pressure range 10 –500 GPa. The material behind the shock front remains in a compressed
and heated state for several microseconds, which is long enough for the atoms in the
material to experience millions of vibrations to make it behave as a fluid.
Profound changes in the physical properties of low-Z materials (i.e., gases and ices)
occur at pressures prevalent within planetary interiors. The dominant molecular species in
gaseous planets are thought to be primarily H2, He, NH3, CH4, H2O, CO2, N2 and C. Under
the extreme conditions (e.g., 5 – 7 TPa) in Jupiter’s chemical interactions, the atomic and
molecular characters of materials are lost and they form dense plasmas (Ichimaru and
Ogata, 1995). In this regime of P, T, theoretical calculations offer the most accurate
scenario.
New observations have been made in hydrogen in the pressure range of
,1 –300 GPa. Measurements at megabar pressures on hydrogen isotopes indicate that
they transform to semi-metal at 150 GPa and low temperature. The results compare well
with the theoretically predicted plasma-phase transition and the reported metallization of
the fluid at similar pressures and high temperatures. Very relevant to the study of
planetary surfaces and atmospheres, observations have been made of new compounds
(clathrates and Van der Waals compounds), EOS and sound velocities under pressure.
However, much work needs to be done in deciphering the physical and chemical
properties of a hydrogen-related system at high pressure and simultaneously at high
temperature.
A knowledge of the mixture of He, H2O, NH3, CH4 and H2 is of critical
importance to progress in our understanding of issues relating to planetary interiors
Cosmochemistry and Properties of Light Element Compounds 47

(Stevenson, 1987). In particular, recent studies of Saturn’s moon, Titan, suggest that its
interior may contain complex mixtures of methane, high-density ammonia and water,
including a possible methane clathrate hydrate (Lunine, 1993). The crystal structure of
clathrate is shown in Fig. 1.4. For hydrogen, the pressure-induced decrease in the
intramolecular stretching frequency is well known. On this basis, at high P – T condition,
a thermally-induced dissociation and concomitant density increase have been suggested
by Hemley (1988). The H2 –He system has been experimentally investigated to 200 GPa
(Loubeyre et al., 1996). Molecular dynamic (MD) calculations of H2 – He predict a lower
temperature of miscibility than obtained earlier by static lattice computations (Klepeis
et al., 1991).
Oxygen forms diatomic molecular crystal and persists in a variety of molecular
phases to #100 GPa. At pressures above 95 GPa, oxygen not only becomes metallic but
also superconducting, and is designated as z-O2 phase.
Simple molecular solids under pressure may transform into a polymeric phase
before transforming to the metallic state. Such polymeric forms have been noted in the
cases of N2 (Mailhiot et al., 1992), CO, diamond (Bundy and Kasper, 1967) and symmetric
H2O (Goncharov et al., 1996). The structures of N2 polymer and diamond at high pressure
can be viewed as identical to the heavier elements in the same periodic group such as P and
Si, respectively, at low pressures.
Under very high pressure, the decrease in band gap leading to band overlap in the
molecular phases leads to metallization (see Fig. 1.6 and the caption). In disordered
systems having no well-defined bands, the mobility edges converge. All these lead to
metallic conductivity in the limit T ! 0 K: But in large planets, the thermal excitation of
electrons at temperatures that prevail within the molecular layer is expected to give rise to
a conducting fluid even before the band-gap closure or dissociation occurs. In this
phenomenon, impurity ionization, induced by pressure and temperature on soluble
components (viz. H2O, NH3, and O2), could contribute conducting electrons from impurity
bands into the gap prior to gap closure. This process can contribute to the enhancement of
the electrical conductivity at shallower depths. Nevertheless, a continuous increase in
conductivity with depth cannot be ruled out, especially under the influence of magnetic
fields. The ferro-electric phenomena are known in large planets, and these have been
discussed in Section 15.2.2, in the context of the ferroelectric property of perovskite-like
structures deep within the Earth.
In the giant planets, the mantles are dense gaseous (hydrogen and inert gases) and
icy (H2O, NH3, CH4, CO and CO2) layers (e.g., Duffy et al., 1994). Under pressure, the
so-called inert gases no longer behave as inert. They form compounds such as He(N2)11,
NeHe2 and Ar(H2)2 (e.g., Loubeyre et al., 1993, 1994). Under 30 GPa pressure, the melting
points of inert gases (e.g., Ar and Xe) will increase steeply, such that when iron melts,
argon and xenon remain as crystalline solids (Jephcoat, 1998). New high-pressure
crystalline compounds have been reported in the H2 – H2O (Vos et al., 1993) and H2 –CH4
systems (Somayazulu et al., 1996). These Van der Waals compounds have been discussed
earlier in Section 1.5.1.
Ammonia is an essential compound from which living cells acquire nitrogen.
48 Chapter 1

1.7.1. Transition pressure

Temperature tends to increase the transition pressure to compensate for the thermal
expansion. Hence, it is not possible to express precisely the transition pressure in terms of
depths within the Earth, because the thermal expansion coefficient at these pressures is not
known.

1.8. Hydrogen

Very high-pressure studies have led to exploration of the metallic phases of light
elements that are fundamentally and astrophysically important, partly because of their
potential as high-temperature superconductors. Among the elements, in condensed form
hydrogen is unique. It forms the only quantum molecular solid, in which molecules freely
rotate even at the lowest temperatures, which, at normal P, T, forms a tenacious insulator.
One of the novel phases described is metallic hydrogen, which is widely believed to have a
very high critical superconducting temperature. Hydrogen has been studied to 300 GPa,
yielding a solid exceeding 12-fold compression. Although the molecular state persists over
this pressure range, pairing of the protons clearly prevails.
Study of the behaviour of hydrogen at megabar pressure has gained momentum in
recent years. Quantum simulations and density functional approaches have focused on
molecular orientation and band structure or the lowest-energy crystal structure. The high-
pressure vibrational spectrum presents the higher energy excitonic transitions not
obtainable by optical spectroscopy.
Studies of hydrogen under high pressure have led to the discovery of a number of
unexpected phenomena, some of which are listed as:
(a) striking intensity enhancement of vibrational modes (Hanfland et al., 1993; Hemley
et al., 1994),
(b) unusually complex phase diagram (e.g., Mazin et al., 1997),
(c) new class of excitations in the solid (Hemley et al., 1997),
(d) characteristic bonds for metallization at megabar pressure (Hemley et al., 1996).
The study of hydrogen under pressure is providing a database crucial for planetary
modelling (Hemley and Mao, 1998).
At low P and T, hydrogen behaves as an insulating molecular solid. Its molecular
structure is shown in Fig. 1.5a along with its elastic behaviour (Fig. 1.5b). At extreme
pressures, hydrogen molecules will dissociate to form a monatomic metal. The hydrogen
bands under pressure are shown in Fig. 1.6. Profound transitions in hydrogen are observed
around 150 GPa.
Collision-induced absorption in the H2 –He system in the giant planets is
responsible for their opacity. The radiative heat transfer and optical opacity as a function
of depth within the planets can be modelled by determining the infrared absorption.
An increase in density increases the collision-induced processes, which cause a
weakly allowed dipole-forbidden internal stretching mode of the isolated molecule to
occur. Such induced processes occur in a dense solid, constrained by the crystalline
symmetry.
Cosmochemistry and Properties of Light Element Compounds 49

Figure 1.5. (a) The crystal structure of molecular H2 at atmospheric pressure. (b) A model for the elastic
behaviour, with strong springs between the atoms in the molecules, and weak springs between the molecules
(Hemley and Mao, 1998, q 1998 American Geophysical Union).

High-pressure synchrotron infrared spectroscopy has allowed measurements to be

extended to above 200 GPa (Hanfland et al., 1993). Mao, Hemley and others of GRL
carried out high-pressure experiments on H2 and D2 at the European Synchrotron
Radiation Facility, Grenoble, up to 105 and 119 GPa, respectively. It was noted that their
structure remained hcp over this pressure range. In shock experiments, hydrogen remains
in thermal equilibrium. Within the time resolution of 3 £ 1029 s, hydrogen undergoes 105
collisions and vibrations, which are more than sufficient to achieve thermal equilibrium.
Thus, the measurements in 100 ns may represent what happens when the time period is a
billion years or so. The time of one vibrational period of a free H2 molecule is

Figure 1.6. The bands of hydrogen under pressure. With increasing pressure the electronic energy bands of solid
hydrogen broaden and eventually they will overlap, making a conducting state; but spontaneous polarization of
the H2 molecules might postpone that or even prohibit it (after Edwards and Hensel, 1999, q 1999 Nature).
50 Chapter 1

0.8 £ 10214 s; this value changes by only a few percent at high pressure. At 200 GPa,
the calculated dissociation fraction is 15%, which is a modest perturbation of the
molecular EOS.

1.8.1. Hydrogen molecular states

The covalent bond between hydrogen pairs is tenacious and persists to pressures
beyond 300 GPa. Three molecular phases of dense hydrogen have been found
experimentally (Fig. 1.7):
Phase I: a high-temperature, low-pressure phase,
Phase II: a low-temperature, high-pressure phase; rotational order is purely quantum
mechanical,
Phase III: a higher pressure phase; rotational order is largely classical (Mazin et al., 1997).
The quantum mechanical simulations reproduce aspects of the ordering for phases
II and III. Like solid hydrogen, dense ice also shows a quantum character of protons
evolving towards a classical symmetric hydrogen-bonded state at high pressure (Benoit
et al., 1998).
At low T and very high P, a symmetry-breaking transition occurs (phase I to II in
Fig. 1.7). At higher pressures, the electronic transitions are manifested when phase III
shows a precipitous drop in Raman vibron frequency and an enormous increase in
infrared vibron absorption (Hanfland et al., 1993) (Fig. 1.10). The boundary between I
and III terminates at a critical point, above which the transition becomes continuous.
This topology is remarkably similar to the Mott transition (Mott, 1990) wherein, at a low
temperature, there is an electron spin ordered antiferromagnetic phase (phase II for

Figure 1.7. Schematic high-pressure phase diagram of H 2 . (o ¼ ortho-(normal) hydrogen, and

p ¼ para-hydrogen) (Hemley and Mao, 1998, q American Geophysical Union).
Cosmochemistry and Properties of Light Element Compounds 51

Figure 1.8. Solid molecular hydrogen at high pressure and (a) the proton distributions obtained by quantum and
classical simulations for phases I, II and III of dense H2. The arrows indicate the molecules to be orientated in a
particular direction, whereas spheres indicate that the molecules are rotating. The quantum mechanical
simulations (Kitamura et al., 2000) reproduce aspects of the ordering observed experimentally for phase II and III.
(b) In a classical simulation, the particle moves unquantized on the potential energy surface (which may be one-,
two- or multi-dimensional). In the quantum simulation, the particle tunnels through the barriers leading to
quantum localization. Depending on the shape of the multidimensional potential surface, the particle can be
localized on one side or the other to give a phase of broken symmetry (as predicted for phase II of hydrogen), or
localized at a high-symmetry midpoint (Kitamura et al., 2000).

hydrogen); at high temperature, a disordered phase (phase I) and, at high pressure, an

electron delocalized phase (phase III) appear alongside a triple point and a critical
point in the phase diagram. The quantum and classical simulations for proton
distribution in phases I, II and III are presented in Fig. 1.8 (read the caption). Both
the Raman and infrared vibron frequencies decrease with increasing pressure above 30
and 140 GPa, respectively (Fig. 1.9). Both of these reflect a gradual weakening of
the bond.
52 Chapter 1

Figure 1.9. Pressure dependence of the Raman and infrared intra-molecular modes (vibrons) in solid hydrogen at
room temperature (from Hemley et al., 1995).

The strong covalent bond of diatomic hydrogen persists in low-density

condensed phases, where the molecules interact very weakly through Van der Waals
forces (Van Kranendonk, 1983). At very high densities, molecular bonding has long been
predicted to give way to a monatonic and presumably metallic lattice (Wigner and
Huntington, 1935). At intermediate densities, intermolecular interactions are expected to
increase.
However, Johnson and Ashcroft (2000) predict that, due to band overlap at a critical
pressure, some bonding electrons move into the conduction band to produce a molecular
metallic solid. But there may be many ways in which the hydrogen molecules can orient
themselves to prevent the transfer of bonding electrons to the conduction band. They also
predict that an insulating molecular state will persist to ,400 GPa, at which point band
overlap occurs and the ever-tenacious covalent bond continues to still higher pressures in a
molecular metallic state (Hemley, 2000).
If intermediate molecular metallic states exist, they may exhibit novel properties
such as very high temperature superconductivity (Richardson and Ashcroft, 1997). In the
fluid and solid phase (phase I) of hydrogen, the intensity of the H – H stretching
fundamental (vibration) increases with increasing density ðrÞ as r2 or as 1=r 6 ; where r is
the average intermolecular separation. The high-pressure solid phase (phase III), stable
above 150 GPa, shows a marked increase in IR absorption (Fig. 1.10). In phase III, a sudden
onset of strong infrared absorption of the hydrogen “stretching mode” indicates a
symmetry-breaking charge transfer. Under pressure, the band gap of hydrogen decreases.
This pressure dependence of the band gap of hydrogen relates to the electronic properties
Cosmochemistry and Properties of Light Element Compounds 53

Figure 1.10. Representative infrared absorption corresponding to the intra-molecular H–H stretching mode
(vibron) of hydrogen (after Hanfland et al., 1993).

in the molecular state (Mao and Hemley, 1994). The effect of pressure on the electronic
properties of molecular hydrogen, specifically testing the prediction of pressure-induced
band overlap metallization, has been a major interest in the physics of dense hydrogen
(Mao and Hemley, 1994). Infrared absorption and reflectivity measurements have
established that hydrogen is transparent in the mid-infrared down to ,1,200 cm21
(0.15 eV) to at least 200 GPa (Hemley et al., 1996) and that the solid remains an insulator
in phase III, at least just above the 150 GPa transition. The effective oscillator frequency
which tracks the band edge can be measured from the dispersion of the refractive index at
lower pressures.
These solid-phase measurements agree well with the shock-wave electrical
conductivity measurements as well as the theoretical calculations beyond the local
density approximations (Mao and Hemely, 1994). At the highest static pressures
(,300 GPa) on hydrogen attained by Mao and Hemely (1994), optical absorption is
evinced in the visible wavelengths.

1.8.2. Vibrational properties

In planetary interior study, the molecular dissociation transition in hydrogen under

pressure draws intense interest. With increasing pressure, the frequency of the
54 Chapter 1

intramolecular vibration is observed to decrease in both Raman scattering and infrared

absorption. This indicates that the molecular bond becomes destabilized with increasing
density. In contrast, intramolecular vibrational coupling increases dramatically with
pressure (Mao and Hemley, 1994) and also the intramolecular interaction, such as charge
transfer, increases. Raman studies reveal that the molecular bond is stable to at least
,250 GPa (at ,77 K) (Mao and Hemley, 1989). Raman scattering study above
250 –300 GPa, however, reveals a loss of the molecular bond reflected by the loss of the
Raman signal.
The marked changes in the infrared (Fig. 1.10) of the intermolecular H –H
stretching modes (vibrons) (Hanfland and Hemley, 1993) with increasing pressure can be
interpreted in a manner analogous to the behaviour of charge transfer at the ambient
pressure, including those exhibiting pressure-induced neutral-to-ionic transitions. Infrared
spectroscopy reveals a three-orders-of-magnitude increase in absorption of the
intramolecular stretching mode at a phase transition found at 150 GPa (Hanfland et al.,
1993). Increased molecular overlap in dense hydrogen leads to symmetry breaking, which
makes possible the charge-transfer states between adjacent H2 molecules. Theoretical
calculations show that such a transition involves a breaking of electronic symmetry in the
proton pairs, a state of spontaneous polarization akin to that in an antiferromagnetic
material. This transition leads to an increase in band gap instead of band closure by
pressure.
Pressure dependence of the IR- and Raman-active vibrons and lattice-mode
frequencies is observed (e.g., 6,200 cm21 band). The splitting between IR and Raman
vibrons is dominated by vibrational (or factor-group) splitting (Hanfland et al., 1993). The
distinction between intra- and inter-molecular bonding creates a gap in the vibrational
density of states, which can be determined as a difference between the lowest vibron and
the uppermost phonon.
In the pressure range 250– 255 GPa, Raman experiments showed an increase of
intensity by a factor of 1.5 – 2.5 of the librational modes with 716 –724 nm excitation
(Goncharov et al., 2002). Optical measurements of solid hydrogen up to a pressure of
320 GPa at 100 K by Loubeyre et al. (2002) reveal that the vibron signature of the H2
molecule persists to at least 316 GPa and, above 320 GPa, solid hydrogen becomes
completely opaque and black. However, above 300 GPa, features characteristic of a direct
electronic band gap are noted. It is predicted by Loubeyre et al. (2002) that, at ,450 GPa,
this direct gap would close and metal hydrogen be observed.

1.8.2.1. Vibrational excitations

Vibrational spectra are crucial in characterizing these compressed states. The
vibrational excitations in the solid consist of vibrons, rotons and phonons (Fig. 1.11),
which, under pressure, show splitting as follows:
Raman vibrons ½Q1 ðJÞ; ðDn ¼ 1; DJ ¼ 0Þ ! multiplet,
Rotons ½So ðJÞ; DJ ¼ 2 ! triplets,
Phonon ½Q1 ðJÞ; ðDn ¼ 1; J ¼ 0Þ ! singlet.
All are characteristic of the hcp structure. XRD measurements were carried out for
phase I to 120 GPa (Loubeyre et al., 1996).
Cosmochemistry and Properties of Light Element Compounds 55

Figure 1.11. Principal vibrational excitations of solid H2 (hcp phase) (Hemley et al., 1998, q 1998 Plenum
Publishing Corporation).

1.8.2.2. Experiments (P > 300 GPa)

Above 300 GPa, the Raman spectrum of hydrogen can be measured accurately.
With pressure, a continuous broadening of the vibron peak is observed. Above 160 GPa,
this becomes more pronounced in phase III. A large span of vibrons seems to be an
intrinsic property of hydrogen. In contrast, the low-frequency libron modes remain sharp
up to the maximum pressure. The pressure shift of librational modes, the phonon modes
and the vibron modes are seen in phase III to 220 GPa (Goncharov et al., 1998).
Continuous pressure shift in phase III suggests it to be of molecular hydrogen, which is
stable up to 316 GPa at least (Loubeyre et al., 2002). Also, the vibron mode proves that the
molecular form persists to the maximum pressure.

1.8.2.3. Vibrons
The Raman vibron spectrum of solid hydrogen is characterized by a single strong
band and this was the first excitation to be measured in the solid at very high pressures
(.10 GPa) in p-H2 and o – p mixed crystal.
The presence of the Raman-active phonon and its continuity with phase I suggests
an hcp-based structure for phase III, the stability of which extends to at least 230 GPa.
There are at least two different phases of molecular H2 in the same pressure range,
depending on the o – p ratio. The frequencies of the IR and Raman bands in phase III
(vibrons, phonons and librons) as well as the transition pressure are seen to be independent
of the P – T path.
The o – p conversion rate in phase I shows an unexpected non-monotonic behaviour
as a function of P, and increases markedly with increasing pressure above 3 GPa. The
transition to phase II depends on the o – p concentration. Mixed crystals transform at lower
pressure to an ordered superstructure, whereas essentially para-rich samples undergo the
BSP transition of the original spherically symmetric molecules. Rich and distinct libron
spectra are observed in phases II and III and evidence for multiple Raman vibrons is found.

1.8.2.4. Phonons
Phonons are the quantum units of molecular vibration in the solid state. The
thermodynamic and elastic parameters, which are important for interpreting seismological
observations, are related to phonons. Theoretical calculations have extended our
56 Chapter 1

knowledge of the full-phonon spectrum of hcp iron. Mao and his global collaborators are
developing a method for determining the phonon density of states of hcp iron to 153 GPa
(.CMB pressure).
The activity of the optical phonons at higher frequency (molecular translational
modes) is characteristic of crystal structure. The Raman active phonons are shown in
Fig. 1.12. The phonon characteristic of hcp structure is observed over the entire range of
stability of phase I, as observed in the case of ortho- and para- ðo – pÞ hydrogen mixed
crystals (Hemley et al., 1990). The intensity of the phonon mode near 900 – 1,000 cm21
becomes extremely weak in phase II for pressures higher than 115 GPa. On the other hand,
the Raman-active phonon mode becomes strong in phase III and it appears to split into
a doublet.
Hemley et al. (1998) documented vibron, phonon, roton and libron excitations in
pure p-H2 as a function of pressure through all three phases I, II and III. Detailed study of
the pressure dependence of o – p conversion in H2 has revealed unexpected changes in
conversion rate, implying a new o – p conversion mechanism at high pressure. The pressure
dependence of the bands in all three phases is continuous in the spectra of the o – p mixed
crystal through phase II (where bands are observed). This arises because structures are all
derived from close-packed types and as such are expected to have similar lattice mode
frequencies. At the lowest temperatures, all molecules are initially in the J ¼ 0 state: p-H2
and o-H2.
During o – p conversion at zero (or very low) pressure, the conversion energy
(E ¼ 171 K/molecule) is transferred to lattice vibrations and is carried away by the creation
of one or two phonons.

Figure 1.12. Low-frequency Raman spectra of phases I, II, and III of molecular hydrogen ( p-H2) (Goncharov
et al., 1998, q 1998 American Physical Society).
Cosmochemistry and Properties of Light Element Compounds 57

1.8.2.5. Rotons and librons

Rotons are characteristic of the molecular quantum crystal and provide information
on the extent of intermolecular and crystal-field interactions.
Substantial changes in the low-frequency Raman spectra of hydrogen are observed
at the transition to phase II at 110 – 120 GPa. The main So ð0Þ Raman multiplet at 350 cm21
transforms into a doublet (Fig. 1.13) and two additional bands appear at lower frequencies.
Hemley et al. (1998) attributed these modes to librons (lattice phonons derived from
restricted rotational motions). Changes in relative intensities are observed at 140 GPa;
these may result from changes in orientational ordering (directions of librational motion)
(Mazin and Cohen, 1995) or another weak phase transition.

1.8.2.6. Hydrogen bridges

The hydrogen bond interactions can be investigated by several methods.
Vibrational spectroscopy observes the frequency of OH stretching motions. A decrease
in the OH stretching frequency can be attributed to weakening of the intramolecular O –H
force constant as, for example, is caused by an increase in the H· · ·O hydrogen bond
interaction. The neutron diffraction method determines hydrogen positions and provides
direct information on the geometry surrounding hydrogen atoms.
Hydrogen bridges, O – H· · ·O, are important structural elements in certain silicate
phases. The energies of OH vibrations are determined by the strength and geometry of the
hydrogen bridge.

Figure 1.13. Effective charge as a function of pressure in phases II and III (Hemley et al., 1997a, q 1997
Europhysics Letters).
58 Chapter 1

In hydrogen bridges (e.g., O – H· · ·O), the low energy shift of the nOH band with
increasing P is due to a decrease of the O· · ·O distance in the hydrogen bond. The
shortening of the hydrogen bridge is either a result of a decrease of the distance between
the octahedral chains, lying parallel to the crystallographic b-axis, and/or decrease of the
angle at which the two chains are tilted with respect to each other (at 1 bar). It is observed
that both the frequency and strength of absorption of the respective stretching vibrations
are mostly dependent on the oxygen –oxygen distance, i.e., the strength of the hydrogen
bridge. Bending vibrations shift to higher wave numbers with increasing hydrogen bond
strength.
Langer et al. (1979) studied the effect of P on hydrogen bonds in opals. The high-
energy shift of the 2,160 cm21 band on pressure increase has also been noted.

1.8.3. Quantum condensate, BEC

Until the late 1970s, hydrogen atoms were considered as the ones that could be
made both dense enough and cold enough for the atoms’ quantum identities to merge
into a Bose –Einstein condensate (BEC). Perhaps the most elegant demonstration of the
wave-like nature of matter occurs at temperatures within a few nanokelvin of
absolute zero, when atoms coalesce into the coherent collective states of a BEC.
Other atoms approaching absolute zero, however, are likely to solidify but thwart the
solidification.
In the 1980s, the MIT group coerced over 100 million atoms of hydrogen through
evaporative cooling into a single condensate, 10 times more than what can be achieved
for other atoms. They caged the atoms in a magnetic trap and lowered the walls of the
trap very fast. Hydrogen trapped when the spin of its electron and that of the nucleus
were pointing in the same direction. Over time, collisions would flip the atoms’ spins
and they would leak out faster. The MIT group speeded up evaporative cooling with a
radio frequency burst that selectively flipped the spins of the hotter atoms so that they
could flee the trap.
The remaining atoms condensed at about 40 millionths of a degree. The high
density of the condensate would force the atoms’ energy levels to become closer, lowering
the frequency of the re-emitted light. For condensates that contain relatively few atoms
(103 – 106), theoretical studies suggest that the macroscopic many-body coherence of these
matter waves can oscillate between collapse and revival. This oscillation between
coherence and incoherence in a BEC has now been experimentally confirmed (Nature,
419, 51– 54, September 5, 2002).

1.8.3.1. Proton quantum tunnelling

An electron – proton system under high pressure evolves into a fully quantum-
mechanical many-body problem that defies an exact theoretical solution. For an
interacting ensemble of particles, most calculations have to dwell on the Schrödinger
equation — the quantum mechanical description of the system. A complete theoretical
solution requires that the nuclei and electrons are treated on the same quantum-mechanical
footing.
Cosmochemistry and Properties of Light Element Compounds 59

In hydrogen bonding, the transfer of protons is of fundamental importance. In this

process, quantum tunnelling can play an important role and has been observed directly in
the solid state. The tunnelling behaviour seen in such studies usually displays the
characteristics of a particle confined in a double-well potential. But proton tunnelling can
also occur in a coordinated fashion that involves many hydrogen bonds simultaneously.
Such a process may significantly affect the properties of linear and circular networks of
hydrogen bonds, which occur in ice and in macromolecules containing hydroxyl groups
(Brougham et al., 1999).

1.8.4. Insulator –metal transition

As early as 1935, the metallic solid state of hydrogen was known and was predicted
to be superconductive. Metallic hydrogen is postulated as a major component of giant
planets and possibly the most abundant material in the Solar System.
Electrical measurements show that, under pressure .230 GPa, hydrogen (solid)
remains as an insulating material at room temperature and below. At such high pressures,
however, nitrogen remains a solid, non-molecular semi-conductor. Sulphur becomes a
superconductor near 100 GPa and remains so to ,230 GPa while xenon transforms to
a metal near 140 GPa.
Metallization of hydrogen, a quantum solid for condensed matter physics, requires
a static pressure above 300 GPa but spectroscopic characterization of hydrogen beyond
250 GPa is hindered because of the problems cited below:
(i) chemical reaction of dense hydrogen with DAC components, such as gasket, pressure
calibrant or DC itself;
(ii) line-broadening due to the non-hydrostaticity of hydrogen;
(iii) strong fluorescence from DAC, resulting in decrease in signal/noise ratio;
(iv) red-shift of the diamond window absorption edge obscuring the spectral features.
From a linear extrapolation of exciton and direct band gap of hydrogen as a function
of density, Loubeyre et al. (2002) observed that the electronic band gap closes at
0.71 mol cm23, which could lead to the formation of atomic metallic hydrogen. The
corresponding pressure of ,450 GPa seems to be a more reliable estimate of metallization
pressure.
Dynamic experiments provide evidence for liquid hydrogen becoming an electrical
conductor at 4,000 K and 140 GPa but no proof for solid hydrogen to be metallized has
been obtained through optical or conductivity measurements. The electrical conductivity
of hydrogen as a function of P, T is of interest to planetologists and also to material
scientists for its potential as a high-temperature superconductor.
The metallic transition and its effects on the EOS at pressures near 100 GPa are
integrals to the appreciation of:
(a) models of many hydrogen-bearing astrophysical objects (Van Horn, 1991), including
Jovian planets (Nellis et al., 1995), extrasolar giant planets (Saumon et al., 1996),
brown dwarfs (Hubbard et al., 1997) and low mass stars (Chabrier and Baraffe, 1997);
as well as to
60 Chapter 1

(b) the design of deuterium – tritium-burning targets for inertial confinement fusion (Nakai
and Takabe, 1996),
The phase space of hydrogen in the vicinity of the finite temperature insulator –
metal transition is difficult to address theoretically. This is because, for a dynamic, strongly
correlated and partially degenerate composite of H2Hþ and electrons (as well as other
components such as H3), no simple approximation is possible.

1.8.5. Solid hydrogen: frustrating metallic behaviour

In 1926, Bernal proposed that solid matter under high enough pressure would
become metallic, permeated with a sea of free electrons. Solid hydrogen was created in the
laboratory just over a century ago. As early as 1935, it was postulated that hydrogen
molecules should break up under high pressure to form a monatomic metallic phase
(Wigner and Huntington, 1935) or the diatomic character may yield to alkali metal
properties (at .25 GPa) (see Section 1.8.5.4).
In 1979, the first solid hydrogen at room temperature was made at a pressure of
5.7 GPa (Mao and Bell, 1979). At 30 GPa, the vibrational frequency of molecular
hydrogen starts to decrease, indicating that the molecular bond is weakening as the
pressure is increased and, at ,150 GPa, a phase transition is noted (Hemley and Mao,
1988). Above 250 GPa, the solid hydrogen begins to darken and becomes opaque,
suggesting metallization. By advances in DAC techniques, solid hydrogen has been
investigated to pressures of ,300 GPa (3 Mbar) over a range of temperatures (Mao and
Hemley, 1994), corresponding to a .12-fold increase in density relative to the zero-
pressure solid. Spectroscopic data provide information on the crystal symmetry from the
distribution of Raman- and IR-active vibrational modes as well as their intensities. Three
solid phases (I, II and III) are observed over this pressure range (Hemley et al., 1998).
A possible metallization of hydrogen under pressure provides new insights into the
nature of giant planets Jupiter and Saturn (both .400 Earth masses). At ultra-high
pressures, spontaneous electronic polarization of hydrogen dimer molecules has been
observed by Edwards and Ashcroft (1997). This arises because an electronic charge piles
up preferentially at just one of the constituent protons. The hydrogen molecule thus
develops a permanent electric dipole moment. In the limit, one would regard hydrogen as
“protonium hydride”, HþH2.
At very high pressures, H2 molecules begin to interact more strongly and the energy
bands widen. At the limit of the zero band gap, one sees an insulator-to-metal transition,
i.e., a metallic hydrogen. The pressure-induced metallization of solid hydrogen can be
viewed in terms of electronic band theory, in which one sees a completely filled valence
band separated by a very large electronic gap (.13 eV) from the empty conduction band
(Fig. 1.6) (Edwards and Hensel, 1997).
Under ambient pressure, elemental hydrogen could only conduct by thermal
excitation of large numbers of electrons from the valence band to the conduction band. But
that would require an enormously high temperature. In dense solid hydrogen, the average
density of electrons is 2 –3 times that of the excellent metal, aluminium. Strange as it is,
solid hydrogen in such a condition remains a stubborn insulator. The dense hydrogen
Cosmochemistry and Properties of Light Element Compounds 61

(Fig. 1.7) can be postulated to have a remarkably complex phase diagram of three phases
(Ashcroft, 1995):
Phase I: An orientationally disordered state — the individual H2 molecules execute
complete rotational motion in addition to the usual molecular vibrations.
Phase II: At 1.5 £ 106 atm (,120 K), the H2 molecules become “frozen” in a random
orientation in the crystal.
Phase III: The highest pressure phase, but showing no metallization. A striking absorption
of IR appears which is absent in phases I and II.
Substantial IR activity takes place in molecules with an intrinsic dipole moment. H2
molecules, when free, have a spatially symmetric electron density distribution between
their two protons and so do not possess an intrinsic dipole moment. Therefore, H2
molecules even in a dense solid state show infrared activity. Phase III, occurring at high
pressure but low temperature, shows strong IR activity (but no metallic behaviour).
With increasing pressure as the band gap continuously decreases, there comes a
point when the ionic states of hydrogen mix (in a small proportion), result in a
hybridized ground state. The pressure-induced electric dipole formed on any one of
these hydrogen molecules will become stabilized so dense hydrogen in phase III is
composed of molecular dipole hydrogen (Edwards and Ashcroft, 1997). With enough
compression, the molecular hydrogen may eventually become fully ionic, namely
HþH2 (protonium hydride). The presence of even partially ionic character in the
ground state of dense solid hydrogen will act to widen the previously narrowing band
gap and hence frustrate the transition to the long-sought metallic state. Thus, solid
hydrogen may possibly never achieve metallic status, thus frustrating Bernal’s
optimistic 1926 generalization.

1.8.5.1. Black hydrogen and metallization

Optical measurements of solid hydrogen up to pressures 220 GPa at 100 K by
Loubeyre et al. (2002) reveal that the vibron signature of H2 molecule persists to at least
316 GPa and, above 320 GPa, solid hydrogen becomes completely opaque and black.
However, above 300 GPa, the feature characteristic of direct electronic band gaps is noted.
It is predicted by Loubeyre et al. that, at ,450 GPa, this direct gap would close and
metallic hydrogen be observed.
Metallic hydrogen, a quantum solid for condensed matter physics, requires a static
pressure above 300 GPa. But spectroscopic characterization of hydrogen sample beyond
250 GPa is hindered because of the problems cited below.
(1) Chemical reaction of dense hydrogen with DAC components such as gasket, pressure
calibrant or DC itself
(2) Line broadening due to the non-hydrostaticity of hydrogen.
(3) Strong fluorescence from DAC, resulting in decrease in signal/noise ratio
(4) Red-shift of the diamond window absorption edge obscuring the spectral features.
For a linear extrapolation of exciton and direct band gap of hydrogen as a function
of density, Loubeyre et al. (2002) observed that the electronic band gap closes at
0.71 mol cm23, which could lead to the formation of atomic metallic hydrogen.
62 Chapter 1

The corresponding pressure of ,450 GPa seemed to be a more reliable estimate of

metallization pressure.

1.8.5.2. > 300 GPa

Above 300 GPa, the Raman spectrum of hydrogen can be measured accurately
(e.g., Laubyere et al., 2002). With pressure, a continuous broadening of the vibron peak is
observed. Above 160 GPa, this becomes more pronounced in phase III. Large widths of
vibrons seem to be an intrinsic property of hydrogen. In contrast, the low-frequency libron
modes remain sharp up to the maximum pressure. The pressure shift of librational modes,
the phonon modes and the vibron modes are seen in phase III to 220 GPa (Goncharov et al.,
1998). A continuous pressure shift in phase III suggests it to be of molecular hydrogen,
which is stable up to 316 GPa at least (Loubeyre et al., 2002). Also, the vibron mode
proves that the molecular form persists to the maximum pressure.
From the absorption edges in solid hydrogen at ambient pressure, an exciton of
10.9 eV and direct band gap of 14.5 eV have been determined. The direct band gap of
hydrogen closes faster under pressure than the band gap of the diamond window and they
would cross under pressure. Above the crossing point, the absorption edge of solid
hydrogen would no longer be observed by the absorption of the diamond window and,
hence, the electronic state of dense hydrogen could be deciphered with confidence. The
direct absorption edge corresponding to an absorption coefficient of 30,000 cm21 is typical
for a direct band gap. The absorption measurements can be translated into an optical
aspect:
290 GPa: hydrogen changes white –yellow – orange
290 GPa , P , 320 GPa: hydrogen changes to red
320 GPa: opaque to visible light.

1.8.5.3. Effective charge: phase III

The charge transfer effect in hydrogen phase III may be quantified (see Hemley and
Mao, 1998) by the effective charge Q p associated with the vibron (Hemley et al., 1997),
defined as

Qp ¼ Qi þ Qd ¼ dD=du;

where D is induced dipole moment, Qi is ionic charge and u is ionic displacement. For a
vibron, Qd ¼ R½dQi ðRÞ=dR; where R is the H – H bond length. The charge is related to the
oscillator strength f as Qp ¼ eðMf =2pmÞ1=2 ; where e is the electron charge and M is
the reduced mass of the vibrational mode. At the highest pressures of the measurements in
phase III (230 GPa), Q p reaches a value of 0.04e (Fig. 1.9). The results indicate strong
interaction between molecules at these pressures, but the magnitude of Q p puts strong
constraints on the hypothesis that phase III is an ionic state consisting of HþH2 ions. The
intramolecular stretching mode (vibron) of hydrogen III is shown in Fig. 1.10. The
magnitude of the absorption represents a significant increase over that found at low
pressures in phases I and II. Notably, in phase I, the absorption increases with the square of
the density ðr2 Þ (Hanfland et al., 1992), the same dependence which is documented for
Cosmochemistry and Properties of Light Element Compounds 63

collision-induced absorption in low-density gases and used in opacity calculations for

planetary atmospheres (Guillot et al., 1994).

1.8.5.4. Solid hydrogen: alkali metal(?) at 340 GPa

Hydrogen, having a single electron, shows characteristics of both the group I alkalis
and the group VII halogens. At low pressure, hydrogen isotopes are halogenous; its
covalent diatomic molecules form insulators. Under pressure, they transform to behave
like alkali metals.
Natoli et al. (1993) determined the crystal structure of solid hydrogen and suggested
that, for solid hydrogen, a transition occurs to a diamond-like cubic atomic phase (a
gapless semi-conductor or a semi-metal) at ,340 GPa. They concluded that this should be
transformed to a metallic simple cubic phase at 650 GPa and then to a metallic body-
centred cubic phase around 900 GPa. Both these cubic phases have one free electron per
atom.
Solid hydrogen, an electrical insulator, was predicted to become an alkali metal
under extreme compression, although controversy surrounds the pressure required to
achieve this (e.g., Loubeyre et al., 1996). Calculations (e.g., Natoli et al., 1993) suggest
that depairing through the destruction of the molecular bond should occur at ,340 GPa,
accompanied by the formation of alkali metal at this pressure (Ashcroft, 1990) or at a much
higher pressure. But recently, Narayana et al. (1998) report that solid hydrogen does not
become an alkali metal even at pressures up to 342 ^ 10 GPa (near to Earth’s core
pressure), which can be achieved using a diamond-anvil cell. This observation evidently
has important implications in the theoretical understanding of the solid-state phase of
hydrogen.
Metallization of solid hydrogen by band overlap is theoretically calculated to occur
in the pressure range between 260 and 410 GPa (e.g., Johnson and Ashcroft, 2000). This is
predicted to happen before breakdown to a mono-atomic solid. However, experimental
study in this pressure range faces a great problem. A large, stress-induced increase in
optical absorption and fluorescence in diamond anvils (Ruoff et al., 1991) obstructs optical
measurements of hydrogen samples and pressure calibration by ruby fluorescence. At
pressures .180 GPa, the ruby fluorescence becomes extremely weak and is masked by
diamond fluorescence.
For an ultra-high pressure study of hydrogen, IR spectroscopy is particularly suited
because of the dramatic increase in vibron intensity in phase III. Hydrogen vibron persists
to very high pressure, indicating that hydrogen molecules remain in tact. Goncharov et al.
(2001) constrained the pressure of transition to the metallic state to 325– 495 GPa.

1.8.6. Ortho- and para-hydrogen

The possible rotational states of the hydrogen molecule are linked to the total of its
nuclear spins employing the constraints of quantum mechanics. This leads to separating
two species of hydrogen: ortho with parallel nuclear spins, and para with opposite spins.
However, the difference in an apparently insignificant microscopic spin property may
manifest macroscopically.
64 Chapter 1

The ortho– para conversion in solid hydrogen at ambient pressure is mainly due to
magnetic dipole– dipole interactions between the nuclei of two neighbouring ortho-
molecules, with the energy released carried away by phonons (e.g., Berlinsky, 1975).
However, recent studies on solid H2 up to 58 GPa have revealed a new conversion
mechanism in which the emerging excitations are coupled to the converting molecules via
electric quadrupole – quadrupole (EQQ) rather than nuclear spin – spin interactions
(Strzhemechny and Hemley, 2000). In the new mechanism, the coupling enhancement
is ensured by high compression and a gap closing. With increasing pressure, the conversion
energy diminishes.
Over the pressure range up to 58 GPa, the EQQ interaction was invoked. The
characteristic dimension of the charge distribution of the hydrogen molecule (derived from
Q ¼ er 2 ) is considerably less than the interatomic separation R such that r=R , 0:25:
Strzemechny and Hemley (2000) considered matrix elements between J ¼ 1 states and
between J ¼ 0 and J ¼ 2 states, for which the hexadecupolar matrix elements are zero.
They first considered J ¼ 2 protons and phonons as energy sinks. The J ¼ 2 proton is too
high for the conversion energy to bridge the gap at pressures below 60 GPa.
The solid consisting of J ¼ 0 molecules ( p-H2) transforms to an ordered (BSP
phase or phase II) near 110 GPa (Lorenzana et al., 1990). The other transformation is
observed near 150 GPa (phase III). It is characterized by an 100 cm21 vibron discontinuity
(both IR and Raman) and an increase by several orders of magnitude of the IR vibron
oscillator strength (Hanfland et al., 1993). The latter possibly arises from symmetry-
breaking charge transfer and orientational ordering (Mazin et al., 1997). In phase III, a
completely different manifold low-frequency excitation is observed (Fig. 1.12)
(Goncharov et al., 1998). Increasing pressure within the phase leads to sharpening of
the peaks and a substantial frequency shift to higher energies. The IR combination bands
originate from zone-centre IR active vibron and Raman-active librons.

1.8.6.1. Ortho –para conversion: quantum solid state

Ortho –para ðo – pÞ conversion of solid hydrogen is an ultimately quantum process.
High-pressure studies help to explore phenomena such as molecular dissociation and
metallization to quantum solid-state effects. o – p conversion in solid hydrogen is known to
pressures up to a few GPa. The rate of o – p conversion obeys the equation:

2C2 dC=dt ¼ dðC 21 Þ=dt ¼ K;

where C is the ortho fraction. The equation requires that two ortho neighbours should be
needed for one of them to convert.
The o – p conversion is promoted by magnetic dipole –dipole interaction between
the nuclei of two neighbouring ortho molecules. In the conversion, the energy released is
carried away by phonons. At zero pressure, the energy is removed by emission of two
phonons. When the pressure is raised to a few GPa, one phonon dominates over the two-
phonon process, owing to the quantum crystal nature of solid hydrogen. The one-phonon
emission corresponds to the conversion of one ortho molecule. However, the conversion
rate diminishes rapidly as the pressure is raised (Strzhrmechny et al., 2002) and reaches
Cosmochemistry and Properties of Light Element Compounds 65

a minimum at ,5 GPa. Beyond 5 GPa, the conversion rate increases rapidly. The
conversion is utilized in probing the phonon DOS in the highly anharmonic quantum
crystal of hydrogen.
Three factors control the o – p conversion rate in a solid hydrogen.
(a) Various magnetic field sources driving ortho molecules to convert to the para state.
(b) The channel for carrying away the energy released from the conversion to the citric
acid cycle.
(c) The spatial distribution of ortho molecules in the sample.
For (b), the three basic energy sinks in channels are: (i) phonons, (ii) J ¼ excitations
(rotons) and excitations within J ¼ 1 manifold (librons).
Since o – p conversion is caused by the interaction with neighbouring ortho
molecules, the rate of variation of the ortho fraction C is proportional to the average
number of nearest ortho neighbours, M:

dC=dt ¼ 2K=12MC or d=dtð1=CÞ ¼ KðM=12CÞ

where K is conversion rate.

Two ortho neighbours are needed for at least one of them to convert to para. When
the ortho distribution is random, then M ¼ 12C (12 nearest neighbours in a hcp lattice). At
low temperature, when diffusion is faster than o – p conversion, the number of ortho – ortho
nearest-neighbour pairs will be substantially larger than the random distribution. This leads
to clusterization of low ortho fractions at low temperatures.

1.8.6.2. Conversion energy channels: EQQ

Conversion mechanisms may include double conversion, excitation in the J ¼ 1
and J ¼ 2 manifolds as conversion energy sinks. Intermediate states are possible from
which conversion energy is dissipated via the strong EQQ interactions. Compression
decreases the conversion energy (gap closing).
The other conversion energy excitations may be from rotational subsystems.
These excitations, including librons (or reorientations within the J ¼ manifold) and
rotons (J ¼ 2 excitations), are controlled by the EQQ interactions scaled as R-5
(Strzhemechny et al., 2002).
A look into the possible lowest energies, related to the EQQ interactions reveals the
following options:
(i) single conversion, no nearest para neighbours, (ii) single conversion, one or a
few para neighbours, (iii) double conversion, no nearest para neighbours, (iv) double
conversion, one or a few para neighbours.
Pressure dependence of the ortho– para conversion rate in solid hydrogen provides
a basis for a possible conversion mechanism, including quantum diffusion. Double
conversion (when both interacting ortho molecules go to the para state) and excitations
within the J ¼ 1 manifold (librons or their analogues in disordered o – p mixtures) serve as
sinks for the released conversion energy.
J ¼ 1 manifold (librons) excitations: Since the conversion energy is the difference
in energy between J ¼ 1 (surrounded by ortho neighbours in the solid) and J ¼ 0 states,
66 Chapter 1

the pressure-related increase of the EQQ interaction must result in a reduction of the
Boltzmann-averaged conversion energy. This is because, at very low temperature, all ortho
molecules are mostly in ground states. Since the ground-state energy of J ¼ 1 molecules
varies from site to site, the conversion energies are distributed over a finite interval, even at
zero temperature.
J ¼ 2 excitations (rotons): In the channel, the other energy-removing sink is the
J ¼ 2 excitation (roton). An ortho molecule converting to the para state can go directly to
the J ¼ 2 manifold. A significant contribution to the conversion energy sink may be
derived from the Hamiltonian, which drives two ortho molecules to the para state
simultaneously, one to the J ¼ 0 and the other to the J ¼ 2 state. During conversion, when
the molecule goes from J ¼ 1 to J ¼ 2; it finds virtually the same molecular field.
Strzhemechny et al. (2002) suggest a new channel of o – p conversion in solid
hydrogen that is efficient at high pressure. This conversion channel with excitations in the
J ¼ 1 manifold serves as the energy sink and is responsible for the strong conversion
enhancement at high compression ðj . 3Þ: Gap closing adds substantially to the
conversion enhancement at high pressure. An abrupt slowing down of the conversion is
expected (by theory) as pressure goes beyond 65 –70 GPa for 65 –75% ortho.

1.8.7. Hydrogen in Jupiter

The planet Jupiter is composed primarily of hydrogen (,90%) with about 10 at.%
helium. Most of the hydrogen is at ultra-high pressure in a fluid metallic state. This state is
responsible for the large magnetic field arising from the fluidized metallic hydrogen within
Jupiter’s mantle. Near the transition, the hydrogen atoms exhibit complex coupled motions
(similar to those seen in ferroelectric materials) as well as a quantum mechanical
intermediate state characterized by proton tunnelling.
The surface conditions of Jupiter are P ¼ 1 bar; T ¼ 165 K: A quiescent boundary
layer might be expected to occur between depths corresponding to 30 –180 GPa. The
boundary is specified to be possibly at 42 GPa. This boundary layer might facilitate
separation of ice and rock to cause abrupt density change. At greater depths, pressures
become sufficient to completely dissociate and metallize hydrogen. Near the molecular
mantle – core boundary, there might be a local minimum temperature. The P and T at the
centre range up to 400 GPa and 20,000 K, respectively. Hydrogen is metallic in the inner
,77% of the radius of Jupiter. In it, the boundary between the molecular hydrogen mantle
and the metallic hydrogen core is estimated to be at 300 GPa (3 Mbar) and 10,000 K at
0.77 of Jupiter’s radius.
The density distributions are calculated from the gravitational moments, which are
very sensitive to the EOS at very high temperature and pressure. The interior is assumed to
be fully convective and adiabatic, and thus obeys an isentropic EOS.
Above 20 GPa and 4,500 K, H2 begins to dissociate and hydrogen becomes
monatomic at pressures of 300– 500 GPa. Dissociation causes the absorption of a few
electron volts per molecule, which results in a considerable drop in temperature. Lower
temperatures are primarily caused, therefore, by the partial and continuous dissociation of
Cosmochemistry and Properties of Light Element Compounds 67

hydrogen into monatomic hydrogen, as observed in the case of nitrogen dissociation

(Nellis et al., 1991).
In Jupiter, He (of ,10 at.%) causes the temperature to increase monotonically with
density and pressure. It is because of the temperatures involved that helium has no internal
energy-absorption mechanism, such as dissociation or electronic excitation. The model of
Nellis et al. (1995) shows that the continuous dissociation of H2 keeps the isentropic
temperatures relatively constant near 4,000 K over a wide range of pressures (and thus
depths) from 30 to 180 GPa. The continuous dissociative phase transition in hydrogen
suggests that no sharp boundary exists at what is now called the core – mantle boundary.
That is, as H2 dissociates continuously to the monatomic phase and eventually metallizes,
there is probably no sharp discontinuity in the relative composition of the molecular and
metallic phases nor in the density (Nellis et al., 1995).

1.8.8. H in terrestrial planets

The presence of a significant amount of water is evidenced in the polar caps and the
subsurface channels of Mars (Malin and Edgett, 2000; Zuber et al., 2000). Large quantities
of hydrous phases are surmised to be present on Europa (McCord et al., 1999).
The incorporation of minor amounts of H in mantle phases may have played
significantly in the development and evolution of the Earth’s hydrosphere. Hydrogen has
considerable effects on the properties of mantle minerals in terms of rheology, phase
relationships, crystal chemistry and dehydration reactions (e.g., Inoue et al., 1998).
The critical point of water is ,3748C and 22.1 MPa. At elevated temperatures
(150 – 3008C), the dielectric constant of water is greatly reduced while the ionic product
increases (Siskin and Katritsky, 1991). At elevated temperature, the solubility parameter of
water approaches that of polar organic solvents. Pressure greatly expands the temperature
stability of the aqueous phase.

1.8.8.1. Hydrogen in the Earth’s minerals

In crustal minerals and fluids, hydrogen is commonly incorporated as OH2 and
H2O, with “free” protons usually considered bound as H2Oþ, H5Oþ þ
2 and NH4 . The OH
2

can be incorporated in the structure by substitution and/or crystallographic shearing

mechanisms. The other mechanism, called hydrogarnet substitution, involves the exchange
of 4Hþ for Si4þ (or 3Hþ for Al3þ) (e.g., Prewitt and Parise, 2000). In addition, hydrogen
can form iron alloys as iron hydride (H2) at the Earth’s core (e.g., Fukai, 1993).
Hydrogen bond occurs as a linear O – H· · ·O unit, which involves the pairing of a
weak O· · ·H and strong covalent O –H linkage. The strength of hydrogen bonding depends
on the O – O distance and the O – H· · ·O bond angle. Effects of pressure on the hydrogen
bond have been studied on Mg(OH)2, Ca(OH)2, etc. (e.g., Catti et al., 1995; Nagai et al.,
2000). Hydrogen bonding can either increase or decrease on compression, depending on
the crystal structure and composition (e.g., Kagi et al., 2000).
High-pressure polymorphism in hydrous phases is likely to be strongly influenced
by pressure-induced variations in hydrogen bonding (Faust and Williams, 1996).
The prototypal case is the symmetric hydrogen-bonded phase of H2O at 60 GPa
68 Chapter 1

(Aoki et al., 1996). Compression of the O –H· · ·O group in this system gives symmetric
bonding with distances of 2.38 – 2.40 Å.
Under pressure, the weakened covalent OH bonds may cause large effects on the
diffusive behaviour or superionic conductivity (e.g., as predicted for subsolidus H2O)
(Cavazzoni et al., 1999). This may provide a possible mechanism for attenuation at seismic
frequencies.
Pressure can induce a disordering of the hydrogen sublattice (Parise et al., 1998),
perhaps arising from H – H repulsion on compression. In some cases, this disordering may
appear as amorphization. Such metastable transition may lead to large-scale phenomena
such as deep-focus earthquakes (i.e., driven by pressure-induced amorphization/dehydra-
tion in serpentine) (Meade and Jeanloz, 1991). But this view has been contested later by
some workers (e.g., Kuroda and Irifune, 1998).
Under high pressure, hydrogen reacts with silicates, oxides and metals and forms
hydrogen silicates and oxides. Metal hydrides have been documented to very high
pressure, and iron hydride at room temperature is seen to be stable up to megabar pressure.
All these observations suggest that hydrogen can exist in substantial quantities in planetary
cores and that hydrides could have formed in the early stages of planetary evolution.
Hydrogen and other light elements depress melting relations significantly (Yagi et al.,
1994); this could have a profound effect on the thermal state of planetary interiors.
A sample of iron placed in a diamond-anvil cell with liquid hydrogen as the
pressure medium shows expansion at 3.5 GPa when hydrogen is forced into an iron
structure. This transition to an iron hydride structure appears to be complete between
9.5 and 14.7 GPa. This structure consists of double hexagonal stacking (ABAC) with two
different crystallographic sites for Fe. The composition is estimated to be FeH0.94. The
presence of this compound accounts for the observed hydrogen embrittlement of iron noted
by metallurgists. For further observation on H in an iron core, the reader is advised to
see Section 14.2.

1.8.8.2. Water in the Earth: D/H ratios

Nominally anhydrous phases may contain very significant amounts of hydrogen and
thus may serve as major sinks for protons in the mantle. The presence of the nominal
amount of 0.01% H2O in the minerals of the transition zone is tantamount to 800 m of
liquid water over the entire surface of the Earth. The presence of H in rocks has a major
effect on melting temperature, strength and elastic properties, and thus may control melt
generation, solid state convection and seismic velocities in the Earth.
The hydrogen contents in olivines from mantle xenoliths range from 10 to
60 ppm wt H2O. The hydrogen content of xenolithic olivines does not attain equilibrium
with water in the host magma during the transportation from the Earth’s mantle to the
surface and is taken as a reflection of the hydrogen condition in the mantle.
The mantle is generally poor in incompatible elements and is generally water-poor
(0.01 – 0.1% H2O by weight). Hence, hydrous minerals cannot be present throughout the
mantle but may be present locally in K-, F-, Ti- and H2O-rich areas or cold regions such as
subducting slabs (e.g., Thompson, 1992). Hydrogen is incorporated into mantle olivine by
coupled substitution as OH2 ion in oxygen positions adjacent to the M-site vacancies
Cosmochemistry and Properties of Light Element Compounds 69

(Kurosawa et al., 1997). Hydrogen can be associated with negatively-charged oxygen

interstitials and its uptake is governed by point-defect mechanisms and coupled
substitution with other cations (e.g., Bell and Rossman, 1992) (see also Section 1.8.8.3
and Chapter 13).
The properties of dense hydrous magnesio-silicates (DHMS), supposed to be
stabilizing in the transition zone and below, are discussed in Chapter 13. For DHMS
phases, the strength of the hydrogen bonding at ambient pressure generally increases as the
pressure, stability and density of the DHMS phases increase (e.g., Faust and Williams,
1996). In these metal hydroxides, where H· · ·H and O· · ·H interactions are well separated,
the competition between pressure-induced hydrogen bonding and H – H repulsive
interactions can be an important issue (Parise et al., 1998, 1999).

D/H ratio in the Solar System. The Sun and planets formed some 4.55 b.y. ago from the
protosolar nebula, a rotating disk of gas and grains largely made of molecular hydrogen
and helium. From the centre to the edge, the disk is believed to have had a homogeneous
isotopic composition.
However, the isotopic composition of water on Earth differs widely from that of the
primitive Sun. A deuterium – hydrogen (D/H) ratio of (149 ^ 3) £ 1026 has been
estimated from the bulk Earth, compared with a solar ratio of (20 ^ 4) £ 1026 deduced
from solar wind implanted into lunar soils (see Robert, 2001). The origin of the Solar-
System water is ascribed to an interstellar process. A theoretical D/H ratio in interstellar ice
(up to 1022) differs markedly from the highest measured value in the Solar System
(720 £ 1026). In the Solar System, two sources of water can be presumed: an interstellar
source and a protosolar source.
Spectroscopic studies of water vapour from comets approaching the Sun have
revealed D/H ratios of (310 ^ 40) £ 1026; substantially higher than that of terrestrial
water (150 £ 1026). (Note: The contribution of cometary water to terrestrial oceans should
be ,10%. Some models predict that the D/H ratio for water condensed at 1 AU should be
close to protosolar value of ,80 £ 1026. Hydrogen extracted at low temperature under
pyrolysis from meteorites exhibits a deuterium-depleted signature: D/H ratio
,80 £ 1026.)
The primitive carbonaceous chondrites are seen to contain two hydrogen carriers:
water in clay minerals and hydrogen in organic compound with macromolecular structures.
Chemically extracted organic matter shows an enrichment in deuterium relative to Earth
with D/H ratios up to (380 ^ 10) £ 1026; in some rare meteorites this reaches up to
(720 ^ 120) £ 1026. This enrichment is ascribed to chemical reactions that took place
earlier to the planet-formation process.
For the Martian meteorites (SNCs), the observed D/H ratio of 300 £ 1026 is
ascribed to the mantle of Mars. If so, the cometary water on Mars should be much higher
than on Earth. On Mars, water is photo-dissociated by the UV flux and the liberated lighter
isotope of hydrogen escapes to space at greater rate than D. Thus, in the Martian
atmosphere, the D/H ratio increases to a value as high as 810 £ 1026 compared with that in
its mantle, amounting to 300 £ 1026, which is higher relative to the Earth. Groundwater,
70 Chapter 1

rainwater, seawater, glacier ice and extraterrestrial water each have a distinctive isotope
ratio, which allows determination of where the water originated from.

1.8.8.3. H/H2O in mantle phases

In garnets, the SiO4 tetrahedron may be substituted by O4H4. In the major phases of
mantle xenoliths such as olivine, orthopyroxene and clinopyroxene, variable amounts of
structural water occur as defects, as shown in the following:

Mantle minerals in xenoliths H2O content Author

Orthopyroxene 200–650 ppm (e.g., Ingrin and Skogby, 2000; Keppler and Rauch, 2000)
Olivine ,100 ppm (e.g., Ingrin and Skogby, 2000; Keppler and Rauch, 2000)

In olivine, the planar features likely involve substitution of OH2 hydroxyl units for
oxygen ions coupled with the formation of Mg2þ or Si4þ vacancies to charge balance (e.g.,
Libowitzky and Beran, 1995). The estimates of H2O contents in minerals from mantle-
derived xenoliths probably represent only the lower estimates of the water present at depth
(Williams and Hemley, 2001).
The solubility of water in olivine increases rapidly with pressure, reaching a
maximum value near 0.12 wt% (1,200 ppm) at a temperature of 1,1008C at pressures
corresponding to 400 km depth (Kohlstedt et al., 1996). Such water affects the electrical
and viscous transport properties of the mantle.

1.9. Water and ammonia in Uranus and Neptune

In Uranus and Neptune, the density profiles suggest that their interiors are
composed mainly of a thick intermediate layer of “hot ices”, predominantly water,
hydrocarbons and ammonia in solar proportions (molar fractions: H2O 56%, CH4 36% and
NH3 8%) (Hubbard, 1981). Above this intermediate layer lies the gaseous atmosphere and
below is the rocky core. Uranus and Neptune have H2O as the major constituent, hence the
properties of H2O under pressure are crucial for modelling their interiors and transport
properties.
Pressure (and temperature) conditions within the icy layer range from 20 GPa (and
2,000 K) to 600 GPa (and 7,000 K) along the planetary isentrope (Podolak and Stevenson,
1995). The computed EOS of water and ammonia, close to the planetary isentrope,
compare well with shock-wave data. The melting curves of water and ammonia run below
all presently accepted planetary isentropes (Hubbard et al., 1995). The isentropes of
Uranus and Neptune are believed to be similar. Along the planetary isentrope, ammonia
and water components of the ice layer are predicted to be electronically insulating up to
300 GPa. Thus, the electrical conductivity in the outer part of the ice layers can only come
from the large proton mobility in the ionic liquid phase.
Shock-wave studies showed evidence for the breakdown of CH4 at 100 GPa,
suggesting that this material may be pyrolysed to diamond or a carbon-rich phase within
Cosmochemistry and Properties of Light Element Compounds 71

planetary bodies. Bernasconi et al. (1995) predicted the existence of new dense
hydrocarbon phases produced from polyacetylene below 100 GPa.
Gravitational data suggest that the mantles of Uranus and Neptune are composed of
materials of a density intermediate between H2 –He mixtures and silicate-iron of terrestrial
planets (Hubbard et al., 1995).

1.9.1. Electrical conductivity: “synthetic Uranus”

At low pressure, molecular water exhibits a low ionic conductivity arising from
dissociated molecules. With pressure, the number of protonic carriers increases
exponentially across the molecular –ionic cross-over. In the ionic regime, all the protons
contribute equally to the conductivity. The ice layer is considered to be the source of the
magnetic fields of Uranus and Neptune measured by the Voyager 2 spacecraft (Ness et al.,
1989). Electrical conductivities in the ice layer of the order of 10 ohm21 cm21 are
necessary to sustain the planetary dynamo mechanism for the generation of such a
magnetic field (Kirk and Stevenson, 1987). The large conductivity may arise from nearly
complete ionization of H2O. At the deeper regions, if pressure metallization of the ice layer
occurs then electronic conduction may also contribute to the dynamo.
Only shock-wave experiments up to 200 GPa (Nellis et al., 1997) have measured
the EOS of water, ammonia and isopropanol — collectively termed “synthetic Uranus”.
The results show that the electrical conductivity in water and ammonia increases
exponentially along the Hugoniot up to ,20 GPa and then levels off. The measured value
of the conductivity of water above 20 GPa (10 ohm21 cm21) supports the planetary
dynamo models. Conductivity data of water and ammonia above 77 GPa are not available
and this delimits modelling of the origin of the magnetic fields of Neptune and Uranus.
Ab initio simulations have been used to determine the high-pressure, high-
temperature behaviour of methane (Ancilotto et al., 1997) and the structural phase
transitions of ice at high pressure (Benoit et al., 1998; Bernasconi et al., 1998). Simulations
at 300 –400 GPa predict a new crystalline phase of ice to display a fast proton conductivity
above 2,000 K (Beniot et al., 1996). This supports the ionic model of conductivity in an
ice layer.

1.9.2. Metallicity(?) of water and ammonia

Deep inside the giant planets, the ice –core boundary is met at 600 GPa and
7,000 K, in close accordance with the presumed planetary isentrope (Hubbard et al., 1995).
At this P, T condition, water and ammonia are predicted to be metallic (water is assumed to
be metallized at 7,000 K /,300 GPa). At 300 GPa, ammonia metallizes when T is around
5,500 K.
According to Cavazzoni et al. (1999), the dynamo generation of the magnetic field
should involve the contribution of the high electronic conductivity due to the metallic
liquid in the inner part of the ice layer and the lower electrical conductivity due to the
proton mobility in the electronically insulating liquid in the outer part of the ice layer.
72 Chapter 1

In other models of the giant planets (Hubbard, 1997), the interior is isothermal at <5,000 K
above 150 GPa, and the ices would remain electronically insulating in the whole planet.

1.9.3. Water: structural order and anomalies

The phase diagram of water is shown in Fig. 1.14. In liquids such as water,
directional attractions (hydrogen bonds) combine with short-range repulsions. Anomalies
in the kinetic and thermodynamic properties of water have been discussed with the results
of the study of translational (e.g., Truskett et al., 2000) and orientational order in a model
(Chau and Hardwick, 1998) of water. The structural anomalous region is bounded by loci
of maximum orientational order at low densities and minimum translational order at high
densities. In this region, order decreases on compression. The anomalies in structural,
kinetic and thermodynamic properties constitute a cascade when the degree of order
increased. The order in water is measured in terms of two parameters such as translational
order parameter, t, and orientational order parameter, q (e.g., Tanaka, 2000) as discussed in
the following:
(i) Translational order parameter, t: This measures the tendency of pairs of molecules to
adopt preferential separations. For ideal gas, t ¼ 0 and for crystals, t ! 1:
(ii) Orientational order parameter, q: This measures the extent to which a molecule and
its four nearest neighbours adopt a tetrahedral arrangement (Paschek and Greiger,
1999), e.g., in hexagonal Ih :
For an ideal gas, q ¼ 0 while, in a perfectly tetrahedral arrangement, q ¼ 1:
Hydrogen-bonding determines both the mutual orientation and the separation between
molecules. Increasing the temperature weakens the coupling between translation and
orientational order.
In liquid water, there occurs a cascade of anomalies (Errington and Debenedetti,
2001). All anomalous states share the topological property whereby the orientational and
translational order are strongly coupled.

Figure 1.14. Phase diagram of water.

Cosmochemistry and Properties of Light Element Compounds 73

(a) Structural anomalies: This occurs under compression when order decreases and it
spreads over the broadest range to densities and temperatures.
(b) Diffusive anomalies: This occurs as diffusivity measured by compression in the region
of structural anomalies.
(c) Thermodynamic anomalies: This occurs when density decreases upon cooling at
constant pressure inside the region of diffusive anomalies.
In addition to the above, liquid water is known to possess anomalies such as:
(i) increase in isothermal compressibility and isobaric heat capacity upon cooling,
(ii) decrease in viscosity upon compression, (iii) polyamorphic transition and (iv) the
possible existence of a second critical point (Mishima and Stanley, 1998a,b).

1.9.3.1. Proton (and oxygen) diffusion in water

In a condition of increasing pressure, an abrupt transition was seen to occur in water
at ,2,000 K, after which the protons become highly diffusive. The proton diffusion
coefficient Dwater
H at 2,500 K (150 GPa) equals 6(^1) £ 1024 cm2 s21. Along the O –O
separations, diffusion occurs by jumps. Simulation study shows that, even around this
temperature, the oxygen atoms vibrate around the bcc lattice positions, leading to a
superionic solid phase of space group Pn 3m: 
Further increase in T(¼ 3,500 K) leads to the melting of the oxygen sublattice.
The diffusion coefficient of oxygen in water at 4,000 K(/150 GPa) is seen as Dwater o ¼
3ð^15Þ £ 1025 cm2 s21 and the increase in proton mobility is Dwater H ¼ 1:8ð^3Þ £
1023 cm2 s21 ; indicating a further transition to a fluid phase.

1.9.4. Superionic solid state

Cavazzoni et al. (1999) found a well-defined superionic crystalline phase to be

present in the phase diagram of water. Their simulation confirmed the experimental
findings, specifying the cross-over from ionic to molecular liquid at a somewhat lower
pressure (,20 GPa) (Nellis et al., 1988). They found that the ionic liquid phase freezes
spontaneously when cooled in the superionic solid temperature region. The oxygens
solidify to an amorphous state, whereas the protons continue to diffuse highly.
The superionic solid state and the ionic liquid state have an electronically insulating
character. The electronic density of states at 300 GPa shows that the electronic gap
decreases from 10 eV at 300 K to about 5 eV at 2,500 K within the superionic phase, and
finally closes at about 7,000 K inside the fluid phase. At lower pressures (150 GPa),
metallization is observed at a higher temperature of 7,500 K (Cavazzoni et al., 1999).
Inside the fluid phase, three behaviours are observed: (i) molecular regime at
low pressure, (ii) a non-metallic ionic regime at intermediate pressure and temperature and
(iii) metallic regime at high pressure and temperature.

1.9.4.1. Ammonia: superionic state

In ammonia (NH3), the superionic solid state and the non-metallic ionic liquid
states are observed as in water, but at lower temperatures.
74 Chapter 1

Covazzoni et al. (1999), through a simulation, endorsed the earlier observed

(Loveday et al., 1996) stability of the pseudo-hcp (P212121) crystalline structure of
ammonia at low temperature in the pressure range of 5– 15 GPa. With increasing pressure,
only the small distortion (from ideal hcp) present in the nitrogen sublattice was observed to
decrease continuously. With increasing fluid pressure, a cross-over from a molecular liquid
to an ionic state was found. By increasing temperature at higher pressures ($ 60 GPa), a
transition from the P212121 structure to a hcp superionic solid above 1,200 K was
observed. In the phase diagram, a triple point was noted between 30 and 60 GPa.
In the superionic phase of ammonia, the protons jump between the neighbouring
molecules and the diffusion coefficient at 2,000 K and 150 GPa is seen to be:
DNH
H
3
¼ 3ð^0:4Þ £ 1024 cm2 s21

1.10. H2 mixtures and clathrates

The gaseous planets are multi-component systems and the properties of the mixtures
of planetary materials are just beginning to be studied. The most important discovery
relates to new high-pressure compounds involving hydrogen and other low-Z elements
or simple molecular materials, particularly the C –H – O – N system.
In the H2 –H2O binary system, a novel type of clathrate with 1 : 1 ratio was
discovered. In this, H2O and H2 form two interlocking diamond networks and the
compound is stable to at least 30 GPa. The compound approaches a symmetrically
hydrogen bonded state at ,40 GPa. Experiments on a mixture of H2O and H2 reveal new
classes of dense clathrates, such as H2·H2O-clathrate (Fig. 1.4), in which diatomic H2 is
preserved in pressures excess of 50 GPa. In the H2 – H2O system, two new clathrates have
been found (Vos et al., 1994), of which one is stable to ,60 GPa.
H2 – H2O clathrate hydrate has been studied up to 60 GPa using X-ray diffraction
and Raman spectroscopy (Vos et al., 1996). In this novel hydrogen –water clathrate, H2O
molecules are arranged as seen in pure ices Ic and VII. Two types of clathrate hydrates have
been reported: C1 and C2. C2 has unusual 1 : 1 stoichiometry and is stable beyond 2.3 GPa.
The structure of C2 consists of two interpenetrating diamond lattices, one for H2O and the
other for H2. Thus, it can be regarded as ice VII. It is also found that the pressure
dependencies of nO – H and dO – H· · ·O are in excellent agreement with those of ice VII, at 1/2
pressure. Again, the relation between nO – H and dO – H· · ·O is very similar for both C2 and
pure ice VII. Hence, C2 can be regarded as a model system for ice VII (at 1/2 pressure),
and C2 phase seems to be the best candidate for studying symmetric hydrogen bonds in
H2O-bearing minerals (Vos et al., 1996).
Although a compound, Ar(H2)2 has been noted, but no measurable solubility of He
in H2 above 100 GPa(/,300 K) has been reported.

1.10.1. H2 – O2 mixture: “Hard Spheres”

At different layers, the large outer planets are believed to contain clouds of solid
condensates and a depth of metastable H2 – O2 compound layers. The cosmic abundance of
Cosmochemistry and Properties of Light Element Compounds 75

H2 and O2 is high and an explosive reaction between them occurs to form H2O. These two
gases, therefore, offer the ideal clean fuel sources. But under pressure (carefully increased),
the explosive reaction can be arrested. In a diamond-anvil cell, the diatomic molecules are
seen to persist by gradual stepwise pressure increase to very high pressures, with no
formation of water. In their experiments, instead of water, Loubeyre and LeToullec (1995)
found a kinematically stable phase consisting of intact diatomic molecules with a
stoichiometry near (H2)4(O2)3.
Such high-pressure Van der Waals compounds or ordered alloys were first
documented in He – Ne mixtures (Vos et al., 1992) such as the formation of He(Ne2)11
compound. Now such compounds are being found in a growing number of binary mixtures.
The driving force for the formation of such compounds is the packing effect, and their
higher density gives a lower free energy. In such binary mixtures, the gas particles are
treated as “hard spheres”.

1.10.2. CH4 –H2 and Ar –H2 systems: Laves phases

Four new solid compounds having H2 : CH4 molar ratios, 1 : 2, 1 : 1, 2 : 1 and 4 : 1

have been characterized and single-crystal X-ray diffraction was employed for crystal
X-ray structure determination. The 1 : 1 compound was stable to at least 30 GPa (Hemley
and Mao, 1998). In infrared investigation, an unusual charge-transfer process at pressures
above 30 GPa was noted by Somayazulu et al. (1999). A similar observation was made in
HP IR study of Ar(H2)2 compound to pressures of 220 GPa (Datchi et al., 1996). New
compounds in the H2 –CH4 and H2 –O2 systems are detected.

1.10.2.1. Laves phases

There are a number of intermediate structures that, while densely packed, do not
easily fit into the conventional classification scheme. Laves proposed a radius-ratio rule to
predict the existence of intermetallic compounds with characteristic structures. When hard
spheres are assumed, the AB2 compounds (intermediate type) should have a radius ratio
ðr2A =r2B Þ equal to 1.225. But crystals of this type show radius ratios in the range
1.05 –1.67. The atomic structures of the Laves phases, formed from packing of tetrahedral
units, illustrate that, in metallic crystals, the atoms tend to fill space as closely as possible,
maximizing the connectivity (coordination and dimensionality).
The Laves phase of CH4(H2)2 is found to be stable at 5.6 –7.2 GPa, with possible
changes in stoichiometry at higher pressures. Substitution of CH4 for Ar also gives another
Laves phase of Ar(H2)2, which is stable to very high pressures (.200 GPa).
For space group D6h (known for Laves phases) with two formula units
(e.g., Ar(H2)2) per cell, one can predict the following Raman (R) or infrared (IR) active
internal modes for hydrogen molecules (vibrons; see Loubeyre, 1991);
2H2 molecules in argon-rich plane: D3d site symmetry, A1g (R);
6H2 molecules in the hydrogen plane: C2v site symmetry, A1g(R) þ E2g(R) þ E1u (IR).
76 Chapter 1

Figure 1.15. Comparison of the Raman spectra of Ar(H2)2 (a) and CH4(H2)2 (b) at different pressure. Raman peak
at 4250 cm21 arises from a small amount of pure hydrogen present in the sample (Hemley et al., 1998b, q 1998
Societá Italiana di Fisica).

A comparison of the Raman spectra of Ar (H2)2 and CH4(H2)2 exhibits two main
vibron peaks (Fig. 1.15). The two different crystallographic positions in the methane
compound are associated with rather different crystal fields relative to Ar.

1.10.3. N2 – CH4: Titan

It has a dense atmosphere of N2 with a few percent CH4 and abundant aerosols. It is
the largest abiotic organic reservoir in the solar system. When N2 – CH4 mixture is exposed
to UV radiation thiolins, brown organic solids form. The laboratory thiolins react with the
atmosphere to produce numerous oxygen-containing groups. In thiolins highly reactive
nitrogen functional groups may be present.

1.11. H2O

1.11.1. Bonding: covalency

The water-bearing asteroids and/or icy planetesimals that were formed near Jupiter
are the most likely sources for the Earth’s water (e.g., Morbidelli et al., 2000). Earth
probably did not accrete much of its water from cometary sources, because cometary water
is very rich in deuterium. To reproduce the Earth’s low D/H ratio, an admixture of D-poor
solar nebula hydrogen is needed.
The chemical formula of water tells us that two hydrogen atoms are attracted to one
oxygen. Structural studies, on the other hand, show that each water molecule has four
neighbours. Each molecule is linked to a neighbour by a hydrogen bond. A hydrogen is
Cosmochemistry and Properties of Light Element Compounds 77

Figure 1.16. Structure of ice shown in side (a) and top (b) views. Intermolecular covalent bonds exist within the
H2O molecule. Intermolecular hydrogen bonds link H2O molecules together (from K. M. Talls, T. H. Courtney,
and J. Wulff, Introduction to Materials Science and Engineering, Willey, New York, 1976, q John Wiley
Interscience).

shared by two atoms — on one side by a covalent bond, on the other side by the hydrogen
bond (see Fig. 1.16).
In answering what holds water together, Linus Pauling proposed that the influence
of the strong “covalent” bonds within each water molecule leaks into the hydrogen bonds
and lends a hand in binding one molecule to the next. Recently, quantum theory has
confirmed Pauling’s view.
Some assumed that the bonds between the molecules are part covalent and part
electrostatic. In a covalent bond between two atoms, each donates one electron to a
shared pair that no longer belongs to either atom alone but occupies a single orbital
common to both. But in water, oxygen holds the electron pair, leaving the positively
charged nuclei of the hydrogen atom exposed. Hence, water molecules are “polar”. But
Pauling and others have predicted that the electrons of the covalent bonds should spread
into the hydrogen bonds. The nature of the hydrogen bond (cooperative) in H2O ice is
shown in Fig. 1.17.
The researchers of the European Synchrotron Radiation Facility (ESRF, Grenoble,
France) directed an intense, needle-sharp beam of X-ray photons with a precisely defined
energy onto a tiny sample of specially prepared ice. One orientation was so chosen that the
hydrogen bonds, held fixed by the crystal structure of ice, were parallel to the incoming beam
and the other part of the hydrogen bonds at an angle to the beam. If the hydrogen bonds are
partly covalent, there would be shared electrons in the bond which would behave as quantum-
mechanical waves and when one energy distribution is subtracted from the other, an
interference pattern would result. This is indeed what was observed by the ESRF scientists.
Electrons spread out in this way because of a purely quantum-mechanical effect known as
delocalization. Electrons seek the lowest possible energy state and, for the covalent bonding
pairs in water, the lowest energy state apparently extends into the hydrogen bonds.
78 Chapter 1

Figure 1.17. The cooperative hydrogen bond in ice (Sharma and Sikka, 1996).

X-ray diffraction data of H2O up to 128 GPa is consistent with a structure based on the
bcc oxygen sublattice. At 25–45 GPa, the O–H stretching vibration undergoes a cascading
series of resonances with other infrared active vibrational modes (Struzhkin et al., 1997).

1.11.2. Hydrodynamics

The first system whose long-wavelength, low-frequency dynamics was given

serious attention was water. The dynamics of water in motion is called hydrodynamics.
Today, the term “hydrodynamics” is used for the long-wavelength, low-frequency
dynamics of conserved and BSP variable in any system. Thus, for example, spin systems
and crystalline solids as well as water have a well-defined hydrodynamics.

1.11.3. H2O –ice structure: “ice rules”

H2O ice transforms from a molecular system to a dense symmetric non-molecular

oxide material (Goncharov et al., 1996). As already stated, in an H2O molecule, an oxygen
atom is covalently (and strongly) bonded to two hydrogen atoms. The molecules themselves
are linked by weaker hydrogen bonds (Fig. 1.16a,b). A combination of the covalent and
hydrogen bonding (Fig. 1.17) is responsible for a rich variety of structures in ices, 14 of
Cosmochemistry and Properties of Light Element Compounds 79

which have already been reported. A new phase of ice, having a density 4 times the density of
normal ice, is stable to at least 210 GPa. In that case, ice is no longer molecular but forms
ionic crystals akin to dense oxides supposedly present in the deep Earth.
In a landmark paper, Pauling (1935) argued that ice could adopt any of the huge
number of molecular arrangements compatible with the Bernal – Fowler “ice rules”. He
calculated the number to be ð1:5ÞN ; where N is the number of molecules — which, for a 1 g
crystal of ice, is a number with 6 £ 1021 digits! This translates to a residual entropy of
0.81 cal K21 mol21, exactly the value found experimentally (within error range). Later, by
neutron diffraction study, hydrogen in the form of deuterium in its structure was located,
and the structure was confirmed. Below ,160 K, H2O molecules of hexagonal ice settle
into an arrangement made from any of the numerous ð1:5N Þ random structures. The
structure of normal hexagonal ice, 1h ; showing the arrangement of oxygen atoms is of
the many possible arrangements of hydrogen atoms, as shown in Fig. 1.18.

1.11.3.1. Reflectance spectra

Infrared reflectivity measurements demonstrate that H2O-ice transforms at 60 GPa
to a non-molecular ionic phase, which persists to at least 210 GPa. Vibrational spectra
strongly suggest a phase having symmetric hydrogen bonds.
Reflectance spectra at .60 GPa show systematic increase of the reflectivity with
pressure in the 600 –2600 cm21 spectral range. These changes are indicative of a transition
from a structure described by “ice rules” (which are characterized by intramolecular
covalent bonds and conventional hydrogen bonds between the molecules) to a non-
molecular or ionic state. The spectroscopic feature of the phase persists to 210 GPa,
although a weak structural (or order – disorder) transition along with a Fermi resonance
cannot be ruled out at 140 –150 GPa (Hemley and Mao, 1998). Reflectance spectra at 85 K
change abruptly; the n3 band becomes much weaker and a new, very strong low-frequency
1R band appears.

1.11.4. Entropy of ice

Bernal and Fowler (1933), in their postulates for the structure of ice, were guided by
the Third Law of Thermodynamics, which states that the entropy of a perfect crystal is zero
at a temperature of 0 K (absolute zero). Entropy is proportional to the logarithm of the
number of arrangements or motions available to the system of molecules, and so zero
entropy means one arrangement (log 1 ¼ 0). However, others (Giaque and Ashley, 1936)
showed that ice still has some entropy at 0 K.
The entropy of ice, which is a measure of molecular disorder of the system, is
expected to be finite at all temperatures down to 0 K. But this finite entropy defies the Third
Law of Thermodynamics (which requires all substances to have precisely zero entropy at
T ¼ 0 K). Explicitly, zero entropy means there is only one equilibrium arrangement of the
systems of molecules. However, experimental evidence now indicates that ice is indeed out
of equilibrium at low temperature. In particular, the hydrogen-ordering dynamics are
prohibitively slow below about 120 K, suggesting that some sort of glassy non-equilibrium
state exists at such low temperatures.
80 Chapter 1

Figure 1.18. The structure of normal hexagonal ice Ih, showing the arrangement of oxygen atoms with one
possible random arrangements of hydrogen. The cubic modification of ice, Ic, has tetrahedrally coordinated oxygen
atoms; the six-membered oxygen rings are repeated in a three-layer (rather than a two-layer) stacking sequence.
Other high pressure phases of ice have different oxygen lattices and are called II…X. The existence of ferroelectric
ice has been known. In that the dipolar water molecules are thought to be largely aligned (Bramwell, 1999).

1.11.4.1. Ferroelectric alignment

Water molecules are dipolar and in an electric field can be partly aligned. There are
some experimental evidences that, in normal ice, (at least partial) ferroelectric alignment
can be induced by an interaction with a substrate (Iedema et al., 1998) or by doping with
impurities (Jackson et al., 1997).
The cubic modification of ice, Ic, has the tetrahedral coordination of oxygen atoms,
but the six-membered oxygen rings are repeated in a three-layer, rather than a two-layer,
stacking sequence. Other high-pressure phases of ice have different oxygen lattices and are
called II, III, IV…X. The stability fields of a major phase in the P – T diagram are shown in
Figs. 1.19 and 1.20.
In normal hexagonal ice (ice Ih), each oxygen atom is coordinated in a tetrahedral
arrangement with respect to the other four atoms (Fig. 1.18). As early as 1933, Bernal and
Fowler proposed that ice Ih has the simplest ordered arrangement — a form of ferroelectric
Cosmochemistry and Properties of Light Element Compounds 81

Figure 1.19. Phase diagram of H2O. The boundaries of stable phases are shown as thin lines. Heavy lines are
melting curves for ice IV and the new phase. At much higher P – T condition the melting of ice VII occurs. The
uncertainties in calculated pressures at the Tm of ice V are less than ^3% (Chou et al., 1998, q 1998 American
Association for the Advancement of Science).

ice in which the molecules are, on average, aligned. Li et al. (1994) presented the structure
of normal hexagonal ice Ih, showing the arrangement of hydrogen atoms. Under pressure,
as the oxygen atoms are forced together, the potential energy changes from a double to a
single well (Fig. 1.21).

Figure 1.20. Phase diagram of H2O as in Fig. 1.19 extended to 6 GPa, showing the fields of high density
amorphous (hda) (from Sharma and Sikka, 1996, q 1996 Elsevier Science Ltd.).
82 Chapter 1

Figure 1.21. Proton density in ice under pressure. As the oxygen atoms are forced together, the potential energy
changes from a double to a single well. A new simulation, however, shows that a form of ice with the proton
midway between the oxygens occurs even before that happens (from Telxeira, 1998, q 1998 Nature).

1.11.4.2. Spin ice

Ramirez et al. (1996) studied the “spin ice”, which can be explained with reference
to a square, triangle and tetrahedron, which are the building blocks of common crystal
structure, as shown in Fig. 1.22. The circles in the figure represent magnetic atoms and the
arrows magnetic spins. In a– c, the spins are coupled by antiferro-magnetic interactions,
favouring antiparallel alignment of neighbouring spins. This is easy in a, but impossible in
b and c. In d, the coupling is ferromagnetic but there is the additional presence of
anisotropy constraining the spins to point either directly into (or away from) the centre of
tetrahedron. The resulting configuration with two spins in and two spins out is a
characteristic of the “spin ice” studied by Ramirez et al. (1996).

Vibrational relaxation time. Woutersen et al. (1998) have published femtosecond IR

measurements of the vibrational relaxation time (t1) of the OH stretch in water and ice as a
function of temperature. They find that t1 (in ps) is nearly constant for the ice phase
(30 – 270 K), but abruptly increases at the solid – liquid phase transition at 273 K and then
increases monotonically up to 363 K. Only a few inorganic condensed-phase systems are
known to display such an inverted dependence of vibrational relaxation time on
temperature (Myers et al., 1997).
Cosmochemistry and Properties of Light Element Compounds 83

Figure 1.22. The building blocks of three common crystal structures (square, triangle and tetrahedron). The
circles are magnetic atoms and the arrows indicate magnetic spins. In (a) –(c) the spins are coupled by
antiferromagnetic interactions, favouring antiparallel alignment of neighbouring spins. This is easy in (a), but
impossible in (b) and (c). In (d) the coupling is ferromagnetic, but there is the additional presence of anisotropy
constraining the spins to point either directly into — or away from — the centre of the tetrahedron. The resulting
configuration with “two spins in two spins out”, is a characteristic of the “spin ice” studied by Ramirez et al.
(1999) (from Harris, 1999, q 1999 Nature).

Solid-state convection in ice: Europa’s case. Jupiter’s ice-rich moon Europa may be
geologically active, evidenced from its sparsely cratered surface. Possibly tidal inter-
actions with Jupiter produce enough heat to maintain a subsurface liquid-water layer.
Galileo spacecraft revealed that Europa has distinctive surface features such as pits, domes
and dark spots (circular to elliptical) ,7 –15 km in diameters, spaced 5 –20 km apart.
These features may be surface manifestations of diapirs, relatively warm localized ice
masses that have risen buoyantly through the subsurface. Such features can be explained
by thermally induced solid-state convection within an ice shell, likely overlying a liquid
water layer (Pappalardo et al., 1998).

H2O (D2O) ice: morphology. The O – H bonds are not easily severed by pressure. Indeed,
H2O ice is stable at over 1 Mbar (at least 128 GPa; Hemley, 1987). Spectroscopic
observation reveals that this solid is veritably not molecular — the hydrogen bonds
84 Chapter 1

approach a symmetrized state (Pruzzan, 1994) and this has to be considered an oxide. The
work done in compression (pressure – volume) is comparable to the chemical bond
strengths, and this is sufficient for rearranging the chemical bonds.
Morphological changes of ice crystals under pressure were investigated by Shichiri
et al. (1994) by mixing of isotope D2O. It was found that when pressure .90 MPa was
applied to H2O melt, the {1010} facets appeared on the periphery of rounded platy crystals
with flat {0001} faces. The six corners became sharp at 200 MPa, but these became
rounded below this pressure range.
Shichiri et al. (1994) also studied the morphological changes of ice crystals in the
[0001] direction by applying pressure at triple points and by mixing the isotopes. At
P , 60 MPa; ice crystals take a circular disc form; at P . 90 MPa; facets appear along the
periphery of the disc and, at 200 MPa, crystals take a hexagonal platy form. Near the triple
point, multi-faceted crystals appear.

1.11.5. Ice Ih, III, IV, V and VI: phase diagram

The manifold ways in which the water molecules may link through hydrogen
bonding give rise to a remarkably rich phase diagram (Aoki et al., 1996; Goncharov et al.,
1996; Mishima, 1996; Harrington et al., 1997) (see Fig. 1.20).
Hobbs (1974), in his book Ice Physics, documented five liquidus phases up to
2 GPa. Of these, four are stable phases (ices Ih, III, V and VI) and one is metastable (ice
IV). Different polymorphs of ice can be identified by their distinct crystal morphologies,
growth patterns and melting curves. The P – T melting relations of these phases are shown
in Fig. 1.19. Recently, Lobban et al. (1998) reported a metastable phase that formed below
0.6 GPa. At still higher pressures, Mishima and Stanley (1998c) suggested the existence of
two phases. These recent results emphasize the contention that much remains to be learned
about H2O, especially the complexity of its nature under pressure. This complexity is
enhanced by the existence of proton-ordered and proton-disordered forms as well as
metastable crystalline and amorphous phases (Mishima, 1996).

1.11.5.1. Ice VI in diamond

An investigation of a H2O phase included in cuboid diamonds (Siberia) by IR
spectroscopy revealed bands at 5,180 cm21 (due to liquid H2O) with a shoulder at
5,000 cm21. This band represents stronger hydrogen bonding arising from the presence of
ice VI formed at 1.9 GPa (Kagi et al., 2000). Heating and pressure experiments reveal the
T – P phase relationships between water, ice VI and ice VII.

1.11.5.2. Ice Ih: stability boundary

Quasi-harmonic lattice-dynamics methods have been used to calculate the
thermodynamic and mechanical stability boundaries of ice Ih under pressure (Tse et al.,
1999). A thermodynamic solid –liquid stability boundary is determined by the equality of
free energies of the two phases, whereas for deciphering the mechanical stability, the
elastic constants at several pressures and temperatures are determined. The temperature
where mechanical instability occurs is always higher than the thermodynamic melting
Cosmochemistry and Properties of Light Element Compounds 85

point. The calculated thermodynamic melting line is seen to meet the mechanical
instability curve at ,160 K where two temperature regimes separate out. Above this
temperature, the water molecules have sufficiently large-amplitude thermal motions and,
consequently, the thermodynamic melting process takes precedence over mechanical
instability. In contrast, at temperatures lower than 160 K, the vibrational amplitudes of
the water molecules are reduced under pressure and the crystal structure collapses due to a
mechanical instability (Tse et al., 1999).
A mechanical instability due to the softening of the elastic modulus C66 ( ¼ C11 –
C12) of the ice structure under high pressure has been proposed in a molecular dynamic
study (Tse, 1992) (see also Fig. 1.25).

1.11.5.3. Proton ordering/disordering: new phase

Water molecules may link through H-bonding to form a rich phase diagram
manifesting six or more solid phases (see Fig. 1.23). In addition, there are proton-
ordered and -disordered forms and metastable crystalline and amorphous phases (e.g.,
Mishima, 1996). Employing in situ microscopy and Raman spectroscopy at 0.7 –1.2 GPa,

Figure 1.23. Raman spectra of the new ice compared with ices, I, II, III, V, VI and water measured in situ in the
diamond cell. Ice I is proton disordered; ice II is proton ordered; ice III is partially proton disordered but has a
proton-ordered form (ice IX); ice V is partially proton disordered. Ice VI is suggested to be partially ordered as
well. Characteristic Raman peaks / bands occur at 192, 490, 3215, and 3410 cm21 for the new phase; 145 and
157 cm21 for ice VI, and 95 and 190 cm21 for ice III. Intensity is given in arbitrary units (from Chou et al., 1998,
q 1998 AAAS).
86 Chapter 1

Chou et al. (1998) documented the existence of an another H2O phase of unusual
crystalline morphology and showing evidence for significant proton disorder.
It shows peculiar growth patterns, distinct P – T melting relations and vibrational
Raman spectra. Its Raman spectra and crystal morphology are consistent with a disordered
anisotropic structure with some similarities to ice VI. Kamb (1973) proposed that pressure-
quenched ice VI is partially ordered with space group Pmmm. In its stability field, the
phase is proton-disordered (space group P42 =nmC) and a fully ordered form (space group
Pn) was also predicted.
For studying ices in the complex region of the phase diagram below 2 GPa, they
used a hydrothermal-type diamond-anvil cell (Bassett et al., 1993). Typically, the water
sample froze at about 2408C and ,600 –800 MPa. By slow cooling, a well-formed crystal
could be grown. The sample was cooled by a stream of cold nitrogen and heated by two
individual heaters. Under isochoric conditions, with bulk sample densities between 1,203
and 1,257 kg m23, up to three different forms of ices were observed to melt: ice V, ice VI
and a new solid phase (^ice IV). Employing the EOS of water (Saul and Wagner, 1989)
and from the density obtained for the isochoric melting of ice V or ice VI, the pressure at
the melting temperature of the new phase was calculated.
The P – T melting relation for the new phase has been represented by the equation
(Chou et al., 1998):

PðMPaÞ ¼ 665:1ð^9:1Þ þ 12:73ð^1:15ÞT þ 0:184ð^0:039ÞT 2 ð0 , T , 308CÞ

The Raman spectra of the new ice phase are presented in Fig. 1.23 (Chou et al.,
1998) and compared with ices I, II, III, V, VI and water measured in situ in the diamond
cell. The new phase exhibits bands in the lower frequency region at 192 and 490 cm21,
which are assigned to transitional and rotational (librational) excitations. In the higher
frequency (O –H stretching) region, a sharper feature at 3,215 cm21 and a shoulder at
3,410 cm21 were observed. The structure in this spectral region is distinct from that of the
other ice phases. The lack of sharp structure in the low-frequency transitional and
rotational excitations indicates that protons (or molecular orientations) are disordered in
this polymorph, as also seen in ices 1h, V and VII. The spectra also show that the new
phase is distinct from both low- and high-density amorphous ice (Hemley et al., 1989).

1.11.5.4. Higher isomorphs: ices VII, VIII and X

At room temperature, the compression of liquid water leads to a tetragonal ice VI
phase at about 1.05 GPa, followed by cubic ice VII around 2.3 GPa. In the absence of P – T
cycling, ice VII usually occurs as a polycrystalline aggregate in DAC. The idealized
structure of ice VII is shown in Fig. 1.24.
For water at 30 and 60 GPa pressures, the configuration is taken as ice VII (Hobbs,
1974) and at 150 and 300 GPa it is ice X (Aoki et al., 1996; Goncharov et al., 1996). In
these polymorphs of ice VII (Fig. 1.24) and ice X, the oxygen atoms form a bcc sublattice.
But these differ in the positions of protons, which are off-centre in ice VII and are at the
centre of the O – O separation in ice X, a new symmetric form of ice.
Cosmochemistry and Properties of Light Element Compounds 87

Figure 1.24. Idealized structure of ice VII. At low pressure the H is covalently bonded (solid line) to an O of the
same molecule and weakly bonded (dashed line) to an O of a neighbouring molecule. The symmetric hydrogen
bonded state is to occur with decreasing O –O distance on compression when the point of maximum probability of
finding the H moves to the middle (solid circle) (Holzapfel, 1972, q 1972 Chemical Society of America).

In ices VII and VIII, the position of hydrogen is asymmetric, around 0.1 nm from
one oxygen and 0.18 nm from the other. Along this O – O line, the potential experienced by
the proton density has two minima and the proton is localized in one of them. These two
minima would collapse into a single one, half way between the two atoms, yielding the
symmetric form of ice X. Figure 1.21 (Teixelra, 1998) depicts the proton density in ice
under pressure. As the oxygen atoms are forced together, the potential energy changes
from a double to a single well but, even before that happens, a new form of ice occurs with
a proton midway between the oxygens.
Ice VII, an unquenchable phase, may be stable in the coolest regions of the
subducting slab.

1.11.5.5. Ice VII: as pressure medium

H2O can be used as a pressure medium. Above 2.3 GPa, H2O transforms to the ice
VII polymorph, which is stable beyond 28 GPa. Reynard et al. (1996) followed the (110)
line of ice VII and used the EOS determined by Hemley et al. (1987) to recalculate the
pressures. A significant broadening of the (110) line-width of ice VII was observed with
increasing pressure: from 120 eV at 2.4 GPa to 200 eV at the maximum pressure. If the
broadening is due to pressure difference across the X-ray spot, the pressure difference can
be estimated from

DP ¼ DE110 =ðdE110 =dPÞ;

where DE110 is the line broadening with respect to the ambient pressure-extrapolated value
(,100 eV) and ðdE110 =dPÞ is the slope of the energy variation with pressure of (110) line of
ice VII.
During decompression, the (110) ice line-width increases further to a constant value
of 240 eV. This effect may be partly due to an irreversible (e.g., plastic) deformation of the
ice lattice or to a reduction of the grain size.
88 Chapter 1

Figure 1.25. Changes in elastic constants of ice VII under pressure (Zha et al., 1998b, q American Geophysical
Union).

Zha et al. (1998b) measured the Brillouin spectra of an aggregate of polycrystalline

sample of ice VII to 40 GPa. Three elastic constants C11, C12 and C44 were resolved for
ice VII to 40 GPa and are shown in Fig. 1.25. The splitting between C12 and C44 increases
with pressure above 10 GPa.
With pressure, a strong increase of elastic anisotropy noted in ice VII (Fig. 1.25)
may be associated with the high strength of ice VII (Hemley et al., 1988) and also with a
dramatic change in interatomic interactions (e.g., hydrogen bonding) over this pressure
range. The EOS of ice VII is reflected in Fig. 1.26.

1.11.6. Supercooled water

When liquid water is cooled fast enough (1068C s21) to avoid crystallization, it
forms glass. Such vitreous water is found as frost on interstellar dust, in dense molecular
clouds and comets are made of it (Jenniskens and Blake, 1994).
Certain anomalies of supercooled water, especially the large increase in
specific heat at low temperature, etc., were explained by Poole et al. (1994) employing
the concept for two forms of glassy water: low-density (LDA) and high-density (HDA)
amorphous ice IV (Mishima et al., 1985). The amorphous ice and the liquid would be

Figure 1.26. Change in EOS of ice VII with pressure (Zha et al., 1998b).
Cosmochemistry and Properties of Light Element Compounds 89

separated by a glass transition. These two forms have very different densities. At 77 K, the
condensation of LDA into HDA occurs with an abrupt decrease in volume of ,22%.
The transient low-density and low-energy structured molecular arrangements may be
stabilized by strong, orientation-dependent interactions such as hydrogen bonding
(e.g., Shiratani et al., 1998).
This concept of liquid –liquid co-existence in a pure substance like H2O is
intellectually challenging. This is because of the ubiquity of supercooled water in the
stratospheric clouds and the glassy water in the interstellar dust, and also for the
technological potential for storing labile biochemicals in supercooled water emulsions or
in water – carbohydrate glasses (Debenedetti, 1998).

1.11.6.1. Amorphous ice polymorphism: high-density and low-density

Pressure-induced amorphization in H2O crystalline ice Ih was first documented by
Mishima et al. (1984). Examples of transformation in other materials are described and
reviewed by Hemley et al. (1988), Williams and Jeanloz (1989), Kruger and Jeanloz
(1990) and Serghiou et al. (1992).
When normal ice (ice I) is compressed under liquid N2 temperature (i.e., 77 K), it
transforms to an amorphous form with high density that is significantly higher than that of
normal ice (e.g., Mishima and Stanley, 1998). When warmed under ambient pressure, this
HDA form transforms to a LDA form. Thus phase transition occurs between two
amorphous states. This kind of polymorphism has been observed in a number of systems
involving minerals, molecular solids and metals under both positive and negative
pressures.
Solid glassy water (also called amorphous ice) can exist when the temperature
drops below the glass transition temperature Tg (,130 K at 1 bar). The solid glassy water
exhibits a disordered structure, much like a liquid. Initial results on the amorphization of
ice under compression at 77 K showed that it turned into an amorphous structure at
10 GPa. It did not transform back into a crystalline state when the pressure was released.
The density remained approximately 26% higher than that of glassy ice obtained by rapid
cooling of water, and hence it was termed HDA. This is called amorphous ice
polymorphism. However, a crystalline form of ice was recovered by heating the pressure-
cycled sample to 152 K via a LDA state above 117 K. This suggests that these two forms of
“amorphous ice” could be different.
LDA, when heated to the transition temperature, Tg, at ,130 K, transforms to a
highly viscous liquid. HDA was first obtained by compressing ice Ih below ,150 K. When
compressed further at still lower temperature, HDA crystallizes to form high-density
crystalline ice.
The polymorphic transition is accompanied by a surprisingly sharp and large
volume change (.20%) when thermodynamic parameters (P or T) change infinitesimally.
The shear modulus of LDA decreases on compression and jumps discontinuously at the
LDA ! HDA transition, and then increases with further compression. Thus, this transition
(LDA ! HDA) seems to be a first-order transition rather than a relaxation phenomenon, in
which case the structure should change continuously.
90 Chapter 1

HDA has a structure similar to that of high-pressure liquid water, suggesting that
HDA is a glassy form of high-pressure water. Pressure-induced amorphization is
conventionally either a “two-phase melting” (crystalline solid becoming liquid) at a
temperature below Tg, or a “one-phase melting” (a solid becoming a collapsed solid)
(Fig. 1.27). HDA could as well be surmised as a collapsed “ill-crystalline” phase. Below
Tg, the melting is “delayed” to higher pressures, causing a shift from two-phase melting to
one-phase melting.
A HDA phase of ice Ih (Mishima et al., 1984) forms at a pressure close to the
extrapolated melting curve of ice, suggesting that it may have a structure similar to that of
dense water. On annealing, HDA ice transforms into a LDA phase with a distinct phase
boundary (Mishima et al., 1985; Mishima, 1994). Extrapolation of thermodynamic data
along the HDA – LDA co-existence line into the liquid region has led to the hypothesis that
there might exist a second critical point for water and the speculation that liquid water is a
mixture of two distinct structures with different densities (Mishima and Stanley, 1998b).
The quasi-harmonic lattice dynamics calculations by Tse et al. (1999) show that
the amorphization mechanism in ice Ih changes from thermodynamic melting for
T . 162 K to mechanical melting at lower temperature.

Very high-density amorphous (VHDA) ice. Ice is famously anomalous, and is found to
have 13 different forms and several amorphous varieties. Recently, Finney et al. (2002)
described a new very dense form of ice. This VHDA ice, 40% denser than ordinary ice, was
originally discovered by Loerting et al. (2001).

Figure 1.27. Two-phase melting contrasted with one phase melting diagram, showing the shift of the melting line
Tm ðPÞ of ice Ih from melting above Tg to one-phase melting below Tg. The experiment marked “exp”, follow the
dashed line.
Cosmochemistry and Properties of Light Element Compounds 91

HDA ice is 30% denser. Each water molecule is surrounded by five nearest-
neighbour water molecules, in contrast to ordinary ice, which has a hexagonal structure
wherein each H2O molecule has exactly four nearest neighbours. If HDA ice is warmed to
,120 K at 1 atm, it transforms into LDA ice, releasing quite a lot of heat. HDA and LDA
may be related to supercooled forms of liquid water. Between LDA and HAD, a series of
amorphous ice forms have since been characterized (Tulk et al., 2002). When HDA is
heated up to ,160 K at 1.15 GPa, VHDA ice is seen to form (Loerting et al., 2001).
The melting point (m.p.) of ordinary hexagonal ice changes with pressure. The
melting curve of ice has a negative slope and ice in fact melts at 2208C under 0.2 GPa
(Klug, 2002).

1.11.6.2. Diffraction study

The structure of liquid water can be described at any given temperature and
pressure by a mixture of the HDA and LDA ice structures (Bellissent-Funel, 1998) but
there is a distinct phase boundary between HDA and LDA ice (Mishima, 1994).
X-ray and neutron diffraction have been performed on both HDA and LDA ice. At
0.77 GPa, the density of liquid water is close to that of HDA ice studied at 77 K. Difference
in the positions and relative amplitudes of the peaks in the range 3 –7 Å in the O – O pair
correlation function between HDA ice and dense liquid water (Okhultov et al., 1992) are
reported. In water at 0.77 GPa, a peak appears at 3.2 Å. A peak at 4.25 Å in liquid water is
in contrast to the peak appearing at 4.65 Å in HDA ice.

1.11.6.3. LDA ice, ice Ih and quenched water: vibrational spectra

The vibrational spectra of ice Ih LDA ice and quenched water also indicate a
structure for LDA ice that differs from that of liquid.
To investigate the short-range order of amorphous ice, the vibrational spectra were
obtained from inelastic, incoherent neutron-scattering (IINS) experiments on LDA ice,
hyperquenched (hq) water, and ice Ih (Fig. 1.28) (Tse et al., 1999). The quenched water
was prepared by rapid cooling (106 degree s21) of 3 mm diameter water droplets on a cold
Cu substrate, which gives a glassy form of water. Rapid cooling avoids the crystallization
that would occur at slower cooling rates. This method provides a very sensitive
measurement of the nearest-neighbour interactions (hydrogen-bond strengths). The band
profile of quenched water is distinctively different from that of LDA ice and ice Ih: the low-
energy edge shifts to a lower frequency. The lower onset frequency indicates that the
average hydrogen-bond strength is less in quenched water than in LDA ice and ice Ih and,
therefore, the local structure also differs.
In contrast to IINS, Raman spectroscopy measures with very high sensitivity the
frequency of a particular localized oscillator, which is determined primarily by the near-
neighbour O –H· · ·O interaction. The frequencies of the uncoupled O –H (or O –D) modes
in LDA ice are quite different from quenched water, indicating that their local hydrogen-
bond interaction is not the same. In addition, the vibrational spectra obtained from Raman
spectroscopy and IINS show that there are distinct differences between LDA ice and
quenched water.
92 Chapter 1

Figure 1.28. The inelastic incoherent neutron scattering (IINS) function for LDA ice, hyperquenched (hq) water,
and ice Ih. Data shown in librational (a) and internal vibration (b) regions illustrate the different spectral responses
of these materials. The arrows indicate the inset energies in (a), and the mean frequencies in (b). The error limits in
the IINS measurements are ^1.1 meV in the vibrational frequency and ^0.2 in the intensity units (Tse et al., 1999,
q 1999 Nature).

The results of Tse et al. (1999) also indicate that the pressure-induced phase
transformation of ice Ih at low temperatures is not a melting transition in the
thermodynamic sense and HAD, therefore, is not expected to have a liquid-like structure.

1.11.6.4. VHDA ice

Ice is famously anomalous, as it is found to have 13 different forms, and several
amorphous varieties. Recently, Finney et al. (2000) describe new very dense form of ice.
This very high-density amorphous (VHDA) ice, 40% denser than ordinary ice, was
originally discovered by Loerting et al. (2001).
High-density amorphous (HDA) ice is 30% denser. Each water molecule is
surrounded by five nearest neighbour water molecules; in contrast to ordinary ice, which
has hexagonal structure, wherein each H2O molecule has exactly four nearest neighbours.
If HDA ice is warmed to ,120 K at 1 atm, it transforms into low-density amorphous
(LDA) ice, releasing a fair amount of heat. The HDA and LDA may be related to
supercooled forms of liquid water.
Cosmochemistry and Properties of Light Element Compounds 93

Between LDA and HAD a series of amorphous ice forms have since been
characterized (Tulk et al., 2002). When HAD is heated up to ,160 K, at 1.15 GPa, VHDA
ice is seen to form (Loerting et al., 2001).

1.12. Deuterium at high pressure: Saturn’s core

By shocking liquid D2 to pressure at and above the metallic transition, a different

regime has been accessed by Collins et al. (1998) using a high-power laser. They used the
Nova laser to shock liquid D2 to pressures between 22 and 340 GPa, spanning the metal –
insulator phase boundary. The shock-wave (SW) pressure P and density r were determined
from measurements of the shock-wave speed VS and the particle velocity VP behind the
shock, using the Hugoniot relations: P ¼ r0 VS VP ; and r ¼ r0 VS =ðVS 2 VP Þ: On the
Hugoniot above the metal – insulator transition regime, the compression must eventually
approach a value of 4 (i.e., r=ro ¼ 4).
In such experiments, the probe laser ðl ¼ 1:064 mmÞ is used for velocity
interferometry, a technique that measures the Doppler shift of light reflected from a moving
surface. The recorded fringe shift is directly proportional to the Doppler shift and, therefore,
to the velocity of the reflecting surface. The measured velocity is VS ¼ lP FðtÞ=2t; where
tð¼ 76:3 psÞ is a delay-time set by the configuration and FðtÞ is the fringe count.
In the experiments by Collins et al. (1998), in addition to VS, the interferometer also
supplied instantaneous measurements of the (single-wavelength) reflectivity of the shock
front. Above 55 GPa, the measured reflectivities were ,60% (which is equivalent to a
metal). The temperature of the shocked D2 was ,0.75 eV, which is much less than 1F ð¼
,15 eVÞ and the ionization potential. Hence, the high reflectivity observed was due to free
electrons produced by density and thermal ionization effects (; pressure ionization in
plasma physics).
At 100 GPa, the SESAME (Kerley, 1980) D2 Hugoniot density is 0.68 g cm23
(when r=ro ¼ 4). At the lowest compression, the laser data agree with the gas gun results
(Holmes et al., 1995). The compression and the laser data above 200 GPa show a trend
toward the ideal gas compression. The flattened Hugoniots for two EOS models such as
those of Saumon and Chabrier (1992) and Militzer et al. (1998) are the result of the
Hugoniot passing through a predicted first-order phase transition from the molecular to the
metallic state. However, others suggest that the metal – insulator transition is a continuous
transition along the Hugoniot.
If hydrogen, which makes up over 90% of Saturn, goes soft at high pressures or
undergoes a phase transition, it could store up energy in the same way that water vapour
stores more energy than liquid water. This stored energy might keep Saturn relatively
warm in its old age. The results may explain why Saturn appears so much younger in age
than the rest of the planets in the Solar System. Under shock from the Nova laser at
Livermore, a drop of deuterium behaves as it should at the core of Saturn. The schematic
arrangement of the experiments is shown in Fig. 1.29. By shining a beam of X-ray through
two windows, the Livermore physicists could track the shock speed and infer the pressure
and the resulting density of the deuterium over 5 or 10 billionth second before the whole
assembly flew apart.
94 Chapter 1

Figure 1.29. The experimental set-up at Livermore. A shock from the Nova laser makes a drop of deuterium
behave as it would at the core of Saturn (Kestenbaum, 1998).

Some models predict that at some high P, T, deuterium should undergo an abrupt
phase transition. Such pressure would free one electron, which would wander from its
atom, inducing others to follow and thereby dissolving the structure. This soup would be
hard to compress since the shock would cause heating and expansion. Computer
simulations, however, predict that hydrogen or deuterium atoms can sometimes link up
into chains bound by borrowed electrons. Such structures may absorb some of the shock
energy without heating the fluid.
Deuterium under pressure conducts like a metal. This may explain the strong
magnetic field of massive planets such as Saturn and Jupiter.

1.12.1. Deuterium in Mars

Allan Hills ALH-84001, a 4.5-b.y.-old meteorite, contains the possible relics of

life on Mars. For life, there must have been water. A way to trace the history of water
on Mars is through the isotope composition of hydrogen. The work of Boctor (1999) at
GL shows that the hydrogen isotope composition of the present-day Martian atmosphere
is enriched in the heavy isotopic composition of hydrogen. The isotope composition of
the present-day Martian atmosphere is enriched in the heavy isotope deuterium by a
factor of 5.2 ðdD ¼ 4; 200‰Þ relative to mean terrestrial ocean water. (Note: ‰ ¼ parts
per thousand.)
The deuterium enrichment in the Martian atmosphere is attributed to isotopic mass
fractionation associated with loss of hydrogen to space through time.
In ALH 84001, three water-bearing phases with the following dD contents have
been identified by the GLW group:
Phosphate (‰) Carbonate (‰) Felspathic glass (‰)
dD ¼ 166 –733 165–1,165 1,073–1,748
Cosmochemistry and Properties of Light Element Compounds 95

The isotopic compositions show that the water is extraterrestrial. The high dD
values in carbonates and glass are due to interaction with hydrothermal fluids that
equilibrated with the Martian atmosphere. The glass experienced impact melting at a
pressure of ,4 £ 105 atm.
In another Martian meteorite, EETA 79001, the dD in impact-melted glasses shows
values as:
Felspathic glass (‰) Mafic glass (‰)
dD ¼ 1,612–2,757 2,023–2,901

From these, the shock pressure is measured to be 6– 8 £ 105 atm.

A positive correlation is seen to exist between dD values and the shock pressure.
Hydrogen loss by impact may have contributed significantly to the high deuterium
enrichments in the impact glasses in both the meteorites.
A discussion on the high-pressure behaviour of hydrogen in space invokes a study
of alkali metals, discussed in Sections 1.8.5.4 and 1.13. The other common volatile
compound, CO2, and its elemental source carbon follow in sequence.

1.12.2. D/H ratios in minerals

Hydrogen isotope fractionation has been investigated in various minerals, such as

muscovite, biotite, amphibole, serpentine, tremolite, epidote, zoisite, clay minerals, etc.
Three types of water are present in the magmas: (i) dissolved water in a silicate
melt, (ii) bounded in hydrous minerals and (iii) aqueous fluid. The distribution of hydrogen
isotopes between hydrous silicates and water depends on the chemical composition of the
octahedral cation site in the hydrous silicates as well as on the temperature of equilibration.
The D/H fractionation factor between dissolved water and aqueous fluid may be defined as
(Haria, 1984):

a ¼ ð1 þ 1023 dDdissolved water Þ=ð1 þ 1023 dDaqueous fluid Þ

The dissolved water in quartz melt is depleted in deuterium while the latter is
enriched in the cases of albite and K-feldspar melts. There exists a difference of several
tens per million in deuterium content between albite and anorthite glasses. This is due to
the difference in structural configuration around the hydrogen atom in the hydrous melts of
albite and anorthite. The site preference of D and H is closely related to the energy level of
the site where D or H resides. In granitic constituents, the D/H fraction (a) values are as in
the following:

a values : quartz ¼ ,1; K-feldspar ¼ .1; albite ¼ .1; anorthite ¼ ,1:

1.12.2.1. D/H ratio in extraterrestrial and subsurface water

D/H ratios (per million ›D values) can hardly help distinguishing between
primordial and recycled water in the deep Earth. There are fundamental uncertainties about
96 Chapter 1

the D/H ratio developed through the escape of terrestrial H and the extraterrestrial inputs to
the hydrosphere.
However, the D/H ratio in mantle-derived magmas, xenoliths and their fluid
inclusion is surprisingly uniform. The D/H ratio lies between 240 and 295 (e.g., Deloule
et al., 1991). Fractionation of deuterium from hydrogen at magmatic temperatures is
minimal and the ›D values of seawater are near 0, while the bulk hydrosphere (including
groundwater and ice) shows ›D values near 210 (Taylor and Sheppard, 1986).
Extraterrestrial inputs have wide D/H (›D) ranges:
Meteorites 2 500 to þ9,000‰ Yang and Epstein (1983)
Comet (Halley) ,990‰ Eberhardt et al. (1995)
Carbonaceous chondrite 2100 ^ 60‰ Lecuyer et al. (1998)

The apparent contract in D/H content of the hydrosphere and the Earth’s interior
can be explained by the fact that the degassed material represents an unquantified
combination of primordial (chondritic) water and recycled water through subduction. The
D/H ratio in hydrosphere is modified by UV-induced photochemistry, cometary inputs,
impact-induced degassing and subduction-included degassing. It is possible that the
Earth’s hydrosphere has evolved towards heavier ›D values (Lecuyer et al., 1998).
(a) During the solidification of the early Earth, the magma ocean of that time could retain
significant amounts of water at depths.
(b) In the Earth’s silicate mantle, primordial water may have existed.
(c) The Earth’s degassing may have produced the present hydrosphere.
(d) The core may contain some hydrogen as iron hydride.

1.13. Alkali metals: Li to Cs

Some 76 of 92 naturally occurring elements are metals in ambient conditions and

more elements manifest metallic properties under pressure. Alkali metals (Group I
elements) under pressure may show characteristics of transition metals. This is effected by
the so-called s ! d transition whereby, through hybridization of the valence electrons, the
d character is brought down into the valence structure by pressure. The alkalis and alkaline
earth metals transform formally to d0 and then to d1 and d2 metals.
Under ordinary conditions, all alkali metals from Li to Cs occur as bcc or close-
packed lattices, typical of nearly-free electron metals. The only exception is hydrogen, in
which protons pair to form H2 molecules and a condensed molecular solid results that is
insulating. Because of its place as the first element in the Periodic Table, it is a Holy
Grail of physics to compress hydrogen by pressure until it becomes metallic like the other
alkali metals.
Neaton and Ashcroft (1999), based on state-of-the-art electronic structure
calculations, show that, under pressure, Li acts like H2, i.e., the nuclei form pairs
producing condensed phases similar to H2. The bcc structure of Li leads to it transforming
to an orthorhombic structure near 50 GPa and, at ,100 GPa, to a molecular semi-metallic
phase. Finally, at very high pressure, it should transform back to a monatomic metal again.
Cosmochemistry and Properties of Light Element Compounds 97

At intermediate pressure, the molecular-like structures are favoured by Li, as are thought to
occur in molecular hydrogen (H2) near its insulator –metal transition.
For sodium (Na), preliminary calculations indicate that similar transformations
should occur at much higher pressure with a much smaller degree of molecular pairing
(Neaton and Ashcroft, 1999).

1.13.1. “Nearly-free electron” behaviour

The nearly-free electron behaviour of the valence electrons of the alkalis is revealed
most clearly by their Fermi surfaces, which describe the boundary between occupied and
empty electronic states. These resemble the sphere experiencing no influence from the ion
cores. The effective interaction of the valence electrons with the ion cores can be described
in terms of weak electron – ion interactions or “pseudopotentials” but, more often, the
alkali metals cannot be handled so simply.
Interactions of the electrons with the lattice vibrations appear to be strong enough to
lead to conventional superconductivity, especially in Li, but this is not seen to happen. This
is an enigma which has attracted many kinds of attempted explanations. Large
modifications of the Fermi surface have been invoked for suppressing the superconduc-
tivity. Under extreme pressure, the electronic structure of Li was calculated by Boettger
and Trickey (1985). They found exceptionally large deviations of the Fermi surface from a
sphere, which is interpreted as an “s – p” transition, in which the lowest-energy electronic
states change from primarily s-like to primarily p-like and produce an effective change in
the chemistry of Li.
Caesium (Cs) also shows a promotion of electrons to unusual valence
configurations. The observed anomalous volume discontinuity under pressure is
interpreted as an “s ! d” transition, by which the electronic d-states become lower in
energy relative to the s-states under pressure (McMahan, 1978). Changes in electronic
structure may also lead to a change in the crystal structure in a remarkable way, such as
phase transitions from bcc to close-packed structures.
While studying Li under pressure, Neaton and Ashcroft (1999) employed the Pauli
exclusion principle, which states that any two electrons cannot occupy the same state.
Under pressure, as the volume available to the valence electrons decreases, this exclusion
effect increases and leads to a strong repulsive pseudopotential causing distortion of the
lattice. In high-symmetry structures where there are multiple electronic states with the
same energy, the energy states will split together with distortion under pressure. This is
the solid-state analogue of the Jahn –Teller effect in molecules, which can always lower
their energy by distortion.
Neaton and Ashcroft (1999) find that, in a molecular-like state, the valence
electrons are excluded from the region between the pairs of nuclei and they instead occupy
states with maximum density in the interstitial regions.
Neaton and Ashcroft’s (1999) calculations predict large changes in the optical
behaviour of Li from a silvery metallic appearance to a transparent or black colour, as
expected for an insulator or semi-metal. If the structures have low symmetry like those in
molecular hydrogen (H2), then there may also be strong infrared absorption, as seen in
98 Chapter 1

hydrogen (Mao and Hemley, 1994). If Li does form molecular analogues of hydrogen
under pressure, then there is the possibility of superconductivity at a relatively high
temperature and other interesting phenomena long sought for hydrogen will manifest. But
because the effects in Li occur at a much lower pressure than in hydrogen, it can perhaps
point the way to creating the Holy Grail of metallic hydrogen under pressure. Finally, such
behaviour will clearly move Li and other alkalis away from being thought of as “simple
metals” (by condensed matter physicists) to being called “interesting metals”.
Next to H2 (and helium) and H2O, the gaseous phase that is geochemically most
significant is CO2, which is discussed in the following sections.

1.13.2. Fermi pressure in lithium isotopes

Lithium gas, composed of 6Li and 7Li, cooled to about 250 nK shows a resultant of
contrasting behaviour: 6Li is a fermion, while 7Li is a boson. The clouds of these two
isotopes mixed together can be imaged differently, as 6Li atoms are affected by Fermi
pressure, while 7Li atoms, being gregarious, are not affected by pressure. The shape of 6Li
cloud arises from similar mechanism that stabilizes white dwarfs against the collapse.
The astrophysicists thus may derive strong enough reasons to look closer to the cold
atoms in the laboratory to reach out to the deeper depths of the cosmos.

1.14. CO2

CO2 is one of the most important gaseous compounds in the Earth and other
planetary bodies. It is found as clathrates on Mars and other planets (Musselwhite and
Lunine, 1995). Also, CO2 is one of the most abundant constituents of the Earth and its
atmosphere. It participates significantly in the evolution of the biosphere, hydrosphere and
also the lithosphere.
CO2 is recycled back into the Earth’s interior at subduction zones, where
metamorphic decarbonation occurs and contributes to arc magmatism, global carbon cycle
and paleo-atmospheric CO2 concentrations. Carbonate is abundant in two main pelagic
marine sediment lithologies: (i) siliceous limestones and (ii) clay-carbonates (marls).
Along high-temperature geotherms, clay-rich marls completely devolatilize before
reaching the depths at which arc magmatism is generated but, for 80– 180 km depth,
little devolatilization occurs for all carbonate-bearing marine sediments.
Significant changes in the mineralogy and mineral proportions occur up to ,8008C
ðP < 3 GPaÞ; whereas there is comparatively little variation above that temperature
(Kerrick and Connolly, 2001). The rise in the mole fraction of CO2, XCO2, up to ,7508C
correlates with consumption of carbonates, whereas the diminution in XCO2 above ,7508C
occurs because of argonite production. Most of the initial CO2 and H2O is subducted
marine sediments which will not be released beneath volcanic arcs and a deficiency is
noted in the amount of CO2 released from volcanoes compared with the amount of CO2
contained within subducted carbonates. Identical imbalance is noted between subducted
versus expelled H2O.
Cosmochemistry and Properties of Light Element Compounds 99

At room pressure, CO2 transforms at 194.5 K to dry ice, crystallizing in space

group Pa3 with C atoms occupying special position 4a and the O atoms occupying 8c.
It transforms to a liquid at 7.5 MPa and to a solid at ,0.5 GPa, which is known as
“dry ice” II. Above 10 GPa, dry ice II transforms to a Cmca structure (Aoki et al., 1994).
A single crystal X-ray analysis of CO2 at 1.0 GPa exhibits a cubic Pa3 symmetry
with cell edge 5.49 Å and a C – O bond length of 1.17 Å (Downs and Somayazulu, 1998).
Four high-pressure polymorphs of CO2 have been identified:
(i) CO2-I: a cubic phase observed below 10 GPa, known as “dry ice”.
(ii) CO2-II: produced at 0.5– 2.3 GPa.
(iii) CO2-III: an orthorhombic Cmca phase.
(iv) CO2-IV: a “distorted phase”, occurring between the stability fields of CO2-I and
CO2-III (Olijnyk and Jephcoat, 1998).
The strength of CO2, especially the orthorhombic phase III, increases with
increasing pressure.
The stability of these polymorphs is outlined below.

1.14.1. Stability of CO2 polymorphs: CO2-V quartz-like

The phase boundary between CO2-I and CO2-III is not well understood. For
example, Raman studies showed the stability of CO2-III to be above 18– 20 GPa (Hanson
and Jones, 1981), but X-ray diffraction studies proved its structure at 10 GPa. Both the
structure and existence of CO2-II and CO2-IV are still not well characterized.
Recent Raman studies showed that CO2-I and CO2-III co-exist over a large pressure
region. Above 30 GPa, CO2-III shows a broad diffraction pattern, which possibly arises
from a large pressure gradient and an increase in the material’s strength (Iota et al., 1999).
Ioto et al. (1999) synthesized an extended solid phase, CO2-V, in a DAC by laser
heating the molecular orthorhombic phase (CO2-III phase) above 40 GPa and 1,800 K. This
new material can be quenched to ambient temperature above 1 GPa. The vibration
spectrum of CO2-V is similar to that of quartz (SiO2) (or coesite), suggesting that it is an
extended covalent solid with C –O single bonds. The correspondence between the mode
frequencies for the SiO2 and CO2 structures was estimated by considering the ratio of the
reduced vibrational masses for the symmetric stretching modes in single bonded
C – O –C and Si – O – Si structures. The angle between the two X –O single bonds was
chosen to match the a-quartz bond angle (104.68).

1.14.2. H2O –CO2 mixture

Frost and Wood (1997) determined the volumes of H2O – CO2 mixed fluids between
0.95 and 2.0 GPa pressure and 1,100 –1,4008C. The volumes measured agree with the EOS
defined by Holloway (1977) (Thermodynamics in Geology, ed. D.G. Fraser, pp. 161 –181)
but, above 0.95 GPa, the measurements are confined to fluid compositions of XCO2 , 0.45
due to the nucleation of CO2 hydrates in the quenched mixtures. Small but significant
excess molar volumes of mixing ðXi . 0:6 cc mol21 Þ occur. Significant non-ideal mixing
occurs for Xi , 0:8 at 0.95 GPa and 1,2008C.
100 Chapter 1

Under high P – T; molecules of CO2 and H2O may form large molecules, which may
assemble into vesicles (seen in cells). These vesicles may be the early structural forms from
which the evolution to life started in the prebiotic Earth (Yearbook GL, 00/01, pp. 26 –27).

1.15. Carbon in space and in the Earth

Carbon is one of the major abundant elements in the cosmos, and 3/4 of the
intersteller (1S) and circum-stellar molecules are carbon-bearing.
Carbon clusters have been observed in the gas surrounding carbon-rich stars, in the
tails of comets, in carbonaceous meteorites and in the interstellar medium by using visible,
IR and FIR astronomy.
The C/H ratio in the Solar System is 237 £ 1026, and in the interstellar media it is
ð225 ^ 50Þ £ 1026 : It is estimated that 215 atoms of C per 106 H atoms are available for
carbon-dust formation in the Solar System, and 85 atoms of C per 106 H atoms in
the interstellar media (Henning and Salama, 1998). Hydrogen incorporation may bring
about smaller crystallite sizes in C grains with disorderly arranged graphitic units.
Carbon in its various polymorphic forms and structures (Fig. 1.30a – c) plays a
major role in the evolution of the interstellar matter (1SM) (Table 1.2). Because of its
ability to form complex species, carbon plays one of the most important roles in the
evolutionary scheme of the Universe, most importantly in the life cycle. It is also the key
element in the evolution of prebiotic molecules. Its role on the origin and the evolution of
life on early Earth through the exogenous delivery (by cometary encounters and meteoritic
bombardments) of prebiotic organics is the most significant.
Because it is the main supplier of free electrons in diffuse interstellar 1S clouds,
carbon contributes to the heating of 1S gas. Emission lines of neutral (CI) and ionized
(CII and CIV) atomic C are important cooling channels for the warm IS gas.
In the interior of stars, carbon is the first of the lighter elements (CNO cycle) that are
exclusively formed. The 3a process bridges the gap between the nuclei masses 4 (helium)
and 12 (carbon). In AGB stars, convection processes transport elemental C and O from
their C –O cores to their surfaces. The total production rate of solid C is ,0.002 M yr21
from C-rich stars (with C/O ratio $1 in their mantles and envelopes).
Carbon has an ionization potential ðE1 ¼ 11:3 eVÞ below the Lyman edge, so C is
almost completely ionized in space, excepting C in dense clouds. Carbon, the lightest of
the Group IV elements, is one of the most versatile, useful and interesting chemical
elements, generating various forms and phases. Usually, carbon atoms in a solid are
electronically bonded in two preferred forms: (1) sp2 hybrid: in which a carbon atom is
bonded to three equidistant carbon atoms with a 1208 angle between each two (e.g.,
graphite) and (2) sp3 hybrid: in which the carbon atoms occur at the four corners of a
regular tetrahedron with one at the centre of the polyhedron (e.g., diamond).
Amorphous or glassy carbon forms represent micro-zonal mixtures of these two types
of bonding.
The ability of C to form hybridized orbits accounts for its rich chemistry
(Table 1.3).
Cosmochemistry and Properties of Light Element Compounds 101

Figure 1.30. Polymorphic forms of carbon: (a) Diamond. (b) Graphite. (c) Buckminsterfullerenes. The inset
shows an order of increasing carbon bond stability of the polymorphs: C60, diamond, and graphite (After A. H.
Hebard, AT&T Bell Laboratories).

Primitive chondritic meteorites contain two types of pre-solar carbon grains: nano-
diamonds and graphitic particles (Huss and Lewis, 1995). Primitive organic materials in the
Solar System have been obtained from carbonaceous chondritic meteorites, interplanetary
dust particles (IDPs) of cometary origin collected in the upper atmosphere of the
Earth and grains from comet P/Halley (Cronin and Chang, 1998). Some carbon
102 Chapter 1

TABLE 1.2
Synopsis of C in space stellar and interstellar materials (ISM) (Source: Henning and Salama, 1998)

Location Atoms and molecules Solids

Carbon-rich circumstellar CO, C2H2, complex hydrocarbons Non-graphitic C with no pronounced

envelopes around red giants and gas-phase PAHs p–pp transition, silicon carbide
and ACB stars
Diffuse ISM Cþ, simple diatomic molecules, Graphitic material with strong p –pp
gas-phase PAHs and C chains transition, carbonaceous solids
with aliphatic hydrocarbons
Dense ISM CO and complex hydrocarbons Carbon-containing ices (CO, CO2
and CH3OH) and coagulated
carbonaceous grains
IS material in primitive PAHs Carbides, graphitic grains, poorly
meteorities graphitized C, C onions and
nano-diamonds

material closely resembles kerogen. It is generally assumed that the kerogen-like material
formed in the early solar nebula.

1.15.1. Fullerites and nano-crystallites

Carbon atoms with sp2 orbitals also make the other newly found form: fullerenes
(buckminister), of which C60 is the archetype (see Fig. 1.30c) discovered in 1985.
After graphite and diamond, fullerite solids make a third type of crystalline carbon
solids (Table 1.4). Small fullerites, such as C60 and C70, are not formed in significant
quantities in the low-density and H-rich outflows of C-rich AGB stars. Mixed
hybridization states lead to curved structures. Examples include fullerites and onion-like
polyhedral particles, in which the bending of the graphite layers is caused by C atoms
having tetrahedral bonds.
Carbon nanotubes make another structure, cylindrical systems composed of closed
shells capped at each end by pentagons. This has potential technological applications in
wide areas.
Carbon nanoparticles have a strong tendency to from agglomerates and even to
coalesce. A systematic broadening of the p – pp absorption feature of isolated
nanoparticles is correlated with an increased clustering. Nanocrystalline C films and

TABLE 1.3
Carbon hybrid orbitals (Henning and Salama, 1998)

Atomic orbitals Resulting hybrid orbitals Example

3
2s, 2px,y,z Four equivalent sp tetrahedral orbitals Alkanes, as in CH4
2s, 2px,y, 2pz Three planar sp2 orbitals and one perpendicular p orbital (2pz) Alkenes, as in C2H4
2s, 2px, 2py,z Two linear sp orbitals and two perpendicular p orbitals (2py,z) Alkynes, as in C2H2
 Cosmochemistry and Properties of Light Element Compounds
TABLE 1.4
sp-hybridized carbon solids (Henning and Salama, 1998)

Material Description of structure Bonding state

Diamond Tetrahedrally arranged C atoms in a cubic structure Four sp3-hybridized orbitals (strong, bonds with neighbouring atoms);
electrons completely localized
Graphite Layers of hexagonally arranged C atoms in a planar sp2 in the layers; weak bonds between layers are metallic
condensed ring system (graphene layers); layers (p electrons delocalized over s skeleton)
are stacked parallel to each other
Fullerite solids Geodesic structure of cage-like spheroids; network of Mixed hybridization state (s orbitals have partly p-orbital character):
12 five-membered rings (pentagons) and a variable
number of six-membered rings (hexagons)
Carbyne Long chains with either conjugated triple bonds or sp hybridization
cumulated double bonds
Nanocrystalline Planar graphic microcrystallites as “basic structural units” Different sp2/sp3 ratios; mixed hybridization states
materials (BSUs) embedded between clusters consisting mostly of
sp3-hybridized C, formation of bent graphene layers observed;
completely amorphous structures with no detectable
BSUs also present

103
104 Chapter 1

particles are characterized by different sp2/sp3 hybridization ratios as well as mixed

hybridization states (curved structures) (see Table 1.4).
The carbon materials such as soot or C black particles, glassy C or hydrogenated
amorphous C films cover a wide range of properties exhibiting different densities, band-
gap energies and chemical reactivity.
For C particles, the astronomically relevant spectral characteristics are dominated
by two strong electronic transitions:
(i) The s –sp interband transitions are centred around 80 nm,
(ii) The p –pp transitions (only in graphite C) are located between 180 and 270 nm.

1.15.2. Carbon polymorphs

Carbon occurs in crystalline form as graphite and diamond, in amorphous forms as

glassy carbon and carbon black and the metastable forms are referred to as “carbynes”
(linear molecules) or chaoite-like. The more recently discovered crystalline forms of pure
carbon molecules are termed “fullerenes”, some forms of which, C60 and C70, are
nicknamed as “buckyballs” and “buckyfootballs”, respectively (see Fig. 1.30c).
Carbon as an element provides inexhaustible insights into the properties of matter.
The physical properties of graphite and diamond are mostly opposite in character. Despite
its softness, the chemical bonds between atoms in the hexagonal graphite structure are even
stronger than in diamond. However, graphite shows the weakest Van der Waals bonds
known in nature and its layers slide easily.
Because of the high cohesive and activation energy, carbon polymorphs typically
exist metastably well in a T, P region where a different solid phase is thermodynamically
stable. For example, while in room conditions graphite is thermodynamically the most
stable form, diamond survives indefinitely. Conversely, except at very high temperatures,
graphite stubbornly persists at pressures far into the diamond stability field. This is also the
case for fullerenes and amorphous carbons.
Between carbon atoms, the binding energy is very large. For instance, in diamond,
the cohesive energy is 717 kJ mol21. This high cohesive energy accounts for the extremely
high melting temperatures (,5,000 K) of the solid forms of carbon. Again, the activation
energy required to transform one solid carbon phase to another is very large. For this
reason, high temperatures are often required to initiate spontaneous transformations.

1.15.3. Carbon in the Earth

In the Earth, the two most common crystalline allotropes of carbon are graphite and
diamond. The amorphous allotrope of carbon “shoot” converts to hard, clear diamond at
pressures exceeding 5 GPa at temperatures above 1,0008C — the condition achieved at a
depth of 150 km. Indeed, diamonds have the deepest known origins of any minerals found
at the surface, having been rapidly ejected from a depth of ,150– 200 km through
explosive volcanic eruption.
The role that carbon had in the fractionation of elements during core formation of
the Earth has been evaluated by Jana and Walker (1997). They determined the distribution
Cosmochemistry and Properties of Light Element Compounds 105

of Fe, Ni, Co, P, Ge, W, Mo and O between molten silicate and liquid metal at 5 –8 GPa
and 2,000 –2,3008C, the conditions corresponding to core formation. Carbon reduces the
siderophile tendencies of P and Ge, with the latter showing reduced silicate partition
coefficients from the C-free metal/ultrabasic silicate system to the C-bearing
metal/ultrabasic silicate system. P becomes lithophile at C-saturation level, while Ni and
Co show a reduction and W and Mo show an increase in siderophility at that level.
As P is depleted in the mantle, core formation probably did not occur at C saturation.

1.15.4. Carbon in high P – T: stability

The phases of pure carbon under high P – T conditions are shown in Fig. 1.31.
The topology of the stability fields of the thermodynamically stable phases is simple. The
diamond –graphite – vapour triple point occurs at 12 GPa/5,000 K.
In this connection, it is important to remember that both graphite and diamond
occur in two crystallographic forms: (i) “hexagonal” and “rhombohedral” graphite, and (ii)
“hexagonal” and “cubic” diamond. But in their respective forms, the chemical bonding
remains unchanged: sp2 for both kinds of graphite and sp3 for both kinds of diamond.
Differences in sequences of layering create different types. For hexagonal types of
graphite and diamond, the sequence is ABABAB…, wherein the third layer of atoms

Figure 1.31. P, T phase diagram for carbon up to extremely high pressures based on theoretical calculations
(Bundy et al., 1996, q Elsevier Science Ltd.).
106 Chapter 1

exactly superposes the first layer. In rhombohedral types for both the polymorphs, the
layering sequence is ABCABC, i.e., the fourth layer exactly superposes the first layer. The
nearly horizontal dashed line at the B region marks the P – T threshold of very fast (,1 ms)
and complete solid –solid transformation of graphite to diamond, which is always cubic.
This cubic diamond is formed from graphite under a pressure above 12 GPa and by pulsed
heating electrically or by laser. The dashed line B –F – G (Fig. 1.31) marks the activation-
energy threshold for the fast transformation of compressed graphite or hex-type diamond
into the thermodynamically stable cubic-type diamond. The enthalpy increase of graphite
from room temperature to melting lies in the range of 92 –115 kJ mol21. The value
115 kJ mol21 is probably the most reliable.
First-principle calculations by Galli et al. (1989) indicate that liquid carbon at
low pressure and 5,000 K should have mainly 2- and 3-fold coordination. There is
experimental evidence that liquid carbon at pressures above the graphite – diamond –liquid
triple point contains a pseudo-binary mixture with graphite-like and diamond-like
linkages. In this, the composition changes smoothly with pressure. On the other hand, the
first-principle molecular-dynamics calculations performed along high-density isochores
indicate that the melting curve of diamond should have a positive slope (i.e., dP=dTm . 0)
to above 100 GPa (Galli et al., 1990). By Clapeyron equation, a solid of smaller volume
than the melt requires dTm =dP to be positive since the entropy of the melt is always
larger than that of the solid. Experiments show that Tm of diamond increases with pressure.
DAC experiments indicate the stability of cubic diamond to extend to at least
275 GPa. Theoretical calculations (Biswas et al., 1987), however, show that the cubic form
would remain stable relative to all the likely metallic forms up to pressures in the range of
1,300 –230 GPa and that the diamond melting line has a positive slope. The difference in
behaviour between carbon and Si and Ge is attributed to the absence of p-electron states in
the 1s2 atomic core which allows the p-character of sp3 bonding electrons in diamond to be
held close to the nucleus (Yin and Cohen, 1983). The “diamond melt” produced at high
pressure in a DAC (Gold et al., 1984) contains a significant amount of sp2 as well as sp3
coordination at a pressure of 30 GPa.
Theoretical extrapolation of the melting line of diamond and the solid – solid phase
boundary between diamond and the (supposed) BC8 metallic phase are presented in
Fig. 1.32. In the Earth’s mantle (up to 135 GPa /2,500 – 3,500 K), the carbon is known to
occur as diamond. In the interiors of the outer planets, Uranus and Neptune, the maximum
P=T are in the range 600 GPa/7,000 K. In such a condition, carbon would exist as e
diamond phase.
New high-pressure carbon phases have recently been observed experimentally
or predicted theoretically. These include n-diamond (Hiria and Kondo, 1991), transparent
pressurized graphite (Utsumi and Yagi, 1991) and BC-8 carbon (Maihiot and
McMahan, 1992).
Graphitization of carbon typically occurs by an annealing process at high
temperature of 2,000 –3,0008C and the better-ordered graphite is formed at the higher
annealing temperatures. In the sp2-type bonds, the bond length changes are typically less
than a few percent at the pressure of 20 GPa.
Cosmochemistry and Properties of Light Element Compounds 107

Figure 1.32. P – T phase diagram for carbon based on theoretical calculations to very high pressures (Bundy et al.,
1996, q Elsevier Science Ltd.; Young and Grover, 1998, q 1998 North Holland).

Raman spectroscopy is very useful to identify and characterize various forms

of carbon. Between 17 and 70 GPa, graphite crystallites are ubiquitous. The 27 GPa Raman
spectrum is characteristic of graphite, which is relatively well ordered, especially for shock
synthesized material but, as pressure and temperature increase beyond 27 GPa, the Raman
band at 2,710 cm21 decreases, indicating increasing disorder in graphite.
Under shock compression, fast martensitic rather than slow re-constructive
transition is observed.
Because of the unusual open structure of fullerenes, the nature of phase transitions
in fullerites is of great significance over a wide range of temperatures and pressures. Shock
pressure in fullerenes is expected to be quasi-hydrostatic within the stability field of
fullerenes because of the weak Van der Waals interactions.
Weathers and Bassett (1987) melted small particles of graphite or diamond in a
NaCl matrix in DAC under 5– 30 GPa and by a pulsed Nd : YAG laser. Weathers and
Bassett (1993) observed that the smaller (,0.2 mm) spherules of carbon were having the
onion-like structure of giant fullerenes, which in essence were hybrid graphite/diamond
structures. The spherules commonly had a polycrystalline diamond core surrounded by an
onion-structured fullerene mantle. This observation may suggest that, in special
conditions, the hybrid fullerene structure can be energetically stable relative to diamond
even at conditions well within the diamond stability field.
1.15.4.1. C, Si and Ge
The phase diagram for silicon ([Ne] þ 3s23p2 structure) and germanium
([Ar] þ 4s23d104p2 structure) are used to construct the hypothetical phase relations for
108 Chapter 1

carbon ([He] þ 2s22p2 structure). Silicon and germanium both assume a nearly close-
packed metallic structure at high pressure. The hybridization of s, p and d electrons is
essential for the high-pressure response of silicon and germanium.
Carbon, with no easily accessible d states and no closed-shell configuration,
remains in the same non-spherical electronic configuration at pressures in excess of
100 GPa. The diamond structure is the closest possible packing for the non-spherical
carbon atoms. It is known that graphite shocked into the diamond stability field exhibited
sound velocities appropriate for solid diamond to 150 GPa at an estimated temperature of
,6,000 K. Thus, the stability of diamond at higher pressure is linked to the electronic
properties of this element.

1.15.5. Carbon-bonding structure

In diamond, each carbon forms strong bonds to four nearest neighbours oriented
at angles of 109.48 along the axes of a tetrahedron. In graphite, the three carbons are
bonded in the same plane at 1208 angles to each other. Overlap of sp2 orbitals makes the
in-plane bonding. This planer structure of graphite is responsible for its lubrication.
Delocalization of electrons throughout the individual planes in the unhybridized p-
orbital makes it a good conductor. In diamond, the valence electrons are strongly
localized in single bonds formed by sp3 hybridization, making it an insulator. The
polymorphic forms of C, diamond, graphite and buckminister fullerenes, were shown
earlier in Fig. 1.30.
The fourth form is composed entirely of sp-hybridized carbon atoms, arranged in a
linear (perhaps slightly stress-bent) geometry. Two resonance structures exist for the linear-
bonded carbons: all double bonds (cumulene) or alternating single and triple bonds
(polyyne).
Carbon clusters smaller than C10 would form linear chains. For odd-numbered
clusters (C3, C5, C7 and C9), all electrons in the molecular orbitals are paired, resulting in
singlet electronic configuration for the ground (lowest) state. For the even clusters (C4, C6
and C8) the two electrons in the highest occupied molecular orbital are unpaired so that the
ground-state configuration is a triplet. Clusters larger than C60 would exist as planar
monocyclic rings.
The chemical bonding in small clusters is mainly cumulenic, wherein the double
bonds exist between neighbours, which are sp hybridized. The average CyC bond length
for linear carbon molecules (e.g., C3, C5, C7 and C9) is ,0.128 nm.
The increased angle strain (relative to the ideal sp hybrid angle, 1808) in a small ring
is off-set by the added stability arising from the pairing of the two outer electrons in
changing from a chain to a ring.
By analysing the rotational transitions that accompany transitions between
different vibrational energy levels, precise information can be obtained about bond
lengths and bond angles, the electronic configuration of the cluster and the dynamics of
vibrational motions.
Cosmochemistry and Properties of Light Element Compounds 109

1.15.6. Graphite and diamond phases

Small diamonds, used in lapping and polishing, are commercially made by

shocking graphite mixed with iron and copper. Ideal graphite has a density of 2.2 and
diamond of 3.2. The bulk of industrial diamond is produced at P, T ranges of 4.5 –6.0 GPa
and 1,400 – 1,800 K. When good crystalline graphite is pressed to 10– 14 GPa at ,1,0008C,
it collapses to diamond, which is mostly a hexagonal-like (not cubic) wurtzite structure.
The hexagonal form is called lonsdaleite.
Using a diamond-anvil cell, Goncharov et al. (1989) observed that, at about
35 GPa, single-crystal graphite transforms to an optically transparent state with the
formation of sp3 bonding. It is surprising that no Raman band of diamond was seen; the
spectral absorption edge was too broad for diamond. The fact that it was transparent to
visible light suggests that it was non-conducting. It is likely that sp3 bonding begins to
develop at about 12– 14 GPa. This bonding removes conduction electrons from the
system, decreases the electrical conductivity and optical reflectivity and increases
the optical transmittance.
The graphite –diamond conversion temperature can be accurately determined. This
is because (a) the graphite behaves as an excellent black body and (b) the temperature at
the conversion ceases to rise when the carbon is converted to the transparent diamond form
and no longer absorbs the infrared radiation from the laser. The dashed line, B – F –G
(Fig. 1.32) marks the threshold of a very fast (i.e., ms –ms) transition of graphite to cubic
diamond. When graphite is converted to diamond, the sample becomes transparent and
absorption in the IR ceases.
The quantitative determination of the free energies of graphite and diamond showed
that their difference at 1 atm/0 K is only ,2.5 kJ mol21 (Rossini and Jessup, 1938). This
difference is insignificant compared with ,712 kJ mol21 of the vapour. Hence, with
respect to free energy, the condensation of carbon vapour to graphite or diamond would be
of nearly equal probability. By control of the reaction path and the nucleation conditions, it
should be possible to condense carbon vapour as diamond under low pressure at metastable
conditions. Nowadays, the chemical vapour deposition (CVD) technique helps the HP
diamond synthesis.
A plasma carbon vapour containing atomic hydrogen and hydrocarbons nearly
eliminates the graphite nucleation in preference to diamond. Controlling the condition,
polycrystalline diamond can be deposited as films, sheets, etc., which are of commercial
utility.

1.15.6.1. Superhard graphite

Opaque graphite under compression transforms to transparent insulator, which is a
hexagonal diamond or such material. Under compression along the C-axis, the bonds in
graphite of bridging carbon atoms convert into bonds, while the bonds of non-bridging
carbon atoms remain unchanged. Thus, the new phase is essentially graphite in structure
but possesses the physical properties of superhard diamond, and it can even indent a
diamond (Fig. 5, in Carnegie Inst. YB, 02– 03, p. 43).
110 Chapter 1

1.15.6.2. Resistivity and phase transition

A graphite crystal compressed in the c direction and monitored in the a and b
directions showed a decrease of resistivity with increase of pressure to ,12 GPa
(Aust and Drickamer, 1963), whereupon the resistivity rapidly increased, suggesting
transformation to a non-conducting form.
Li and Mao (1994) showed that the electrical resistivity of polycrystalline graphite
and amorphous carbon decreases with increasing P at ,19 GPa.
The P-dependence of the resistivity ðd In r=dPÞ is determined as 20.37 GPa21 for
the polycrystalline graphite and 20.039 GPa21 for the amorphous carbon. A phase
transition, interpreted as transformation to the hexagonal diamond phase, is observed in the
polycrystalline graphite at ,20 GPa but not in the amorphous carbon. These results
support the assumption that this phase transition is martensitic in nature.

1.15.6.3. Pre-solar nano-diamonds

A comparison of supernova explosions (SN) is invoked for the formation of pre-
solar diamonds. In the SN of type 1, the disruptive explosion results from the sequestering
of matter from a companion star by a dense white dwarf.
In a SN of type II, a single star, usually a supergiant in the evolutionary sequence,
undergoes a gradual and then a catastrophic collapse as core fuels in sequential layers are
expended. On final implosion and rebounding, the outermost envelopes are blasted into the
interstellar media. It is in these outermost expanding and cooling envelopes of type I and
type II SN that nano-diamonds nucleate, perhaps a month or so after the SN event. SN
shock waves reach velocities (.600 km s21) well above the threshold of 6.8 km s21
needed to form diamond. Pre-solar minerals in the assemblage of diamond, graphite, SiC,
Si3N4, corundum, spinel and other carbides are from multiple stellar sources that were
assembled as solids into the early solar nebula. Diamonds formed from CVD process or SN
shock waves were fairly abundant in the early Solar System. “The early Earth must have
been bathed in pre-solar diamonds, and this record may still exist in seeds that add to the
glitter of the cratonic microdiamond” (Haggerty, 1999).

Pre-solar carbonaceous meteorites. Diamond grains have been reported from cosmic
dust and meteorites. The diamonds in the Murchison meteorite are considered to be pre-
solar in origin. (For an interesting discussion, see Encyclopaedia Britannica Yearbook of
Science and the Future, 1989).
However, the rare-gas composition of minute diamond grains in some chondritic
meteorites have prove to be of interstellar (pre-solar) origin, having probably formed by
vapour condensation in stellar atmospheres. Anderson et al. (Astron. Astrophys. 1998, 330,
1080 –1090) extracted tiny diamond grains from the 1969 Allende meteorite. These grains
were probably condensed from a stellar outflow before the birth of the Sun. Such primitive
meteorites are time capsules, incorporating dusty grains, which remain unchanged since
the earliest days of the Solar System. They analysed the diamonds’ spectral and absorption
characteristics at various wavelengths and predicted the spectrum of a plausible candidate
star, a carbon-rich red giant.
Cosmochemistry and Properties of Light Element Compounds 111

1.15.6.4. Terrestrial occurrence

Diamonds in the Earth’s mantle, occurring in igneous rocks such as kimberlite or
deep metamorphic rocks, called ultra-high pressure (UHP) rocks, are formed under intense
pressure and temperatures (see also Section 1.23.8).
Graphitized diamonds have been described in garnet clinopyroxenite from the
Beni Bousera and Ronda peridotite massifs (e.g., Pearson et al., 1996). Some of these
peridotites contain up to 15% carbon. Several of the graphite octahedra contain faceted
inclusions of garnet and clinopyroxene. Their occurrence constrains the depth of origin for
pyroxenites to be in the diamond P – T stability field (.4.5 GPa and ,1,1008C).
Anomalously light carbon isotope values (d13C ¼ 216 to 227.6‰) for the graphite
suggest the crystallization of diamond from subducted kerogenous carbon.
More than 100 diamond crystals ranging from 0.1 to 0.5 mm have been recovered
from heavy mineral separates from two ophiolitic peridotite bodies in Tibet (Bai et al.,
1993). Associated mineral separates include SiC (moissanite), octahedral olivine and
serpentine crystals.
In another form, polycrystalline black diamond, carbonado, occurs within
sediments in Brazil (Bahia) and Central African Republic (Ubangi). But no carbonado
diamond has ever been reported from an in situ rock. However, carbonado can form in
nature by the process of carbon subduction, shock metamorphism or irradiation of organic
matter.
The Cambrian continental collision zone in northern Kazakhstan shows the
presence of diamonds in a matrix of metamorphosed sediments, which were subducted to
depths of 125 km or more and returned to the surface without losing their isotopic
characteristics.

Crustal materials in diamonds. Crustal materials in mantle rocks and in diamonds of deep
origin suggest transportation of crustal materials into the mantle by the process of
subduction.
The crustal signature in a mantle-derived diamond has been offered by a number of
mineral inclusions in diamonds. Even the presence of staurolite, a common mineral in
metamorphic products of terrigenous sediments of pelitic composition, has been reported
as inclusions in diamonds from kimberlite, occurring close to the margin of Kaapvaal
craton (Allsopp and Roddick, 1984). The staurolite-bearing diamond from Dokolwayo
Kimbalite (Swaziland) also shows the presence of coesite.
The coesite-bearing diamond assemblage of Dora-Maira formed at 3.6– 3.7 GPa at
700 –8008C. But staurolite needs to be isolated from coesite for its survival, and this
happens when in its early history it was enclosed in garnet and later was carried deep into
high-pressure diamond stability field. But Dokolwayo staurolite inclusion had grown with
the crustal material in the subduction zone, and later was incorporated in the mantle
diamond stability field.

Inclusions: lower-mantle minerals. Diamonds may contain tiny inclusions of lower-

mantle materials, which provide insights into the mantle chemistry, diamond formation
112 Chapter 1

and mantle dynamics. The majority of these inclusions are ferropericlase, the rest being
enstatite and Ca-silicates mostly perovskite in structure (McCammon, 2001).
Experiments have shown that Ni is preferentially incorporated into ferropericlase at
high pressures, leaving Mg-silicate depleted. Upper mantle enstatite normally contains
0.1 –0.2 wt% Ni, but enstatite (formerly Mg-silicate perovskite) found in lower-mantle
inclusions contains less than 0.02% Ni (Kesson and FitzGerald, 1991).
Besides the high-P majoritic garnet, the other transition-zone inclusion in diamond
is a garnet-structured Na pyroxene – Mg pyroxene solid solution that is stabilized at a
minimum of 16.5 GPa (,475 km) (Wang and Sueno, 1997). Green tetragonal garnet is of
lower-mantle origin (Harris et al., 1997). The lower-mantle mineral assemblage in
diamond is ferropericlase (Fe, Mg)O plus MgSiO3, resulting from the breakdown of
ringwoodite: g((Mg, Fe)2SiO4) ¼ (Fe, Mg)O þ MgSiO3.
A HP phase of CaSiO3 is also reported as an inclusion in diamond (Harte and
Harris, 1994). CaSiO3 and (Mg, Fe)SiO3 phases are predicted to have the dense high-P
cubic or pseudocubic(?) structure of perovskite (CaTiO3).
It is discussed elsewhere in this book that experimental results have shown that
REEs are enriched in Ca-silicate perovskite lower mantle inclusions relative to
ferropericlase and Mg-silicate perovskite.
Lower-mantle phases included in diamond may also be quartz (low-pressure
conversion product of stishovite) and the tetragonal almandine pyrope phase (TAPP;
Harris et al., 1997), which has an open crystal structure. These garnets are Al-rich. Also,
one Mg-silicate perovskite grain is reported to be 5 times more enriched in Al than the
usual concentration (Harte, 1999). An occurrence of an inclusion of ferropericlase highly
rich in iron possibly indicates its origin near the core –mantle boundary. Some inclusions
of Ca-silicate perovskite, enriched in trace-element content several-fold above
the primitive mantle (PRIMA) levels and positive Eu anomalies, suggest an enrichment
through a subducted source (Kesson and Fitz Gerald, 1991).
With respect to nitrogen concentration, all lower-mantle diamonds are noted for
their low concentration and are classified under type II. In contrast, 98% of upper-mantle
diamonds belong to type I. However, in the former case, nitrogen shows a high degree of
clustering, implying a long residence time in the lower mantle with high temperature. The
lower-mantle diamonds show a high degree of plastic deformation, complex growth
pattern and etching features (Stachel et al., 2000).
The evidence of lower-mantle diamonds provides perhaps the strongest evidence
that the 660 km boundary is penetrable. Diamonds with lower-mantle inclusions are
reported from cratonic (.1,600 m.y. old) stable interiors of continental plates.
Diamonds from kimberlites often contain inclusions of forsterite, pyroxene, garnet
and coesite silica (not stishovite). These are formed at depths of about 100 –300 km
(,3.5– 10 GPa) and temperature ,1,0008C. Most diamonds are over a billion years old.

Nitrogen inclusion. Besides the mineral inclusions, natural diamonds are mostly flawed
and coloured. Diamonds of type 1a contains several ppm of nitrogen atoms in the form of
coalesced groups, which show a band at 1,280 cm21 in infrared absorption, but produce no
EPR signal. In contrast, the type Ib diamonds contain isolated nitrogen atoms in place of
Cosmochemistry and Properties of Light Element Compounds 113

carbon atoms. This type shows EPR signals and infrared bands at 1,130 and 1,343 cm21.
Heating (1,600 –1,9008C at .5 GPa) can make the isolated nitrogen atoms coalesce and
produce the type 1a diamond.

Kimberlite clan rocks (KCR): ages and occurrences. Metasomites are considered
essential to the bulk chemistries of kimberlite clan rocks. Varying ages of KCRs have been
reported from the following localities.
1 Ga : Africa, Brazil, Australia, Siberia, India, and Greenland
, 450 –500 Ma : White Sea, China, Canada, S. Africa, and Zimbabwe
, 200 Ma : Botswana, Canada, Swaziland, Tanzania
80 –120 Ma : South, Central, and West Africa, Brazil, Canada, India, Siberia, U.S.A.
, 50 Ma : Canada, Tanzania
22 Ma : Ellendale in N.W. Australia

Continental cratons: Orpa kimberlites. Diamond distribution is linked to the distribution

of continental cratons. A close relation may be observed between diamond-growth
episodes and the continental craton stabilization. Continent margin accretion via
subduction helps the development of continental keels.
Orpa kimberlite is host to one of the Earth’s largest and most productive diamond
mines and is situated on the northwest side of the ancient continental core making up
southern Africa called the Kaapvaal craton. Orpa shows well-exposed kimberlite,
diamondiferous eclogite xenoliths in kimberlites and diamonds having inclusions of
sulphide and silicate minerals.
The sulphides can be classed into two groups based on Re –Os isotope data: One is
2,900 m.y. old (with diamonds having inclusions of eclogitic material) and the other is
990 m.y. old. These suggest two episodic growth periods of diamonds separated by about
2,000 m.y. The continental core stabilization about 2,900 m.y. ago was accompanied by
oceanic crustal under thrusting to make eclogite (Shirey, 1999).

Shock-produced lonsdaleite and SiC. Shock-produced diamond and lonsdaleite (the

hexagonal polymorph) were first observed in explosions. Diamonds have also been
found in association with several Russian impact craters and in K/T boundary impact
ejecta.
In the impact melts of the Reis crater (S. Germany), hexagonal diamond
(lonsdaleite) plates and cubic diamond in association with silicon carbide (SiC) have
been reported. These phases can be formed by CVD from the hot vapour cloud of the
impact. Vaporized carbon-bearing rocks may deposit cubic diamond and silicon carbide
(moissanite). Hence, their occurrence is diagnostic of hypervelocity impact. Similarly,
carbonados in sediment may result from the “heavy bombardment” period in the early
Earth, rather than from mantle-derived diatremes.
114 Chapter 1

As stated earlier, similar nano-diamonds have been found in meteorites, condensed

in stellar atmosphere. The diamonds reported in meteorite and ureilites were regarded as
products of shock pressure on carbon present in the meteorites.
Diamonds in clays in the K/T (Cretaceous – Tertiary) boundary provide unambigu-
ous evidence of a large-scale impact. Polycrystalline diamonds up to ,1 cm in size and
having appreciable amounts of lonsdaleite are seen in the impactites of Russian and
Ukrainian craters (Masaitis, 1992).

Diamond in alkaline fluids. Natural diamond forms in the P – T range of 5 –6 GPa and
900 –1,4008C. In alkaline carbonate melts, diamond can form at this P – T range (Harlow,
1997). Considering the abundance of carbonates in diamond-bearing rocks of magmatic
(Haggarty, 1986) and metamorphic origin, as well as the aqueous carbonaceous
composition of mantle fluid (Schrauder and Navon, 1994), Pal’yanov et al. (1999)
suggested that alkaline carbonate-fluid melts represent the most likely medium for natural
diamond formation.

CMSCHO system: XCO2. Ogasawara et al. (1997) investigated the CaO –MgO – SiO2 –C–
O2 – H2O (CMSCHO) system for understanding the paragenetic relations of diamond
and associated phases in eclogite and ultramafic rocks, and also f O2 –T – P stabilities
of diamond, coesite, enstatite, forsterite, graphite, magnesite, and dolomite. The low f O2
conditions perhaps allow diamond-bearing assemblages to occur along with carbonate
rocks. This and other calculated diagrams reveal that, with XCO2 lower than 0.1 and
a geothermal gradient less than 58C km21 (208C kbar21), diamond is stable at
T , 1; 2508C:
Thermodynamic calculations show that a partial pressure of CO2 is essential if
diamond is to be stable in an environment containing substantial amounts of Fe2þ and Fe3þ
ions. The necessary condition for diamond preservation is an accumulation of volatiles
under mantle conditions. A relatively cool mantle below a compressed crust may be the
most favourable location for accumulation of the high amounts of volatile matter that
appear to be necessary for diamond preservation. A vast majority of Africa’s diamonds
have their source in the interior regions of compressional stresses.

Ultrahard diamond. GL scientists deposited microwave plasma CVD-grown single-

crystal diamond (. 1 nm) on type Ib {100} synthetic diamond. Annealing at 2,0008C and
5 –7 GPa transformed the CVD diamond to an ultrahard (160 GPa) transparent crystal,
which is harder than both type Ia (natural) and polycrystalline diamonds.

1.15.7. Carbon isotopes

In terrestrial diamonds, the carbon isotope composition shows bimodality. Most

diamonds are poorer in 13C by ,5‰ compared with typical carbonate rocks. Carbon
isotope compositions are usually reported as parts per thousand (‰) deviations from Pee
dee belemnite (PDB), a calcium carbonate standard. These values are denoted as d13CPDB.
Cosmochemistry and Properties of Light Element Compounds 115

The d13CPDB value of 25‰ is found in basalts and carbonatite magmas. This value
probably represents the isotopic composition of mantle carbon.
Low 2d13C diamonds (around 20‰) often contain inclusions of minerals
commonly found in eclogite; these low 2d13C diamonds represent materials recycled
from the Earth’s surface.
Neoproterozoic glacial deposits usually have negative d13C values, attributed to a
collapse of biological activity in surface oceans caused by global glaciation (termed
“snowball” earth) (e.g., Hoffman et al., 1998). The cold climate recorded by low d18O and
low dD may not have coincided with the deposition of high d13C carbonates.

1.15.7.1. Oxygen and carbon-12 (C-12) evolution on the Earth

Before 3.5 Ga, the Earth’s atmosphere contained almost no oxygen but when
cyanobacteria evolved they probably first photosynthesized generating oxygen as a by-
product. Fossil evidence of cyanobacteria dates back to 3.5 b.y. (Western Australia) and
oxygen remained in trace amounts in atmosphere until about 2.5 b.y. ago (see, for a review,
Copley, 2001).
For photosynthesis, organisms prefer to use carbon dioxide that contains lighter
C-12 isotope. Thus, fossil organic matter contains high levels of C-12 compared with those
formed from inorganic materials which contain more C-13. Rocks formed around 2.2 b.y.
ago show a peak in C-12 levels, indicating that burial of organic carbon increased at this
time. Since within 30 m.y. from the time cyanobacteria evolved, the atmosphere has
contained a very high oxygen content. As oxygen appeared in the atmosphere, it started
forming a layer of ozone, which blocked out much of incoming solar UV light. Oxygen in
the atmosphere promoted the production of sulphates in the ocean.

1.15.7.2. 12C/13C ratios: interstellar

The 12C/13C ratio of around 90 may have been common in interstellar settings
before 4.6 GPa. Stars in our galaxy have 12C/13C ratios that are either .65 or ,65.

1.15.7.3. Raman line: P, T calibration

Raman spectroscopy of 13C diamond has been developed as a new method for
measuring pressure at high temperature in diamond-anvil cells. This method is useful for
measuring pressure at high temperatures, especially in samples that are corrosive to oxide
pressure sensors, such ruby and Sm : YAG.
The vibrational spectrum of cubic diamond itself is very simple: the expected
vibrational modes are:

Gvib ¼ T2g ðRÞ;

which gives rise to a single first-order peak in the Raman spectrum at 1,332 cm21.
The Raman spectrum of diamond has been studied to ,40 GPa (e.g., Aleksandrov
et al., 1986).
116 Chapter 1

In 13C diamond, the temperature dependence of Raman line has been measured up
to 1,329 K and is given by (Bassett et al., 1982):

nðTÞ ¼ 1; 294:5 2 0:022T 2 3:8 £ 1026 T 2

where n is the frequency and T is in kelvin.

This calibration using 13C diamond chips helps to measure pressure accurately
(^0.3 GPa) up to 16 GPa at temperatures as high as 1,200 K for samples in corrosive
environments.

1.15.7.4. 14C diamond: elastic moduli

Ramdas et al. (1993) showed that the elastic moduli of 14C diamond are ,0.5%
higher than the corresponding values for natural diamond, suggesting that 14C diamond is
harder than natural diamond. Ramdas et al. (1993) proposed a simple model incorporating
the zero-point motion and the anharmonicity of the interatomic forces; this explains
quantitatively the increase in the elastic moduli and the decrease in the cell-edge length (a).

1.15.8. Optical behaviour of diamond: flow and pressure-luminescence

Diamond is prized for its properties relating optical transparency in the UV to IR,
electrical insulation, thermodynamic stability, unique hardness and strength. Theoretical
electronic-structure calculations predict that diamond should be stable to at least
1,200 GPa (12 Mbar) hydrostatic pressure but in uniaxial load it may transform at lower
stresses (400 GPa). Diamond is seen to manifest plastic flow at 172 GPa pressure.
Synchrotron X-ray imaging and stress measurements by Hemley et al. (1997) of
diamond-anvil cell gaskets revealed large elastic strains at the diamond tip at a pressure
of 300 GPa. The diamond bent 168 over a distance of 300 mm without failure.
The diamond anvils for the pressure experiment may also manifest optical features
due to band-gap closure. Nitrogen in type Ia diamonds gives rise to a variety of absorption
and luminescence bands in the visible and ultraviolet. Even under low stresses, light
yellowish-brown colouration is commonly produced.
The lowest energy post-diamond structure predicted for carbon is BC8 and is
calculated to be stable at 12 TPa: experimental studies to date indicate that diamond is
stable to at least 400 GPa (Bundy et al., 1996). With increasing pressure, diamond-anvil
tips change from transparent to yellow to brown, eventually becoming deep red. A striated
pattern, due to plastic deformation of the anvil tip, was also observed (Mao et al., 1979).
Spectral study using absorption and fluorescence (Mao and Hemley, 1991) indicated that
the red colour was caused not by the absorption in the blue alone but also by intense
luminescence in the red.
Absorption measurements show that there is a considerable red shift in the
absorption edge at 100 GPa; at a pressure of ,300 GPa, absorption throughout the visible
range is observed. At ,300 GPa, the broad luminescence peak markedly increases in
intensity to a factor of 103 and a sharp red peak at ,2 eV (620 nm) appears.
Cosmochemistry and Properties of Light Element Compounds 117

Figure 1.33. Raman spectrum of diamond at the sample–anvil interface. (a) Zero pressure. (b) 160 GPa (solid N2,
295 K). (c) ,300 GPa (solid neon, 77 K). The 590 cm21 band is observed in nine experiments with H2, D2, Ne,
Csl, NaCl and SiO2 samples above 150 GPa. The band shows resonance enhancement with blue-green excitation,
and is independent of temperature from 77 to 295 K. Weaker Raman bands at ,900 and 2,000 cm21 have also
been observed. In addition, fluorescence bands appear in this region, but they can be distinguished from Raman
peaks by varying the laser excitation wavelength (from Mao and Hemley, 1991).

The intensity of the broad-band luminescence is directly correlated to absorption in

the blue. With increasing pressure, the optical absorption edge shifts from ultraviolet to
red. The changes are reversible and are associated with intense luminescence peaks at
3.0 –2.2 eV under 458– 514 nm radiation.
Raman scattering measurements at ,150 GPa (Mao and Hemley, 1991) show
evidence for new structural transitions, associated with large macroscopic deformation. At
pressures of 150– 300 GPa, a new Raman peak appears at 590 cm21 (Fig. 1.33; see the
caption), which is responsible for yellow fluorescence. The tips of the diamonds also show
stronger luminescence after loading.
The appearance of new Raman bands may indicate new stable semiconducting
(or insulating) phases of carbon. At much higher pressure, a gradual closing of the band
118 Chapter 1

gap between valence and conduction bands may be responsible for the red shift of the
absorption edge.
Pressure-induced electronic changes in deep-level impurity centres may be
responsible for the luminescence. Localized vibrational modes at these impurity centres
may cause the Raman line (Seal, 1984). The new band which appears at ,2 eV (620 nm) is
manifested as a brown colour and luminescence. This may be due to dislocation arising
from deformation but this new Raman feature shows close similarity with one-phonon
density of states of diamond, which has a broad peak centred at 600 cm21 (Dolling and
Cowley, 1966). Such a correlation could imply a breakdown in crystalline selection rules,
perhaps resulting from growth of defects (stacking faults) at the anvil tips associated with
macroscopic flow of the diamond, or from the formation of a new high-density amorphous
form of carbon (Mao and Hemley, 1991).
Density-functional theory calculations for diamond under large uniaxial strains
predict that the band gap closes at stresses of 400– 450 GPa (Nielsen, 1986). In the metallic
state, the T2g phonon is predicted to become unstable. The loss in intensity of this band may
be related to this predicted instability.

1.15.9. Carbon clusters

Carbon clusters of less than 10 carbon atoms and linear structures containing
10 –20 atoms (C10 – C20) are energetically favoured. Carbon atoms greater than 30 favour
aromatic and fullerene-related carbon cages (Curl, 1993).
In interstellar and circumstellar shells, carbon molecules can form fullerenes such
as C60 and C70 (Kroto et al., 1985). Becker et al. (1999) found C60 and C70 fullerenes as
well as remarkably stable clusters of C100 – C400 in the acid residue of the Allende
meteorite, a carbonaceous chondrite.
Carbonaceous chondrites contain interstellar grains — such as diamond, silicon
carbide (SiC), graphite and carbon-based molecules (perhaps fullerenes, C60, etc.) — that
were formed in stellar atmospheres, mixed into the solar nebula and incorporated into the
meteorite matrix (Zinner et al., 1995). Polycyclic aromatic hydrocarbons (PAHs) may also
form in circumstellar environments (Curl, 1993). The diffuse interstellar bands may also
contain diamonds, graphite, SiC, PAHs, fullerenes and fulleranes. Presence of these
compounds may indicate that these were present in the early solar nebula or occurred as a
component of pre-solar dust (Becker and Bunch, 1997).
Fullerenes and higher-carbon clusters may have been important on the early Earth
for providing the essential element of life, carbon.

1.15.9.1. Charged carbon clusters: low-P diamond

Charged carbon nuclei clusters, which are expected to form in the gas phase, are
suggested by Hwang et al. (1996) to be responsible for the formation of diamond in low-P
synthesis. When the carbon cluster is sufficiently small, the capillary pressure built up
inside the cluster can be high enough to make diamond more stable than graphite.
Cosmochemistry and Properties of Light Element Compounds 119

The chemical potential of carbon between diamond and graphite is shown to be reversed
when the size of the carbon cluster is sufficiently small.
The gas-activation process produces charges such as electrons and ions, which are
energetically strong heterogeneous nucleation sites for the supersaturated carbon vapour.
These lead to the formation of charged nuclei clusters. Once the carbon clusters are
charged, the surface energy of diamond is reduced by electrocapillarity while that of
graphite cannot be. This is because diamond is dielectric and graphite is conducting.
The number of atoms in the carbon clusters that make diamond more suitable than
graphite is estimated to be ,104 at 1,200 K and 2,700 Pa (calculated from the reported
values of the surface energy data; Hwang et al., 1996). The role of gas activation is
probably to decrease the surface-energy ratio of graphite to diamond and thus to increase
this number, inducing the dominant formation of diamond.

1.15.9.2. C-nanotubes
A single-walled carbon nanotube (SWNT) is a single graphite sheet rolled
seamlessly into a tube. In this structure, which manifests both 2D and 1D properties, the
electrons on the surface can only move forward or backward (;1D conductor and takes
more energy to spin about the axis). Hence, they move almost without scattering, i.e.,
without loss of energy (Frank et al., 1998).
At low temperature, only the low-frequency 1D vibrations are excited, consisting of
two transverse vibrational modes (longitudinal and twisting modes). Consequently, the
heat capacity shows linearity with temperature but, at higher temperatures, the higher
frequency modes are excited to form standing waves along the tube axis. On further
excitation, vibration of the nanotubes manifests 2D character resulting in a quadratic
temperature dependence.
The measurements of heat capacities as a function of temperatures show a smooth
curve, which in fact represents a quantized phonon spectrum of the nanotube as it
progresses through the change in dimensionalities with temperature (Hone et al., 2000). In
carbon nanotubes from 0 to 150 K, the soft interlayer vibrations contribute to the heat
capacity and eventually saturate. In this range, the T-dependence is cubic, characteristic of
a 3D solid. Hence, the dimensionality cross-over occurs from 3D to 2D.

1.15.9.3. Fullerenes
Small nanometer-size carbon clusters in the 40 –300 atom mass range are formed
when a laser puffs off vapour from a carbon rod and is cooled by expansion through
a supersonic nozzle. Significantly, it is seen that the C60 cluster is outstandingly stable,
narrowly followed by C70 and other species.
The great stability of C60 and others may be due to the truncated icosahedral cage
made up of hexagons and 12 pentagons (the shape seen on a soccer ball) with carbon atoms
at each intersections. Geometrically, it can be deciphered that 12 pentagonal “defects”
cover a planar hexagonal array of any size into a quasi-icosahedral cage as C60, C240 and
C540. Here it should be remembered that onion-layered giant fullerenes are observed to
form in the stability field of diamond (Bassett and Weathers, 1990).
120 Chapter 1

C60 and C70 are soluble in benzene. C70 compresses under pressure at room
temperature to an “amorphous” form, much like graphite, and decompresses reversibly
back to a C70 crystalline structure, whereas the “amorphous” C60-derived material
decomposes to a more dense diamond-like form.
Laser desorption (linear time-of-flight) mass spectrometry (LDMS) can be used to
detect fullerenes (C60 and C70) and any higher-molecular carbon clusters. Fullerenes are
extractable in an organic solvent such as toluene. LDMS analyses of the toluene extract of
Allende meteorite revealed peaks at Cþ þ þ þ þ
60, C70, C74 and C76 –C96 and a more prominent
þ þ
high-mass envelope extending from C100 to C250 (see Fig. 1.34; Becker et al., 1999). The
insoluble carbon component of carbonaceous chondrites (C1, CM, CO and CV) is a
“macromolecular material”. However, in the Allende meteorite, the principal insoluble
component is elemental carbon, which differs from graphite (sp2-hybridized carbon) in
having a poorly ordered structure.

C60 and C70 fullerene

C60 fullerene. A pressure .0.5 GPa stabilizes the Pa3 structure with the hexagon centre
facing the double bond in the adjoining molecule. With increasing pressure, the
intermolecular distances decrease. The pressure dependence of the distance between the

Figure 1.34. LDMS mass spectra for the TCB extract of the Allende meteorite. LDMS spectra showing a small
peak for Cþ þ þ
60 and a range of remarkably stable large carbon clusters between C100 and C250 (clusters were observed
to Cþ
400) (Becker et al., 1999, q 1999, Nature).
Cosmochemistry and Properties of Light Element Compounds 121

surfaces of two adjacent C60 spheres can be estimated from the Murnaghan EOS
(Murnaghan, 1974) using the bulk modulus and its pressure derivative (for Murnaghan
EOS, see Sections 2.6.1.2, 5.6.3 and 5.8.7.3).
For solid C60, this distance reduces from 3.04 Å at P ¼ 0 to 1.89 Å at P ¼ 22 GPa:
At such close distances, a sizeable fraction of double bonds can open and link covalently
with carbon atoms in the neighbouring molecules. When this happens in a random fashion,
a crystalline to amorphous transition takes place.
The simultaneous application of pressure and temperature aids the stabilization of
the free rotor face centred cubic (fcc) phase in C60. This is possible because the
transformation temperature to the frozen Pa3 phase increases at the rate of 100 K GPa21.
The C60 molecules either freely rotate or at least execute ratcheted motions. This helps the
double bonds in the adjacent C60 molecules to come close together for the (2 þ 2) cyclo-
addition and for polymerization reaction to take place. A model of the fullerene structure is
shown in Fig. 1.30c.
The C60 fullerene molecules have a high-symmetry truncated icosahedral
structure, which is very stable despite the hollow cage with large strains in p-bonds
and inherent structural “defects” on the five-membered rings. The exceptional
stability of C60 is central to the concept of developing new materials such as
lubricants, ultra-strong fibres, hard materials and high-temperature superconductors
(Hebard et al., 1991).
C60 fullerite crystal orders as a fcc structure with very weak Van der Waals
interaction between molecules. This causes a large compressibility at low static pressures.
C60 fullerenes are stable in the 13 –17 GPa pressure range. The ball diameter at 17 GPa is
estimated to be 6.35 Å. In C60, the reduction in the CyC bonds at 17 GPa may be limited to
the maximum of 2%. The interatomic C – C distance is then 1.51 Å, which is similar to the
bond lengths in C60. The two bonds lengths in the C60 cluster are 1.46 and 1.40 Å and in
graphite the CyC length in the plane is 1.42 Å. The 17 GPa Raman spectrum indicates
mixed phases of graphite and C60. At 27 GPa, the C60 to graphite transition is nearly
complete.
Yoo and Nellis (1991) shock-compressed C60 powders quasi-isentropically and
quenched from pressure in the range 10 –110 GPa and found the onset of a fast (,0.5 ms)
re-constructive transformation to graphite occurring near 17 GPa (Fig. 1.35). The graphite
recovered has a crystal planar domain size of about 100 Å. Above 50 GPa, a continuous
transformation to an amorphous state is observed. The C 60 ball is extremely
incompressible and its molecular structure essentially remains unchanged at moderately
high pressures. However, the centre to centre distance between nearest-neighbour C60
molecules collapses from 10 Å at ambient pressure to 7.82 Å at 17 GPa. This is
comparable to the C60 diameter of 7.1 Å (Kratschmer et al., 1990). Interfullerene C – C
distances at 17 GPa are then expected to be close to the CyC bond length in C60 and
in graphite.
Under these conditions, graphite can be formed through p-electron rehybridization
(a fast process) without involving a great deal of atomic rearrangement (a slow diffusive
process). That is, when the distance between C atoms on adjacent fullerenes approaches
122 Chapter 1

Figure 1.35. The P, T paths of the multiple-shock experiments of C60 overlaid on the phase diagram of carbon.
Each set of symbols represents the sequence of P, T points calculated for each multiple-shock experiment. The
solid lines represent equilibrium phase boundaries and the dashed line represents the kinetic line for the direct
graphite to diamond conversion. The phase boundary between diamond and liquid carbon is obtained from the
theoretical calculation by van Thiel and Ree (1989) (from Yoo and Nellis, 1991, q 1991 AAAS).

the C – C separation within a fullerene, bonding arrangements probably change from

intrafullerene to interfullerene in such a way as to collapse the balls into planar graphite.
When C60 is subjected to both high-pressure and high-temperature treatments, the
double bonds of neighbouring molecules form a polymerized chain. However, under
different conditions of P, T, fcc C60 is seen to transform to an orthorhombic structure
consisting of one-dimensional polymeric chains or to a rhombohedral or tetragonal
structure consisting of two-dimensional polymeric layers. Beyond 20 GPa, C60 transforms
to an amorphous carbon phase.
Raman spectroscopy and optical reflectivity studies revealed that solid C60 at
pressures .22 GPa undergoes an irreversible transformation to an amorphous carbon
phase while a rapid non-hydrostatic compression transforms it to a cubic diamond
structure.
C60 in C-nanotubes: Pulsed laser vaporization of graphite in the presence of
certain metallic catalysts produces both carbon nanotubes and C60 molecules. The
single-walled carbon nanotubes contain the closed carbon shell. Measurements of these
endofullerenes suggest that many of them are C60 molecules. The endofullerenes
coalesce into longer capsules under the action of an electron beam. (Note: Boron nitride
(BN) nanotubes, very similar to the carbon nanotubes, have different physical
properties.)
Cosmochemistry and Properties of Light Element Compounds 123

C70 fullerenes. In C70, the second most abundant fullerene, there are two orientational
ordering temperatures. One corresponds to the free rotor phase going to the long axis-
oriented rhombohedral phase and the other corresponds to the completely frozen
molecules in the monoclinic phase. Application of pressure increases the transition
temperature of fcc to rhombohedral transformation temperature at a high rate of
300 K GPa21 and the rhombohedral to monoclinic at a much lower rate of 50 K GPa21.
Beyond 20 GPa pressure, C70, like C60, transforms to an amorphous phase. It can also
be photopolymerized but to a much lesser degree than C60. Reversible amorphization of C70
at pressures beyond 20 GPa have been reported. In C70, the transition c ! a occurs as a
function of pressure and is reversible. High-pressure, high-temperature treatments of C60 are
seen to undergo a polymerization reaction, while C70 samples at best dimerize under similar
P and T. This difference is due to the topochemical constraints specific to the C70 molecule in
that only the double bonds on the polar caps are reactive, whereas the cyclic double bonds on
the equatorial belt cannot react without affecting the symmetry of the molecule.
It must be noted here that for the amorphization to occur, the two fullerene
molecules must be closely juxtaposed so that a sizeable fraction of the carbon atoms
change their coordination number from three (as in C60 or C70 molecules) to four, i.e., sp2
to sp3 hybridization. For the polymerization reaction to occur, parallel configurations of
CyC double bonds of the neighbouring molecules should be brought adjacent to each other
so that the (2 þ 2) cyclo-addition occurs in a coordinated manner.
For solid C70, the two adjacent C70 molecules approach each other from 3.39 Å at
P ¼ 0 to 2.67 Å at P ¼ 20 GPa when the Raman spectrum shows signatures of
amorphization. At this larger distance, the degree and strength of covalency may be
much weaker for C70 compared with C60. This may be the reason why amorphization is
reversible in the case of C70.
In C70, there are serious topochemical constraints which inhibit polymerization but
the double bonds at the polar caps come close together (2.23 Å). Near the polar caps, the
reactive bonds are five double bonds. Molecular orbital calculations show that the cyclic
bonds at the equatorial belt constitute the HOMO levels, which cannot polymerize without
affecting the symmetry of the C70 molecules.
The appearance of new modes and the splitting of the existing IR modes at HP HT,
with a corresponding decrease in lattice constants, would lead one to speculate about C70
dimer formation under HP HT. Such dimerization destroys the inversion symmetry of the
molecule. It is known that when inversion symmetry is broken, Raman modes are observed
in IR spectra and vice versa. Indeed, some of the new modes observed in the IR spectra
correspond to Raman modes seen in C70 at low temperature.

Low-carbon fullerenes: pentagon road and fullerene road. In 1990, Kratschmer, Lamb,
Fostiropoulos and Huffman (KLFH) (Kratschmer et al., 1990) caught the chemical world
off guard by reporting an easy synthesis of C60-buckminister fullerene. The KLFH
synthesis is a single-step procedure — strike an arc across two graphite electrodes and
large quantities of fullerenes are produced.
124 Chapter 1

The arc-generated vapour of atoms and dimers initially condense to form linear
carbon chains containing 3– 20 atoms. The next stage in fullerene growth involves a
transition, somewhere between C10 and C20, of carbon chains into monocyclic ring
structures. Two growth mechanisms have been proposed to explain the KLFH fullerene
synthesis:
(a) The pentagon road — in this (above C60) chemistry, the hexagons and pentagons
predominate. In such clusters, the number of pentagons is always minimized and so
these structures tend to remain open as they grow. The dangling bonds at the edge of
the carbon clusters provide reactive sites for cluster growth.
(b) The fullerene road — in this, the hexagons and pentagons dominate around C30 and
the dangling bonds are minimized. The adjacent pentagons are the points of highest
strain energy for a closed fullerene. Thermodynamic arguments favour the fullerence
road whereas kinetic arguments favour the pentagon road.
Manalopoulos and Fowler (1996) predicted a fullerence road, starting at C24
(which has only one structural isomer) and terminating at C60. Most fullerences smaller
than C60 are characterized by many possible structural isomers. The fact that bulk amounts
of C 36 can be prepared by the KLFH synthesis provides evidence that the
thermodynamically favoured pathway (the fullerene road) dominates bulk fullerene
production (Piskotti et al., 1998).
NMR study on C36 has proved D6h symmetry, i.e., a 6-fold rotation axis, consistent
with the fullerene road. The C36 ðD6h Þ form is characterized by the minimum possible
number of shared pentagon – pentagon bonds (Fowler and Manolopoulos, 1995). Heath
(1998) has shown that C60 is surrounded by seven of its smaller C36-fullerene cousins. C36
is golden yellow in solution whereas C60 is blue/purple. C36 has the minimum possible
number of shared pentagon – pentagon bonds (helping the growth of C60 along the fullerene
road).
D6h structure of C36 should be characterized by a triplet ground electronic state.
The mixing of s- and p-bonds may well predict a singlet D6h ground state. The C36
molecule is characterized by a large amount of curvature-induced bond strain, hence
isolation of smaller fullerenes is highly improbable.
The C50-fullerene with D5h structure seems to be the most stable C50 isomer and it is
also a dead-end on the lowest-energy pathway on the fullerene road (Manolopoulos and
Fowler, 1996).

Magnetic behaviour of carbon allotropes. A flow of currents is postulated around the

carbon ring system of graphite in response to an applied magnetic field. C60 possesses
magnetic properties common to both the other C allotropes with a weak diamagnetism, as
in diamond, but with other ring currents resembling graphite.

1.15.10. Organic minerals in meteorites: shock loading

Carbonaceous chondrites show enormously complex organic chemistry. Studies

by applying NMR spectroscopy of protons, 15N and 13C nuclei of meteoritic organic
residue of various carbonaceous chondrites, e.g., Orgueil, Murchison and Tagish Lake
Cosmochemistry and Properties of Light Element Compounds 125

(GL Yb, 01-02) reveal macromolecular organic material composed of a wide range of
organic (aromatic and aliphatic) functional groups, including those containing oxygen
(e.g., organic acids, ketones, alcohols and ethers). Each meteorite exhibits enormous
differences in the relative abundance of each functional group. Such variation possibly
reflects different processing histories in the early solar nebula.
Collisional processes and attendant shock effects in the formation of many
meteorites have been evaluated by the study of their minerals — their texture, chemical
and isotopic characteristics. Thermodynamic considerations provide important guide-lines
in understanding shock processes.
While the shock behaviour of many minerals and rocks is well described (e.g.,
Stoffler et al., 1988), comparatively much less is known about the shock-loading
modification of organic matter.
Organic molecules present in meteorites may fragment and decompose or
recombine into secondary products due to high stresses and elevated temperatures. Tingle
et al. (1992) shocked Murchison chondrite samples up to 36 GPa and found incipient
devolatilization at 20 GPa and 70% loss at 36 GPa, the latter probably reflecting
decomposition of amines, loss of elemental S and modification of abundances of aliphatic
and aromatic hydrocarbon components. Experiments have been conducted on the effects of
impacts on carbonaceous compounds such as diamond, graphite, fullerenes and pthalic
acid. From the K/T boundary, extraterrestrial amino acids and fullerenes have been
identified (e.g., Heymann et al., 1994). Fullerenes have been found in the 1.85-Ga-old
Sudbury impact structure and in the craters on the Long Duration Exposure Facility
(di Brozolo et al., 1994).
Some fragments of Canyon Diablo meteorite recovered from Arizona crater seen to
have suffered shocks of ,13 GPa. The event was typical of asteroidal impactors striking
Earth at ,15 km s21. The impactors carried organic compounds, including amino acids,
which could have survived the impact. The intact asteroidal and cometary organic matter
may have contributed to the origin of life on the Earth (e.g., Cyba and Sagan, 1992).
Murchison meteorite revealed a racemic population of amino acids containing
many of the compounds utilized by life on Earth as well as many others alien to terrestrial
biology (e.g., Cronin et al., 1988). From the Murchison chondrite, chiral non-protein amino
acids have been reported (Cronin and Pizzarello, 1997).
Peterson et al. (1997) subjected amino acids to shock impact over a pressure range
of 3.5 – 32 GPa to study the extent of amino acid destruction, racemization and conversion
to secondary amino acids. They observed racemization but some residual optical activity
remained in the amino acids surviving shocks up to 32 GPa. Their results reveal that,
within the microsecond time frame of impact, two of the following three processes
occurred as the impacting pressures increased from 3.5 to 32 GPa:
(i) increase in destruction of the parent amino acids,
(ii) decrease in retention of optical activity (i.e., racemization) among amino acid
enantiomers, and
(iii) formation of secondary population of amino acids.
A better survival of the shock pressure was noted for the dicarboxylic acids,
glutamic acid and aspartic acid; all suggest a selectivity in the order of destruction.
126 Chapter 1

TABLE 1.5
Variation in amino acid abundances (ng AA/mg sample) with peak shock pressure (GPa) in aluminium (Al) and
stainless steel (SS) target containers (Peterson et al., 1997)

Pressure (GPa) Metal Alpha ABA NORLEU PRO ASP PHE GLU ISOVAL Alpha AIBA

Allende
0.000 4,000 4,700 4,000 4,400 3,900 4,400
3.50 Al 2,500 3,400 2,300 3,300 7,000 3,200
6.50 Al 2,400 3,200 1,800 2,600 4,700 3,400
6.80 SS 2,700 3,000 2,100 2,900 5,400 3,300
10.0 Al 960 840 650 2,100 2,700 740
10.2 Al 550 860 320 610 1,600 410
12.0 SS 2,600 3,200 2,000 3,400 5,700 2,500
15.6 SS 2,100 2,400 1,400 1,500 3,200 1,000
18.4 SS 300 490 91 320 1,100 150
19.3 SS 270 320 140 120 670 240
25.0 SS 15 19 6 54 22 31
25.2 SS 16 13 6 28 55 19
29.8 SS 2 7 4 6
Murchison
0.00 27,000 23,000 18,000 20,000
18.3 SS 4,200 4,300 4,300 8,000
30.4 SS 130 25 31 180
ABA, aminobutyric acid; Pro, proline; ASP, aspartic acid; PHE, phenylalanine; GLU, glutamic acid; ISOVAL,
isovaline; AIBA, a-amino isobutyric acid.

Secondary product formation became apparent at 19 GPa. The abundance of surviving

amino acids in shock pressures (3.5 – 30.4 GPa) in various target containers is shown in
Table 1.5 (Peterson et al., 1997).

1.15.10.1. Amino-acid racemization: chirality retention

Optical activity can be compared with amino-acid abundance, both of which
decrease with increasing pressure. Decreasing retention of optical activity corresponds to
increasing racemization. In larger objects, a higher degree of amino-acid decomposition
and racemization would occur. Simple porosity tends to amplify thermal effects of shocks
at grain boundaries where amino acids are dispersed.
If extraterrestrial amino acids were chiral, some remnant of their chirality would
also have survived. This possibility arouses considerable interest in the studies of intact
asteroidal and cometary organic matter that might have led to the origin of life on Earth
(Chyba and Sagan, 1992).
Most of the natural amino acids occur in the L -enantiomeric configuration and so do
the amino acids in the living systems. However, on death and decomposition, the surviving
amino acids become slowly transformed to D -enantiomers (the mirror image of L ), thus
losing optical activity and approaching enantiomeric equilibrium with equal numbers of
both (total racemization).
During shock, the effects of temperature are inseparable from pressure: as the
pressure goes up, so too does the temperature within the sample. Thus, the P=T relationship
Cosmochemistry and Properties of Light Element Compounds 127

in shocks causes inter-conversion of L - and D -enantiomers. The quantity, 1 2 D /L , obtained

from measurement of the ratio of D - to L -enantiomers for an amino acid at a particular shock
pressure, measures the fraction of initial optical activity or racemization. At the outset,
1 2 D /L is unity and when the racemization is complete this quantity equals zero.
As peak pressures increase, less retention of optical activity is observed. In Allende
and Murchison mineral matrices, excess L -amino acid enantiomers continue to be
observed up to shock pressures of 25 GPa where less than 1% of the original compounds
remain. In an amino-acid molecule highly stressed by shock, bond homolysis occurs at the
asymmetric a-carbon and recombination of the radical fragments can reform the original
molecule as the L -enantiomer.
The fact that amino acids (except ASP) after $99% decomposition can retain
$60% of their chirality (in Allende, for instance) suggests that fragment mobility is not
facile under shock conditions.
During shock experiments, water from clay minerals is released. In the Murchison
meteorite, incipient water loss occurs at 11 GPa, leading to complete devolatilization at
30 GPa (Tyburczy et al., 1986). However, water released during shocks exerts little
influence on racemization. Nevertheless, amino-acid racemization under pressure has been
noted in bones, teeth, shells and wood where water is absent (e.g., Bada, 1991).
If the Canyon Diablo event was typical of asteroidal impactors striking Earth at
,15 km s21, some small fraction of the organic compounds, even amino acids present in
such projectiles, would have survived intact. If the extraterrestrial amino acids were chiral,
their chirality would also have survived the impact.

1.15.10.2. Vitrinite maturation

Some experimental studies have been carried out on the pressure dependence of
vitrinite maturation. The results showed that applied pressure suppresses vitrinite
reflectance values. Addition of H2O to the starting material was seen to enhance the
pressure effect.
A kinetic model of vitrinite reflectance should therefore account for P (with T and
time) for the accurate interpretation of geothermal histories of deep sedimentary basins and
low-T/high-P metamorphosed regions.

1.16. Nitrogen

In diatomic nitrogen, the triple bond has the greatest binding energy of all molecules.
At low P, T, nitrogen forms molecular crystals in which these strong bonds
co-exist with weak Van der Waals interactions between molecules, producing an insulator
with a large band gap (Freiman, 1997). Under increasing pressure, intermolecular
interactions increase and the molecules eventually dissociate to form monoatomic metallic
solids (similar to hydrogen). In theory, N2 can be transformed to a non-molecular framework
or polymeric structures at pressures between 50 and 94 GPa. Non-molecular N2 is
semiconducting to at least 240 GPa, when the energy gap decreases to 0.4 eV. At 300 K, the
transition from insulator to semi-conductor starts at ,140 GPa. This transition also
exhibits a large hysteresis with an equilibrium transition estimated to be around 100 GPa
128 Chapter 1

(Eremets et al., 2001). The large hysteresis is accompanied by a large volume change:
DV < 35% (calculated).
Nitrogen is semiconducting in the post-molecular phase, while hydrogen is
predicted to become metallic on dissociation. (Note: Theoretical studies indicate that
insulating (semiconducting) and metallic phases are very close in energy.)
Modern HP techniques have progressed to create novel high-energy density
materials—which can be used in future as fuels and propellants.
The phase diagram of nitrogen up to 900 K and 100 GPa shows strongly interacting
(e.g., polynitrogen) diatomic molecules. Above 100 GPa numerous transformations are
observed including a polymetric nitrogen. A novel salt, nitrosonium nitrate (NOþNO2 3 ),
was produced by laser heating of NO2 and N2O4 mixture under pressure. Even below
180 K the NOþNO2 3 species is seen to persist in atmospheric pressure, making it an
intriguing high-energy material.

1.17. Sulphur

Sulphur gases in the early Earth underwent both biological and non-biological
fractionation. Micro-organisms that fed on sulphur preferred the lighter of the elements
three isotopes.
Elemental sulphur is an even-valence, wide band-gap insulator under normal
pressures but at about 90 GPa it becomes metallic (Luo et al., 1991), rather
superconducting metal with a transition temperature of 10.1 K. Moreover, Tc increases
linearly with pressure up to 160 GPa, where the second structural phase transition occurs
and Tc increases to 17 K. Thus, highly compressed sulphur has the highest Tc of any
elemental solid measured to date (Hemley and Ashcroft, 1998).
Elemental sulphur, an insulator, transforms at a pressure of 90 GPa to a body-
centred orthorhombic (bco) metallic phase, which also manifests superconductivity with
Tc of 10 K. At 160 GPa, it transforms into a superconductor with a record high Tc of 17 K
(Luo et al., 1993).
For discussions on other native elements, such as noble gases K, Fe, Au, etc., see
their respective sections.
32
S:34S. Microbes preferentially metabolize 32S; consequently, the fractionation
34 32
of S/ S isotopes becomes significant. These are easily measured in pyrite-bearing
rocks. The isotope enrichments are transported by aerosol particles and are metabolized
by microbes, and finally get fixed in pyrites. The measurement of the four stable isotopes
of sulfur, 32S, 33S, 34S, and 36S, in a single grain of pyrite identification can be made of
the microbial activity and the atmospheric photolysis by cosmic UV radiation.
The ratio of three isotopes 32S, 33S, and 34S may show anomalous fractionations of
three sulfurs with respect to different depositional environments, represented by black
shales and dolomitized limestone. Large changes in isotope ratios may probably be caused
by combination of two factors involving aerosols and microbes (i) differential dilution of
native sulfur in atmospheric aerosols by seawater sulfate and (ii) different sulfur isotope
fractionation by different populations of microbes.
 129

(B) Terrestrial Planets

This page is intentionally left blank
Cosmochemistry and Properties of Light Element Compounds 131

1.18. Early geochemical evolution

The standard models of planet formation envisage the gravitational collapse of a

rotating interstellar cloud which will form a star or star system (binary stars) depending
on the angular momentum of the collapsing system (Safronov and Vitijazev, 1986;
Witherill, 1990). Coalescing planetesimals and planetary embryos formed around the
gravitating star.
The terrestrial planets evolved from the proto-solar cloud by the process of
accretion of planetesimals of composition similar to carbonaceous chrondrite. The process
involved stochastic accretion of planetesimals that increased in size and decreased in
number with time. In only about 105 years, the planetesimals attained the size of the Moon,
Mercury and Mars, and the Earth accreted to its present mass (6 £ 1027 g) in about 2 £ 108
years (Wetherill, 1990).
In the proto-solar cloud, refractory lithophile elements, viz. Ca, Al, Si, Ti, Sc, REE,
etc., condensed very early. Because of their high temperature of vaporization, they would
resist volatilization during giant impacting. In the planetary bodies undergoing fractional
crystallization, refractory lithophile element ratios would vary depending on the relevant
crystal/liquid partition coefficients.
The terrestrial planets experienced giant impacts in the later stages of their growth.
One or several giant impactors (products of stochastic accretion) induced rapid spins on
these planets and large-scale melting, perhaps leading to a complete melting of small
bodies comparable to Mercury in size. Because of this melting, core formation took place
by gravitational segregation as the proto-planets grew.
Earth accreted most of its mass through catastrophic impacts of large planetisimals
during its early phase. The Moon was formed from the accumulation of dust clouds
generated by a giant impact of a Mars-size body on the proto-Earth, when both the
impactor and the Earth already had a metal proto-core. Numerical modelling indicates that
much of the impactor core many have ploughed its way through the mantle and merged
with Earth’s core. Concurrently, the kinetic energy buried deep within the Earth would
have been sufficient to melt a large portion of the silicate mantle (Melosh, 1990). It could
also have been that the metal from the proto-planets was excavated, remixed and
re-equilibrated with the molten mantle.

1.18.1. Chondritic character of terrestrial bodies

It is presumed that the primitive composition of the terrestrial planets, including the
Earth, is that of the devolatilized solar nebula. The solar abundance of the refractory
elements is reflected in the composition of undifferentiated C1 chondrites, which are
considered to represent the chemistry of the primitive solar nebula (Anders and Greeves,
1989). This chondritic Earth model was propounded by many (Liu, 1982; Anderson and
Bass, 1986). However, some emphasized that the Mg/Si ratio of the terrestrial planets is
more representative of the solar nebula value than that of the C1 chondrites. The chemistry
of Sun is seen be richer in iron and calcium than the C1 chondrite. Hence, Earth may be
richer in these elements than what are seen in C1 chondrites.
132 Chapter 1

The Cr/(Cr þ Al) ratio for carbonaceous chondrites falls in the range of 0.10– 0.13
(the solar abundance value is 0.13). This value is also supposed to represent the bulk
mantle of the Earth. This is concluded on the logic that, in the nebula, Al and Cr condensed
as oxides (thermodynamic calculations of Grossman, 1972) and did not go into solution
with metallic iron (or Fe-alloy). The core-forming accretion of metallic iron, therefore, did
not take in Al and Cr. Thus, the Cr/(Cr þ Al) ratio of 0.10 –0.13 for chondritic and solar
abundances should also be true for the bulk mantle of the Earth. On this basis, there should
be Cr-rich (Cr/(Cr þ Al) .0.10– 0.13) complement of the Cr-poor upper-mantle materials
(Cr/(Cr þ Al) , 0.05) such as lherzolite.
In the chondritic Earth model, the lower mantle is considered to be more silica-rich
than the upper mantle and is composed primarily of (Mg, Fe)SiO3 perovskite, thus
suggesting two-layer mantle convection.
However, Ringwood (1979) advocated a whole-mantle convection of the
homogeneous mantle of “pyrolite” composition, which is a olivine –pyroxene rock
capable of producing basaltic magma and peridotite residue on partial melting.

1.18.1.1. Chemical differentiation: siderophile elements

About 93% of the Earth’s earliest mass was primitive chondritic material accreted
under reducing conditions, when siderophile elements were almost entirely scavenged into
the iron-rich metal that segregated to form the proto-core. Subsequent to that, about 7% of
the Earth’s mass, more oxidized and rich in volatiles as noted in the upper mantle, was
added. The mantle became re-enriched in highly siderophile elements (HSE) and
moderately siderophile elements (MSE). Subsequently, segregation of small amounts of
metal and sulphide to the core stripped the silicate mantle of HSE. (Note: Siderophile
elements are divided into two groups: HSE — Ex., Os, Re, Ir, Ru, Pt, Rh, Au and Pd; and
MSE — Ex., P, Fe, W, Co, Ag, Ni, Sb, As, Ge and Mo.)
HSE (except Au) are refractory and are hardly fractionated out by cosmochemical
processes. Indeed, HSE occur in the mantle in the same relative proportions (Foster et al.,
1996) as in the primitive Solar-System material (; the carbonaceous chondrites). HSE
have very low solubility in silicates, solids or melts, especially at the low f O2 required
for core segregation.
In the homogeneous accretion model, the metal segregation to form the core was an
equilibrium process. Experiments determining the high P – T metal partitioning
coefficients revealed that the abundances of most of the MSE in the mantle can be
accounted for by metal –silicate equilibrium in a magma ocean at around 28 GPa and
2,200 K (Righter et al., 1997).
In homogeneous accretion, core segregation must have occurred under conditions
where partition coefficients for all HSE were essentially the same, thus leaving them
unfractionated in the mantle. Isotopic data (Foster et al., 1996) show that Archean mantle
had a chondritic Re/Os ratio.
Segregation of siderophile metals into the core occurred simultaneously with
accretion and this was followed later by addition of (,10% of the Earth) a more oxidized,
volatile-rich, lower-temperature nebular component. This established the relatively high
abundances of the moderately siderophile elements W, Co and Ni.
Cosmochemistry and Properties of Light Element Compounds 133

Depletion of siderophile metals in the mantle is well known. The depletion of the
highly siderophile elements (Au, Ir and Pd) is especially stronger in relation to the weaker
depletion of the moderately siderophile elements (Mo, W, Co and Ni). However, a core-
growth curve based upon Pb isotopes shows evidence of extraction of Pb from the
mantle and its influx into the core. This evidence implies that about 85% of the
core formed as a continuous process in the first 50– 200 m.y. of the Earth’s history
(Allégre et al., 1982).

1.19. Accretionary evolution of the Earth

The current thinking is that, during its accretionary growth stage, the early Earth
formed over a period of tens of million of years and most of the mass arrived in sizes much
larger than the Earth’s Moon and possibly even larger than Mars. The Earth today is still
growing in mass through impacts with comets and asteroids, albeit very slowly. The early
growing planet was impact-melted and an immiscible iron-rich liquid aggregated and
descended under gravity pull to form the core (a consequence of Rayleigh – Taylor
instability).
For its early history, Wetherill (1990) modelled the Earth as one of the coalescing
planetesimals, which formed planetary embryos. Subsequent mixing of asteroidal material
with the proto-Earth was very likely. This model also lends support to heterogeneous
accretion of the Earth (Wänke and Dreibus, 1988; O’Neill, 1991a,b).
In the first stage of accretion, the material was mostly of refractory elements while,
in the second stage, addition of moderately volatile material took place. This was followed
by the third stage with addition of mantle abundances of PGE, S, Se, Te and C. These
elements later sequestered towards the Earth’s core. Such an addition of late-stage veneer
would account for the mantle abundances of the PGE, S, Se, Te, C, etc., and a global-scale
melting of the mantle.
Moon formation: At the final stages of accretion, impacts of large planetesimals
were likely to produce substantial melting of the Earth. Earth’s Moon is considered to have
been produced from a giant impact of a Mars-size body on the Earth. The impact of such a
large planetesimal released enough energy to melt the proto-Earth which, notwithstanding,
was nearly 50% molten (Stevenson, 1987) and had a temperature of about 4,000 K before
the Moon-forming impact. The rock vapour having the chemistry of the undifferentiated
mantle of the Earth, reached a temperature of 1,600 K. The refractory vapour constituents
condensed to form the Moon.
Hadean history: The first billion years (early Archean) of the pristine Earth can be
summarized as below:
Year before present (Ma) Time from the zero year Event
4,566 0 Early condensation of refractory materials
4,565 1 m.y. Formation of planetesimals
4,555 11 m.y. Igneous activity in planetesimals
4,500 66 m.y. Moon formed
4,450 116 m.y. Core segregation and atmospheric outgassing
(continued on next page)
134 Chapter 1

Year before present (Ma) Time from the zero year Event
4,430 136 m.y. Final accretion of the Earth
3,800 766 m.y. Oldest known rocks; earliest life?
3,500 1,066 m.y. Oldest known fossils

The working model for the Hadean history of the Earth, therefore, involves five
stages:
(a) a process of accretion of planetesimals of chondritic composition,
(b) a giant impact to melt and devolatilize the early mantle of the Earth,
(c) a gravitational separation of immiscible iron and silicate liquids to form the core and
the mantle,
(d) an equilibrium crystallization of the mantle from the bottom up (with fractionation in
the upper 150– 300 km only or not at all),
(e) a partial melting of the upper mantle to form oceanic and continental crust, and a
complementary residual upper mantle depleted in Si/ Mg and Fe/ Mg ratio with respect
to the lower mantle.

1.20. Compositional characteristics of the Earth

The compositional models of the Earth are based on the evidences of (i) the seismic
profile of the Earth, (ii) the compositions of the primitive meteorites and the solar nebula
and, (iii) the chemical and petrological models of peridotite basalt-melting relationships.
Compared with the chondritic meteorites the Earth is more strongly depleted in the more
volatile elements. Indeed, there is no group of meteorites that has a bulk composition
matching that the Earth (McDonough and Sun, 1995).
The two major divisions of the Earth are the metallic (core) and the silicate (crust
and mantle) portions. These are divided into sub-regions, based on seismic velocity
features such as crust, upper mantle, transition zone, lower mantle, D00 zone, outer core and
inner core (Fig. 2.3). The upper mantle consists of olivine, two pyroxenes and garnet, and
has a pyrolitic composition.
As of today, the opinion of the majority of geochemists largely converges in
stipulating the following main three geochemical characteristics of the Earth:
(a) The Earth follows the compositional trends of chondrites, particularly carbonaceous
chondrites. (Note: If stable isotope fractionation is considered, the Earth was closer to
enstatite chondrite than to other meteorites.)
(b) For major elements, the mantle is compositionally homogeneous, i.e., the upper and
lower mantle have the same composition.
(c) The Earth’s core is made of 80 – 90% Fe– Ni alloy, mixed with lighter elements to
reduce its density by 10 –15% from the density of pure Fe –Ni alloy.
Issues related to these features can be discussed in view of the following premises.
The continental crust constitutes a small mass fraction of the total silicate portion of
the Earth (,0.5%). Therefore, its extraction from the mantle cannot significantly alter the
concentration of the major elements of the residual mantle. This is a consequence of a
Cosmochemistry and Properties of Light Element Compounds 135

simple mass-balance criterion. The compositional fields of major crustal rocktypes can be
expressed in terms of a pseudo-ternary SiO2 –Al2O3 –MnOn/2 system (see Fig. 11.1 where
M ¼ K, Na, Ca, Mg and Fe2þ). The phases appearing show bonding characterized by
NBO/T, where NBO is the non-bonding oxygen.
As regards (b) above, the pressure – temperature –density relationships of mantle
mineral assemblages are consistent with a compositionally uniform pyrolite model of the
entire mantle (Chopelas and Boehler, 1989). However, some claim that the lower mantle is
richer in iron and/or silicon with respect to the upper mantle. This is based on a comparison
of density or seismic velocity profiles with profiles derived from mineral physics
experiments. However, it is now known that the Earth’s mantle is anomalously depleted in
silicon.
The PRIMA is defined as the average mantle composition before the extraction of
the continental crust. Its chemistry needs to be uncontaminated by the complex chemical
machine of plate tectonics involving extraction and re-injection of partially melted
material as well as recycling of oceanic and continental crust.

1.20.1. Magma ocean generation and crustal fractionation in early Earth

Mantle differentiation on a global scale can hardly be established from the available
geochemical signature. The relatively chondritic proportions of the refractory lithophile
elements in the fertile peridotites and the primitive Nd and Hf isotopic compositions of the
early Archean komatiites from Barberton Mountain Land, South Africa (e.g., Grau et al.,
1990) do not endorse any early global differentiation. Indeed, no geochemical evidence for
global-scale differentiation of the mantle is presently evident.
However, any such evidence of differentiation in Hadean times might have been
subsequently erased by rapid whole-mantle convection (McDonough and Sun, 1995). It is
believed that a magma ocean existed at the beginning of Earth’s history (e.g., Abe, 1993).
Minerals crystallized from the magma ocean would have different compositions from the
host magma. Deviations in element partitioning from unity between minerals and co-
existing melt led to compositional evolution in the melt upon crystal segregation, creating
a heterogeneous distribution of elements in the mantle. This segregation would control the
spatial distribution of elements in the Earth’s interior.
The models of planetary accretion involve late occurrences of giant impacts (e.g.,
Wetherill, 1986), which may cause temperatures to increase to melt the whole or part of the
Earth (e.g., Benz and Cameron, 1990). This creates a magma ocean. At temperatures of
3,000 –4,000 K, the phase relationships of carbonaceous chondrite (Agee, 1993) suggest
that the Earth’s mantle would be completely molten up to about 25 GPa, the pressure
where the liquidus phase changes from majorite to silicate perovskite ^ magnesiowüstite.
(Note: For a chondritic mantle composition, silicate perovskite will make up .80% of the
lower-mantle mineralogy.)
Impacts perhaps generated a magma ocean which continued crystallizing at
extremely high temperatures of the dry solidus. The crystallization of a magma ocean is
controlled by melting relations and is limited by the mantle solidus, which can be
estimated from available experimental data on melting in simple and complex systems.
136 Chapter 1

The KLB-1 natural peridotite solidus is considered to be representative of the mantle

solidus for a peridotitic mantle. Olivine is a major constituent of the mantle and is the
liquidus mineral in peridotite mantle to depths of about 400 km (e.g., Zhang and Herzberg,
1994). This is succeeded in quantum by pyroxenes and garnets. The fractionation of the
garnet layer at high pressures produced a garnet layer in the transition zone. The data on
partitioning of elements between garnet and ultramafic liquid allow 10– 30% of garnet
fractionation (e.g., Herzberg and Gasparik, 1991), which places a limit on the depth of the
magma ocean to ,500 km.
Garnet produced early in the fractionation process has a higher pyrope content,
higher melting temperatures and is deposited at great depths. Garnet produced in the final
stages of fractionation approaches the composition and the melting temperature of the
enstatite – jadeite join. Melting experiments on natural peridotites and komatiites identified
the conditions for the origin of the partial melts, which produced komatiites in the
Archean. The origin of these komatiitic melts is at the intersection of the P=T trajectories
of adiabatically ascending mantle plumes and the dry solidus of a peridotitic mantle
(Gasparik, 1992).
The Archean geotherm is based on the melting of the oldest volcanic rocks such as
Mg-rich peridotitic komatiite of 3.3 Ga. The KLB-1 solidus indicates that volcanism
started immediately after the solidification of the magma ocean and that the solidification
was completed at 400 km depth.
In the final stages of solidification, crystallization of residual melts enriched in
Na, Fe and K incompatible elements and volatiles produced a pyroxene-rich layer at
300 –400 km depths (Gasparik, 1992).

1.20.1.1. Early crust

All the present ocean crust was created during the last 200 m.y. For the location
of earlier plate boundaries, one must search in the continental rocks. In some sites,
ancient plate boundaries have been recognized within the continents as geological
discontinuities or as deformed zones. Remnants of ancient deep oceans occur at
elevated heights as ophiolites trapped along sutures in the continents. An early, highly
mobile microplate tectonic phase progressed through a more stable, largely intra-
cratonic, ensialic, mobile belt phase to the modern microplate tectonics phase that
involves large, rigid lithospheric plates (Goodwine, 1981) with modern continents and
ocean basins. Historically, Precambrian continental crust bears 85% of the Earth’s
geological record.
The early tectonics processes involving heat must have been different from the
modern ones because the higher magnitude of heat and the lesser thickness of the lithosphere
have changed over this long period of time. Indeed, komatiites show higher melting
temperature than present-day lavas. The ophiolites and blueschists that are associated with
modern converging boundaries are absent from early Precambrian terranes.
The present-day plate tectonics perhaps emerged after a major thermal event that
produced extraordinary bodies of anorthosite, typically formed on the Moon followed by
the Earth and other planets. The tectonic style of a planet is closely related to its internal
thermal history.
Cosmochemistry and Properties of Light Element Compounds 137

Radiometric ages of the oldest samples of the Moon and meteorites are close to
4.6 b.y. but the oldest age of terrestrial rocks does not go beyond 3.8 b.y. Therefore, there
is a gap in knowledge of the rocks formed during the first 800 m.y. of the Earth’s history.
Possibly the dynamic processes in the early Earth have destroyed that history. For clues,
however, we can look to the neighbouring planets, which still bear the imprints of
geological records. Space missions have offered the insight that the terrestrial planets have
been heavily bombarded by cosmic objects during the period from 4.2 to 3.9 b.y. ago.
The mission results also indicate that the lithospheres of the Moon, Mars and
Mercury are continuous spherical shells, much in contrast to that of the Earth showing
broken lithospheric plates recycling into the interior. The tectonic features of the
lithosphere of Venus are yet less known. The lithosphere may be regarded as the thermal
boundary layer above the convecting cell in the Earth’s mantle.
In this context, it should be remembered that the tectonic style of a planetary body is
controlled by its thermal regime. Because temperature changes viscosity, which again
controls the drag in the convective cycle, the critical viscosity, or base of the lithosphere,
has to be at greater depth in a small planet than in a large one. Consequently, the tectonic
styles would differ from one planet to another. Thus, Mercury, Mars and the Moon, all
smaller than the Earth, should have a thicker lithosphere and a lower degree of melting.

Crustal evolution. The crust and lithosphere respond to a changing geothermal regime.
The modern oceanic crust, accounting for three-quarters of the surface area, is transitory
and is consumed by subduction to the mantle. The unconsumed sialic fraction accretes to
the growing continental masses.
The unidirectional development of the continental crust is caused by (a) the
geothermics of the Earth, (b) melt fractionation and cratonization and (c) oxygen level
increase in the atmosphere.
(a) Geothermics: The heat production in early earth about 4.5 b.y. ago was about
53 pW kg21, which declined to the present-day level of 11 pW kg21. This is measured
from the contents of the isotopes of K, U and Th.
(b) Crustal fractionation and cratonization: About 3 b.y. ago, the Archean crust probably
constituted less than 12% of the present-day continental crust. The cratonized area was
perhaps still smaller. During the late Archean and early Proterozoic periods (3.0 –
2.0 b.y. ago), there was an accelerated rate of crustal fractionation and cratonization.
The growth of stable cratons apparently crossed the threshold about 2.5 b.y. ago.
(c) Oxygen level increase: During the Proterozoic period, the proliferation of biological
growth and photosynthesis resulted in a dramatic increase in O2 level in the
atmosphere (10% of the present level about 700 –600 m.y. ago).
More than 3.8 b.y. ago, chemical evolution was the dominating process.
Thereafter, oxygenation and filling of O2 sinks in the form of banded oxide iron
formation initiated oxygen build up. The onset of red-bed formation and the O2
shielding of the intracellular process heralded the eukaryotic cells. Thus, the period
2.7 –2.0 b.y. ago markedly typifies a critical period in Precambrian crustal history of
the Earth.
138 Chapter 1

The Earth’s crustal landforms are dominantly modified by mountain-forming

orogenies. Orogenic evolution is thermotectonic and is episodic but it is not clearly known
why it is so episodic. This may have arisen from periodic changes in the convective system
within the mantle. A linear convergence of plates develops linear mountain belts
manifesting the thermotectonic events. Again, the thermally affected domains are intruded
by magmatic rocks but, strangely, remain tectonically undeformed. Cratons are bordered
by mobile belts; these developed principally between 2.8 and 2.5 b.y. ago, with the process
of stabilization involving massive granitoid intrusions. This orogenic process marks the
boundary episode of the Archean Proterozoic period. Some of the features of the major
well-identified peaks of orogenies are stated below:
1.6 b.y. ago (Hudsonian Orogeny);
1.2 – 0.9 b.y. ago (Grenvillian Orogeny);
0.8 –0.5 b.y. ago (Pan-African Orogeny): major crust-forming event, affected Gondwana-
land continents.

Precambrian rock types. Precambrian time is divided into three major periods: Archean
(.2.5 Ga), early Proterozoic (2.5 –1.5 Ga) and late Proterozoic (1.5 –0.6 Ga).
The principal Archean rock associations are granite-greenstone and high-grade
gneiss. The oldest rocks recorded are in West Greenland (.3.8 Ga), formed after the major
episode of asteroid impacting. The main rock types of this age are metamorphosed
granitoids, mafic intrusions and clastic and chemogenic sediments. Rocks as old as
3.5 –3.0 Ga are seen in S. India, S. Africa and Western Australia.

Greenstones. The greenstone belts (3.5 –1.8 Ga) are deformed elongated metavolcanics
and metasedimentaries accompanied by granites, etc. These belts are linear to curvilinear,
steeply inclined, plutons-intruded, volcanic-rich isoclinal keels. Major greenstone belts
belong to the age group of 2.7 Ga. The volcanics are dominantly tholeiitic, but include
komatiites (primitive Mg-rich mafic volcanics) and calc-alkalic (evolved volcanics).
Komatiites extruded at high temperature, ,1,7808C. Most komatiites extruded mainly
during 3.3 Ga. The absence of komatiites intrusion in the post-Archean period suggests a
marked change in thermal regime at the Archean– Proterozoic boundary. The tholeiitic
basalts make up the dominant greenstone component. Prevolcanic granites and
magmatized margins are common associates of greenstone belts.
In the high-grade metamorphic terrains of the Precambrian period, the commonest
rocks are generally granulite to upper amphibolite facies. But they contain the layered
igneous complex and the remains of early volcanics and sediments. The low-grade
greenstone belts commonly lie alongside the Archean high-grade terrains. These high-
grade terrains are regarded as vertically uplifted lithosphere, which was once deeply
buried.
By 3.0 Ga, epicratonic basins formed locally, manifesting the formation of stable
crust at this stage. The sedimentary basins in Swaziland and S. Africa offer such an
example.
Cosmochemistry and Properties of Light Element Compounds 139

During 2.8– 2.5 Ga, Archean orogeny culminated with major new crustal additions,
cratonization and stabilization of the crust developing the characteristics of the oncoming
Proterozoic.
The early Proterozoic (2.5 –1.5 Ga) crust featured thick sedimentary –volcanic
accumulation, rich uranium – gold place deposits, large banded iron formations (BIFs) and
major Pb – Zn– Cu deposits in shales. The other features included red-bed layered mafic
intrusions, diabase dyke swarms and nickel and platinum deposits (possibly associated
with impact metals). During this period, the stable cratons bordered by mobile belts
constituting super continents were moving as units.

87
Sr/ 86Sr ratio reflecting rock evolution. The ratio of 87Sr to 86Sr in carbonate rocks
offers the clue to understanding the sudden transition from a mantle-dominated crust to the
stabilization of continental shelves. In early Proterozoic time, recycling of continental and
mantle flux expressed an increase in 87Sr/86Sr ratios to ,0.708, which is much above the
mantle growth line. Similar increases are noted in the ratios of K2O/Na2O, La/Yb and the
abundances of rare-earth elements.
Low initial 87Sr/86Sr rations, seen in Archean volcanic rocks and granitoids, may be
regarded as originating from massive additions to the crust of mantle-derived material.
Ophiolites and blueschist (high-pressure) rocks are absent in Archean terrains. The near
absence also of high-pressure kyanite and eclogite in Archean rock suggested a hotter
(steep thermal gradient), thinner lithosphere model lacking subduction (Benioff) zones.
Rifting and sagging led to the partial melting of the old crust, leading to copious
tonalite – granite intrusions. This is reflected in the low 87Sr/86Sr ratios.

1.21. Fluids within the Earth

1.21.1. Water in the Earth

If H is considered to be the dominant light alloying component in the Earth’s core

(e.g., Badding et al., 1992), then as much as ,3% of the planet’s mass is elemental
hydrogen. This corresponds to a hydrogen content 103 times that present in the Earth’s
hydrosphere. The Earth’s primordial water content was governed by (a) the chemistry,
size and temporal distribution of accreting objects, (b) the efficiency with which the
early planet lost water through impact-induced effects or hydrogen escape, (c) the
efficiency of the early volcanism (or magma ocean circulation) in degassing water
from the interior of the planet and (d) the degree of chemical interaction between iron-rich
and hydrated material during planetary accretion and core formation (Williams and
Hemley, 2001).
The materials accreting the Earth spanned from iron-rich proto-planetary cores to
icy cometary objects. The primary sources of water in the proto-Earth were chondrites
(carbonaceous) and comets. Primitive carbonaceous chondrites (C1-type) contain ,10%
water. Comets are ,30 –60% water. Ordinary condrites contain ,0.2% water. (Note:
Earth’s hydrosphere contains water only ,0.02% of the mass of the Earth.) The D/H ratios
140 Chapter 1

of comets do not, however, support any concept for derivation of the hydrosphere from
cometary impacts. During the accretion process, volatilization of gases was produced by
pervasive melting and vaporization accompanying a giant (lunar-forming) impact
(e.g., Wetherill, 1990). Simulations indicate that ,30% of Earth’s radius could be
vaporized during this impacting event (Comeron and Benz, 1991).
The solubility of water in silicate melts increases from near zero at atmosphere
pressure 0 . 10 wt% near 1 GPa (; 30 km depth) and the solubility of H2O in mantle
melts at depths corresponding to 30 –42 GPa is also this high (Sumita and Inoue, 1996).
The solubility of water in silicate melt is effectively infinite and water initially dissolved
within a terrestrial magma ocean will likely be retained at depth in the solidifying planet.
The retention of primordial water at deep Earth is indicated by the 3He degassing
from ridges and hotspots. This non-radiogenic isotope continues to be degassed even now
at the rate of about 1.5 kg yr21 (Jambon, 1994).
Interactions of water with mantle minerals produce major topographic features
such as the reform of continents by hydrous melting (Campbell and Taylor, 1983) and
the formation of mid-oceanic ridges through dehydration-induced strengthening
of oceanic upper mantle by partial melting (Hirth and Kohlstedt, 1996). Degassing of
the Earth’s solid mantle was the prime mechanism for the development of the
hydrosphere.
Water played a significant role in generating voluminous mantle melting and in
the genesis of continental flood basalts and the komatiites, a major ultramafic rock type of
Archean (e.g., Parman et al., 1997). Even a small amount of water (,0.4%) can play a
pivotal role in the generation of voluminous continental flood basalts (Turner and
Hawkesworth, 1995). Hence, a small degree of hydration of the Earth materials can
produce phenomena of considerable geodynamic and petrologic significance.
The presence of significant amounts of water is evidenced in the polar caps and in
sub-surface channels in Mars (Malin and Edgett, 2000; Zuber et al., 2000). Large
quantities of hydrous phases are surmised to be present on Europa (McCord et al., 1999).
The different reservoirs for water on the Earth have been discussed by Williams and
Hamley (2001).
Subduction of oceanic lithosphere into the mantle is the major process for sinking
water to deep depths. Peacock (1990) infers that 8.7 £ 1011 kg of H2O is subducted per
year. From this, about 1.4 £ 1011 kg yr21 is returned through arc magmatism. If this rate is
valid, then 7.3 £ 1020 kg of water would be subducted every billion years, which means
half the amount of the current hydrosphere is subducted every billion years.
The continental lithosphere mantle is water-enriched. Deeply exhumed ultramafic
complexes are seen to contain high-pressure hydrous phases (Scambelluri et al., 1995)
while the mantle-derived mica and amphibole-bearing xenoliths are observed in
continental volcanics (e.g., Dyar et al., 1993). Water-rich fluid inclusions in diamonds
indicate the presence of deeper subcontinental metasomatic fluids (Schrauders and
Navon, 1994).
Near-primary basalt compositions, experiencing little fractional crystallization,
show water contents ranging from ,700 to ,6,000 ppm. This is in considerable contrast to
the basalt from back-arc basins wherein H2O content lies between 1.0 and 2.9 wt%
Cosmochemistry and Properties of Light Element Compounds 141

(Sobolev and Chaussidon, 1996). The studies in cosmogenic nuclide 10Be in arc magmas
offer evidence for these being carried to depths of at least 100 –150 km.

1.21.2. Water in the magmatic processes

Water plays a significant role in the magmatic process, particularly in the following
ways:
(a) Water participates in the generation and differentiation of magmas in various planetary
environments.
(b) Dissolved in mantle minerals, water lowers the melting temperature and the viscosity
of the rocks.
(c) Water in high abundance determines many of the physical properties of magmas and
eruption dynamics.
(d) Water in the hydrosphere of the Earth contributes to the continuance of active plate
tectonics.
The planetary water content may be diminished either by emanative loss to space
through chemical dissociation, etc., as from Mercury and Venus, or may be frozen to solid
ice trapped inside the crust, as on Mars. Earth’s surface ambience favours the stability of
water in the liquid state. Water from the surface of the Earth is carried into the interior by
plate subduction and serves as a formidable lubricant facilitating plate tectonics. At the
subduction zones, water-rich explosive volcanism is the direct evidence for the presence of
water in deep depths of the subducted plate. Some part of the subducted water comes out
through volcanism and the rest is carried to still deeper parts of the mantle to lubricate plate
tectonics.
Certain Hawaiian volcanoes (Koolau, Mauna Loa and Mauna Kea) are reported
(E. K. Haiiri, GRL Yb 98-99, p. 117) to contain components of recycled oceanic crust and
sediment that had been carried deep into the mantle during an ancient subduction episode.
These components had interacted with water near the Earth’s surface. The melt inclusions,
trapped in solid crystals formed at depths in magmas long before the eruption, show
moderate water contents and hydrogen isotope ratios. The H2O and Cl data on melt
inclusions reveal that many magmas are contaminated by seawater-derived components
prior to eruption, a process which changes their original Cl, H2O and H isotope
compositions.
Magmatic water in Mars: The surface morphology of Mars offers scope for
speculation of an ancient ocean but chemical analyses of Martian meteorites of igneous
origin offer evidence for low magmatic water content. However, a study on pyroxene
present in Shergotty meteorite — a basaltic achondritic mass ejected 175 m.y. ago from
Mars — revealed that water contents of pre-eruptive magma on Mars could have been up
to 1.8% (McSween et al., 2001). The presence of enriched soluble trace elements in the
inner cores of pyroxene (formed at depths) corresponding to the outer rims (formed near
the surface) implies, however, that water could have been present at depth but was largely
lost in pyroxene as it was carried upwards by magma ascent. The ascending magmas could
possibly have delivered significant quantities of water to the Martian surface in late recent
times to produce the erosional features observed in the surface morphology.
142 Chapter 1

1.21.3. Fluids in the lower crust: granulites

The rheological properties of the lower crust depend much on the fluids present in
it. Fluid inclusion and phase-equilibria studies of lower crustal granulites have shown that
fluids with low water activities, due to the presence of dissolved component such as CH4,
N2, CO, CO2 and chlorides, are present at least episodically in the lower crust.
Markl and Bucher (1998) have reported the occurrence of a solid-salt solution
(NaCl –KCl) found together with chlorine-rich amphibole and biotite in lower crustal
granulites. However, results of a number of studies have also led to suggestions that the
lower crust was devoid of a free fluid phase during most of its history.

1.21.4. Mantle fluids

The nature and the presence of fluid phases in the mantle have been known from the
studies of: (i) fluid inclusion in a host of xenoliths (e.g., Schrauder and Navon, 1994),
metasomatic alteration products (e.g., Dyar et al., 1993) and (ii) seismic evidence for
partially molten zones in the 300 –400 km depth range in the upper mantle (e.g.,
Revenaugh and Sipkin, 1994).
The volatiles such as H2O and CO2 significantly affect the physical properties of the
mantle and the melting processes. Modelling the geochemical cycles for volatiles requires
consideration of the mantle part of the cycle because subduction processes potentially
return volatiles to the mantle and also because the mantle is degassing at divergent plate
boundaries and at intra-plate settings. The potential stability of hydrous and carbonate
minerals in the mantle has obvious implications for this modelling and this leads us to the
question of the temporal evolution of the surface hydrosphere (Luth, 1997).
Hydrous silicate melts within the mantle serve as the major reservoir of water. Such
melts in the deep upper mantle remain neutrally buoyant (Rigden et al., 1998) and may
assume high densities (Tyburczy et al., 1991). Hence, these may have an infinite residence
time in great depths of the upper mantle, the transition zone and even the lower mantle.
Hydrated silicate melts and hydrous fluids may be fully miscible in the upper-mantle
conditions (Shen and Keppler, 1997).
In the sub-continental mantle above dehydrating subducting slabs, free water can
occur at P – T above the stability field of non-alkalic amphiboles (.,3 GPa ,90 km
depth) (e.g., Watson et al., 1990). In the shallow upper mantle, if free water is present, there
would be a significant magnetotelluric signature. The electrical conductivity of water
increases dramatically at the mantle P – T condition, consistent with the presence of proton
carriers (Cavazzoni et al., 1999; Chau et al., 1999).
Considerable debate and controversy continue over the extent of the solubility of
water in the mantle and the kindred of chemical species that are created (e.g., Bell and
Rossman, 1992a; Bai and Kohlstedt, 1993; Young et al., 1993). Water can have substantial
effects on the mechanical properties of minerals (e.g., Mackwell et al., 1985) and may also
influence the melting behaviour. Thus, any water-rich phases could strongly affect the
rheological behaviour of the upper mantle. The Earth’s upper mantle is believed to be
Cosmochemistry and Properties of Light Element Compounds 143

composed mainly of (Mg, Fe)-rich silicates, most of which contain small amounts of water
as structurally bound OH (Bell and Rossman, 1992b).
In the uppermost mantle, the dominant mineral, (Mg, Fe)2SiO4 olivine, contains
only small amounts of water (3,975H/106Si) (Miller et al., 1987). The most water-rich
olivines are found to be those with a high Fe content. At greater depths, 40 –550 km,
olivine transforms to the b-phase of (Mg, Fe)2SiO4 with the wadsleyite structure
(see Section 6.4), which has been shown to contain much larger amounts of water
(10,000– 65,000H/106Si) (Young et al., 1993). There are a host of OH-bearing phases in
the upper mantle which are discussed in detail in Chapter 13.
Cratonic peridotites can form as residues of partial melting with variable H2O
contents at depths of about 200 km (Ohtani et al., 1996). Hydrogen can occur in diamonds
as fluid inclusions of H2O and also as structural defects (OH defects). Ices VI and VII can
be present as inclusions at confining pressures of ,2 GPa (Kagi et al., 2000). Likewise,
pressure-solidified CO2 has been reported from natural diamond (Schrauder and Navon,
1993). Such fluid inclusions in diamonds provide prima facie evidence for the presence of
hydrous fluids in regions of pressure greater than ,5 GPa (,150 km depth).
In the sub-continental mantle above dehydrating subducting slabs, free water can
occur at P – T above the stability field of non-alkalic amphiboles (.,3 GPa ,90 km
depth) (e.g., Watson et al., 1990). In the shallow upper mantle, if free water is present, there
would be a significant magnetotelluric signature. The electrical conductivity of water
increases dramatically at the mantle P – T condition, consistent with the presence of proton
carriers (Cavazzoni et al., 1999; Chau et al., 1999).

1.21.5. Atmospheric noble gases in mantle melts

The Earth’s atmosphere is presumed to have been derived from the emission of
gases from the mantle through partial melting. In such artial melting, noble gases are also
expelled from the solid material and dissolved into the resulting melt because of their large
atomic size and chemical inertness. Mantle degassing is invariably accompanied by
degassing of noble gases, which serve as excellent tracers not only for the study of mantle
degassing but of the evolution of the atmosphere as well. Noble gases are known to be
accommodated in “holes” in silicate melts. Pressure reduces the sizes of holes and thus
decreases the solubility of noble gases in the melts.
Measurements of the isotopes of the noble gases should be able to constrain models
of the Earth’s birth and evolution. Solar-type noble gases in the mantle (Honda et al., 1991)
have been evidenced from many sources. Indeed, the 3He/4He ratio in MORB (e.g., Sarda
et al., 1998) and oceanic island basalts is one order of magnitude higher than that of the
Earth’s atmosphere.
Long residence time and ineffective mixing may explain the survival of ,2-b.y.-old
chemical and isotopic (e.g., He, Ne, 40Ar, etc.) heterogeneity. (Note: Whole-mantle
convection would cause efficient mixing while, in layered convection, a thermal boundary
layer would separate the upper and lower mantles. The layer would effect the chemical and
isotopic isolation of the lower mantle.)
144 Chapter 1

1.21.5.1. Inert gases: solar-like?

Samples from the Earth’s mantle show neon isotopes of a solar-like composition
(e.g., Pepin, 1998). The presence of solar-like argon in the mantle should yield lower-than-
air 38Ar/36Ar ratios. Solar-like Ne isotopes have been reported in MORB (e.g., Sarda et al.,
1998) and oceanic island basalts (e.g., Craig and Lupton, 1976). Pure solar neon has the
ratio 20Ne/23Ne ¼ 13.8.
Laser measurements of 4He and 36Ar in a volatile-rich MORB glass have been
undertaken and a solar-like elemental abundance pattern for all rare-gas species in the
mantle has been inferred.

1.21.5.2. Ar, Kr and Xe

However, later workers (e.g., Moreira et al., 1998) showed that there is no solar
argon in the MORB source or, rather, solar argon represents only a negligible constituent
of the upper-mantle non-radiogenic argon budget (Kunz, 1999). It could be that most of the
36
Ar and 38Ar present in the mantle may have been re-injected from the atmospheric
reservoir into the mantle, possibly by subduction.
Xe is depleted by a factor of 20 in the atmosphere of Earth and Mars relative to the
other rare gases Ne, Ar and Kr; this is known as the “missing xenon problem”. This Xe
depletion in the atmosphere may be due to its incorporation in the rocks and minerals
(possibly under pressure). In contrast to rocks, the polar ice is not a good sink for Xe
on Earth.
Jephcoat (1998) reports that heavier noble gases (Ar, Kr and Xe) become solid in
the mantle and these could be stored in the deeper mantle or in the Earth’s iron core
without degassing. These gases are also seen to be solidified at room temperature at high
pressures and their solid forms have been shown to melt at high temperatures (several
thousands degree kelvin) as pressure is raised (e.g., Jephcoat and Besedin, 1996). Jephcoat
found that both the melting curves and the densities for xenon and krypton are well above
the estimated temperature and density of the lower mantle. Thus, Ar, Kr and Xe should be
mostly in solid form in the mantle. The solid Kr and Xe have much higher densities than
silicates throughout the whole mantle and the calculated density of Xe even substantially
exceeds that of iron in the Earth’s core. Jephcoat (1998) suggests that the solid xenon
particulates would segregate in the Earth’s deeper regions, leaving the mantle depleted in
xenon and, possibly, in part, in krypton and argon as well.

1.21.5.3. Ar solubility
In their studies on noble gas geochemistry, Chamorro Perez et al. (1998) observed
that argon solubility in silicate melts first increases almost linearly with pressure
(following Henry’s law) up to ,5 GPa (;150 km); beyond this pressure, the solubility
abruptly decreases by more than an order of magnitude (Fig. 2 in Chamorro Perez et al.,
1998, p. 353). Below 150 km, argon in the solid mantle is not released into the magma.
Therefore, most of the mantle at this depth remains “undegassed”. From this depth (i.e.,
150 km), the solubility of Ar decreases sharply. The concentration of argon in mantle-
derived materials (such as volcanic modules or rocks) is generally less than 1 ppb.
Cosmochemistry and Properties of Light Element Compounds 145

Seismologists are prone to believe that degassed mantle corresponds to the

seismologically identified upper mantle above 670 km. Thus, we get an example wherein
the seismological layering in the mantle (i.e., 670 km) conflicts with geochemical layering
(occurring at 150 km).

1.21.5.4. The “missing xenon problem”

Relative to other noble gases, xenon is found to be depleted in the Earth’s mantle,
the mantle concentration being extremely low (,1 part per trillion (ppt), 10212) (Ozima,
1998). An explanation can be found in the results of Jephcoat’s (1998) investigation, which
shows that, in most of the mantle, xenon (in particular) is in the solid form, as are krypton
and argon (but at respectively greater depths). The density of xenon is higher than that of
the surrounding mantle, and even that of the core. In consequence, the solid xenon
particles, in particular, are driven deeper into the mantle. This process can account for the
“missing xenon problem” — the depletion of xenon relative to other noble gases in the
Earth in comparison with its relative abundance in meteorites.
Atmosphere-like compositions have been reported for non-radiogenic isotopes
(e.g., Ar, Kr and Xe) from MORB and oceanic island basalts. No clear excess 129Xe or
131 – 136
Xe has been identified. [Note: Excess 129Xe is the decay product of the extinct
nuclide 129I (half-life, 16 m.y.) and excess 131 – 136Xe is the fission product of both 244Pu
(half-life, 83 m.y.) and 238U (half-life, 4.47 b.y).]
Recently, however, many samples from Hawaii and Iceland have offered clear
evidence of excess 129Xe and 131 – 136Xe (Trieloff et al., 2000). All these evidences suggest
that the sources of these oceanic island basalts had atmosphere-like components of heavier
noble gases.
The proposal of Trieloff et al. (2000) implies that atmospheric noble-gas
components had been incorporated into the mantle by subduction into the two mantle
reservoirs represented by MORB and OIB. However, the mantle Ne might have been the
product of Ne – B, which is suggested to be a pristine component in the mantle.

1.22. Potassium budget in the Earth’s mantle

1.22.1. K-feldspar/Hollandite

In the Earth’s mantle, potassium is known to be hosted by a number of phases,

dominated by K2Si4O9 wadeite, KAlSi3O8 hollandite and phlogopite. The melting
temperatures of the phases are lower than the solidus temperature of the mantle peridotite.
The presence of these phases may cause the attenuation of seismic waves.
K-feldspar is a stable host for K only up to ,7 GPa and 1,5008C (Yagi et al.,
1994) and it is only rarely found in eclogitic xenoliths and diamond inclusions
(Schulze and Helmstaedt, 1998). The higher P-phases, K2Si4O9 wadeite and KAlSi3O8
hollandite, are stable at high P but have never been recovered in natural samples.
146 Chapter 1

1.22.2. Phlogopite

Phlogopite, because of its occurrence in mantle-derived xenoliths (cf. Dawson,

1980) may serve as a potential host for K and H2O in the Earth’s upper mantle. Most
models of undepleted mantle lherzolite contain ,0.03 wt% K2O (Allègre et al., 1995).
Luth (1997) reported the results of studies of the phlogopite diopside system under
pressures up to 17 GPa. In this system, phlogopite becomes unstable with increasing
pressure, breaking down to K-richterite, which again breaks down to phase X (a K-bearing
hydrous phase), such that a K-rich phase co-exists with clinopyroxene to 17 GPa.
Clinopyroxenes contain #1.3 wt% K2O in the following assemblages at different
pressures:
(i) at 3 –5 GPa: phlogopite þ clinopyroxene ^ olivine ^ liquid
(ii) at 7 –9 GPa: phlogopite þ clinopyroxene ^ olivine ^ liquid þ garnet
(iii) at 11 GPa: clinopyroxene þ olivine ^ liquid þ garnet ^ K-richterite
(iv) at 14 –17 GPa: clinopyroxene þ olivine þ garnet þ phase X
In these assemblages, K is partitioned into hydrous phases or liquid rather than into
clinopyroxene. By inference, phlogopite (or its higher-pressure breakdown products) may
serve as the primary host for K in the mantle, if H2O is present, and any co-existing
clinopyroxene should have a very low concentration of K (Luth, 1997).

1.22.3. Clinopyroxene

Pyroxene may be stable down to ,400– 500 km and may carry a substantial amount
of K. K could have been flushed outwards with “basaltic” liquid during early
differentiation of the Earth. However, a significant quantity of K may enter a S-rich
core at high pressure (Goettel, 1974).
On the basis of both natural and experimental studies, Luth (1997) suggested that
the major host for potassium in the Earth’s mantle is clinopyroxene, which is also the host
for lead.
Some experimental studies have produced K-rich clinopyroxene at high pressure
(e.g., Luth, 1992; Edgar and Vukadinovic, 1993). With increasing pressure, a large ion like
Kþ or Naþ will behave more like relatively smaller ions Ca2þ or Mg2þ. A smaller Naþ
at high pressure would provide a mechanism for size averaging of Kþ to fit into M2
of diopside.
K-rich clinopyroxene, found as inclusions in diamonds, results from high pressure
and high K activity. Clinopyroxene may thus be the fundamental solid host for K in most of
the upper mantle and should be a good source for K activity in the mantle (Harlow, 1997).

1.22.4. K2O in mantle solidus: seismic attenuation

Much like H2O and CO2, the content of K2O largely reduces the dry solidus
temperature of peridotite up to 4 GPa (e.g., Takahashi and Kushiro, 1993). Except for
clinopyroxene, wadeite-type K2Si4O9 and hollandite-type KAlSi3O8 are the most likely
candidates for K-accommodating phases above 4 GPa. (see section 7.2.2).
Cosmochemistry and Properties of Light Element Compounds 147

The melting temperatures of these phases, especially wadeite-type K2Si4O9, are

lower than the solidus temperature of dry peridotite (e.g., Takahashi, 1986) by more than
2008C. Thus, the mantle solidus may be depressed below the melting point of these
potassic aluminosilicate phases under dry conditions and it may be possible to generate a
small degree (,1022 – 1023) of partial melting in the hot region such as upwelling plume
in the upper mantle. Such a small degree of partial melting produces attenuation of seismic
waves — more so when the interfacial energy between the solid – melt contact is low. This
concept may help in detecting the plume in the mantle seismic tomography.
For discussion on K in the core, see Section 1.30.

1.23. Mantle isotopes

A modern trend in igneous geology is to integrate classical petrology with trace-

element and isotopic systematics. This combined approach helps in understanding the
chemical evolution and differentiation of the Earth and other terrestrial planets. The bases
involved are as follows:
First, the determination of the age of the rocks is carried out using parent – daughter
element pairs, “blocked” in various mineral phase or even in glasses.
Second, light element stable isotopes, such as of H, O, C, S, etc., show the effect of
fractionation persistently present in magmatic conditions. This behaviour is widely
employed in geothermometry and volatile loss studies.
Third, isotopes heavier than Z . 20 are not fractionated at high temperature and,
hence, heavy isotopes are used as tracers for genetic affinities, source materials, etc.
Commonly, the long-lived radioactive decay series is used as tracers.
Measurement of C, H and O isotope ratios of co-existing minerals provide insights
into protoliths, geothermometry, fluid sources and fluid –rock interactions.

1.23.1. Sm – Nd ratios

The isotopic composition of the PRIMA is normally estimated from the Nd value of
meteorites and Nd – Sm isotopic correlation. It is presumed that 4.55 Ga ago, when the
Earth’s mantle was first formed, it had the 147Sm/143Nd ratio equivalent to the Solar-
System average, which effectively corresponds to that in carbonaceous chondrite.
In the Earth, partial melting of the mantle shifts this ratio by a slight increase in the
concentration of Sm in the melt compared with that in the residual solid. With the passage
of time, the 143Nd/144Nd ratio in the residual mantle tends to change faster than in the
chondritic reference value. (Note: 144Nd is the non-radiogenic reference isotope of Nd.) In
the melt, the reverse is observed. Sm –Nd isotope systematics of the ocean-ridge basalts
have revealed that the upper mantle beneath them underwent melt extraction more than
once in earlier times (prior to 1.5 Ga). The complementary melt fraction resides in the
continental crust, although modified by later processes of intra-crustal activity.
148 Chapter 1

1.23.2. Eu anomaly

In plagioclase, the selective substitution of Ca2þ by Eu2þ results in depletion of

divalent europium while the concentrations of trivalent europium and other rare-earth
elements remain unchanged in the melt from which plagioclase is crystallized. The
evidence of Eu depletion in lunar mare basalts offers the clue that the lunar mantle
experienced an early Eu depletion by its sequestration in the plagioclase-rich rocks,
forming the lunar highland terrains.

1.23.3. Sr, Nd and Hf

The mantle regions depleted in incompatible elements have low Rb/Sr ratios and,
therefore, low 87Sr/86Sr ratios coupled with high 143Nd/144Nd and 176Hf/177Hf ratios. On
the other hand, the continental crust, enriched in incompatible elements, shows the
opposite isotopic signatures.

1.23.4. Osmium

It is seen that the PRIMA value of 187Os/188Os (or 187Os/186Os) is identical to the
meteorite value (Meisel et al., 1996). This implies that the Earth’s mantle has the same
Re/Os ratio as in meteorites. This evidence would enable one to surmise a high degree of
meteorites to have fallen after the main process of accretion and core formation of
the Earth.
Basalts have much higher (up to a factor of 100 or more) Re/Os ratios than their
mantle sources so the growth of 187Os/188Os in the oceanic crust is quite rapid. Therefore,
Os isotopes should act as excellent tracers for the recycled basaltic material in the mantle
(Fig. 1.36; read the caption).
The typical values of 187Os/188Os for mantle peridotites are ,0.13.
The mafic shallow upper-mantle xenoliths obtained in central Arizona show high
187
Os/188Os values ranging between 0.216 and 1.51. Combined Os and Sr, Nd, Pb data
indicate that these xenoliths were formed from recent interaction between old mafic
materials in the lithosphere and high 87Sr/86Sr, low 143Nd/144Nd melts of proterozoic
lithospheric mantle.
The data from central Arizona (Esperanca et al., 1997) demonstrate the utility of the
Re –Os system in defining mantle separation ages. The discussion on Re –Os is continued
in the following section.
187 187
1.23.5. Re and Os

The rhenium isotope 187Re (1/2 life, 41.6 b.y.) decays to 187Os and the platinum
isotope 190Pt (1/2 life, 449 b.y.) to 187Os. Since crust is derived from melting of mantel, it
will have high Re/Os and Pt/Os ratios. Similarly, crystallization of the inner core will leave
the outer core with high Re/Os and Pt/Os. Osmium is present in the mantel only at ppb
level but is highly concentrated in the Fe – Ni– S mineral phase and Os –metal alloys.
Cosmochemistry and Properties of Light Element Compounds 149

Figure 1.36. The high Re/Os in the subducting slab produces high 187Os/188Os over ,108 yr through slab
subduction and recycling the oceanic island basalt (OIB). The subduction of slab composed of MORB and
sediments introduces inhomogeneities, stirred by hot mantle convection, when these become buoyant and produce
OIB. This recycling process is thought to be typically ,1 b.y. By the time the plume melts to produce OIB it has
“aged” isotopically and has higher 187Os/188Os than the surrounding mantle (Halliday, 1999, q 1999 Nature).

Because of the strong affinity of Os for such phases, diffusion is inoperative in mixing Os
isotopes over a distance in the mantle. This is in contrast to Rb – Sr and Sm –Nd
radiometric systems which are usefully employed in the study of the melting history of the
Earth’s mantle.
In the mantle regions from which partial melts have been extracted, the 187Os/188Os
ratios are low. High 187Os/188Os ratios have been reported in lavas from Siberia and
Hawaii (Brandom et al., 1998) and also in oceanic basalts. Meibom and Frei (2002) used
these isotope data to derive the partition behaviour of Re, Os and Pt during core formation,
which was underway within 250 m.y. of Earth’s formation. Early inner-core formation
implies rapid initial cooling of the Earth (Carlson, 2002).

1.23.6. U – Pb and Re – Os ratio

Long-lived radioactive isotopes such as 238U and 187Re produce changes in

the isotopic compositions of their daughter elements (Pb and Os) in the mantle.
These parent/daughter ratios are altered by processes such as melting and low-temperature
interaction with water.
Subducted slabs of altered basalt and surface sediments have U/Pb and Re/Os
values that differ from those in the mantle that produced the original MORB material. The
Pb isotopic compositions of OIB are highly variable and consistent with time-integrated
high U/Pb, confirming the basic model that OIB is derived from old, recycled slabs
150 Chapter 1

(Hofmann and White, 1982). By the time the plume melts to produce OIB, it has “aged”
isotopically and has higher 187Os/188Os than the surrounding mantle.
High Re/Os in the subducting slab produces high 187Os/188Os over ,109 years.

1.23.7. Isotopes in MORB and hotspots

The ultimate cause of mid-plate “hotspot” volcanoes like Hawaii is a narrow plume
of hot mantle rising from the boundary between core and mantle. However, the Hawaiian
volcano seems to contain a small component of material derived directly from the Earth’s
metallic core.
Chemically, the hotspot volcanic lavas are distinct from those of oceanic ridges.
This difference arises because the hotspot magmas are contributed to by the melting of the
subducted crustal plates at the convergent plate boundaries. The crust shows much higher
abundances of samarium and its daughter neodymium. Consequently, the hotspot basalts
contributed by molten slabs show higher contents of Sm and Nd compared with MORB.
[Note: The MORB volcanics are presumed to be produced by ,10% melting of peridotite
mantle (e.g., Kinzler and Grove, 1992).]
As the continental crust is highly enriched in incompatible elements like K, U, Th,
etc., the mantle is depleted in these elements. The crust and mantle are assumed to be
formed by differentiation of bulk silicate earth (Hofmann, 1988). But the enriched crust is
insufficient to balance the depleted peridotite mantle to account for the whole mantle (see
Table 1, Helffrich and Wood, 2001).
Enriched OIB contain isotopic evidence of contributions from old subducted
oceanic crust (HIMU) and from continental crust and lithosphere (EM-1, EM-2) (see also
Section 1.26.1.2).
The current rate of subduction of 3 km2 yr21 means ,18 km3 of oceanic crust is
being returned to the mantle each year. Assuming a constant rate of subduction for 4 Gyr,
one can calculate that the mantle contains 5% recycled oceanic crust, 45% recycled
“sterile” mantle and ,0.3% recycled continental material. However, when a higher (,3.6
times) heat production is assumed at 4 Gyr ago, then the entire mantle may be seen to have
been through the cycle of partial melting, oceanic crust formation and sediment recycling.
From this perspective, the mantle of today would be a heterogeneous mixture of
“sterile”, highly depleted mantle containing patches of oceanic and continental crust of
different ages.
Modern MORB comes from a “marble-cake” mixture of sterile mantle and recycled
oceanic crust. Large peridotite bodies tectonically emplaced in the continental crust still
exhibit a wide range of metre- and centimetre-scale heterogeneities.
Nearly all Indian Ocean MORB have distinctly high 208Pb/204Pb, 87Sr/86Sr and
208 p 206
Pb / Pb values relative to Pacific and Atlantic MORB. This suggests a mapable
isotopic “flavour” of the mantle along some regions, namely, south of the equator.
Hotspots, in general, contain only a limited range of isotopic compositions but most have
distinctive isotopic “flavours”.
The silicate crust and the mantle are low in such isotopes as those of osmium,
rhenium and platinum. These isotopes strongly concentrate into the iron metal of the core
Cosmochemistry and Properties of Light Element Compounds 151

rather than in the silicate minerals. Iron meteorites also show a high concentration of
these elements.
187
Re decays to 187Os (1/2 life ¼ 42 b.y.), and 190Pt decays to 186Os (1/2 life ¼
450 b.y.). Hotspot lavas have higher relative abundances of 187Os than does the upper
mantle (Hauri and Hart, 1993). By sufficiently precise measurements, Bardon et al. (1998)
showed that the 187Os excesses in hotspot lavas are accompanied by correlated excesses in
186
Os. The observed covariance of 186Os and 187Os in hotspot lavas is postulated to arise
from mantle to which #1% core material was added. Lead, like osmium, is concentrated in
the crust (compared with the mantle). The hotspot lavas tend to show non-radiogenic lead
isotopic variations which suggest that core input to hotspot sources must be very much less
than 0.3% (Widom and Shirey, 1996). Thus, it is also very possible that hotspot volcanoes
reflect the direct transfer of heat from core to surface (Bardon et al., 1998).
The mantle is presumed to contain the following isotopically extreme mantle
components, which contribute to the usually mixed sources of oceanic basalts (Zindler and
Hart, 1986):
(1) HIMU has the highest 238U/204Pb ratios and the lowest 87Sr/86Sr of any OIB, almost as
low as MORB Examples: St. Helena, Azores, etc.
(2) EM-1 (“enriched mantle 1”). Examples: Tristan and Pitcairn hotspots.
(3) EM-2 (“enriched mantle 2”) has the highest 87Sr/86Sr, and relatively high 207Pb/204Pb
ratios. Examples: Samoa and Societies hotspots.
Depleted MORB mantle (DMM) is represented by depleted MORB samples with
the highest 143Nd/144Nd and the lowest 87Sr/86Sr, 206Pb/204Pb, 207Pb/204Pb and 208Pb/204Pb
ratios.
The 3He/4He ratio in all parts of the Earth is decreasing with time owing to the
production of 4He from U and Th. Very old subducted mantle, which underwent partial
melting in the distant past, would have had a relatively high 3He/U ratio and a high
3
He/4He ratio through Geologic time because of the low rate of 4He production.
The Cambrian continental collision zone in northern Kazakhstan shows the
presence of diamonds in a matrix of metamorphosed sediments, which were subducted to
depths of 125 km or more and returned to the surface without losing their isotopic
characteristics.

1.23.8. Isotopes in UHP rocks

The products of ultra-high pressure metamorphism help us understand the physical

and chemical processes at convergent margins of the colliding continental plates. In these
processes, the surface rocks are subducted down into the upper mantle and returned to the
surface. In UHP terrains, the continental crust retains distinctive mineralogical, chemical
and isotopic properties to depths of at least 100 km. Stable isotope geochemistry gives
unequivocal evidence that the protoliths of UHP metamorphic rocks were at the Earth’s
surface prior to continental collision and subduction. Measurements of the isotope ratios of
oxygen, hydrogen and carbon in co-existing minerals help answer important questions
relating to UHP such as: (a) To what extent were UHP rocks subducted and exhumed as
152 Chapter 1

coherent structural units? (b) Were the protoliths of UHP rocks ever near the Earth’s
surface? and (c) What was the interaction between the fluids of subducted crustal rocks and
the upper mantle?
During Triassic continental collision and subduction, the marbles in Dabie-Sulu
experienced coesite – eclogite facies metamorphism at ,7008C/3 GPa. The impure
marbles reflect that they are surface rocks, e.g., continental platform passive margin
sediments, that have been subducted into the upper mantle to depths of 80 km or more and
subsequently exhumed.

1.23.8.1. 18O isotopes: non-equilibration

Oxygen enriched in 18O is produced by low-temperature interactions between a
rock and liquid water. Unusually low values of d18O and dD in eclogites and other Chinese
UHP rocks and also in Neoproterozoic geothermal areas in which cold-climate meteoric
waters altered basalts, granites and associated sediments have been recorded. Data from
UHP terrains in China, Italy and Norway show that dissimilar rocks failed to equilibrate
isotopically on a centimetre to metre scale. Residence in the upper mantle had no
discernible metasomatic effect on the stable isotope composition of crustal rocks
subducted during continental collision (Rumble, 1998). The discussion continues in
Section 11.6.3.4 with respect to 18O distribution (see also Fig.1.37)
The Triassic collisional terrain in eastern China manifests an interesting cold-
climate signature inferred from depleted 18O and deuterium isotopes. 18O/16O analyses and
zircon studies reveal that a cold climate existed 750 m.y. ago, coincident with glacial
deposits for the “Snowball Earth” era.
In marbles, the isotopic composition ranges are:

d18 OVSMOW ¼ þ7 to þ 25 and d13 CVPDB ¼ 0 to þ6

High d18O values (þ25) reflect pristine protoliths, while the low 18O values
reflect contamination by meteoric water during diagenesis and dolomitization
(Rumble et al., 2000).

1.23.8.2. d18O: an example for isotope separation

Ultra-high pressure (UHP) metamorphism (.2.5 GPa) corresponding to depth
limits of 80 – 90 km are normally manifested with high d18O. All the published data either
cluster around the mantle d18O value (þ5 to 6%), or are close to the oxygen isotope
composition of the country rocks (e.g., Sharp et al., 1993). For example, UHP eclogites
from the Western Gneiss region of Norway exhibit d18O values ranging from þ4 to þ8%
(Agrinier et al., 1985). The minerals in the UHP white schists from Dora Maira massif,
western Alps, show oxygen isotope composition in the range from þ4.5 to þ8.6% (Sharp
et al., 1993).
However, unusually low d18O has been reported from the UHP metamorphic rocks
from Sulu Terrain, Eastern China (see Fig. 1.37). The eclogite and quartzite interlayers are
found to have negative oxygen isotopes compositions. In the mineral separates, the d18O
values range from 210.4 to 27.3%, which are the lowest values reported from
Cosmochemistry and Properties of Light Element Compounds 153

Figure 1.37. Variations of d18O-isotope data for VHP minerals from the Sulu-Dabie terrane and other typical
crustal metamorphic rocks and various waters (modified after Rumble, 1998).

metamorphic rocks (Yui et al., 1995). Possibly these rocks acquired such low oxygen
isotope compositions before they underwent UHP metamorphism. A hydrothermal system
involving meteoric waters can be responsible for the 18O-depletion processes. The parent
rocks (pre-UHP) such as eclogite and quartzite were subducted to a great depth during
continent – continent collision, forming UHP minerals with concomitant local oxygen
isotope redistribution. The negative oxygen isotope compositions also imply an extremely
limited isotopic exchange between the subducted rocks and their surroundings during the
UHP metamorphism.
The UHP rocks perhaps resulted from collision of the Sino-Korean and Yangtze
cratons during 200 – 240 Ma (Eide et al., 1994) while the protoliths of these rocks are
154 Chapter 1

Precambrian (Li et al., 1993). The P – T conditions for these UHP rocks were estimated to
have ranges: P ¼ 2:0 – 3:5 GPa and T ¼ 550 – 8908C (Enami et al., 1993).
The eclogite mainly consists of garnet, omphacite and phengite (with minor rutile,
kyanite, epidote, quartz and talc); the quartzite contains quartz, phengite, kyanite, garnet,
opidote and rutile. Both rocks are foliated and phengite mainly develops along the foliation.
Porphyroblastic epidotes with inclusions of coesite pseudomorphs cut the foliation,
suggesting that UHP deformation occurred before the growth of epidote in the coesite
stability field. Stability of the coesite/diamond-bearing UHP rocks require abnormally low
geothermal gradients (,78C/km), which can be attained only by subduction of a cold
oceanic crust with pelagic sediments or old continental crust (Schreyer, 1988).
Crystal rocks range widely in d18O values but are generally greater than zero.
Mantle rocks are nearly homogeneous at þ5 to þ6% d18O. Ocean water is zero in both
d18O and dD. Meteoric waters, such as rain, ground water, snow and ice, have negative
d18O and dD values (see Fig. 1.37).

1.24. 3He/4He reservoirs

Partial melting at mid-oceanic ridges causes rocks to lose almost all of their U and
Th, which produce 4He over geologic time, causing 3He/4He to decrease. But when the
rocks first became oceanic lithosphere, the 3He/4He was frozen at higher values. Because
3
He is lighter than air, it is continuously lost to space from the atmosphere. Relative to the
atmospheric 3He/4He ratio (RA ; (3He/4He)atm ¼ 1.4 £ 1024), continental crust has low
ratios (3He/4He < 0.01RA) because it is enriched in Th and U, which produce 4He during
decay. MORB have rather uniform values of (8 ^ 1)RA (Allègre et al., 1995) and OIB
shows a range from 5 to 30 RA (Hanan and Graham, 1996). Plumes might rise directly from
the high 3He/4He reservoir through the upper low-3He/4He layer (Porcelli and Wasserburg,
1995). Otherwise, helium might get into the plume from the lower mantle (O’Nions and
Tolstikhin, 1994).
A combined lithospheric and crustal recycling can explain the high 3He/4He
reservoir and the 1 – 2 b.y. old reservoirs. Slabs penetrate to the CMB and the
high-3He/4He reservoir is contained in two large megaplumes. From a uniformitarianist
view, this reservoir would be recycled oceanic lithosphere (see Tackely, 2000, and
Section 2.11.2.3).
 155

(C) Heavy Element Compounds

This page is intentionally left blank
Cosmochemistry and Properties of Light Element Compounds 157

1.25. Ferrous metals in rocky planets

On the crustal surface of terrestrial planets, the abundance of iron and titanium is
the highest compared with other transition metals. Since these elements contribute
dominantly to the reflectance spectra, their coordination sites determine the nature and
position of the reflectance bands. Comparing the spectral profile of the reflection of
sunlight from planetary bodies with those from laboratory reflectance studies of rocks and
minerals, one can determine the modal mineralogies of regoliths on these bodies. These
remotely-sensed reflectance spectra are obtained through Earth-based telescopes or
detectors aboard the orbiters.
Earth: In comparison with the cosmic abundance, the crust of the Earth has less Fe
and Ti. This is because of their fractionation during the formation of Earth’s iron core,
mantle evolution and the leaching of these elements from crustal rocks by weathering
processes.
A large concentration of iron occurs in the form of BIF as sedimentary deposits seen
as layers of iron oxides and oxyhydrites. The other major deposits of transition elements of
first series are notably Cu in porphyry calc-alkaline rocks, Ni in ultramafic rocks, and Mn
in sub-aqueous fissures. These are commonly concentrated to form deposits in such zones
as hydrothermal veins, deeply eroded igneous rocks, etc. However, compared with the
cosmic abundance of Fe, the Earth’s outer part contains less of it because of chemical
fractionation through the core formation.
Moon: The Apollo and lunar mission samples show that lunar minerals and glasses
contain high concentrations of Fe and Ti, which exist in low oxidation states as Fe2þ, Ti3þ
and Ti4þ. Significant amounts of Cr2þ, Cr3þ and Mn2þ have been seen in some minerals.
Mars: The Viking Lander analysis of the surface regolith of Mars using XRF
showed iron ions present are predominantly Fe3þ.
Venus: Venera and Vega missions indicated that the chemistry of the Venusian
surface is similar to that of the Earth.
Asteroids: Vesta and Ceres are essentially composed of basaltic achondrites such as
eucrites and diogenites, having Fe2þ, Ti4þ (þ Ti3þ(?)) in their mineral phases.
Mercury: An essentially silicate planet, possibly containing a low amount of
transition metal ions in low oxidation states.

1.26. Element distribution in mineral system

The early empirical Goldshmidt (1937) rules for distribution of elements based on
cation size and charge have proved to fall short in accounting for many of the observations
on the patterns of distributions of trace elements in minerals.
The relationship between the distribution coefficients for trace elements and their
ionic radii appear anomalous for transition-element ions of first series (e.g., Cr3þ, V3þ and
Ni2þ). Crystal-field effects play a more important role than ionic radii in accounting for the
distribution of most transition elements in minerals and rocks (Schwartz, 1967; Roster and
Bouge, 1983; Dahl et al., 1993). It will be seen later in this section that this distribution
pattern is better explained by the crystal-field stabilization energy (CFSE) values for their
158 Chapter 1

respective sites. The trends in the geochemical distribution and transition metal ions can be
qualitatively interpreted by crystal-field theory (see Section 1.27).
To explain the element distribution and partitioning in a crystal/melt system,
various “rules” have been proclaimed. These rules and crystallo-chemical arguments have
greatly helped in interpreting the distribution of trace elements in crystal structures and
also in suggesting the locales for mineral occurrences. Thus, a knowledge of the valence
and ionic radius of a trace element would at once enable one to assess the likely crystal
structure that would accommodate the element. This is exemplified by some common
observations, such as:
(a) The Mg2þ and Fe2þ ion sites are easily occupied by Ni2þ, Co2þ, Liþ and Sc3þ in the
structures of olivine, pyroxene, etc.
(b) Ca2þ ion in the calcic pyroxene (diopsides, etc.), anorthite, apatite and eudialyte are
easily replaced by Mn2þ, Na1þ and R3þ (R ¼ lanthanides) ions.
(c) Kþ ions in K feldspars, zeolites and K-bearing Mn4þ-oxides (e.g., cryptomelane and
nomanechite) are substituted by Rbþ, Tlþ, Ba2þ and Sr2þ ions.
(d) OH2 ion in amphiboles, micas, tourmalines and topaz is substituted by F2 ion.
However, from all these, a generalization to understand the behaviour of transition
elements has been found to be unsatisfactory. The reasons for the failures of several criteria
have been elucidated by many workers.
The ionic radius criteria of Goldshmidt (1937) suffer from the drawback of the
concept that the ionic radius of an ion extends from the nucleus to the outermost orbital
occupied by electrons. The very nature of the angular wave function of an electron, which
approaches zero asymptotically with increasing distance from the nucleus, indicates that
an ion has no definite size. That is why along a metal –anion bond zero contours could
rarely be deciphered in the electron density maps prepared from XRD study.
Cations in the structures of ferromagnesian silicates are commonly located in
distorted coordination sites, where the M –O distances differ significantly. For example, in
the forsterite structure, M –O distances vary from ,205 to ,221 pm. If oxygen is assigned
to an ionic radius of 140 pm, the ionic radius of Mg2þ in forsterite varies from 65 to 81 pm
in the M1 and M2 sites. In enstatites, the M –O distances range still more widely from
199 to 245 pm, i.e., by about 50 pm.
This wide variability in M– O distances in silicate minerals would control the entry
of the substituting cations (i.e., relative enrichment) based on their ionic radii. The
geometries and dimension of coordination sites of an individual major cation change from
one mineral to another, as they do within a particular mineral structure. Hence, for
explaining the element partitioning in mineral structures, the ionic radius criterion appears
too simplistic and can hardly explain the trend in the distribution of trace metals in the
cogenetic suite of rocks.

1.26.1. Partitioning of elements

1.26.1.1. Siderophile elements

Siderophile elements (e.g., Fe, W, Co, Ni, Mo, Re and Os) preferentially partition
into metallic phases (see also Section 15.17.1.3)with respect to the silicate phases.
Cosmochemistry and Properties of Light Element Compounds 159

If the segregation of the Earth’s metallic core was an equilibrium process, either before
(within parent bodies), during or after accretion, then the abundance of siderophile
elements left behind in the silicate portion of the Earth should reveal the conditions of
equilibrium (e.g., Ringwood, 1979).
Experimental determination of the distribution of siderophile elements between
liquid metal and liquid silicate at 10 GPa and 2,0008C demonstrates that it is unlikely that
siderophile element abundances were established by simple metal – silicate equilibrium.
This also indicates that segregation of the core from the mantle was a complex process
(Hillgren et al., 1994). Earth’s mantle is known to have excess abundances of siderophile
elements such as Co, Fe, Mn and Ni. The partition coefficients of these elements were
invoked to explain their enrichment in the mantle (Ramamurthy, 1991). But, more
effectively, the changes in the nature of bonding brought about by magnetic collapse at
high pressure can also affect the partition coefficients (KD) of some siderophile elements.
Thus, electronic changes such as magnetic collapse under pressure could have important
implications for the geochemistry and physical properties of the mantle and the core.
At low pressure (atmospheric and 0.5 GPa), the partitioning behaviour of many
siderophile elements has been determined as a function of oxygen fugacity ( f O2),
temperature (,1,200 –1,6008C) and composition (e.g., Capobiano and Amelin, 1994;
Holzheid et al., 1994). Ohtani et al. (1991) investigated the partition coefficients for V, Cr,
Mn, Fe, Co and Ni between Mg – perovskite, magnesiowüstite, and solid metallic iron at
very high pressures (25 –27 GPa) and temperatures of 1,700 and 1,9008C. Hillgren et al.
(1994) have determined partition coefficients at high pressure and temperature (10 GPa,
2,0008C) for Fe, Ni, Co, W and Mo. They show an important decrease in both Dmet/sil Ni and
Dmet/sil
Co and increase in D met/sil
W with respect to atmospheric pressure data.
Using rigorous thermodynamics, Capobiano et al. (1993) have concluded that the
metal – silicate partition coefficients could either increase or decrease with temperature,
depending on the valence state and thermodynamic properties such as the enthalpy and
entropy of formation of the oxide species. Phase transition (of oxides) to metallic phases
would also influence the electrical and thermal boundary conditions of the Earth’s mantle
and contribute to the dynamic processes such as plumes that give rise to volcanic islands
like Hawaii (Manga and Jeanloz, 1996).

1.26.1.2. Incompatible elements

The bulk partition coefficients of the most highly incompatible elements increase in
the approximate order: Ba < Rb # Th , U < K , La , Ce. This sequence and some
derived ratios produce the leftward descending branch in the MORB. The pattern is shown
in Fig. 1.38 (read the caption). These are normalized to primitive-mantle concentrations in
average continental crust, average MORB, average Mauna Loa, Hawaii and three types of
OIB: average Tristan and Inaccessible islands representing EM-1 and Tubuai representing
HIMU islands. (after Hofmann, 1997). In the Tubuai pattern, a relative depletion for Rb
and Ba is seen. HIMU basalts are also of similar characteristic and are distinguished from
EM-type basalts. This corresponds to the low Rb/Sr of HIMU sources, inferred from their
87
Sr/86Sr ratios, which indicate a “depleted heritage” for both MORB and HIMU sources
(see Section 1.23.7).
160 Chapter 1

Figure 1.38. Concentrations of selected trace and major elements, arranged in the order of ascending compatibility.
These are normalized to primitive-mantle concentrations, in average continental crust, average MORB, average
Mauna Loa, Hawaii, and three types of OIB: average Tristan and Inaccessible islands representing EM-1 and Tubuai
representing HIMU islands (After Hofmann, 1997, q 1997 Nature Publishing Group).

The patterns for ORB and the OIB differ by their different enrichments (or
depletions) in incompatible elements, however, with respect to their positive Nb and
negative Pb anomalies. The continental crust has opposite Nb and Pb anomalies. A fourth
type of OIB, called EM-2, is similar to continental crust. Except for EM-2, all types of
oceanic sources have similar Nb/U which are significantly higher than PRIMA Nb/U. OIB,
EM-2 and Chile ridge MORB uniquely lie on mixing trajectories between DMM or HIMU
mantle and average continental crust (probably in the form of mantle-recycled sediments).
This observation may consequently lead to the following conclusions:
(1) Sources of MORB and OIB are linked.
(2) The OIBs do not represent the PRIMA reservoirs.
(3) The island arc magmatism and continental crust formation are both linked to plate
tectonics, generating these.
(4) The presence of recycled continental crust in OIB sources can be tested by tracer
studies.

1.27. Transition metals in magmas: CFSE

In one of the most thoroughly investigated trace-element distributions during

magmatic crystallization in Skaergaard magma, a relation was found to exist between the
order of uptake of transition metal ions from the magma and the relative CFSE of the
cations in octahedral coordination (Williams, 1959).
Cosmochemistry and Properties of Light Element Compounds 161

The structural environments of transition-metal ions in magma at high temperature

resemble those of quenched silicate glass. In glass structure, the cations occur in various
coordination environments and are coordinated to bridging and non-bridging oxygens
bound to silicon in the [SiO4] network and, therefore, they experience a wide range of
crystal fields.
In a granitic magma, the bridging oxygen atoms predominate but, with
decreasing Si/O ratio, as happens towards basaltic melt composition, the proportion
of non-bridging oxygen ions increases. The short-range ordering of transition-metal ions in
their coordination sites in a variety of silicate glasses are determined by spectroscopic
techniques such as optical/crystal field (see Burns, 1993), Raman (Cooney and Sharma,
1990) and 57Fe Mössbauer (cf. Mitra, 1992, Chapters 7 and 8).
A host of important observations have emerged from such silicate studies.
Some of the interesting ones relate to some major transition-metal ions and are cited
below:
(a) Fe2þ ions: occur predominantly in octahedral sites in basaltic glasses. However,
tetrahedral ones are seen in some orthosilicate glasses.
(b) Mn2þ ions: occur in octahedral, tetrahedral or distorted non-centrosymmetric
sites, with decrease in non-bridging oxygens the fraction of terahedral or non-
centrosymmetric distorted sites increases. Thus, in orthosilicate glasses, the
tetrahedral [MnO4] clusters occur.
(c) Ni2þ ions: occur mainly in distorted octahedra, also in network-modifying tetrahedral
and trigonal bipyramidal (5-fold) sites, e.g., in Ni2SiO4- and CaNiSi2O6-bearing
glasses. However, under high pressure, Ni2þ and Co2þ become less compatible in
igneous fractionation processes.
(d) Cr3þ, Mn3þ and V3þ ions: occur in glasses mostly in octahedral sites whereas Fe3þ
ions occur in network forming (tetrahedral) sites.
For transition-element ions the CFSE differences between crystal and glass (e.g.,
basaltic melt) structures, presented as DCFSFcrystal – glass, decreases (rather, becomes less
negative). The order for divalent cations occurs as: Ni2þ . Co2þ . Fe2þ . (Mn2þ) and,
for trivalent cations: Ti3þ . Cr3þ . V3þ . Mn3þ . (Fe3þ). This has already been
discussed in Section 1.26.
During the partial melting of the subducted oceanic lithosphere, Cu and Mn
preferentially enter into the melt structure with distorted sites with high CFSE, leaving the
crystalline silicate structures because of the lower CFSE present in the latter. In a subsequent
stage, when fluid phases appear with H2O, CO2, HCl, HF and other volatiles, Cu and Mn
selectively partition to the fluid phases from the magma. The metals are thus transported
hydrothermally through the porphyry copper system as Mn2þ, Cuþ or Cu2þ, halide or
bi-sulphide complexes. At still lower temperature, interactions with groundwater lead to the
formation of insoluble sulphide, oxide and metallic copper, as seen in porphyry deposits.

1.27.1. Principles of metal distribution in magmatic differentiation

Early in the history of the Earth, large-scale fractionation of solid silicates and
metal phases from silicate melts took place. This process was accompanied by a strong
162 Chapter 1

partitioning of a host of elements, including transition elements (e.g., cobalt and nickel).
In the modelling of the chemical evolution of the terrestrial planets, viz., Earth, Mars,
Venus and Mercury, the pressure-induced coordination changes in silicate melts have to be
taken into account. For this reason, distribution of these elements has received
considerable attention in constraining the models of the chemical evolution of the Earth
and the Moon (e.g., O’Neill, 1991) although in these models, only the partition coefficients
at low pressure have been determined.
However, the relative efficiency and importance of Newtonian versus non-
Newtonian flow and brittle versus ductile deformation in (a) the segregation of melts form
their source regions, (b) the fractionation of phases during magma evolution and (c) the
magmatic systems as plutonic or volcanic igneous rocks and juvenile fluids are poorly
known. An approach to understanding segregation transport mechanisms of silicate melts
following partial melting has been made through experimental investigations.
With a dynamic approach, the time scales associated with differentiation processes
such as crystal fractionation, magma mixing and assimilation can be constrained by a
knowledge of atom mobilities within melts. Melt viscosity is of particular importance for
understanding the evolution of Earth’s early history.
In magmatic melts (granitic and basaltic), network-modifying 4-, 5- and 6-fold
coordination sites predominate and cations go into these sites. But most cations prefer only
the octahedral sites in crystals. Hence, during magmatic crystallization, partitioning of
cations in a crystal takes place in octahedral sites while, in the magmatic melt, they occur
in all sites. (Note: For crystal-field theory, see Mitra, 1989; Burns, 1993; Chapter 3.)
The baricentre of 3d orbital energy levels is seen to be largely independent of the
coordination number of a transition-metal ion in a compound. Computed molecular orbital
energy level diagrams reveal that electrons in core orbitals (1s, 2s, 2p, 3s and 3p) and also
the non-bonding oxygen 2p molecular orbitals have comparable energy levels for iron in
different coordination sites.
On the basis of the estimated energies of site preference between 6-fold (octahedral)
and 5-fold (trigonal bipyramidal) coordination, one can very well see the reason for a
strong preference for octahedral sites for a good number of cations (e.g., Ti3þ, V3þ, Cr3þ,
Fe2þ, Co2þ and Ni2þ) and strong enrichments of a few cations (e.g., Mn3þ, Cr3þ and Cu2þ)
in trigonal bipyramidal sites in a melt. The latter are 3d4 and 3d9 cations whereas the
former constitutes all other distributions of electrons in 3d orbitals.
In a magma, the Maxwell – Boltzmann distribution law dictates that the ratio of
cations in octahedral and tetrahedral sites, no/nt, is related to CFSE as

no =nt ¼ exp½2ðEo 2 Et Þ=RT

where Eo and Et are the octahedral and tetrahedral CFSE (above a reference level, Uo).
Unfortunately, CFSE data for transition-metal cations in magma at high temperatures are
not available. Notwithstanding this, some workers projected the room temperature
spectroscopic data to magmatic temperatures, presuming that the relative orders of
octahedral site preference energy (OSPE) remain almost constant in the range 25– 1,2008C
and obtained satisfactory results.
Cosmochemistry and Properties of Light Element Compounds 163

With increasing polymerization of the melt, the transition element partitions more
into the crystallizing minerals. Thus, in a crystallizing magma, where all sorts of
coordination sites are present, the transition elements partition preferentially more into the
octahedral sites, as represented by

ð½4 Mnþ þ ½5 Mnþ þ ½6 Mnþ Þmagma ! ð½6 Mnþ Þcrystal

As a result of polymerization at late-stage differentiation, the equilibrium should be

driven more to the right in granitic melt compared with the basaltic melts. This accounts
for why the transition elements become scarce in residual granitic melts. The lattice
energies of these compounds display a fairly smooth variation across the series of ions,
irrespective of crystal structures.
The CFT, supported by the transition-state theory of reaction rates, demonstrates
that cations less susceptible to substitution reactions are those with 3d3, low-spin 3d6 and
3d8 configurations, such as Cr3þ, Co3þ and Ni2þ, which acquire high CFSE in octahedral
coordination. These factors contribute to the enrichment of Cr, Ni and Co in laterite
deposits, where the other elements undergo gradual depletion. Evidently this explains why
Ni, Co, Cu, etc., substituting Mn4þ in [MnO6] octahedral sites in Mn-module minerals,
become stabilized to form important sources of such strategic metals in the oceanic
nodules.

1.27.2. Transition-element partitioning in mineral systems

Silicate melts and aqueous fluids are two of the principal vehicles for mass and
energy transport in the Earth’s interior. At high pressure and temperature, major elements
become significantly soluble in these (e.g., Manning, 1994). As a consequence, the major
and also minor (trace) elements experience partitioning between the fluid and melt.
The OSPE as well as DCFSEoct-glass measures the relative affinities of cations in a
magma for octahedral sites in silicate minerals vis-à-vis in glass. Accordingly, in
crystallizing silicate minerals, the uptake of transition elements increases in the order:
Ca2þ, Mn2þ, Zn2þ , Fe2þ , Co2þ , (Cu)2þ , (Cr)2þ , Ni2þ for bivalent cations, and
Sc3þ, Ga3þ, Fe3þ , Ti3þ , V3þ , Co3þ , (Mn)3þ , Cr3þ for trivalent cations. The
cations in parentheses produce Jahn – Teller distortions and are stable in more deformable
sites and, consequently, crystallize in late-stage minerals.
The crystal-field theory can be employed to help explain the fractionation process
operative in the partial fusion of the mantle or of the subducted oceanic lithosphere and in
the patterns of metallogenesis of transition elements, e.g., Cu, Mn, Co, Ni and Cr. Thus, the
enrichment of Ni and Cr in certain lherzolites, Cu and Mn in porphyry copper deposits, Cr
in olivines enclosed by diamonds from kimberlites and Ni in peridotites (alpine) and
laterites can be explained by crystal-field theory (Burns, 1993). The high value of
DCFSEcrystal – glass of Ti3þ accounts for its prevalence in the basalts on the Moon, where
much lower oxygen fugacites existed compared with those in the magmas of the Earth to
stabilize trivalent titanium.
164 Chapter 1

1.27.2.1. Ni2þ and Co2þ: pressure partitioning

In a deep magma ocean, iron equilibrates with molten silicate at an average
pressure of ,28 GPa (ø80 km depth). This can account for the depletion of many
siderophile (metal-loving) elements in Earth’s upper mantle, most notably of Co and Ni
(Li and Agee, 1996). Keppler and Rubie (1993) showed that, at high pressure, the
coordination of Co2þ and Ni2þ ions in silicate melts changes. Also, the partitioning of
these species into the melt phase increases markedly, such that the crystal/melt and
metal/melt partition coefficients decrease by more than an order of magnitude at lower-
mantle conditions.
At atmospheric pressure, Co2þ seems to be in tetrahedral coordination in silicate
glasses (Keppler, 1992). The environment of Ni2þ is greatly distorted from an ideal
octahedral geometry to a possible five-coordinated one (Galoisy and Calas, 1992) or it may
even occur in a tetrahedral site. The CFSE of Co and Ni in melts is much smaller than in
most minerals and, hence, any pressure-induced coordination change will strongly affect
the partition coefficients of these elements.
Keppler and Rubie (1993) measured the crystal-field spectra of albite (NaAlSi3O8)
glasses doped with 1 wt% NiO or CoO and quenched from melts at up to 2,9008C and
10 GPa. The crystal-field spectra of Co2þ-doped samples after deconvolution show the
peak components to be due to tetrahedral and octahedral Co2þ. At high pressure (10 GPa),
about 50% of the Co2þ appeared to move to octahedral coordination. When the tetrahedral
and octahedral components were subtracted from the 10 GPa spectrum, only a small
residual of 15,000 cm21 was left. This residual arises because the peak of the tetrahedral
species around 15,000 cm21 moves to only slightly higher wave numbers when the sample
is quenched from high pressure.
A similar change in the coordination in Ni2þ-doped albite was observed under
pressure. The band positions of octahedral Ni2þ move, indicating an increased electrostatic
field strength around the Ni ion at high pressure. These structural changes are quenchable
and hence not due to simple elastic deformations. The main absorption peak of Ni2þ shifts
from 21,650 cm21 (1 atm) to 22,680 cm21 in the glass quenched from 10 GPa. The other
bands also show similar effects. A weak band, arising possibly by a spin-forbidden
transition of octahedral Ni2þ (3A2g(F) ! 1Eg(D)), is seen to be enhanced by distortion
under this quenched pressure.
A change from tetrahedral to octahedral coordination in a melt under pressure will
increase the CFSE of Co2þ by about 29 kJ mol21; for Ni2þ, this increase is 61 kJ mol21
(Keppler, 1992). This effect will stabilize both Co2þ and Ni2þ in silicate melts relative to
crystalline silicates as well as relative to a metal phase. If a complete change from
tetrahedral to octahedral coordination occurs, crystal – silicate melt and metal –silicate
melt partition coefficients will be reduced by one (for Co) to two (for Ni) orders of
magnitude at 1,500 K. This is also seen to hold good for Co2þ at ,20 GPa (Keppler and
Rubie, 1993).
Ni2þ and Co2þ at high pressures will behave as less compatible and less siderophile
than are usually known (at low pressure). Some olivine/melt partitioning study at high
pressure (McFarlane and Drake, 1990) also revealed similar behaviour.
Cosmochemistry and Properties of Light Element Compounds 165

1.27.2.2. Cr3þ, Ni2þ, Fe3þ and Ti4þ ions

Cr3þ and Ni2þ ions are enriched in earliest formed minerals such as chromite and
olivine in a crystallizing magma. The dominant product of Cr3þ ion fractionation is the
phase chromite (FeCr2O4), which, in disseminated form, may fractionate with olivine in
dunite ! peridotite sequence.
Cr3þ and Ni2þ, having very large octahedral site-preference energy, would be
strongly biased in favour of crystalline sites, resulting in their preferential expulsion from
the melt. The Ni2þ ions enter the octahedral M1 sites in olivine, while Cr3þ ions occupy the
octahedral B sites in crystallizing spinels and this effectively brings about a large
precipitation of chromite. The abundance of Cr3þ and Ni2þ may serve as an index of partial
melting. The concentration of these ions in certain lherzolite xenoliths was postulated to
represent the refractory residuum resulting from partial melting of the mantle (Burns,
1973). In comparison, Mn2þ, Fe3þ and Ti4þ ions, having zero CFSE for octahedral sites
(OSPE ¼ 0), remain dispersed in the melt in tetrahedral, distorted octahedral and other
sites. Therefore, in the residual lherzolite xenoliths, these cations are impoverished.
However, high oxygen fugacities lead to the formation of magnetite, causing a depletion of
Fe, Ti, V, etc.
The fractionation path of a basic ! ultrabasic melt, initially crystallizing olivine
and chromite, may deviate from the normal course of crystallization due to contamination
of siliceous (often granitic) material derived from partial melting of the roof of the
intrusion. This hybrid melt precipitates chromite initially, followed by pyroxene (in place
of olivine), manifesting a petrologic sequence: peridotite ! chromitite ! orthopyroxenite
(Irvine, 1975). However, contamination mechanism lowers the actual temperature of the
magma, reduces its liquidus temperature and may cause chromite to crystallize alone. This
process also leads to a later-stage deposition of magnetite, sulphide and even platiniferous
and auriferous bodies.

1.27.2.3. Cr2þ, Ni2þ and Co3þ ions

Cr21 ion. Cr2þ ion in natural minerals was first reported from lunar samples (Boyd and
Smith, 1971), where very low oxygen fugacity has persisted since the Moon formed.
Under very low oxygen-fugacity condition, Gasparik (1981) synthesized a number
of Cr2þ-bearing phases from a system MgO – Cro – Cr2O3 – SiO2. Angel et al. (1989)
synthesized Cr2þ-bearing pyroxene (Mg1.425Cr0.611Si1.964O6) and found by X-ray structure
refinement that the majority of Cr was localized at the more distorted octahedral M2 site,
which usually accepts more transition metal cations than the M1 site (Burns, 1970; Ghose
et al., 1975). A large expansion of the average M2 – O bond length occurs from 2.144 Å
(Ohashi, 1984) to 2.209 Å, when Cr occupies the M2 site (r(Mg2þ) ¼ 0.72 Å,
r(Cr2þ) ¼ 0.80 Å; Shannon, 1976). The ionic radius of Cr3þ in octahedral coordination
is 0.615 Å, compared with a value of 0.72 Å for Mg2þ but the spectroscopic observations
on this synthetic enstatite are consistent with the chromium ion as divalent one.
Chromium (#0.4% Cr2O3) is reported in olivine crystals in the diamonds of
kimberlites. Cr2þ replaces Mg in olivine structure. Low oxidation states of certain
transition metal ions, including Cr2þ ion, may be stabilized at very high P, T field in the
mantle (Burns, 1976). The Jahn – Teller effect in the distorted sites in mantle minerals
166 Chapter 1

(viz. in olivine, and M1 and M2 sites in pyroxene) further enhances the stability of the
occurrence of Cr2þ in the deep upper mantle. In the lower mantle, Cr2þ may be stabilized
in distorted Mg sites of MgSiO3 perovskite phase.

Ni21 ion. Because of its high octahedral site-preference energy, Ni2þ ion preferentially
occupies the octahedral sites in a melt but, at high temperatures when thermal energy
competes over the octahedral CFSE, Ni2þ enters other network modifying tetrahedral and
5-fold coordination sites. Mg2þ ion shows preference for octahedral sites but it does not
attain the additional stabilization energy through crystal-field effects as secured by a
transition metal ion like Ni2þ. For this reason, Mg2þ ion gets more evenly distributed in the
melt in its network-modifying sites, resulting in ðDSfus ÞMg . ðDSfus ÞNi :
In a magmatic crystallization, Ni2þ ions are enriched in the octahedral sites in the
crystalline lattice of olivine in preference to the network-modifying tetrahedral and
trigonal bipyramidal coordination sites in the melt (Galoisy and Calas, 1991). Also, at a
later stage of magmatic fractionation when the melt becomes more siliceous, the Ni2þ
equilibria in olivine , melt shift strongly in favour of the crystallizing olivine phase due
to decreased CFSE attained in more highly polymerized silicate liquids.

Co31 ion. Ni and Co cations in bivalent states co-exist in silicate and oxide phases,
crystallizing from a magma. In seawater, however, they mostly occur as trivalent species.
It is stated elsewhere that Co3þ in a low-spin configuration occur in many oxides, e.g.,
spinel, corundum and perovskite. Low-spin Co3þ ion has ionic radius (54.5 pm), which is
very similar to Mn4þ (53 pm) and, consequently, becomes entrapped and enriched in
Mn-nodules.

1.27.2.4. Ni –Co partitioning

Earth’s upper-mantle samples manifest a high abundance of Ni and Co and also a
chondritic Ni/Co ratio. A heterogeneous accretion model has garnered much support to
explain this but laboratory studies on the partitioning of these elements between the metal
and silicate phases up to pressures of 27 GPa (Ohtani et al., 1991; Hillgren et al., 1996;
Gessman and Rubie, 1998) predict that the mantle should have a much lower abundance of
these elements. This is because of preferential partitioning of these elements into the
iron core.
However, Li and Agee (1996) and Righer et al. (1997) showed that, in the upper
mantle, the chondritic Ni/Co ratio could have been preserved from an early sulphur-rich
magma ocean.
With pressure, the siderophile character of Ni and Co decreases strongly. Studies on
Ni, Co-partitioning between metal, silicate liquids and Mg –wüstite at lower pressures
showed that the ratio D Ni/D Co is nearly independent of f O2 and T (Capobinaco and
Amelin, 1994). But metal – perovskite partitioning does not fractionate Ni and Co and
would provide a high abundance of both elements in the mantle. Also, liquid – solid iron
partitioning of Ni and Co is close to unity.
If the bulk Earth is chondritic with respect to refractory elements and if the mantle
(with Ni content of 0.2 wt%) is chemically homogeneous, the partition coefficient
Cosmochemistry and Properties of Light Element Compounds 167

observed by Tschauner et al. (1999) implies drastic chemical disequilibrium between core
and mantle.

Ni –Co segregation in the upper mantle. Tschauner et al. (1999) experimented to

determine the partitioning between metal and Mg– perovskite. They used the laser-heated
DAC and secondary ion microprobe to measure the partitioning behaviour of Co and
Ni between Mg – perovskite and Fe-rich metal at lower-mantle pressures of
(,80 GPa ;,1,900 km). Their results indicate that metal – silicate equilibrium would
cause an enrichment of these elements in the lower mantle relative to the upper mantle.
Siderophile elements such as Co, Ni, W, Re and Au partitioned strongly into the Fe-rich
metal, which is known to segregate to form the Earth’s core.
It is because of this process of partitioning that these elements became depleted by
several orders of magnitude in the Earth’s silicate upper mantle relative to their abundance
in the early Solar System, as inferred from some chondritic meteorites. The Co/Ni ratio in
the Earth’s upper mantle is seen to be nearly the same as in chondrites; the ratio of their
partition coefficients (D Ni/D Co) is near unity. Evidently, these elements were not
fractionated from one another during core formation.
The relative abundances of Co and Ni in the upper mantle can be attained only by
equilibrium between molten silicate and molten metal at about 28 GPa and 2,300 K. A
pressure of 28 GPa corresponds closely to the depth at which a peridotitic upper mantle
transforms to a lower mantle, dominated by perovskite. Above this transition, the melting
curve becomes very steep (Zerr et al., 1998) (see also Section 1.18).

Ni –Co segregation in the lower mantle. During the final segregation stage, metal may
have re-equilibrated to some degree with the major minerals of the lower mantle, silicate
perovskite (,85%) and magnesio-wüstite (,15%). The experimental results of Tschauner
et al. (1999) indicate that, at very high pressures, the partition coefficients for Co and Ni
both approach unity and that the D Ni/D Co ratio is also unity throughout the lower mantle.
It may be assumed that a Co- and Ni-enriched lower mantle has remained
chemically isolated for the past 4.5 b.y. Since magnesio-wüstite probably constitutes about
15% of the lower mantle, a knowledge of metal –magnesio-wüsitite partitioning is
important for any model of core formation. In such a model, the possible reaction of the
outer core with lower mantle minerals throughout Geological time has to be taken
into consideration.

1.27.2.5. Plutonic rocks and metal concentration

Subsurface igneous rock bodies manifest various forms — from dykes cross-cutting
the country rocks to large masses (.103 km3) of plutons, typified by granitic rocks, which
apparently crystallized at depths of 0.5– 10 km or more within the continental crust.
These plutons are hosts of important mineral deposits such as Cu, Mo, Sn and Ag
deposits (e.g., Chile, Bolivia and SW United States); Cr and Pt group metal deposit
(e.g., S. Africa, Zimbabwe and Montana (U.S.)). The U minerals tend to be concentrated as
sulphides and oxides in plutons of gabbroic composition. Cu – Mo – Sn –Ag sulphides occur
as a dispersed phase within a wide range of igneous rocks from diorite to granite.
168 Chapter 1

1.27.2.6. Highly siderophile elements

The elements like Re, Os, Ir, Ru, Pt and Pd are called highly siderophile elements
(HSE) because of their extremely high affinity for Fe metal relative to silicates. These
elements can be determined using isotopic dilution method and negative thermal ionization
mass spectrometry (N-TIMS: Re and Os) and multicollector, inductively coupled plasma
mass spectrometry (MC-ICP-MS: Ir, Re, Pt and Pd).
All the HSE, except Pd, condense at temperatures higher than many major nebular
condenses such as forsterite, enstatite and Fe –Ni metal. Therefore HSE may present the
earliest processes in the nebula.
Such data obtained by Horan et al. (2003) when coupled with corresponding
187
Re – 187Os isotopic data of Walker et al. (2002) constrain the nature and timing of
chemical fractionation of these elements in the early solar system. The Re/Os ratios and
187
Os/188Os ratios m carbonaceous chondrites in general are about 8 and 3%, respectively,
lower than the ratio for ordinary and enstatite chondrites.
The fractionation of refectory Re and Os may have occurred within a narrow range
of high-temperature condensation of these elements in the solar nebula.
Pd condenses at a somewhat lower temperature, just below the condensation
temperatures of magnesia silicates and Fe– Ni metal.
The CK chondrite, Karoonda, is unusually enriched in Os and depleted in Re
compared to other HSE. The Re –Os isotopic systematic of this meteorite indicates that this
large fractionation was caused by an early solar system process (Walker et al., 2002).
Karoonda, like the CV3 chondrites, is also characterized by a depletion in Pd relative to
other HSF (except Re).
Several of the H chondrites have significantly fractionated HSE abundances, with
enriched Re and Os relative to Pd. The H3 chondrites, Sharps and Dhajala, have similar
abundances and ratios to the equilibrated H chondrites.
In the ordinary chondrites generally there is a decreasing HSE content (average)
with decreasing metal content (H . L . LL).
Both EL and EH chondrites are more enriched in Pd, compared to the more
refractory HSE, than are ordinary chondrites of the CO and CV groups. More metal-rich
(EH) samples have lower absolute abundances of HSE than the more metal-poor (EL)
samples.
These systematics are opposed to those of ordinary chondrites in which HSE
abundances are positively correlated with Fe abundance and metal content. In both
carbonaceous chondrite groups and ordinary chondrite groups, the more refractory HSE
(Re, Os, Ir, Rn, and Pt) were fractionated relative to Pd.

HSE in carbonaceous chondrites. Carbonaceous chondrites are characterized by 7 –8%

lower Re/Os than ordinary and enstatite chondrites. This fractionation resulted from an
early, high-temperature fractionation in the solar nebula. The lower Re/Os in carbonaceous
chondrites may possibly result from oxidation and heating losses of Re at lower
temperatures in the early formed components of carbonaceous chondrites (Walker et al.,
2002). However, the differences between condensation temperatures of Re and Os may be
only a few Kelvin.
Cosmochemistry and Properties of Light Element Compounds 169

The Pd/Ir ratios for carbonaceous chondrite group may decrease in the order
C1 . CM < CO . CV. The carbonaceous chondrite group (other than C11) are enriched
in most refractory HSE (Re, Os, Ir) (e.g., Jochun, 1996). In the carbonaceous chondrite
groups, the refractory lithophile elements are progressively enriched in roughly the same
order CV . CO < CM . C1 as the refractory siderophile elements. It is suggested that
the siderophile and lithophile excesses were contained in the same nebula component
(Wasson and Kallemeyn, 1988).

HSE in ordinary chondrites and enstatite chondrites. HSE in ordinary chondrite groups
decreases in the order H . L . LL. The Re/Os (and 187Os/188Os) ratios are higher than the
carbonaceous chondrite groups. EH chondrites have distinctly high relative Pd
abundances, and generally unfractionated abundances of the more refractory HSE (Re,
Os, Ir, Ru and Pt). Refractory HSE are depleted relative to Ni, suggesting depletion in HSE
elements relative to Fe metal, yet the silicon-normalized abundances of the refractory HSE
are similar to C1, despite their extreme depletions in refractory silicates.
These are moderately volatile elements with condensation temperatures of 918 K
(Ga), 1,157 K (As) and 1,225 K (Au) (Wasson, 1985), all of which apparently behaved as
siderophile elements under the reducing conditions of the location at which enstatite
chondrites formed.
EH chondrites experienced some process capable of frastanating the more
refractory HSE (Re, Os , Ir Ru and Pt) from the moderately volatile siderophile
elements (Pd, Au, As and Ga). The data of Horan et al. (2003) suggest the depletions in the
refractory HSE occurred between the condensation temperatures of Pt (1,411 K at 10–
4 atm or 100 Pa) and Pd (1,334 K). Within this temperature range major nebula
condensates were likely formed involving forsterite, enstatite and Fe metal.
In EL chondrites, the less refractory HSE elements (e.g., Pd, Au, As, and Ga) are as
much enriched as the more refractory Re, Os, Ir, and Pt. They are similar to C1
concentrations when normalized to their Ni elements. HSE fractionation processes in the
EL and EH chondrites are the same but in EL chondrites they are much less.

Conclusions. Based on HSE characteristics each of the three major chondrite classes can
be discriminated. In these three classes the Re/Os fractionation occurred at high
temperatures in the solar nebula condensation process occurring prior to the condensation
of major silicate and metal phases.
In enstatite chondrites (particularly in EH chondrites) the Pd enrichment indicates a
lower temperature fractionation between moderately volatile siderophile elements and the
more refractory siderophile elements (Re, Os, Ir, Ru and Pt).

Osmium. Rhenium (Re) and Osmium (Os) class of trace elements are referred to as the
highly siderophile elements, which mostly reside in the Earth’s metallic core. But these
may remain in the silicate portion of the Earth’s mantle adequately to make up
concentration of the order of several parts per billion. Mass exchange between the core and
the mantle may also take place to a significant extent.
170 Chapter 1
187
Re decays to 187Os (1=2t ¼ 42 b.y.) producing changes in 187Os/188Os ratio
significantly (Shirey and Walker, 1998). This decay process since the Earth’s formation
4.56 b.y. ago has increased the 187Os/188Os ratio by 33% in the composition of the bulk
silicate Earth (BSE, bulk Earth minus core), while the crustal resources and the mantle,
contributed by partial melting processes, have led to variations in 187Os/188Os ratios over
several orders of magnitude. The Os isotope variability in the mantle is used to examine its
chemical differentiation and its modulation by convective motion.
Partial melting produces a unique signature in the Re –Os isotope system. During
partial melting Re is in the residue, whereby the Re/Os ratios in the mantle residue are
much lowered, retarding the growth of radiogenic 187Os. On the other hand radiogenic
187
Os rapidly accumulates in the crustal materials. The heterogeneity of Os isotopes in the
convecting mantle provides the important constraints on the origin of mantle hotspots and
the distribution of chemical reservoirs in the deep Earth.

1.28. Ca – Al and Mg – Si proportionation in the mantle

The average Ca/Al weight ratio for all classes of chondrites is seen to be remarkably
constant (,1.09 ^ 0.02; Zindler and Hart, 1986). This ratio is not affected even by severe
melting events. Therefore, the bulk composition of a completely molten Hadean mantle
should have the chrondritic Ca/Al ratio because (a) both of these elements would be largely
excluded from the core and (b) this ratio should not be affected by a giant impact that
caused massive melting and volatilization. As the mantle crystallization in the primitive
Earth proceeded upward from down below, the crystallization of the lower mantle would
control the composition of the upper mantle.
When the average values for xenolith bulk compositions (Carter, 1970) are plotted
on Mg/Si versus Ca/Si or Mg/Si versus Al/Si, these plot along with the chrondritic trends.
The intersections give a PRIMA composition with a forsterite content in olivine of Fo88.6
and Ca/Al ratio of 1.11 (essentially the chrondritic ratio).
On the basis of compositional trends of lherzolite from Kilbourne Hole (New
Mexico), USA, the composition of the primitive upper mantle is calculated to have a
strictly chondritic Ca/Al ratio of 1.11 (Carter, 1970). This implies that the magma ocean
that produced the upper mantle crystallized at equilibrium or experienced only a small
amount of fractionation. The non-chondritic Mg/Si, Mg/Ca and Si/Ca ratios of the mantle
are attributed to volatilization that accompanied a giant impact.
The Earth is depleted in volatile elements compared with those in C1 chondrites.
Giant impacts would be expected to cause modest depletion of both Mg and Si and greater
depletion of Si than Mg. The primitive upper mantle, when compared with C1 chrondrites,
is depleted in both Mg and Si relative to refractory lithophile elements such as Ca and Al.
C1 chrondrite has Mg/Ca and Si/Ca weight ratios of 11.4 and 12.3, respectively (Anders
and Grevesee, 1989) whereas, in the primitive upper-mantle composition, these ratios are
10.5 (8% depletion) and 9.4 (23% depletion), respectively.
In the lower mantle, the two major phases, Mg –perovskite and magnesio-wüstite
(Ito and Takahashi, 1987; Zhang and Herzberg, 1994), occur in the approximate proportion
of 80% perovskite and 20% Mg – wüstite. Crystallization of these yield for the remaining
Cosmochemistry and Properties of Light Element Compounds 171

upper-mantle magma ocean a ratio of Ca/Al of 2:1. Additional crystallization of majorite-

garnet in the lower part of the upper mantle (i.e., the transition zone) would increase the
Ca/Al ratio of the uppermost mantle still further (Drake et al., 1993).
On the basis of thermoelastic data, it can be argued that the lower mantle is enriched
in Si/Mg and Fe/Mg relative to the upper mantle (Hemley et al., 1992; Ita and Stixrude,
1992). However, this has been disputed by later workers (viz. Wang et al., 1992, 1994).
The PRIMA probably no longer exists beneath the present crust. The residual xenoliths are
more depleted (lower in Si/Mg and Fe/Mg) than the primitive mantle. This depletion by
partial melting led to the formation of the overlying crust.
If the original accreted material was chrondritic and the global magma ocean
existed, then the present mantle data suggest that there was no major fractionation during
crystallization. Alternatively, the chemical heterogeneities produced during crystallization
(Agee and Walker, 1990; Miller et al., 1991) must have disappeared by connective mixing
of the crystallized mantle. On the basis of the observation of the compositional trends of
the lherzolite from the Kilbourne Hole (Carter, 1970), one may conclude that the primitive
upper mantle, exclusive of volatile elements, is unfractionated (^2%) relative to
chrondrite. The fractionation by percolation of melt from the upwardly advancing
crystal/liquid layer would be important only in the upper 150 –300 km.

1.28.1. Critical ratios

The ratios of the elements which do not enter the core, i.e., are lithophile in
character, are the same in the bulk Earth (BE) as in the mantle. The Si/Mg ratio in the
primitive mantle (PRIMA) offers a key ratio for estimating the composition of the Earth
(see Table 1.6). This ratio, (Si/Mg)PRIMA, has been obtained as 0.945 by Allègre et al.
(1995). The value of (Si/Mg)PRIMA varies from 0.93 to 0.96 (Allègre et al., 1995b). The
other such ratio, (Ca/Al)PRIMA, for terrestrial peridotites is ,1.24 while meteorites have
this ratio uniformly as 1.07 ^ 0.04 (Brett, 1976).
In the pyrolite model of the upper mantle, the Ca/Al wt ratio is 1.27, significantly
lower than the value to be obtained from crystallization of the Mg – perovskite,
Mg –wüstite and majorite but higher than the average chondrite ratio of 1.09.

TABLE 1.6
The critical ratios for the primitive mantle (PRIMA) and the bulk Earth (BE)

Ratios Values References

(Si/Mg)PRIMA 0.93–0.96 Allegre et al. (1995)

0.945 (average) Allegre et al. (1995)
(Ca/Al)PRIMA ,1.24 (terrestrial peridotite) Brett (1976)
1.07 ^ 0.04 (meteorites) Brett (1976)
(Mg/Fe)PRIMA 3.92 ^ 0.03 Allegre et al. (1985)
(Fe/Al)PRIMA 2.0683 ^ 0.011 Allegre et al. (1995)
(Si/Mg)BE 1.1– 1.05 Allegre et al. (1995)
(Fe/Mg)BE 1.8– 1.5 Allegre et al. (1995)
172 Chapter 1

(Mg/Fe)PRIMA: The value of this is taken as (Allègre et al., 1995) 3.92 ^ 0.03,
derived from the value of the atomic ratio Mg/(Fe þ Mg) ¼ 0.9 (Ringwood, 1979, and
others). This gives (Fe/Al)PRIMA ¼ 2.0683 ^ 0.011.
The value of (Si/Mg)BE varies between 1.1 and 1.05, depending on whether it is
calculated from carbonaceous or C1 chondrites. The Fe content of the core depends on the
(Fe/Mg) ratios of the bulk Earth and of the mantle, and on the Mg concentration of the
mantle. The (Fe/Mg) ratio of the bulk Earth compatible with the (Mg/Al) ratio observed in
peridotite varies at most from 1.85 to 1.5 (Allègre et al., 1995).
In MORB and OIB, the ratios of Nb/U (<47), Ta/U (<27) and Ce/Pb are similar but
their values are much higher than those of the continental crust and island arc volcanics (by
factors of 4 – 5) and those estimated for the primitive mantle (by factors of 1.5 –2.5)
(Hofmann et al., 1986). The primitive mantle ratio of Nb/U and the mean value of Nb/U are
30 and 10, respectively, for both the continental crust and island arc rocks.
The relative constancy in the ratios of the elements such as P/Nd, Ti/P, Ti/Pd, Ba/W
and Sn/Sm in peridotites, komatites and basalts of various ages indicates that there had
been negligible exchange between the core and mantle throughout the geologic record of
the last ,3.5 GPa. A depletion of the Archean mantle in incompatible elements is evident
from the high Sm/Nd ratios of even the oldest Archean rocks. A complementary enriched
reservoir formed by differentiation of the mantle prior to 3.8 GPa is required to balance this
depletion. Because even the komatitic source regions were depleted, the enriched reservoir
had to be located at depths greater than 200 –300 km, which is the depth of origin of the
oldest Barberton komatiites (Herzberg, 1992).

1.29. Core differentiation / heterogeneous accretion

The core constitutes ,30% of the Earth’s mass. The Earth’s core is an important
boundary condition for the behaviour of the entire globe. The pressure there reaches over
300 GPa and the temperature is ,6,000 K (about the surface temperature of the Sun). Core
differentiation has been regarded by O’Neill (1991) as an equilibrium process. The core
material might have equilibrated at low to intermediate pressure in reducing conditions,
since core formation could not have occurred at a very high pressure. The evolution of the
oxidation state of the mantle may be a consequence of the differentiation of the core and
the oxygen evolved during reduction of silicates to elemental silicon could migrate upward
and progressively modify the oxidation state of the mantle (O’Neill, 1991). The late stages
of accretion would have occurred in oxidizing conditions (e.g., Wänke et al., 1984).
In the heterogeneous accretion model, the core formation is assumed to have began
with the sequestering into the core of almost the entire siderophile element inventory of the
highly reduced devolatilized, volumetrically dominant component of the proto-Earth. In
the subsequent period, however, more oxidized, volatile-rich siderophile components were
incorporated into the core.
The core formation continued until 4.5 Ga (4.5 £ 109 years) ago. After this, a late-
stage veneer of chondritic material was added to the Earth’s mantle. In the early Archean
(.3.8 Ga), the Earth’s mantle became nearly as depleted in LREE as the source regions of
Cosmochemistry and Properties of Light Element Compounds 173

modern MORB. In contrast, the continental crust was enriched in LREE and grew
episodically in volume.
A significant partitioning of elements such as S, O, Si, Ni and Co between core-
forming iron-alloy and mantle silicate minerals and the partitioning of C and S between
liquid and solid iron (in the inner and outer core) have been observed. The abundances of
Ni and Co in the Earth’s upper mantle are best explained by equilibrium core – mantle
differentiation in a global deep-magma ocean on the primitive Earth.
The core’s density deficit is explained by the presence of lighter elements,
predominantly sulphur. Three new high-pressure Fe –S compounds have been investigated
(by GL) up to 21 GPa and at between 950 and 1,4008C (see Chapter 14).
182
1.29.1. W fractionation and Hf/W ratio

1.29.1.1. Hf/W in early history

Geochronology relies mostly on the radioactive decay of Hf-182 to W-182 (half-
life, 9 £ 106 yr), a process that provides a clock for timing the birth of terrestrial
planets.
The molten iron, trickling through the mantle to from the core, carried tungsten,
which is the hafnium decay product. Hafnium, however, stays on in the rocky mantle; its
“decay clock” effectively resets to zero. Chondritic meteorites (which constituted the
early Earth) show no differentiation to form the core, hence they reflect the full record of
Hf to W decay. By checking their isotopic clocks against those in the mantle rocks, it is
revealed that the Earth’s accretion was essentially complete by 30 m.y. (rather than
60 m.y.).
The time of the last chemical equilibrium between the core-forming melt and the
silicates and oxides (making the proto-mantle) may be determined by using hafnium –
tungsten chronometry. 182Hf decays to 182W in a half-life of 9 m.y. The Solar System,
while forming from the collapse of an interstellar cloud, received nucleosynthetic material
expelled from neighbouring stars. Amongst these were perhaps the short-lived radio-
isotopes like 26Al and 182Hf. Meteorite data now suggest that there must have been an
unexpectedly high level of 182Hf in the early Solar System.
The accretion of the Earth and Moon within the solar nebula is thought (Carlson
et al., 1988) to have taken 50– 100 m.y. but the time involved for the Earth’s core
formation is controversial. Meteorite chronometry based on the 182Hf– 182W system
seems to resolve the debate (Lee and Halliday, 1995) on the premise that a strong
fractionation of hafnium from tungsten would be induced during the segregation of metal
core from the silicates. Hafnium is a lithophile (stone-loving) element whereas tungsten
is a siderophile (iron-loving) element. Therefore, there is a strong fractionation of Hf/W
between stony and iron meteorites and between silicate mantle and metallic core of
a terrestrial planet.
During the core formation, most of the tungsten goes into the core because of its
siderophillic nature and most of the hafnium stays in the silicate portion (the mantle).
Tungsten isotope measurements in iron meteorite are expressed as a ratio of 182W/184W (to
cancel out any chemical effects that shift absolute abundances by large amounts yet leave
174 Chapter 1

isotopic ratios unchanged). Meteorites such as pallasites and eucrites show 182W/184W
ratios which are complementary to the “irons”.
The relationship between W and Hf isotope compositions in the bulk Solar System
or in chondrites (CHOND) can be expressed as (Lee and Halliday, 1995):

 182   182   182   180  

W W Hf Hf
184
¼ 184
þ 180 184
ð1-3Þ
W CHOND W BASSI Hf BASSI W CHOND

where BSSI is the initial value for the bulk Solar System. (182Hf/180Hf)BSSI is primarily a
function of the amount of freshly synthesized r-process Hf introduced to the solar nebula
before its collapse.
The above equation (1-3) defines a linear relationship between 182W/184W and
180 184
Hf/ W that may be obtained by measurements on undisturbed minerals of early Solar-
System materials such as chondrites.
Lee and Halliday (1995) analysed both Allende and Murchison (primitive
undifferentiated meteorites) for 182W/184W ratios, which show higher values than in
iron meteorites. They saw a clear 184W deficit in both compared with iron meteorites,
much in agreement with previous results. They also found a similar isotope ratio for a
lunar mare basalt, suggesting that the lunar formation timing was similar to Earth’s
core formation timing. This is compatible with the ideas of a giant impact origin for
the Moon. The W isotopic composition of the silicate Moon is chondritic and is similar
to the Earth’s. The Moon may have formed from mixed debris created by a giant
impactor that hit the Earth’s PRIMA with a glancing blow after the terrestrial core
formation (Ringwood, 1992). The Earth’s core formation might have been preceded by
core-forming events in the large planetesimals that accreted onto the Earth. The Earth’s
core formation and the formation of the Moon must have occurred at least
62 ^ 10 m.y. after the iron meteorite formed. Jacobson and Harper argued for a
short (,15 m.y.) interval of core formation based on W isotope composition of Toluca,
an iron meteorite.
Gaber and Goldstein (1995) suggested an accretion and core formation interval of
80 ^ 40 m.y. for the Earth, using U – Th– Pb systematics. Heterogeneous accretion
(Wänke, 1981) is generally considered to be a multi-stage process with the bulk silicate
Earth (BSE) inventory of W entirely dominated (.99.9%) by late chondritic additions
(Newsom and Sims, 1991). The Hf/W ratio of the lunar mantle is probably similar to that of
the BSE given the similarity in W/U ratio (Newsom, 1986).
Recent studies (Lugmuir and Maclasore, 1995) of the 53Mn – 53Cr system (half-life
3.7 m.y.) indicate that the Cr isotope composition of the Earth and Moon are similar but
distinct (less radiogenic) from chondritic compositions.
However, the multi-stage character of planet accumulation makes it difficult to
define the timing of core formation. Again, there are many non-isotopic considerations
involved in core – mantle partitioning that might help constrain the physical circumstances
of core formation.
Cosmochemistry and Properties of Light Element Compounds 175

1.29.2. Core:Re/Os

About 99% of the Earth’s budget of Re and Os resides in the core. Consequently,
the Re/Os and 187Os=188Os ratios of the bulk core is the same as those of the bulk Earth and
carbonaceous chondrites.
40
1.30. K in the core
40
K is regarded as the radioactive heat source in the Earth’s core. The heat
production due to 40K in the Earth’s core is 4– 9 TW (TW ¼ 1012 W), which is a
substantial input for the present CMB heat flux. The existence of a core dynamo and the
geomagnetic field for the past 4 b.y. is possible with this additional heat. 40K has a half-life
of 1.25 b.y. and its decay must have contributed more to the heat in the early Earth (4.6 b.y.
ago) than today. Radioactive decay of 40K is responsible for ,10% of the heat loss of the
Earth. Heat is responsible for the convection and also for the magnetic field generation.
The amount of K in the core depends on the core’s content of O and S.
At the core of Mars, the 40 K heat production amounts to 0.1 –0.4 TW. This heat
could have contributed to the dynamo, driven by thermal convection, in the early history
of Mars.
This page is intentionally left blank
 177

Chapter 2
Petro-Tectonic Features of Terrestrial Planets

2.1. Introduction

All the terrestrial planets including the Earth’s Moon are presumed to show three
major physical divisions with respect to density and elastic/seismic behaviour. These
divisions are termed the crust, the mantle and the core.
The two major divisions of the Earth below the crust have sub-divisions: the mantle
has an upper and lower mantle separated by a transition zone, and there is a liquid
outer core and solid inner core. These represent the principal discontinuities. The
inner-core boundary (ICB) is identified with the freezing of the liquid-core iron alloy
and the 670 km (1,6008C) discontinuity is identified with the post-spinel transition.
Each layer is characterized by a difference in elasticity and density arising out of structural
as well as electronic changes.
As discussed in the following sections, the seismological models are based
on velocity – depth profiles determined from the travel – time –distance curves for
seismic waves and on periods of free oscillations. These models employ density,
pressure and elastic moduli as functions of depth. A preliminary reference
Earth model (PREM) representing the global seismological model has been ini-
tiated by the International Association of Seismology and Physics of the Earth’s
Interior. The other two models are related to the compositional and thermal
properties.
Compositional models are constrained by the seismological models presenting the
density and velocity profiles. These, in turn, are related to the geotherm and the thermal
boundary layers, allowing or forbidding convective layering.
Thermal models are based on experimentally determined thermodynamic
parameters, such as the global heat flux and geomagnetic variations. The temperature –
depth profile (geotherm) also delineates the compositional models.
The seismological, compositional and thermal Earth models of the Earth are
discussed in the following.
Seismological model: This principally involves measurement of the velocities of
P and S waves, from which determinations are made of the depth dependence of
pressure (P), density ðrÞ; bulk modulus ðKÞ and viscosity ðmÞ: Calculations are made
for the equation of state (EOS) (employing Birch law, Adams – Williamson relation,
etc.), the velocity profiles and the free oscillations.
178 Chapter 2

Thermal model: This primarily involves determination of temperatures at depths in

consideration of the anchoring temperature points at 400 km and at the ICB. Calculations
are made for the adiabatic gradient (using the Bullen parameter), viscosity of the mantle,
heat flow and geomagnetic variations.
Compositional model: This involves determination of the values of the bulk
modulus (K), viscosity ðmÞ and thermal expansion with respect to temperature and
pressure. The T; P phase diagrams are constructed and the melting temperature (Tm) and
bulk composition are stipulated.
The average density of the mantle immediately below the Mohorovicic seismic
discontinuity at the base of the crust is , 3.3 g cm23, which corresponds to a mineralogy
predominantly of olivine, pyroxene and a minor aluminous phase. The upper mantle is
considered to be of peridotite mineralogy but could in places be composed of eclogite,
a bimineralic assemblage of garnet þ jadeite (Na – Al-rich) pyroxene with a density of
3.3 –3.8 g cm23.

2.2. The Earth models

2.2.1. The PREM model

Knowledge of the nucleosynthesis in stars, the cosmochemical distribution of

elements and the composition of meteorites allow us to stipulate more accurately the
constitution of the interior of the Earth. The PREM is constructed on the basis of the
seismic (travel time of compression and shear waves) and the astronomical information
about the total mass and moment of inertia of the planet PREM, the most widely used
Earth’s model, is the work of Dziewonski and Anderson (1981). In this, r (the density) and
VS and Vp (the shear and compressional velocities) are presented as polynomials in radius
(thus defining the EOS). The variation of density, pressure and temperature with depth is
presented in Fig. 2.1.
The major density contrasts are seen at the base of the crust (60 – 100 km) and the
mantle – core boundary (2,800 km). Figure 2.2 presents the two discontinuities in the
transition zone at 400 and 670 km, delineating the upper and lower mantle; the pressures at
these discontinuities would be approximately 12 and 24 GPa. The pressure near the centre
of the Earth is around 360 GPa. [Note: A cubic anvil press can reach pressures up to
250 GPa (corresponding to the lower mantle) with a simultaneous temperature of 1,800 K,
whereas a diamond anvil can reach about 500 GPa, corresponding to the pressure at the
core of Jupiter.]
The major seismological discontinuities at the depths of 200, 400, 670, 2,890 and
5,150 km (Fig. 2.3) divide the Earth into radially symmetric shells. These discontinuities
represent the boundaries between (i) the uppermost mantle and transition zone, (ii) the
upper and lower mantle, (iii) the mantle and core and (iv) the outer and inner core,
respectively. The contrast in properties across the Earth’s primary structural boundaries is
presented in Table 2.1.
Petro-Tectonic Features of Terrestrial Planets 179

Figure 2.1. Density, pressure, and temperature variation in the Earth’s interior. Density and pressure are from
the Preliminary Reference Earth Model (PREM). Temperatures are based on thermodynamic parameters
determined from seismic profiles by Brown and Shankland (1981) (Navrotsky et al., 1992, q 1981, Royal
Astronomical Society).

The densities (r) at the boundaries as determined are:

(a) below the crust ¼ 3.33
(b) base of the mantle ¼ 5.5
(c) at the centre ¼ 12.97

Figure 2.2. Elastic wave velocities and densities for the crust and mantle of the Earth.
180 Chapter 2

Figure 2.3. Density and velocity changes with depths and the associated mineralogical changes (from
Bukowinski, 1994), q 1994 American Geophysical Union.

At the core – mantle boundary (CMB), the estimated temperature and density are
taken as T0 ¼ 3; 500 K and r0 ¼ 5:55 g cm23.

2.2.2. Seismological models

Seismological models are made with five functions: radius, the velocities of P and S
body waves VP and VS, the density and the attenuation factors in compression and shear
TABLE 2.1
Contrast in properties across Earth’s primary structural boundaries

Propertya Earth’s surface Core–mantle boundary Inner–outer core boundary

3
Density, Dr (mg/m ) 1.0 –2.6 4.3 0.6
Seismic wave velocities (km/s)
Compressional, VP 0.7 –2.3 5.7 0.6
Shear, DVS 0.0 –3.2 7.3 3.4
Electrical Conductivity, Ds (S/m) .0 –102 103 –104 0
a
Global average based on the preliminary reference Earth model (Dziewonski and Anderson, 1981) and a combi-
nation of geomagnetic observations and laboratory measurements of electrical conductivity (e.g., Merrill et al., 1996).
Petro-Tectonic Features of Terrestrial Planets 181

The inverse problem is solved simultaneously for elastic and anelastic parameters.
Perturbations are also taken into account to fit the data. For the uppermost 200 km, elastic
anisotropy is considered.
The seismic shear velocity V s displays the greatest sensitivity to T
perturbations, especially when inelastic effects are included, whereas the bulk-sound
velocity (V0) is an order of magnitude less sensitive to the temperature perturbations.
The anomalous lower mantle shows an unusually high ratio of shear velocity to sound
velocity.
The other waves, such as surface waves, are designated as Raleigh waves and Love
waves. In the former, the displacement of the surface is partly in the direction of
propagation and partly vertical. For the latter waves to form, there must be a superficial
layer resting on another layer, the velocity of S waves in the upper layer being less than in
the lower. The displacement in these waves is entirely horizontal and at right angles to the
direction of propagation.
Seismic tomography studies have shown that ½ðd ln Vs Þ=ðd ln Vp ÞP < 2
throughout the lower mantle (e.g., G. Robertson and J. Woodhouse (1995). Geophys.
J. Int., 123, 85).

2.2.2.1. Elastic constants

The elastic constants that are measured are E, s; m and K. The Young’s modulus, E,
Poisson’s ratio (change in diameter versus change in length in, say, a stressed wire), s; and
bulk modulus, K, are called the elastic constants of a material.
The elastic property calculations along a 1,4008C adiabat, adopted from Duffy and
Anderson (1989), give an excellent match for the absolute values for model S- and P-wave
velocities on the upper (shallow) side of 410 km. The mantle may be considered as
behaving like a Poisson solid. The ransom perturbations in velocity have an exponential
correlation function.
From a knowledge of the density distribution, the values of elastic constants can be
computed. Equations (2-11) and (2-12) give m and K directly. From these, it is possible to
compute the distribution of Poisson’s ratio, s; as:

s ¼ ð3K 2 2mÞ=ð6K þ 2mÞ ¼ ðVp2 2 2Vs2 Þ=2ðVp2 2 Vs2 Þ ð2-1Þ

Poisson’s ratio (see Fig. 2.4) of the inner core is high ðs ¼ 0:44Þ but that of the outer
core is still higher ðs ¼ 0:5Þ; which is expected for a dense liquid. A high Poisson’s ratio at
high pressure can be explained in terms of the second-order theory of elasticity.

2.2.3. Petrological models

Sampling of upper-mantle rocks is obtained from: (a) xenoliths in kimberlite-

and basalt-hosted magmatic rocks, (b) tectonically emplaced sections of the mantle such
182 Chapter 2

Figure 2.4. Variation of Poisson ratio (s) with depths (z) in PREM model (from Poiriar, 1991).

as ophiolites and orogenic peridotites, (c) serpentinized peridotite fragments dredged from
mid-ocean ridges (MOR) and (d) mineral inclusions in diamonds from mantle rocks.
These are composed primarily of three minerals: olivine, orthopyroxene and
clinopyroxene. The other minerals, which can be present as minor phases, are: plagioclase,
spinel, garnet, homblende and phlogopite mica. The vol% of major minerals at depths are
presented in Fig. 2.5 (after Ita and Stixrude, 1992) and Fig. 2.6.

Figure 2.5. The volume percentages of minerals at depth in the Earth for pyrolite (adapted from Ita and Stixrude,
1992, q1992 American Geophysical Union).
Petro-Tectonic Features of Terrestrial Planets 183

Figure 2.6. Mineralogical volume p.c. at different depths (km) in the mantle (piclogitic composition) (after
Weidner, 1986).

Peridotites with , 60% olivine, 20% clinopyroxene and 20% orthopyroxene are
called lherzolites. These often represent the “fertile mantle”, which, on partial melting,
produce basaltic magma seen in MOR. Much of the mantle is thought to be “fertile” and may
be equivalent to “pyrolite” composition, stipulated by Ringwood (1975). Pyrolite is assumed
to be a mixture of 80% harzburgite (olivine þ orthopyroxene peridotite) and 20% basalt.
Figure 2.5 presents the inventory of the phase proportions of an idealized pyrolite
mantle. Above and below the 410 km discontinuity line, the proportions of majorite garnet
and clinopyroxene change significantly. At shallower levels, clinopyroxene reaches a
value of ,28% and garnet decreases to ,12%. Majorite garnet reaches its maximum value
of ,40% at ,500 km depth below which it decreases as CaSiO3-perovskite becomes
stable (see also Fig. 2.5).
Chemically, the mantle is modelled in the following two ways:
(i) pyrolite mantle model: the mantle chemistry corresponds to peridotite plus basalt,
with no chemical differences between the upper and lower mantle.
(ii) piclogite mantle model: a chemical difference exists between the two mantles, the
lower one being richer in silica and iron compared with the upper one. The calculated
volume fraction of mineral phases for a piclogite mantle composition as a function of
depth is shown in Fig. 2.6 (after Weidner, 1986). The norm values of piclogite are
compared with those of pyrolite in Table 2.2 (Weidner, 1986).
Anderson (1989) opined that (i) the lower oceanic lithosphere is eclogite (richer in
garnet), rather than pyrolite, (ii) that the low-velocity zone (LVZ) in the mantle is enriched
rather than depleted, (iii) that the source region for the mid-oceanic ridge basalt (MORB) is
eclogite rather than peridotite and (iv) that the subducted ocean lithosphere sinks no deeper
than 670 km. But now it is postulated, as discussed in later sections, that the slabs may
penetrate up to the CMB.
184 Chapter 2

TABLE 2.2
Upper-mantle models (from Weidner, 1986)

Oxide Pyrolite Piclogite

Norm (wt%)
MgO 40.3 21.0
FeO 7.9 5.7
CaO 3.0 7.0
SiO2 45.2 48.9
Al2O3 3.5 14.4
Na2O 0.0 3.0
Mode (wt%)
Olivine 61 16
Orthopyroxene 15 3
Clinopyroxene 10 16
Garnet 14 36

Mantle stratification is implicitly evidenced by the discrepancy between the

composition of mantle xenoliths and chondritic meteorites. A major part of the chondritic
mantle must be made of pyroxene and garnet (unless the excess silicon is present in the
core). An olivine-rich upper mantle requires a silica enrichment at greater depths if
differentiation is to be operative in the chondritic mantle.

2.2.3.1. Pyrolite model

The primitive mantle material called pyrolite was initially presumed to have a
composition constituted of one part of basalt and four parts of dunite. The pyrolite
model of mantle composition is derived from the complementary melting relation-
ships of basalts and peridotites and contains , 60% olivine by volume. The phase
chemistry of pyrolite at 70 and 135 GPa is presented in Table 2.3. A portion of
a diagram showing the mineralogic make-up of a pyrolite mantle is shown in Fig. 2.5
and Fig. 2.6.
The model suffers an uncertainty because it is based on the properties of samples
derived at shallow (#200 km) depths and on the assumption that the mantle as a whole is
chemically homogeneous.
Li et al. (1998) compared the velocity – depth profiles to 660 km depth for a pyrolite
model. For depths in the upper mantle below 200 km (for which an adiabatic geotherm
applies), the pyrolite model velocities agree with the seismic data both in absolute values
and gradients. Also, Gaherty and co-workers concluded that a pyrolite olivine content
matches the seismic observations.

2.2.3.2. Piclogite model

Eclogite is a high-pressure derivative of metabasalt and shows almost no
anisotropy, which is a characteristic feature of the upper mantle. Eclogites, pyroxenites
and other lithologies are found in association with peridotites and these are derived through
Petro-Tectonic Features of Terrestrial Planets 185

TABLE 2.3
Representation phase chemistry of pyrolite at 70 and 135 GPa

70 GPa 135 GPa

MgPv Mw CaPv MgPv Mw CaPV

SiO2 56.9 n.d 54.0 55.4 n.d. 51.2

TiO2 0.2 n.d. n.d. 0.5 n.d. tr.
Al2O3 5.6 1.2 0.7 5.0 0.8 1.0
Cr2O3 0.5 1.1 n.d. 0.5 0.8 n.d.
MnO n.d. tr. n.d. n.d. tr. tr.
FeO 4.7 23.0 1.9 4.5 24.0 2.2
MgO 31.1 71.5 3.7 32.9 71.9 6.1
NiO n.d. 0.5 n.d. n.d. 0.4 n.d.
CaO 0.6 n.d. 37.7 1.2 n.d. 36.6
Na2O tr. 2.1 n.d. n.d. 1.6 0.9
K2O n.d. n.d. n.d. n.d. n.d. 1.8
Mg# 92.2 84.7 92.8 84.2

Analytical uncertainties are about 1% relative for all elements. A portion NiO inventory is unaccounted for. It has
probably been reduced to Fe –Ni alloy to stabilize minor amounts of ferric iron as required by the major bases.
n.d., below detection; tr., qualitatively discernible at levels near the detection limit. Mg# ¼ 100 MgO/(MgO þ
FeO) molar.

the recycling of oceanic crust into the mantle or through precipitation of pyroxene ^
spinel ^ garnet during the ascent of magmas to the surface. The calculated mineralogical
composition of piclogite mantle is shown in Fig. 2.7.

2.2.4. Convection model

The seismic and density parameters as observed of the lower mantle may be seen to
be compatible with two compositional – thermal models:
(1) a pyrolite composition at the relatively low geotherm temperatures appropriate to
whole-mantle convection; and
(2) a pyroxene –perovskite composition at higher geotherm temperatures appropriate to
layered-mantle convection.
Protagonists of layered convection — generally geochemists — point to the
differences in trace element, isotopic and rare H gas abundances between ocean –
island and mid-oceanic ridge volcanic rocks (e.g., Turcotte et al., 2001). Geophysicists
favouring whole-mantle convection use the pattern emerging from seismic tomographic
imaging.
Reconciliation of two models are attempted by arguing that inhomogeneities may
arise from incomplete stirring and from ponding of subducted material at the base of the
upper mantle followed by avalanches of material into the lower mantle (Tackley, 1993).
Large volumes of subducted material may remain in the transition zone (400 – 700 km)
because of buoyancy.
The models of circulation in the mantle are shown in Fig. 2.8. In the “two-layer
convecting model” the layers are kept separate by an intrinsic compositional density
186 Chapter 2

Figure 2.7. The volume fraction of mineral phases for a piclogite mantle as a function of depth (after Weidner,
1986, q1986 Springer).

contrast or by a negative P –T slope for an isochemical phase transition at 670 km

(< 660 km; Christensen, 1995). In whole-mantle convection models, subducted litho-
sphere penetrates down below the 670 km boundary and most plumes originate from the
CMB (at 2,900 km depth).
However, the two most discussed LM compositional models are: (i) the perovskite
model, composed almost entirely of Mg0.88Fe0.12SiO3 with other minor phases, and (ii) the
pyrolite model, composed of magnesiowüstite and MgSiO3 perovskite, wherein iron is
preferentially partitioned into the magnesiowüstite.
The density of the mantle minerals versus pressure shows, when plotted, that the
density of these minerals converge at high pressures. Knittle et al. (1986) found that the
adiabatically decompressed density of the lower mantle is incompatible with the high-
temperature, zero-pressure density of a pyrolitic lower-mantle assemblage, but is
compatible with that for a perovskite mineralogy.
However, in addition to the requirements of adiabaticity and chemical heterogeneity,
zero-pressure constraints on the composition of the lower mantle rely on the validity of the
adiabatic extrapolation of the seismic properties and also on about 2-fold extrapolation of
perovskite’s thermal expansivity. In addition, an adiabatic extrapolation of the lower
mantle to zero pressure and the extrapolation of perovskite zero-pressure properties to
high temperature may approach, or even cross, polymorphic phase boundaries in
perovskite. Thus, zero-pressure constraints on the lower-mantle composition cannot
discriminate between pyrolite and pyroxene composition models of the lower mantle
(Knittle et al., 1986).
The major chemical components of the upper mantle are transformed primarily as a
mixture of (Mg, Fe)O magnesiowüstite and (Mg, Fe)SiO3 perovskite. Observed variations
Petro-Tectonic Features of Terrestrial Planets 187

Figure 2.8. The whole mantle (a) and layered convection (b) models.

in seismic velocities and densities may lead one to postulate chemical stratification
between the upper and the lower mantle. This may imply convecting layers and a relatively
large thermal inertia of the Earth, in contrast to an isochemical mantle model which
implies whole-mantle convection and lower thermal inertia.

2.2.4.1. Mantle convection at Archean –Proterozoic transition

Geological records of late Archean eon (3 –2.5 b.y. ago) show profound geological
changes, such as rapid growth of continents, deposition of large volume of cratonic
sediments, and cratonization of continental crust with widespread emplacement of K-rich
granitic rocks. This period witnessed change in atmospheric chemistry and the volume of
the oceans, and ended with the Huronian glaciation (earliest documented). This transition
is also marked by a strong increase in geomagnetic field, and the ocean volume changes
with mantle degassing and regassing.
Breuer and Spohn (1995) showed through calculation that in this period in late
Archean the two-layer convection must have broken down leading to tectonic and volcanic
activities which resulted in geological and paleoenvironmental changes.
188 Chapter 2

2.2.4.2. Mantle Raleigh number and flush instability at late Archean

The spinel – perovskite phase transition boundary at 670 km depth, separating the
upper and lower mantles, plays a crucial role in the Earth’s dynamics. It did so even in the
late Archean eon (3 – 2.5 b.y. ago). This phase boundary appears to be stabilized by two-
layer convection at high Raleigh numbers (# 107). Convection currents could easily run
through the boundary and the convection becomes whole-mantle type.
With gradual cooling of the mantle the Raleigh number decreases in the lower
mantle, and the hot lower mantle material shoots up into the upper mantle. Models
explaining plate tectonics since early Archean presume two-layer mantle convection,
established soon after the accretionary formation of the Earth.
The crucial recycling rate is proportional to the area of the continental crust and the
plate speed. The heat flow at the bottom of the crust is proportional to the crust production
rate and is typically a few tens of percent of the oceanic heat flow.
After an evolution through about 1.5 Gyr (from 4.5 b.y. ago) the beginning of rapid
continental growth occurred and flush instability also seems to have occurred. For this
onset of instability the Raleigh number reached the crucial value by way of rapid heat
exchanges between upper and lower mantle layers. At the end of the Archean about 50% of
the continental crust was formed with nearly the present thickness (Breur and Spohn,
1995).
The rapid cooling of the lower mantle through instability enhanced the core
cooling and the vigour of the convection in the core, leading to increased dynamo
activity.
An increase in upper mantle temperature by 200– 400 K and an increase in heat
flow by a factor of three may have caused cratonization of continental crust by intracrustal
differentiation through partial melting of the lower mantle.

2.3. Physical parameters

The density distribution in the Earth acts on the seismological models. To

understand this, we need to argue based on the following known parameters:
Mass of the Earth ðMÞ ¼ 5:974 £ 1024 kg
Mass of the mantle ¼ 4.11 £ 1024 kg
Earth’s mean radius ðrÞ ¼ 6; 371 km
Earth’s mean density ðr
Þ ¼ 5:515
Crustal average density (rav) ¼ 2.4 –3.3
Density at the CMB ¼ 1.07 £ 104 kg m23
Density at the ICB ¼ 1.33 £ 104 kg m23
Moment of inertia about its rotation axis (I) ¼ 0.33 Mr2
That the Earth is layered inside can also be inferred from the fact that, had the Earth
been homogeneous (not layered), its moment of inertia would have been higher to the
extent of 0.4 Mr2, rather than the observed value of 0.33 Mr2.
Petro-Tectonic Features of Terrestrial Planets 189

2.3.1. Parameter changes with depths

Let us evaluate the change in physical parameters of the Earth with depth (2 z) with
respect to its radius, r. Assuming a spherically symmetric Earth in hydrostatic equilibrium,
the pressure change at radius r (r ¼ 6; 371 km 2 z) is
dP ¼ 2rg dr
where r ¼ density, g ¼ acceleration due to gravity at radius r and
ðr
22 22
gðrÞ ¼ Gmr ¼ 24pGr rr 2 dr
0

where m is the mass of the sphere of radius r and density r and

G ¼ 6:67 £ 10211 m3 kg21 s22 is the gravitational constant.
The pressure – depth profile follows immediately by integration of the above
equation. From the velocity – depth profiles (Figs. 2.2 and 2.3) and the normal mode
frequencies, seismological models yield density and elastic moduli inside the Earth.
Assuming hydrostatic compression at depth z, pressure P is given by
ðz
PðzÞ ¼ rðzÞgðzÞdz
0

where rðzÞ and gðzÞ are, respectively, the density and acceleration due to gravity at depth z:
When we consider the gravitational potential, U,

dU=dr ¼ 2g ¼ dP=dr r21

In terms of Laplacian spherical coordinates, the relation of U becomes

72 U ; 1=r 2 d=dr½r 2 dU=dr

Again, the bulk modulus K, when adiabatic (with no exchange of heat), is related to
r and P as:
dK=dP ¼ d ln K=d lnr ¼ n ¼ K 00
When n < 4; it satisfies the second-order Birch –Murnaghan equation (see Section 5.8.7.1).
Pressure derivative with respect to density is proportional to density
dP=dr ¼ Cr
where C is a constant. Pressure at the Earth’s centre (Pc) by PREM model is found to be
3.64 Mbar while that at the deepest bottom of the Pacific ocean is , 2 kbar.
The liquidus temperatures (T ) in the convecting mantle material at different depth
pressures are:
5 GPa ¼ , 1,3008C (Jeanloz and Morris, 1986)
13 GPa ¼ , 1,4508C (Katsura and Ito, 1989)
23 GPa ¼ , 1,600 – 1,7008C (Ito and Katsura, 1989)
190 Chapter 2

2.3.1.1. Lithosphere
Rheological structure. Rheologically, the crust may be divided into upper elastic and
lower viscoelastic layers. As early as 1960, Griggs and his co-workers showed that upper
crustal rocks are relatively strong and yield by fracture but, above , 0.5 GPa confining
pressure (and 5008C), nearly all rocks deform by flow (Griggs et al., 1960). The uppermost
elastic layer, which does not creep significantly over Geological period, demonstrates that
possibly large-density anomalies existed in early Precambrian terrains. There is an elastic
layer , 10– 30 km thick (depending on geothermal gradient) overlying a ductile layer,
which may behave as a viscoelastic material with stress-dependent viscosity decreasing
with depth. Walcott (1970) showed that the upper fifth of the lithosphere behaves as an
elastic body and the lower part behaves as a viscoelastic body of viscosity 1023 N s m22,
which is higher than the 1020 N s m22 for the underlying asthenosphere. The viscosity of
olivine rock is more than two orders of magnitude less than that of dry olivine (Hirth and
Kohlstedt, 1996).
In regions where the geothermal gradient is high, the elastic layer should be
thinner and the stresses are correspondingly greater than average. The lateral density
contrast at a continental margin of aseismic type gives rise to tensional stress in the
continental crust. The resultant deviatoric stresses are distributed throughout the
continental crust, reaching a maximum at middle crustal depths. The maximum
deviatoric stresses occur in the middle crust for the elastic model but occur in the upper
crust for the viscoelastic model. The stress amplification by the rheological layering of
the lithosphere occurs due to stresses produced by body forces as well as those produced
by boundary forces (Kusznir and Bott, 1977).

Viscosity and strain rate (1̇). A generalized model for the “strength” of the Earth at depths
has been developed by Hopper and Buck (1993). Strength is measured by the level of stress
required to maintain a strain rate of 10214 s21.
The vertical averages of a wide range of possible rheological profiles
for the continental lithosphere probably obey a relation of the form (England and
McKenzie, 1982).
t
ij ¼ BE_ 1=n21 1_ij ð2-2Þ
where 1_ij is the ijth component of the strain rate, assuming independence of depth.
The spatial derivatives of deviatoric stress, which balance the gravitational forces,
are, therefore, dominated by spatial derivatives of either the constant B or the unit tensor of
strain directions 1ij =1;
_ rather than the strain rate magnitude 1: _ n ¼ 3 is appropriate for
relating the depth-averaged deviatoric stresses tij and strain rate 1ij (Sonder and
England, 1986).
Assuming B ¼ 1:3 £ 1012 N m22 s1/3, the viscosity of the lithosphere is approxi-
mately 1022 Pa s (Table 2.4). p
The strain rate ð1Þ _ 2; effective viscosity of the lithosphere ½h ¼
_ ¼ E= p
ð1=2ÞBE1=n21  and vertically averaged deviatoric stress t½¼ B1_1=n 2ð1=n21Þ ; calculated
for B ¼ 1:3 £ 1012 N m22 s1/3 and n ¼ 3; are shown in Table 2.4 (England and
Molnar, 1997).
Petro-Tectonic Features of Terrestrial Planets 191

TABLE 2.4
_ effective viscosity ðhÞ and deviatoric stress ðtÞ in the lithosphere (see text) (England and Molnar,
Strain rate ð1Þ;
1997)

1_ (s21) h (Pa s) t (MPa)

214 21
10 1.1 £ 10 22
10215 5 £ 1021 10
10216 2 £ 1022 5
10217 1.1 £ 1023 2

The viscosity, inferred for the strain rates in central Asia (10215 – 10216 s21), is
close to 1022 Pa s. The stress drop of about 107 Pa in continental earthquakes is a
lower limit on the strength of continental upper crust. The viscosity below the
lithosphere is 1020 – 1021 Pa s (Haines, 1998). The viscosity, ,1022 Pa s, appropriate to a
typical strain rate of 10216 – 10215 s21, is higher than the viscosity of the convecting
upper mantle by one – two orders of magnitude [Note: On this basis, the continental
lithosphere of Asia may be regarded as belonging to the fluid portion of the solid
Earth relative to the other small fractions of the Earth acting as rigid plates (England and
Molner, 1997).]
The viscosity of mantle increases from , 3 £ 1020 Pa s in the upper mantle to
22
10 Pa s in the deep mantle. The increase in viscosity with depth increases circulation
(residence) time with lesser mixing, which helps slab penetration deep into the lower mantle.
England and Molnar (1997) obtained spatially compatible strain rates for the upper
crust in central Asia from geological observations. They could relate the spatial derivatives
of the strain rates in the upper crust to the horizontal forces associated with spatial
variations of the gravity potential energy of the whole lithosphere arising from the
topography of Tibet (England and McKenzie, 1982). They estimated the overall viscosity
of lithosphere as 1022 Pa s.
Lame parameter l and density r of the oceanic lithosphere down to the upper
mantle of depth 213 km have been determined as:
0– 8 km depth (layer 1): l ¼ 35 GPa; r ¼ 2; 900 kg m23 (purely elastic)
8– 62 km depth (layer 2): l ¼ 65 GPa; r ¼ 3; 400 kg m23 (purely elastic)
62 –213 km depth (layer 3): l ¼ 75 GPa; r ¼ 3; 400 kg m23 (not purely elastic)
Although the base of the asthenosphere is not well defined seismically, a large
viscosity increase probably occurs at a depth of 200– 250 km because of the pressure
dependence of the viscosity of olivine (Karato et al., 1993). For the lithosphere, a Maxwell
rheology is taken conforming to the deformation of olivine in diffusion creep regime
(Karato et al., 1993).

Core viscosity. Given the physical conditions of the Earth’s core, theoretical
calculations for liquid iron predict a viscosity of 1022 Pa s (Wijs et al., 1998). The
kinematic viscosity n ¼ m=r of iron, where r is the fluid density, is comparable with
the value for water.
192 Chapter 2

Strain rate ð1Þ:

_ COH effect. Strain rate is a function primarily of temperature (T ), stress
(s) and water content (COH); the boundary must be a surface in a three-dimensional (3D)
space (T – s – COH) defined by a fraction f ðT; s; COH Þ ¼ K; where K is a constant.
Seismological observations show that the asthenosphere is a typical oceanic upper
mantle (for which COH < 800 ppm H/Si; Hirth and Kohlstedt, 1996) and s < 1– 10 MPa
has a fabric corresponding to type-A, having VSV . VSH anisotropy beneath MORs.

2.3.2. Parameterized PREM model: EOS

Since PREM is a parameterized model, it implicitly imposes an EOS of its own on

the data to be fitted (for EOS see Section 5.8.7).
Stacey (1998) emphasizes that PREM assumes a homogeneous lower mantle but
the Bullen seismological homogeneity parameter, h; does show small deviations from
unity, which, however, can be attributed to the independent parameterizations of r and the
seismic velocities (Stacey, 1997). When K 0 ð¼ dK=dPÞ is considered, the effect of
parameterization becomes very serious. Indeed, no proposed EOS matches the K 0
tabulation of PREM. Thus, when the third-order Birch equation is used, it leads to K 00 ¼
3:8 (Jackson, 1998), a value which is absurd for any lower-mantle mineralogy.
Stacey (1992, 1995) found the linear relationship between rigidity, m, K and P as:

m m  P

¼ 1 2 K 01 ð2-3Þ
K K 0 K

in which K 01 ¼ ðdK=dPÞP1 (K 01 for the lower mantle is taken as a fixed constant, 1.425
(Stacey, 1998)). The parameter K 01 gives a fixed end point to the EOS as an adiabat
matching the Earth itself.
In equation (2-3), the elastic constants m and K are affected by the parameterization
of PREM in the same way and their ratio is unaffected by the EOS implicit in the PREM
model.
The lower mantle is close to adiabatic (e.g., Stacey, 1997). Thus, an EOS
obtained by fitting an Earth profile gives adiabatic quantities. That is, the bulk modulus and
all higher derivative parameters in the equation are adiabatic, so that K ; KS ¼
rðdP=drÞS ; K 0 ; K 0S ¼ ðdKS =dPÞS and so on, where subscript S indicates constant
entropy.
For the lower mantle, Stacey and Isaak (2000) find that ðdK 0 =dTÞP¼0 is
(6 – 23) £ 1025K 21. The behaviour of this higher-order derivative necessitates a
reconsideration of the lower-mantle EOS fitting by requiring a higher value of K 0SO (K 0S
at P ¼ 0) than usually considered.
The EOS study on CaSiO3-perovskite (pvs) showed elastic properties which
make this phase an invisible component of the lower mantle. Hence, the amount of CaSiO3
can be determined using PREM (Hema and Suito, 1998). However, plane-wave
pseudopotential study (theoretical) revealed that CaSiO3 pvs is an elastically visible
component in the lower mantle (Karki and Stixrude, 1999).
Petro-Tectonic Features of Terrestrial Planets 193

2.4. Seismic model: discontinuities

The simple seismological model considers the Earth divided into radially
symmetric, homogeneous and isotropic shells separated by seismic discontinuities, as
discussed below (see Fig. 2.9).

Figure 2.9. The Earth’s crust and mantle layers (modified from Madon, 1992, q 1992 Academic Press, Inc.).
194 Chapter 2

2.4.1. Seismic discontinuities: Moho to the D00 zone

At about 35 km depth, the discontinuity that was first investigated by

A. Mohorovicic in 1909, is known simply as “Moho”, which is commonly assumed to
lie between 10 and 50 km (Fig. 2.9).
The upper mantle extends from Moho to a depth of 670 km. It is characterized
by rather low P- and S-wave velocity gradients. Its uppermost part of about 70 km
thickness is called the lithosphere, characterized by high-velocity seismic waves
implying solidity and high strength of materials. Below the lithosphere is the
asthenosphere (i.e., a zone of weakness, showing low velocity in seismic waves
especially low S-wave velocity), which extends to a depth of 220 km. This low-velocity
zone may be caused by incipient (, 1%) partial melting of materials or may be due to the
vibration of pinned dislocations in the constituent minerals (such as olivine, pyroxene,
etc.) (Mitra and Bhattachariya, 1982). Therefore, this LVZ is not due to chemical or
phase change, as are seen in other cases.
Around hot spots, MORs and continents seismic tomographic studies reveal large
lateral velocity heterogeneities at depths of several hundred kilometres (e.g., Su et al.,
1994). This seismic signature may be the result of a combination of chemical, thermal and
anisotropic effects.
The seismic velocity changes at 400 and 670 km depths are due to phase changes
such as: (Mg,Fe)2SiO4 a-olivine ! b-phase at 400 km depth (crust – mantle boundary),
followed downwards by b-phase to g-spinel transition and (Mg,Fe)SiO 3-
perovskite(rv) þ (Mg,Fe)O –magnesiowüstite at 670 km depth (transition zone) (see
Fig. 2.9). The S-wave velocity changes due to these transitions are shown in Fig. 2.10
(after Duffy and Anderson, 1989). The mantle below 660 km is marked by a sharp
rise in the density and transmission speed of seismic waves. The g ! pv þ mw
transition occurs at P and T (, 24 GPa and 1,800 – 1,900 K) to approximately 660 km
depth.
The 670 km discontinuity also represents a general change in silicon– oxygen
coordination from 4 to 6. By an experiment using a laser-heated diamond-anvil cell under
hydrostatic conditions, where P and T can be monitored precisely, Chopelas and Boehler
(1992) observed that, at 24 GPa (, 670 km depth), the transition takes place at a
temperature of ,1,900 K. The major discontinuities at 400 and 670 km reflect
density increases of about 6 and 10%, respectively, and the seismic velocity
increases more rapidly. The implied density change with depth can be related to the
phase changes.
The lower mantle extending from 670 to , 2,900 km is characterized by a gradual
change in seismic velocity and density with depth. However, significant lateral variations
in seismic velocities have been uncovered with possible weaker discontinuities (e.g., Bina,
1998b).
At a depth of , 2,800 – 2,900 km, a layer known as the D00 layer (see Section 2.13.3)
is characterized by a decrease in wave velocities. This layer interacts with the liquid outer
core (see Section 2.14).
Petro-Tectonic Features of Terrestrial Planets 195

Figure 2.10. Shear wave velocity versus depth. The S-wave velocity jumps are associated with the olivine to
b-phase, and b-phase to g-spinel transition. The velocities were calculated along a 1,4008C adiabat (following
Duffy and Anderson, 1989) [Source: Agee, 1998, q1998, Mineralogical Society of America].

2.5. Thermal structures of the Earth’s mantle

The Earth’s thermal structure is constructed from laboratory studies on phase

equilibria, melting and thermodynamics at extreme P; T conditions. In recent years, there
has been an appreciable quantum of work on phase transition and melting of materials
relevant to the lower mantle and core. LA large lateral temperature gradient in the lower
mantle is suggested from lateral velocity anomalies observed in seismic tomography and
the volume – sound velocity systematics. The temperature profile of the Earth is presented
in Fig. 2.11.
The thermal gradient measured at shallow depths is about 10 times as high as the
adiabatic gradient in the mantle; this has been derived from thermodynamics and the
distribution of seismic velocities.
Wolf and Bukowinski (1987) plotted a range of possible geotherm temperatures for
the lower mantle along with their adiabatic extrapolations to zero pressure. The lower
temperature geotherms would correspond to a chemically homogeneous mantle and
whole-mantle convection, while the relatively higher-temperature geotherms would be
associated with chemical stratification and layered convection (Brown and Shankland,
1981; Richter and McKenzei, 1981).
There are two major thermal boundary layers in the lower mantle and in the core.
The high-pressure minerals of the transition zone transform to the denser phase
196 Chapter 2

Figure 2.11. Suggested temperature profile of the earth (modified from Madon, 1992, q 1992 Academic
Press, Inc.).

(Mg,Fe)SiO3-perovskite phase plus (Mg,Fe)O –magnesiowüstite phase. This transform-

ation occurs at a pressure of about 24 GPa (Fig. 2.12) corresponding to a depth of 660 km
(rather than 670 km), where a large jump in the seismic velocity is observed. Seismically,
this transition is very sharp to within a few kilometres. The second is the inner core – outer
core boundary, where the temperature is the same as the melting temperature of the core
material, which is mainly iron. If the mantle convects as a whole, these fixed regions and the
adiabatic gradients in the outer core and in the lower mantle would then constrain the
temperature increase across the CMB.
The temperature of the CMB is poorly known and is calculated to be 4,300 K
(de Wijs, 1998). This temperature is linked to the melting temperature of iron, which
is, however, affected by the presence of light elements whose property is constrained
by density arguments (see Section 14.2.2).

2.5.1. Temperature – depth relation

The temperatures in the Earth at different depths ðzÞ determined by different

workers are presented below (Table 2.5). Near the surface, the temperature gradient with
depth is approximately 208C/km and the gradient becomes 108C/km at a depth of 40 km. In
general, the temperature gradient in the upper mantle is nearly constant to138C/km. The
pressure gradient is also about 1/3 kbar km21. A schematic temperature profile of the Earth
was presented earlier in Fig. 2.11.
Petro-Tectonic Features of Terrestrial Planets 197

Figure 2.12. Pressure-temperature diagram for the upper mantle. The fields for plagioclase, spinel-and garnet-
facies lherzolite are shown at the top of the mantle (e.g., “Plagioclase-L”). Abbreviations: Mj, majorite; Mg-Pv,
magnesium perovskite; Ca-Pv, calcium perovskite; Mw, magnesiowüstite. Solidus and liquidus are shown for an
anhydrous peridotite (modified from Philpotts, 1990).

The approximate P; T change with depth may be considered as in the following:

Near the surface: 12 bar 8C21
At the depth of 40 km: 25 bar 8C21
Upper mantle: 250 bar 8C21
Between the Moho and the uppermost mantle, the gradient is 40 –50 bar 8C21. In
the lower mantle, the pressure gradient increases to 1/2 kbar km21 and the temperature
gradient does not exceed 18C. Hence, the parameter change is affected by 1/2 kbar 8C21
change. Thus, in the lower mantle, the compressibility dominates over the thermal
expansion; however, the compression as a result of phase transformation predominates.
By temperature estimation at different depths, the following observations have been
made:
(i) In the lithosphere, the temperature rises rapidly and the heat transfer is by conduction.
Here, the geotherm can be derived from the thermobarometric measurements of rock
198 Chapter 2

TABLE 2.5
Temperatures inside the Earth (in Kelvins)

Depth Sources
z (km) W72 V80 A82 SS84 BS85 BM86 P86 W87 IK89

100 1,273 1,450

150 1,473
350 1,400
371 1,662
655 1,600
6712 1,830 1,970 1,873
671þ 1,980 2,300
1,300 2,800
2,571 3,300 2,814
2,8862 2,937 3,000 2,773
2,886 þ 3,637 3,800 3,573 3,800 3,800
5,156 4,676 5,000 5,000 5,000 6,600
6,371 4,805 5,000 6,600

Source: W72: Wang (1972b); V80: Verhoogen (1980); A82: O. L. Anderson (1982); SS84: Spiliopoulos and
Stacey (1984); BS85: Brown and Shankland (1981); BM: Brown and McQueen (1986); P86: Poirier (1986); W87:
Williams and Jeanloz (1987); IK89: Ito and Katsura (1989).

samples such as mantle xenoliths. The diamond– graphite stability curve intersects
the continental geotherm at 130 km depth. In Fig. 2.13 (Aoki, 1984), oceanic and
continental geotherms along with the P –T stability fields for quartz –coesite,
graphite – diamond, coesite stishovite and forsterite modified spinel are shown.
(ii) A temperature of ,1,100 to ,1,2008C is assumed where the geotherm approaches
the melting point curve of mantle minerals (e.g., at asthenosphere) (see Fig. 2.11).

Figure 2.13. Pressure– temperature stability fields for quartz–coesite, graphite–diamond, coesite–stishovite and
forsterite–modified spinel, and coesite and continental geotherms (from Aoki, 1984).
Petro-Tectonic Features of Terrestrial Planets 199

(iii) The geotherms at 400 and 670 km depths are considered as 1,500 and 1,6008C,
respectively. Across the D00 layer, a temperature difference of 8008C is presumed.
(iv) Presuming 1,6008C at 670 km and using a seismological adiabatic gradient, a
temperature of , 3,3008C (; m.p. of iron) on the CMB is reached.
(v) All the above anchor the geotherm at , 4,7008C at the boundary between the outer
core and the inner core.
In the thermal boundary layers (viz. crust – mantle and core –mantle) heat is
transferred by conduction, mostly because of high viscosity and the small thickness of the
boundary layer.
Within the mantle, heat is transferred by convection because of the high value of the
Rayleigh number ðRa ¼ 107 Þ compared with the critical value ðRaCr , 103 Þ: The mantle
can be considered as a convecting fluid with very high viscosity (1021 – 1022 Pa s21).
The heat flow reaching the surface from the interior is ,80 mW m22 and the
cooling rate is ,100 K m.y21.
At the top of the upper mantle, the thermal expansion is compensated for by
lithostatic compression. At great depths, the compression due to high pressure dominates
over the expansion caused by the very slowly rising temperature. In the depth interval of
40 km (,10 kbar) to 320 km (110 kbar), the coefficient of lithostatic compression, b; is
estimated to be around 0.5– 0.7 (£1026) bar. Consequently, the densities of the pyroxenes,
olivines and garnets that make up the upper mantle increase monotonically by 5 –7%
during subsidence from depths of 40 –320 km. With depth, the thermal expansion is
slowed down as a consequence of a decrease in the temperature gradient with depth. At the
top of the upper mantle, a compensation is reached between expansion and pressure
compression.
In the transition zone depths of 320 – 400 km, the phase transitions such as
pyroxene ! garnet and olivine ! b Mg2SiO4 spinel (Ringwood, 1975) involve about
10% density increase and create the seismic discontinuity. At greater depth, compression
predominates.

2.5.1.1. Heat sources and heat flow

As already stated, the mean value of the surface heat flux from within the Earth is
about 80 mW m22. The sources of heat are outlined below.
(a) The primordial heat: In the early stage of gravitational accretion, heat was
generated from the impact of particles, which partly melted the growing Earth. There was,
in addition, a contribution from short-lived, now extinct, radioactive elements such as 26Al
(half-life of 7 £ 105 a).
At a later stage, when differentiation ensued and droplets of liquid iron or iron–
sulfur eutectic trickled down to form the core, gravitational energy was again released in
the form of thermal energy. The core formation was relatively rapid, having been
completed in about 0.5 b.y. after the formation of the body of the Earth.
(b) Radiogenic heat: The main source of heat in the mantle is from the decay of
radiogenic isotopes to their daughter products, viz.:
235
U !207 Pb; 238 U !206 Pb; 232 Th !208 Pb and 40
K !40 Ca; 40A
200 Chapter 2

Radiogenic heat production in the mantle accounts for no less than 2.3 £ 103 W,
which constitutes about 60% of the total output of heat.
(c) Other sources of heat: Latent heat exsolved from external phase transitions
(olivine ! spinel), frictional dissipation in the convecting mantle and tidal dissipation in
the solid Earth (more so when the Moon was closer to the Earth prior to 400 Ma) contribute
to a lesser extent to the heat budget of the Earth. Crystallization of the solid inner core
released latent heat, which added to that generated by the gravity falling of the solid iron to
the inner core.
For effecting the convection, the heat flux received by the lower mantle from the
core Qc is < 3.6 £ 1012 W, which approximates to about 10% of the total heat output of the
Earth. Current global heat loss is 44.2 £ 1012 W (44.2 TW) (Pallack et al., 1993), which is
contributed by radioactive decay of 40K, U and Th and from secular cooling of the Earth.
K/U ratios in igneous rocks range about 12,500. At present, the inventory of radioactive
elements produces 20 TW or about 45% of the total heat flow. The heat reaching the
surface yields a radioactive heat contribution of 24.9 –32.9 TW (Van Schmus, 1995).
Parametrized convection models of the Earth’s interior cooling yield a secular cooling
rate of 30– 100 K G yr21 (e.g., Christensen, 1985). This loss produces heat fluxes of
5.9 –20 TW.

2.5.2. Thermal anomalies

2.5.2.1. Upper mantle

Tomographic studies of upper-mantle heterogeneity generally show good
correlation with surface tectonic features and can be used to infer the depth extent of
the thermal anomalies that are associated with MOR and hotspots. Satisfactory correlation
between S-wave velocity, bathymetry and basalt chemistry were observed to exist beneath
the Mid-Atlantic Ridge at depths of 100– 200 km. The temperature variations at these
depths were estimated to be in the range of 100 –300 K (Zhang et al., 1994).
The global heat-flow patterns correlate well with seismic tomographic maps at
shallow depths, implying that seismic shear velocity variations at these depths are strongly
dependent on temperature. The possibility that temperature in parts of the upper mantle
may be sufficiently high to produce partial melts at depths greater than 300 km has
received support from a shear wave reverberation study, and sampling the mantle from
beneath the Sea of Japan (Revenaugh and Sipkin, 1997).
Short-period array data revealed that the 660-km discontinuity is depressed by
20 –30 km and the 410 km discontinuity is elevated by ,15 km beneath the subduction
zones. These together imply an average temperature difference of 300 –400 K between the
subduction zone and the normal mantle (Vidale and Benz, 1992).

2.5.2.2. Lower mantle

If the mantle is compositionally layered, a thermal boundary layer must exist at or
near the 660 (^10) km discontinuity that will exert a profound influence on temperatures in
the lower mantle.
Petro-Tectonic Features of Terrestrial Planets 201

The lower mantle (670 –2,890 km depth) is the largest single region of the Earth’s
interior, making up 55% of its volume. As such, it dominates the processes of mass,
momentum and energy transport in the deep interior and hence may have a substantial
influence on the planet’s thermal and chemical evolution.
The melting curve of perovskite, the dominant phase in the lower mantle,
represents an upper bound to the temperature in this region. This, therefore, places
important constraints on the thermodynamic and rheological properties of the lower
mantle. Extrapolation of the melting curves (to 32 GPa) of MgO and (Mg,Fe)O
indicates that these materials melt at 3,500 – 5,000 K at lower-mantle pressures (Zerr
and Boehler, 1993) but molecular-dynamics calculations of MgO melting temperatures
are higher by several thousands kelvin (Cohen and Gong, 1994).
The thermoelastic data for magnesiowüstite and silicate perovskite suggest that
there could be an enrichment of silicon in the lower mantle (e.g., Zhao and Anderson,
1994). Also, an enrichment of the lower mantle in iron is required for the higher
temperatures that would be expected in the deep interior of a chemically stratified planet.
Numerical simulations of convection and two-dimensional and axisymmetric
calculations showed that the presence of the endothermic 660 km boundary promotes
layered convection (e.g., Solheim and Peltieer, 1994). When the Clapeyron (P – T ) slope of
the phase transition is included, downwellings are found to be impeded at the boundary
while rising plumes are unaffected (Liu, 1994) Episodic flow across the 660 km
discontinuity can also produce temperatures in the transition region that are cooler than the
surrounding mantle due to the temporary stagnation of cold subducting material in the
region (Weinstein, 1992).
Sound velocity measurements at high P and T under shock compression have
shown that the compressional velocity – temperatures scaling relation at deep mantle
pressures (,100 GPa) is , 5 times smaller than ambient pressure values (e.g., Duffy and
Ahrens, 1994). From this, it has been estimated that long-wavelength velocity
heterogeneities in the deep lower mantle correspond to root-mean-square thermal
anomalies of ,150 K. The potentially important role of anelasticity in velocity –
temperature scaling relations has been discussed by Karato (1993).
Studies on thermal anomalies in relation to seismic tomography have shown that
large-scale anomalies found in the lower mantle have outer temperatures of 400 K above
the surrounding mantle and thermal anomalies in excess of 1,000 K near the plume centre
(viz. Cadek et al., 1994). In studies of diffracted and reflected waves in the D00 layer,
Wysession et al. (1993, 1994) argue that thermal anomalies as large as 400– 1,000 K might
explain seismic velocity perturbations near the CMB. Thus, thermal anomalies are most
likely to exist in the deep lower mantle.

2.5.2.3. Thermal structure of the core and CMB

The high temperatures in the interior provide the driving force of the Earth’s
convective engine and are ultimately responsible for the vigorous geophysical activity on
the surface of the planet. The P– T slopes of the phase transitions are responsible for the
major seismic discontinuities. For example, the inner core –outer core boundary (IOB) at
202 Chapter 2

5,150 km depth (330 GPa) also manifests such a discontinuity separating the liquid outer
core from the solid inner core (see Fig. 2.3).
The temperature of the IOB has been estimated by Boehler (1993) as
4,850 ^ 200 K and the temperature at the CMB at 2,891 km depth (135 GPa) is about
4,000 K. Brown and McQueen (1986) estimated an ICB temperature of 5,800 (^ 500) K on
the basis of Hugoniot temperatures determined by thermodynamic integration and the
CMB temperature was noted as 4,000 (^ 500) K. Later, Ahrens et al. (1998) estimated the
core-side temperature at the CMB to be 3,930 (^ 630) K. Wijis et al. (1988) performed
simulations at thermodynamic states corresponding roughly to the CMB and the boundary
between the molten outer core and the solid inner core. Estimates of T at the ICB range
from 4,000 to 8,000 K and that at the CMB from 3,000 to 4,500 K. The pressure at the ICB
is accurately known to be 330 GPa (Poirier, 1991).
Using the data compiled by Anderson and Ahrens (1994) for the EOS of liquid iron,
Wijs et al. (1998) found that the density at the ICB state is 1.33 £ 104 kg m23 (and that T
and P at the CMB state are 4,300 K and 1.07 £ 104 kg m23, respectively). Their simulation
also showed that the structure of high-P molten iron, as described by the radial distribution
function, gðrÞ; is close packed with a coordination number of slightly over 12. This means
that the liquid structure also closely resembles that of hcp crystal.
From the melting curve of FeO to 102 GPa, Knittle and Jeanloz (1991) inferred that,
at the CMB pressure, the melting temperature of FeO exceeds that of Fe by 1,000 –
2,000 K. Assuming oxygen as the major alloying component in the outer core, a CMB
temperature of 4,800 ^ 500 was inferred.
Using the dislocation theory, Poirier (1986) calculated a melting temperature of
6,160 ^ 250 K for pure iron at 330 GPa and estimated that the temperature of the IOB is
5,160 –5,660 K. At the oceanic island basalts (OIB), the upper bound to the melting
temperature of iron is considered to be 6,500 K but, when the contribution of the alloying
light elements (T-depressants) are considered, the temperature at the centre comes down to
5,700 K (Anderson, 1993).

2.5.3. Adiabatic gradient

If we assume that the geothermal gradient in the mantle is adiabatic, we get

T=T0 ¼ ðr=r0 Þg ð2-4Þ
where r is the density and g is the Grüneisen parameter for the mantle (for the lower
mantle, g < 1 is assumed).
The geotherms corresponding to MORB adiabats have been determined as 1,3008C
(Nisbet et al., 1993). This adiabat intersects the olivine –b-phase transition at 1,3958C and
13.7 GPa (410 km). In this transition, the reaction is exothermic, while the g-spinel to
perovskite þ Mg-wüstite phase transition is an endothermic reaction and a temperature
decrease should occur across 660 km. However, the phase boundaries representing
equilibrium reactions should run in either direction (up low pressure or down high
pressure) with the perturbing P and T and the sign (þ ve or -ve) of the changes in enthalpy
and entropy would change accordingly. This also signifies that the olivine to b-phase
Petro-Tectonic Features of Terrestrial Planets 203

transition is exothermic and transformation of g-spinel to perovskite þ Mg-wüstite is

endothermic.
Plate tectonics and geomagnetism offer evidence for large-scale convective
motions within the Earth. In a convecting system with a high Raleigh number, constant
velocity and without internal heating, the temperature profile lies along an adiabat
(Turcotte and Schubert, 1982). Non-adiabatic gradients are restricted to boundary layers.
The adiabatic temperature gradient throughout the lower mantle is assumed to lie between
0.25 and 0.308/km (Anderson, 1982). The adiabatic gradient at high shear pressure,
ðdT=dPÞs ; decreases systematically with compression.
The average temperature in the lower mantle will be . 2,000 K and may increase
systematically with depth due to adiabatic compression. The adiabatic temperature rise in
the lower mantle falls between 450 and 6508. The absence of superadiabaticity yields
temperatures at the CMB ranging from , 2,400 to 2,650 K (Boehler, 1996).

2.5.3.1. Deviations from adiabaticity

In regions such as boundary layers, hot rising plumes and cold descending slabs,
large deviations from adiabaticity occur. In boundary layers, where heat is transported
mainly by thermal conduction such as in the lithosphere or across the CMB, the
temperature gradients are much steeper than adiabatic.
An increase in viscosity with depth, due to a high melting gradient, tends to make
the lower mantle super-adiabatic (Stacey, 1977). Seismic tomographic studies show large
lateral heterogeneities in the seismic velocities in the lower mantle (Su and Dziewonski,
1991; Woodward et al., 1993; Fukao et al., 1992). The causes of heterogeneities are: (1)
change in chemical composition, (2) mineral anisotropy in convective flow and (3)
temperature changes due to the cool descending material and hot rising plumes (Duffy and
Ahrens, 1994; Cadek et al., 1994).

2.5.3.2. Heat flow and plate tectonics

Planetary tectonics is a geological machine fuelled by the flow of heat out of the
planet. In the absence of this machine operating, the heat from the interior will be held in
by an encircling shell of immobile rock and the high temperatures of the interior will melt
large amounts of the mantle. The resulting abundant magma will rise here and there to
form a thick, possibly lumpy, crust as happens, e.g., on Venus.
Plate tectonics, on the other hand, helps a planet to run cool and produce a much-
differentiated crust. This efficient heat loss keeps the interior cool and the amount of
mantle melting low. The resulting crust is therefore thin. For further discussion, see the
following section.

2.6. Elastic parameters of the Earth’s interior

The geophysical properties of the Earth’s interior can be evaluated by the study of
the elastic properties of minerals under high pressure and temperature. The elastic stiffness
coefficients provide fundamental insight into the nature of atomic forces in solids. While
204 Chapter 2

the atomic structures of a vast range of substances have been explored at high pressure, the
complete set of elastic stiffnesses are known for only a few, usually simple, minerals.
Knowledge of the individual elastic moduli is necessary for calculating acoustic
wave velocities and their orientational dependence. Comparison of laboratory measure-
ments of acoustic velocities with seismic velocities in the mantle has been recognized as
the most direct means of determining the mineralogy of the deep, inaccessible regions of
the planetary interiors. The density and seismic velocities of key mantle minerals provide
constraints on the mineralogy.
Seismological studies employing stacking techniques are providing increasingly
detailed models of impedance contrasts and velocity gradients in and near the mantle-
transition zone (Shearer, 1996).
The velocity properties of seismic waves are controlled by the bulk modulus
(incompressibility) and rigidity parameters. The bulk modulus (K ) of a crystal is the
cumulative effect of the bulk moduli of component polyhedra of individual sites. The latter
are determined by crystal-structure refinement at elevated pressures by X-ray diffraction,
spectral study and other techniques. This is discussed further in Section 5.6.
X-ray diffraction experiments at high pressure can constrain the bulk modulus
(e.g., Knittle, 1995) but these measurements typically provide no information on material
response to shear deformation. However, such information for Earth materials is critical
for constraining mantle composition and structure.

2.6.1. Stress and strain

A study of the Earth’s material at depth demands an understanding of the

mechanical behaviour of solids (and volatiles) under pressure accompanied by
temperature. Under pressure, the fundamental relationship involves stress and strain.
Broadly, stress specifies the nature of the internal forces operating within a body, while
strain relates the change in size with the force, F: According to Hooke’s Law, stress and
strain are linearly related (see Glossary).
The ratio of stress and strain defines the Young’s modulus, E. If the volume V is
reduced by an amount dV, the relationship between the stress (P) and the strain ðdV=VÞ is
P ¼ KdV=V; where K is the bulk modulus (or incompressibility).
The stresses on the point P (assumed to be located at the origin of the coordinates
x; y; z), along the yz plane may be denoted by pxx ; pxy ; and pxz (see Glossary). The first
subscript x indicates that the plane is perpendicular to the x direction, i.e., the yz plane,
while the second subscript denotes the direction of the stress component. Similarly, the
stress components along the zx and xy planes may be defined. The pxy ; pyz ; pxz are called
shear stresses. It can be shown that pxy ¼ pyx ; etc. This equality of cross-shears reduces the
nine components of stress at a point to six. These stresses are called principal stresses.
In a real fluid, the principal stresses are equal to each other and the stress is
hydrostatic. For a rigid body, there are two kinds of displacement: rigid body and
deformation. The change in size (and also shape) causing deformation is a measure
of strain.
Petro-Tectonic Features of Terrestrial Planets 205

At a point, the state of stress can be represented by nine components Pij ðij ¼ x; y; zÞ
and the strain is also defined by nine components 1ij ðij ¼ x; y; zÞ: Of the nine components
of strain, 1xx ; 1yy and 1zz are the simple expansions or contractions, while 1xy ; 1yz ; … are
shear strains. The two arrays of nine numbers specifying the state of stress and strain at any
point present the tensor quantities. As noted earlier, of the nine components of stress and
strain at any point, only six are independent.

2.6.1.1. Strain tensor

A second-order Taylor expansion enables one to infer Hooke’s Law, which linearly
relates stress and strain and may be expressed as hK ¼ SKi sI ; where h and s are the six
components of the Lagrange strain and stress tensors, respectively, and S is the matrix of
the elastic coefficients.
Notations according to the Voigt’s convention are used, indexed and summed. As
the pressure is applied hydrostatically, the s stress tensor is reduced to
ð2p; 2p; 2p; 0; 0; 0Þ: Therefore, the symmetrically inequivalent strain components are

h1 ¼ 2ðS11 þ S12 þ S13 Þp ð2-5Þ

and

h3 ¼ 2ðS31 þ S32 þ S33 Þp ð2-6Þ

In a trigonal case, S31 ¼ S32 :

For a cell deformation, the strain-tensor component may be directly related to the
parameters of the deformed cell (Catti, 1985) and, when no change in symmetry occurs, the
following relation holds:
aj 2 aj0
hj ¼ ð2-7Þ
aj0

where aj0 and aj refer to the jth edge of the cell before and after deformation, respectively.
The simple relationships among cell parameters, strain-tensor components and compliance
coefficients enable one to relate linear compressibilities to linear combinations of Sij terms.

2.6.1.2. Zero-pressure bulk modulus, K0 : Eulerian strain

The zero-pressure bulk modulus, K0 ; is defined as the inverse of the summation of
Sij coefficients. When i and j indexes range between 1 and 3, its determination directly
follows form boa and boc, as

1
K0 ¼ ð2-8Þ
2boa þ boc

The coefficients boa and boc are determined through first-derivatives versus pressure.
The zero-pressure bulk modulus may also be estimated by least-square fitting of the Birch
(1978) EOS, which is based on the expansion of the internal energy in powers of the
206 Chapter 2

Eulerian strain, f, defined as

"  #
1 V0 2=3
f ¼ 21 ð2-9Þ
2 V

where V0 is the ambient pressure volume and V is the volume. The normalized pressure ðFÞ
is

F ¼ P=3f ð1 þ 2f Þ5=2
where P is the measured pressure. As an example, the normalized pressure as a function of
Eulerian strain ( f) for phase E (a hydrated Mg-silicate formed under pressure) is discussed
in Section 13.4.7 (see Fig. 13.12).
Third-order truncation of Birch– Murnaghan equation works out to the following
expression:
"  # ( !)
2=3
3 V0 2=3 3 0 V0
P ¼ K0 21 x 1 þ ðK 0 2 4Þ 21 ð2-10Þ
2 V 4 V

where K0 and K 00 correspond to zero-pressure bulk modulus and its first-derivative,

respectively.

2.6.2. Seismic waves

In an earthquake, three types of waves may be propagated from the focus: two body
waves, P and S and the surface waves. P is a longitudinal wave like sound waves, while S is
a transverse wave like light waves.
P-wave: The longitudinal wave moves like sound wave and the velocity of it is
given by
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
4
r ; VP ¼ K 2 m ð2-11Þ
3
where r is the density, K ¼ bulk modulus (incompressibility) and m is the rigidity.
S-wave: The transverse wave moves like light waves and its velocity is given by
p
VS ¼ m = r ð2-12Þ
when m; the rigidity, is zero, VS becomes zero. That is, a shear wave cannot travel through
a material of zero rigidity.
From the above, two relations we can see that
Vp2 =r ¼ K=r 2 ð4=3ÞVS2 ¼ VB2 ð2-13Þ
where VB is the bulk (sound) velocity.
Wang (1970) suggested that an estimate of the density in the upper mantle (depth
< 1,000 km) may be obtained from the bulk-sound velocity, VB, involving the relationship
between VB ð¼ ðVp2 2 ð4=3ÞVs2 Þ1=2 Þ and density r (see also Section 5.2.1).
Petro-Tectonic Features of Terrestrial Planets 207

At different stations, the arrival times of seismic signals of the same seismic shocks
are recorded. These help determine the travel time of the disturbance as a function of the
distance. At a distance of about 1058, both P and S waves seem to fade out although waves,
apparently of P type, are found for the range 142– 1808. The range 105 – 1428, referred to as
a “shadow zone”, occurs due to diffraction around the boundary of the core. Seismic waves
travel over oceanic ridges with different velocities in different directions. Waves
perpendicular to the ridges travel ,5% faster than those parallel to the ridges. This may be
due to the alignment of crystals attained during the convection process. Olivine shows
,25% difference between velocities in the slowest and fastest crystal directions.

2.6.2.1. P- and S-waves in seismic discontinuities and in the core

Elastic P- and S-waves travelling down to meet a surface of discontinuity may be
reflected and refracted. Thus, an incident wave may generate again a P- and a S-wave
(called PP- and PS-waves) so an incident S-wave can also generate both P- and S-waves
(called SP- and SS-waves). To indicate an upward reflection at the core boundary, the
symbol “c” is used while “K” is used to denote the path of a wave which is refracted
through the outer core. Thus, SKP is an S-wave that has refracted into the core and
refracted back into the mantle as a P-wave.
Equation (2-12) shows that the velocity of transverse S-waves is proportional to the
square root of the modulus of rigidity of the medium. Using tidal and seismic data in
the mantle, Takeuchi (1950) showed that the rigidity of the outer core is less than 1/40th of
the rocks of the mantle. This very small rigidity suggests it to be fluid.
In the outer core, no travel of S-wave is observed. This is because a non-rigid or
liquid body cannot transmit S-waves. The symbol I is used for a segment of P-type in the
inner core and the symbol I is also used for the paths of S-type wave in the inner core. Thus,
PKIKP is a P-wave that has passed through the inner core, remaining a P-wave throughout
its path (Fig. 2.14). The symbols used for possible reflections and refraction in the major
discontinuities in the mantle and core are shown in Fig. 2.14 (Song, 1997). BC represents
what traverses the bottom of the outer core and DF that samples the top of the inner core.

Figure 2.14. Ray paths and travel times as functions of distance for seismic PKP waves. Distance is measured by
the angle subtended at the Earth’s center by the source and the station locations. These PKP waves were used in
numerous body wave studies of the anisotropy of the inner core (Song, 1997).
208 Chapter 2

But the BC – DF travel time anomalies arise from inner-core anisotropy, which corresponds
most nearly to hcp model. The BC – DF travel time residuals, signifying anisotropy, are
reduced in the quasi-eastern hemisphere (40 –1808E) as compared with the other
hemisphere (Tanaka and Hamaguchi, 1997). Small-scale heterogeneity has also been noted
(Creager, 1997).

2.6.2.2. Minor discontinuities (reflective)

Recent emphatic efforts for fine seismological details cover the horizontal
variations (a few percentage in magnitude) in wave velocities (determined from seismic
tomography) and the kilometre-sized undulations of the interfaces between layers.
A generalized model of the Earth’s onion structure is based on thermodynamic
equilibrium, which is static in nature. In general, phase transitions, and hence the
corresponding seismic discontinuities, occur over a finite depth interval. The seismic
reflection studies offer a combined measurement of the sharpness and impedance contrast
(velocity and density jump) of the discontinuities.
Reflections from some minor discontinuities (i.e., other than 400 or 670 km
discontinuities) may be observed under special circumstances when the local curvature of
the discontinuity provides optimal focusing of the reflected waves (Lees et al., 1993).
Thackley et al. (1993) suggest that the pattern of flow in the transition zone could be
perturbed enough to create reflective discontinuities. Similar ideas are now being
considered for the seismic properties of the Moho, the interface between the crust and
mantle. Thin lithological or textural layering induced by large-scale lateral shearing in the
transition zone might produce strong coherent reflections.

2.6.2.3. Wave velocities in the lower crust

The occurrence of 7.1 –7.8 km s21 velocity layer in the lower crust – upper mantle
region under various structural provinces has been interpreted as being transitional in
nature. Compressional and shear wave velocities (VP and VS) at the lower crustal pressure
(,1 GPa) of granulites and eclogites, formed as products of phase transitions of basaltic
crust, have been investigated by many (e.g., Manghnani et al., 1974). Granulite, an
intermediate between basalt and eclogite, is one rock type that exhibits this velocity range.
Both density and seismic velocity increase gradually throughout the garnet – granulite field
during the transformation of gabbro to eclogite (e.g., Green and Ringwood, 1972). Ito and
Kennedy (1971) postulated that two sharp density increases occur at the basalt – garnet
granulite and garnet granulite –eclogite boundaries. The intermediate garnet –granulite in
the lower crust shows velocities of VP in the range 7.1 – 7.8 km s21.
Most of the granulites and eclogites studied by Manghnani et al. (1974) showed M



values of , 22. Birch’s relationship for these and previously studied basalts ðM , 22Þ was
VP ¼ 21:85 þ 2:87r (where r 2 ¼ 96%). For VS, the best fit of the data for granulites and
eclogites ðM
, 22Þ was VS ¼ 20:33 þ 1:40r (where r 2 ¼ 88%). These experiments
revealed that the elastic properties of garnet granulite with M
, 22 are compatible with
those of the 7.1– 7.8 km s21 crustal layer.
Petro-Tectonic Features of Terrestrial Planets 209

2.6.2.4. Wave velocities in lower mantle: T effects

The relationship between shear and compressional velocities at high pressure offers
a means of reconciling the seismic tomography and mineral physics. The parameter n
defining this relationship is

n ¼ ðd ln VS =d ln VP ÞP ð2-14Þ

Seismic tomography studies have shown that ½ð›lnVS Þ=ð›lnVP ÞP < 2 throughout the
lower mantle (e.g., Robertson and Woodhouse, 1995).
The elasticity measurements at high T and ambient P ðP ¼ 0Þ for Mg2SiO4, Al2O3
and other minerals show all values of n ranging between 1.2 and 1.6. Under the Carribbean,
the upper part of the lower mantle shows the value of n as 1.7, which possibly increases
with depth.
If the lower mantle is at or near the solidus, the value of n is high. Dziewonski et al.
(1986) reported that P- and S-wave perturbations at constant radius are correlated, leading
to n values ranging between , 2.0 and 2.5 in Earth’s lower mantle. In deep Earth, n $ 2:0;
with increasing pressure the value of n increases.
Semi-empirically, the P- and S-wave velocities in the perovskitic mantle at related
temperature can be calculated using (Karki and Stixrude, 1999):

VS ðTÞ ¼ VS ð0Þexp½aT=2ð1 2 GÞ ð2-15aÞ

 
VP ðTÞ ¼ VP ð0Þ exp aT=2½1 2 a2 dS 2 4=3b2 G  ð2-15bÞ

where

VB ð0Þ V ð0Þ
a¼ ; and b ¼ S ð2-16Þ
VP ð0Þ VP ð0Þ

Here, quantities evaluated at zero temperature are taken from the first-principles
calculations (Karki and Stixrude, 1999) and the thermal expansivity a and the Anderson–
Grüneisen parameters, dS and G are defined as:
 
1 dKS
dS ¼ 2 ð2-17Þ
aKS dT P
 
1 dG
G¼2 ð2-18Þ
aG dT P
which represents average values over the temperature interval (Anderson, 1995).
A lower bound for the temperature dependence of the shear modulus is
 
K dG
G¼ T ¼ 2:6
G dP T

There is some theoretical evidence that G decreases with increasing pressure. The
shear modulus may be assumed to depend only on volume.
210 Chapter 2

The dependence of a on density may be described by

  2k 
d TO r
a ¼ a0 exp 21 ð2-19Þ
K r0
The experimental data for MgSiO3 composition give the values (Wang et al., 1994):

a0 ¼ 2:4 £ 1025 K21 d TO < 4:3

Results for Mg0.9Fe0.1SiO3 yield much higher values (Stixrude et al., 1992)

a0 < 4:2 £ 1025 K21 d TO < 7:0

Fe-bearing samples may be biased by their metastability (Bina, 1995).
For MgSiO3 perovskite, the values are (Jeanloz and Knittle, 1989):

a < 4 £ 1025 K21 ; and lð›K=›TÞP l < 5 £ 1022 GPa K21 :

2.6.2.5. Crustal plates and earthquakes

Continents cover , 3% of the Earth’s surface and consist of buoyant material,
# 300 km thick, that exist on the Earth’s surface for billions of years. The continents may
cyclically assemble and reassemble together to form super-continents. Large continents
may break up through the upwelling mantle flow with the development of tensile stresses
of the order of 100 MPa.
The Earth’s crust is composed of a dozen plates, which are driven by the convection
and move against one another at a rate of a few millimetres per year. For example, the
Californian San Andreas fault system depicts the boundary between the Pacific and North
American plates that move sidewise at a rate of , 5 cm per year.
The oceanic lithosphere behaves as a small number of rigid plates, with tectonic
activity confined mostly at the margins of the plates. The continental lithosphere is less
rigid and the deformation spreads laterally over hundreds to thousands of kilometres.
Mostly, the deformation of the brittle upper crust is being driven from below by a flow of
the upper-mantle part of the lower lithosphere. The observed short-term deformation of the
brittle layer reflects the steady viscous strain rates in the lower lithosphere.
New crust forms from upwelling basaltic lava along the MOR. One such is the Mid-
Atlantic ridge along which the North American and Eurasian plates continuously diverge
from one another. From the magnetic signatures of past magnetic pole reversals that are
frozen into oceanic crust along MOR, the rates of deformation averaged over several
million years can be estimated. The other result of plate tectonic activity is manifested by
the Himalayan belt, where collision of the Indian and Eurasian continental plates compress
and deform the crust along the plate boundary. The convection pattern under the
Himalayas, the Pamir and Hindu Kush implies that a sinking mantle block under these
generate intermediate focus earthquakes. These earthquakes are mainly due to phase
transformation.
Along the plate boundaries, deformation is distributed across a broad zone and
occurs intermittently, often in the form of deadly earthquakes. For example, the San
Petro-Tectonic Features of Terrestrial Planets 211

Andreas fault system is a 100 km wide zone consisting of several active faults. The
collisional zone between the Indian and Eurasian plates is more than 1,000 km wide and
contains faults and folds resulting from the 40 m.y. old head-on collision of the two
continents. Although most seismic waves are generated along plate boundaries, localized
deformation within the plate (intraplate) may generate earthquakes. The development of
space geoelectric methods has offered a powerful means to measure the motion and
deformation of crustal plates.
In 1906, the San Francisco earthquake west the of San Andreas fault moved
northward by about 6 m. By precise electronic distance measurement (trilateration), a
measurement of slow deformation that build up stresses between earthquake events has
become possible and a broad zone of elastic strain across the San Andreas fault system is
revealed. An earthquake cycle may operate in a region (e.g., , 50-year cycle in Assam, NE
India), wherein the quiescent period in between the quakes is the period for gradual build
up of elastic strain.
If, at a point within the Earth, the stresses are not too high, elastic or plastic
deformation will take place. When stress is built up for a long time, fracture may develop,
releasing stress and propagating elastic waves, which are body waves like P and S and the
surface waves. Beyond the elastic limit, the rocks may yield to produce a fracture, which at
depths may serve as the focus of an earthquake.
Earthquakes may begin with slow pre-seismic creep within a localized nucleation
zone on the part of a fault. An earthquake can cause an acceleration of deformation caused
by the flow of hot and somewhat viscous rock below about 10 km and an accelerated creep
in neighbouring fault segments, in response to the rupture, is developed.

Depths and energy of earthquakes. Most (, 80%) earthquakes occur within the depths of
a few tens of kilometres, but the rest (, 20%) occur at depths of , 670 km, where the cold
upper 50– 100 km of the lithosphere in the form of a slab dives into the mantle to
compensate for the upwelling along the ocean ridges (see Fig. 2.15). The subduction of an
oceanic crust with sediments below the continental granite/basalt is shown in Fig. 2.15.
The distribution of earthquake frequency with depth shows on exponential decrease
to a minimum at , 300 km, followed by a second peak at 550 –600 km and then an abrupt
shut-off before 700 km. For shallow earthquakes, frictional sliding is important. The
shallowest earthquakes occur by normal brittle shear fracture.
Faulting instability develops long before failure. As the partially altered oceanic
lithosphere is carried downward, dehydration of serpentine and other hydrated phases
occurs (down to , 300 km depth) with release of fluids. This triggers fluid-assisted faulting
(Meade and Jeanloz, 1991). Devolatilization of subducted sediment could contribute to
seismicity along the tops of subducted slabs. The continuous nature of devolatilization is
comparable with the spread of earthquake hypocentres along individual subduction zones
(Kirby et al., 1996) but the interior of the slab remains cold enough to inhibit the
olivine ! spinel transition which could occur at the depth of , 300 km.
In deep subduction zones at pressure equivalent to 700 km, olivine fails transfor-
mation to spinel, decomposing directly to perovskite þ oxide without producing any
212 Chapter 2

Figure 2.15. Subduction of a slab of an oceanic crust with sediment and the associated phenomena (after Ida,
1987, q 1987 American Geophysical Union).

earthquakes. Such decomposition reaction cannot run fast enough to radiate seismic energy.
Earthquakes do not occur at greater depths because there are no more phase changes.
An empirical formula for quantifying the earthquake energy has been suggested
by Richter as
Log E ¼ A þ BM ð2-20Þ
where M is the Richter magnitude, E is the total radiated energy and A and B are constants.
When E is measured in ergs, the values of the constants are commonly taken as A ¼ 11:4
and B ¼ 1:5: A devastating earthquake releases an energy of ,1025 ergs.

2.6.2.6. Strain transients and earthquakes, co- and post-seismic

Mechanics of earthquakes may be studied as related to strain transients, i.e.,
temporal variations in rates of deformation within plate-boundary zones and their
relationship to earthquakes and plate motions. Even small stress perturbations (a few tens
of millibars) can affect seismic activity.
Co-seismic strain is manifested by fault rupture during an earthquake. Co-seismic
changes in the local static stress field may trigger seismicity. Co-seismic strain
accompanying large earthquakes can affect the state of stress on nearby faults and
adjacent segments of the same fault, thereby increasing or decreasing the probability of
earthquake occurrence on those neighbouring fault systems. However, the long-term
effects of stress diffusion following large earthquakes can have larger effects on nearby
Petro-Tectonic Features of Terrestrial Planets 213

faults than the co-seismic strain, but this occurs over time scales of decades rather than a
few years.
Post-seismic strain transients arise from the diffusion of stress surrounding a
zone of earthquake rupture and from slow fault slip following an earthquake. The
devastating Kobe, Japan earthquake of 1995 occurred from seismic slip along the fault
in the Nakai trough, influenced by the post-seismic strain accumulated since the two
great earthquakes that occurred in 1944 and 1946. It is important that train transients
that precede and perhaps even trigger earthquakes are recognized for earthquake
prediction.
Earthquakes are followed by a sequence of aftershocks that decay with time.
Post-seismic deformation may cause an increase in stresses at the hypocentre of the
earthquake. Within a few years of an earthquake, post-seismic deformation may
change stresses in the upper crust by at least as much as the stress released by the
earthquake itself.
Spatial and temporal distribution of aftershocks may reflect the stress changes
occurring during an earthquake (co-seismic deformation). An interplay between Coulomb
stress and the time-dependent evolution of frictional properties of faults can trigger
earthquakes within about 100 km for days to months after the main shock.
The lower crust and upper mantle are viscoelastic materials that cannot bear
significant shear stresses and tend to yield with time. Hence, the lower crust relaxes and
allows the crust above to deform co-seismically through time. The viscoelastic relaxation
of the lower crust (Deng et al., 1998) continuously increases the Coulomb stress (as noted
in the Hector Mine hypocentre by Freed and Lin, 2001). Viscoelastic relaxation transmits
post-seismic stresses to great distances. Thus, co-seismic stresses may provide the driving
force for subsequent post-seismic deformation. The post-seismic stress change in the crust
is due to delayed slip on fault surfaces (frictionally controlled afterslip), time-dependent
deformation concentrated along shear-zones (viscoelastic creep) or other processes
(Hearn, 2001). US National Science Foundation is seriously initiating an “Earth Scope” for
providing detailed measurements of 3D post-seismic surface displacements and strains.
However, all data need to be collated to achieve an improved view on the earthquakes
cycles and near-real-time estimates of the build up and release of stress on active faults.

Strain diffusion to faulting and earthquakes. In deformation experiments on faulting,

Green (1993) observed that faulting instability develops long before failure During
loading, large numbers of tiny tensile microcracks form parallel to the greatest
compressive stress. Then the microcracks cooperatively organize to initiate shear fracture
before failure occurs catastrophically. The motion of plates can lead to deformation at the
interaction boundaries. The deformation results in the slow building of stresses, which
ultimately can be released catastrophically to cause earthquakes. Studies using borehole-
strain meters have revealed that these failures can result on the same faults that have
caused regular destructive earthquakes. The proximity to failure can perhaps be learnt from
the slow deformation of the Earth’s surface through strain diffusion from great
earthquakes, volcanic eruptions or spreading events at the separating plate boundaries.
214 Chapter 2

The propagation velocity of deformation is some 100 km per year near the source
but at greater distances it is only a few kilometres per year. Analysis of this slow
deformation has allowed determination of the viscosity of the crust and of the uppermost
mantle for Japan, California and Iceland.
The slowly propagating strain pulse is capable of unlocking favourably oriented,
highly stressed faults through a reduction the forces, pressing the two walls of the fault
together. This causes the fault slip earthquake. As stated earlier, the highly destructive
1995 earthquake in Kobe, Japan was likely triggered by strain diffusing from great
earthquakes in 1944 and 1946. For any particular fault in a fault system, the likelihood of
failure can be postulated. Again, certain fault orientations experiencing strain pulse may
experience clamping, even though recorded history shows them to be active. The great
subduction events of northeast Japan show a magnitude of about 8 but, in most of the
plates, the motion energy is locked and is released as slow, non-destructive events.
However, it is commonly observed that slow multilayer deformation occurs following a
large earthquake since the crust slowly adjusts to the change in the stress field brought
about by such an event. The San Andreas Fault system shows rupturing on average every
22 years.
A short-term triggering can also be the beginning of a previously unrecognized
longer-term (perhaps 5-year) trend, consisting of a persistently elevated, but decaying,
micro-earthquake rate. Changes by such earthquakes can be widespread, essentially by
sensitization of a large area to small stress perturbations.
Even environmental sources of stress such as tidal loading, precipitation and
barometric pressure change may cause environmental stresses of the order of tens of
millibars, which are capable of triggering small seismic events.

Water-reservoir pressure and crustal deformation: Roseland reservoir. (a) The pressure
exerted on the crust by the weight of water in a reservoir may induce crustal deformation
that precedes earthquakes. A seismic slip should be preceded by an accelerating seismic
slip near the hypocentre of an impending earthquake.
The Roseland reservoir in French Alps is regarded as a natural laboratory for such
studies. Water-level variations behind the Roseland dam are seen constantly to accompany
the bursts of radon gas, changes of electric potential and departure of ground tilt with
respect to the assumed linearity of rock elasticity. Under a reservoir, the crustal rocks may
respond non-elastically to the loading. However, the induced stresses and pore pressures
are not likely to be much greater than 0.1 MPa for each 10 m of water depth. At low stress,
non-linear behaviour of rock mass seems to occur.
Fluid flow past ions adsorbed on rock surfaces produces an electric field, termed a
“streaming potential”, which changes with pressure gradient and permeability (Gouniaux
and Pozzi, 1995).
(b) Radon-content and electrical field-moderate earthquakes in Greece have been
predicted from variations in the local electric field (Varotsos et al., 1993). A precursory
decrease and increase of radon concentration before the 1978 Izu-Oshima earthquake in
Japan has been well documented.
Petro-Tectonic Features of Terrestrial Planets 215

A relationship between radon and electrical anomalies and the deformation of

crustal rocks is well indicated.

2.6.2.7. Precursors to earthquakes

Silver and Wakita (1996) have listed many plausible pre-earthquake “strain
indicators”. Changes in radon emission or in electrical or magnetic fields may represent a
natural amplification of pre-earthquake deformation. The conductance of water or gas in a
rock fracture is proportional to the third power of the fracture’s aperture (Committee on
Fracture Characterization and Fluid Flow, 1996). Fluid flow past ions, adsorbed on rock
surfaces, produces an electric field termed a “streaming potential”, that varies with
pressure gradient and permeability (Jouniaux and Pozzoi, 1995).
Although induced stresses and pore pressures are unlikely to be much greater than
0.1 MPa for each 10 m of water depth, crustal rocks may respond inelastically to loading
by a surface reservoir. The loading rate or fluid flow rate may influence the rheology of
brittle, inelastic rocks. When the level of the Koyna reservoir (India) rose faster than
12.2 m per week (Gupta, 1983), earthquakes larger than magnitude 5 occurred. Similarly,
due to an abrupt decrease in the reservoir-filling rate at Nurek reservoir (Tadjikistan), a
large earthquake occurred (Simpson and Negum-atullaev, 1981). Raising or lowering of
sub-surface fluid pressure could also trigger micro-earthquakes. Increased pore-fluid
pressure counteracts the compressive stress on fault planes in rock, reducing the shear
velocity for slip (Roeloffs, 1999).

Bore-hole strain meter. In seismogenic zones, many kinds of strain instrumentation

are employed these days for helping to sense earthquake possibility. These
involve electronic distance meters (EDMs), GPS receivers, creep meters (which
measure motions locally across the fault), bore-hole strain meters and bore-hole
seismometers.
Bore-hole strain meters (Sacks-Evertson) have been installed in a number of
different active areas, particularly in Japan, California and Iceland. These high-resolution
data allow identification of new processes, such as slow earthquakes in seismogenic zones,
and changes in magma reservoirs during episodes of volcanic activity, which are not
detectable by other techniques. Sufficient aseismic slip would produce near-surface
deformation detectable by a bore-hole strain meter. A 5-magnitude earthquake 10 km deep
produces near-surface strain of about 1027 a site 5 km from its fault plane. Strain increases
30-fold for each unit increase in magnitude but falls off as the 3rd power of distance from
the source.
Linde and his co-workers (1999) in Iceland have analyzed strain changes associated
with a moderate (magnitude 5.8) earthquake in Southern Iceland. The pre- and post-
earthquake slow strain changes are possibly caused by magma movement, possibly related
to the dike growth whereas in Chile, the 1960 great earthquake was caused by slow
increase of shear stress in the seismogenic zone triggering it. Bore-hole strain meters,
together with tillmeters and seismometers, are installed in deep drill holes (.1,000 m)
below the seafloor to have an insight into the processes of plate motion and earthquake
generation.
216 Chapter 2

Precursors to ScS and PcP. Precursors to core-reflected shear (ScS) and compressional
(PcP) waves have been observed in numerous locations. However, the two are uncorrelated
and one may be present without the other. The image of Earth’s lower mantle has, thus,
emerged as uncorrelated discontinuities in shear and compressional velocities with
substantial topography. A review by Lay et al. (1998) elaborates on such local
discontinuities in seismic impedance. They show that the precursors to ScS and PcP are
produced by gradients in shear-wave speed associated with large-scale heterogeneity in the
lowermost mantle. Such a gradient may form the wavefront and produce a distinct arrival-
time difference between the direct S wave and the ScS waves; when the gradients are weak,
no precursors are produced.
Moreover, PcP precursors are rarer than ScS precursors and variations in
compressional-wave speed are not necessarily correlated with variations in shear speed.
Perhaps, PcP and ScS precursors are produced by the same small-scale scatter that are
needed to explain PKP precursors (Clearly and Haddon, 1972).
Interestingly, there seem to exist thin (,50 km), ultra-low-velocity zones
(ULVZ) (Garnero and Helmberger, 1996) and anisotropy (e.g., Vinnik et al., 1998)
near the CMB, discussed in latter sections. These thin and slow regions at the base of
the mantle are reminiscent of the heterogeneous crust on the top of the mantle. This
can also be explained as a result of differentiation of the mantle (like the crust),
interaction between the mantle and core or of deep burial of the fragmented crustal
subduction slab.

Seismic hazard potential. Great earthquakes occur where the oceanic lithosphere plunges
deep into the Earth’s mantle One such great earthquake occurred in Chile in 1960 and in
Alaska in 1964. In such cases, the entire plate boundary ruptures.
In the last decade of the 20th century and at the advent of the 21st century, India
witnessed five major earthquakes culminating in the devastating Bhuj earthquake on 26th
January 2001. The basement rocks of peninsular India pushes towards the Himalayas and
slides beneath it in the process of collision and the records show that one or more major
earthquakes are overdue.
Global positioning system (GPS) studies indicate that India and southern Tibet
converge at 20 þ 3 mm yr21. The southern part of the Tibetan plateau absorbs 80% of the
convergence energy and undergoes strain. The surrounding Himalayan region absorbs the
remaining 20%. Through a century, the process accumulates energies of potential slips of
,2 m. Further down in the south, this energy is naturally of far less magnitude but, over a
long time scale, this is not negligible, as illustrated by Bhuj earthquake in 2001. In the
Himalayas, the potential slip accumulates entirely as elastic rather than inelastic strain,
which permanently deforms the rock. Deformed terraces in the Himalayan foothills
indicate an advance of (21 þ 3) or (21-3) mm yr21 (from GPS studies) in South Nepal
during the past 10,000 years. However, about 10% the strain could be inelastic, which
needs to be released in the form of an earthquake. This release of strain may have
contributed to the great earthquakes in the Himalayan region in the years 1803, 1833, 1897,
1905, 1934 and 1950 (the largest inter-continental earthquake in recorded history). The
1950 Assam earthquake defined a rupture zone , 200 km long. An average slip of , 4 m
Petro-Tectonic Features of Terrestrial Planets 217

(the slip equivalent of 1934 earthquake) is presumed to occur during great earthquakes.
Since the 1950 Assam earthquake, the population growth has increased the number of
fatalities. The population of Kutch has increased by a factor of 10. For this reason alone,
while 2000 fatalities occurred in 1819, the number rose to 19,000 in 2001. Indeed, Bhuj
earthquake occurred in the heightened seismic hazard zone. A repeat of the1905 Kangra
event would evidently lead to a fatality figure of 20,000.

2.6.3. Acoustic and ultrasonic wave-velocities

The staff members of the Department of Terrestrial Magnetism are developing a

new multi-channel acoustic emission data collecting system to detect acoustic emissions at
high pressures and temperatures. This new system will allow testing of the physical
mechanisms of deep earthquakes.
Ultrasonic wave velocities in a large number of samples of crystalline rocks have
been determined by many workers, e.g., Miller and Raab (1997). T dependence of
Poisson’s ratio (at 500 MPa) has been noted at high P (# 500 MPa). The velocity drop is
mostly caused by partial melting.
Knoche et al. (1997) have developed a method for determining the acoustic-wave
velocities in polycrystalline samples at P and T relevant to the upper mantle. Using this
technique, compressional and shear waves can be measured simultaneously. Acoustic
velocities measured in situ in the laboratory can be compared directly with observed
seismic velocities without the use of large extrapolations of P and T that had been
necessary so far. They determined the velocities of compressional waves in hot-pressed
polycrystalline olivine up to 10.5 GPa and 1,773 K.
In ultrasonic interferometry, a burst of high-frequency signal is applied to the
piezoelectric transducer. The resultant elastic wave, propagating inside the sample,
produces a series of echoes. The travel time of an elastic wave through the sample is
measured by observing constructive and destructive interferences among overlapped
echoes as a function of frequency.
The melting of shock-induced high-pressure phase assemblage of olivine,
Mg2SiO4, provides the upper-bound constraint to the temperature at the base of the
Earth’s mantle of ,4,300 K (at a pressure of 133 GPa).
In the lower mantle, MgO seems to have at a thermal expansion coefficient at
2,000 K of 1 £ 10250 K21, a decrease by a factor of 3 of that occurring at high temperature
and zero pressure. These data, along with the high-velocity Hugoniot sound velocity data,
can explain why the 0.1% long wavelength (104 km) variation of compressional wave
velocity in the lower mantle occurs as a consequence of lateral temperature variations in
the range of , 1008C. These temperature differences in the Earth’s lower mantle may be
responsible for driving the mantle convection.

2.6.3.1. Ultrasonic velocities and Q in porous rocks

In the pressure range of 20– 600 MPa, Liu et al. (1997) determined the attenuation
(Q) and velocities of P- and S-waves in silicate rocks at ultrasonic frequencies. Velocities
and Q-values exhibit a steep, non-linear increase with increasing confining P due to
218 Chapter 2

progressive closure of microcracks. Q increases steeply ’ to the foliation as P increases;

the values of Q at high P are higher than those observed parallel to foliation. The reverse is
true for the velocities.
The simultaneous measurements of velocities and Q provide information on the
mechanisms of attenuation, especially for an assessment of whether the predominant
damping mechanism is in shear or in compression.

2.6.4. Subcrustal stress fields: ore localization

The subcrustal stresses are caused primarily by the mantle convection, which may
also cause intraplate tectonics. The sub-crustal stresses exerted by mantle convection can
be calculated, to a good approximation, from the harmonics of geopotential. The
compressional and tensional stresses under the crust, caused by the mantle convection, are
of the order of 108 dyn cm22.
Peridotites from typical rift-zone environment or xenoliths from OIB show stress
magnitudes of , 1– 10 MPa, whereas peridotites from collision zones (low temperature,
high strain rates) show much higher stresses (, 50 –300 MPa; Jin et al., 1998).
The strains in the lithosphere cause deformation or become localized in fractures.
The rocks of the Earth’s mantle, the deep plastic region below the Earth’s elastic crust,
must be churning slowly in vast convection cells under the lithosphere. The tensional
stresses, exerted by the upwelling mantle convection flows under the crust, seem to be
related to such uplifted features as the Tibetan uplift.
The tensional stress fields of the crust often serve as locales for mineral and metal
deposition. Probing the Earth from space, generating gravity data can help to calculate the
tensional stress regimes in the crust in which tension-related ore concentration might have
occurred. Studies on tectonic forces from satellite-derived gravity data have revealed a
subcrustal stress system which provides a unifying mechanism for uplift, depression,
rifting, plate motion and ore formation (e.g., in Africa and China; Liu, 1977, 1978).

2.6.4.1. Gravitational field models: degree harmonics and mantle flow

Gravitational field models derived from satellite tracking and surface gravity data
help derive the forces in the Earth’s mantle. From such models, the regions of crustal
compression and tension have been delineated covering almost all continents. Regions of
high stresses may occur in isolated regions far from plate boundaries and manifest seismicity
of high magnitude. Historic large earthquakes, such as the Beijing-Tangshan (1976) and
Latur (1993) earthquakes, offer examples occurring in such regions. The most impressive
series of measurements in the space programme has been the development of successive
degree harmonics placed in the model of the Earth’s gravity field.
The low-degree harmonics for n # 12 may reflect a large-scale mantle-flow system
(Runcorn, 1967). The high-degree harmonics for n $ 13 may result from a short-
wavelength convection system (Liu, 1977).
The non-linear nature of mantle convection and the wide range of viscosities
present formidable obstacles to a full knowledge of convection patterns. Satellite
and gravity measurements of the non-hydrostatic harmonics of the geopotential
Petro-Tectonic Features of Terrestrial Planets 219

(Smith et al., 1976) have furnished valuable information with data for the development of a
convection and stress system in the upper mantle of the Earth. The crustal
stress field exerted by the convection currents under the African plate has been established
by Liu (1977) up to the 25th degree harmonics of the geopotential.
The stress fields thus derived may be suitable for interpretation of the seismicity,
volcanicity, kimberlite magmatism, ore concentration and tectonic and magnetic features
in Africa.
Upwelling mantle flow seems to explain the tensional features associated with the
East African Rift System. This system may be the expression of a “hot line” in the upper
mantle, above which volcanism, seismicity and rifting occur intermittently. This system,
stretching along the Red Sea Rift and the line of Mt Cameroon, seems to be the site of
penetrative convective upwells. Therefore, the volcanics of this region are predominantly
basaltic and derived from the mantle.
Collisional mechanism involving Indian and Eurasian plates is invoked for the
evolution of the Himalayas. However, the pressure imposed by this collision is largely
compensated for by the flow of material in the lower crust and upper mantle.

2.6.5. Tools for sub-surface studies

Earth’s sub-surface is studied by measuring bulk properties such as seismic

velocity, electrical conductivity, thermal conductivity or magnetization. The tools
employed involve not only elasticity theory, quantum mechanics, rock deformation and
fracture mechanics, but also theories of heterogeneous and porous media, fluid flow and
percolation theory, surface chemistry, electric and dielectric theory, heat transfer and the
theory of magnetic materials. Fluids in the crustal part largely influence such properties of
rocks as acoustic properties, electrical conductivity, dielectric properties, thermal
conductivity and magnetic properties of bulk rocks.
The major experimental apparatus for mantle and core studies comprise (i) piston-
cylinder presses, multi-anvil and multi-stage presses, and (ii) diamond-anvil cells. The
latter method with laser heating devices extends the scope of high-pressure investigations
on the phase relationships within the mantle to the CMB and even beyond.

2.6.5.1. GPS in tectonic studies

Geodetic measurements help determine the crustal dynamics. Very long baseline
interferometry (VLBI) and GPS made possible studies of motions on a global scale. GPS
has become the affordable tool of choice for studies of regional deformation along active
plate boundaries. Thus, in the early 1990s, the development of the GPS system led to a
revolution in crustal deformation studies. Although inchoate, the synthetic-aperture- radar
interferometry promises to revolutionize crustal deformation research.
The principle of GPS geodesy is based on measuring the distance or range to several
GPS satellites from the time it takes for a signal to be transmitted from the satellites that
orbit the Earth at , 20,000 km altitude to a receiver. To achieve the precision necessary for
most crustal deformation studies, it is imperative to model the effects of signal-propagation
delay in the ionosphere and the troposphere. For this, data from a worldwide network of
220 Chapter 2

continuously running GPS sites are used. GPS receivers collect data continuously 24 h a
day. Receivers equipped for measuring signals from GPS satellites (US Department of
Defense) allow the determination of 3D positions accurate to several metres in a few
seconds. GPS should be instrumental in monitoring many Earth processes, such as plate
tectonics, the earthquake cycle, the transfer of angular momentum between the solid Earth
and the atmosphere, the rebound of the solid Earth in response to the unloading of the weight
of glaciers which melted after the last Ice Age and the change in sea level on a global scale.
By using interferometric images of satellite radar data, collected by the Earth
Resource Satellite-1 (ERS-1) before and after the 1992 Landers earthquake in California,
scientists were able to image the deformation surrounding the rupture in very fine detail.
The plates may be internally deforming and localized zones of active deformation may
exist within plates. Such processes of active deformation also operate on the Indian sub-
continent. Space geodetic measurements have confirmed the existence of anomalous strain
transients building up in the periods following an earthquake. An integrated GPS network
and a number of synthetic-aperture-radar interferometry satellites would permit close
monitoring of this deformation and a better understanding of its causes and consequences.

2.6.5.2. Mars global surveyor (MGS)

From the orbital spacecraft 400 km above the surface of Mars, the MGS recorded
the surface elevations employing a laser altimeter This recorded the subtle bobbing of the
spacecraft as it passed over areas of greater or lesser gravitational pulls. The combined
topography –gravity data expose a large range of thickness of the Martian crust. Beneath
the South Pole, it is 75 km thick and at the North Pole it is only 35 km. The MGS, carrying
a thermal-emission spectrometer, has revealed the southern highland to be basalt-rich and
the northern lowlands to be andesite-rich. Water-rich mantle magmas can only produce
andesite. Martian meteorites and the orbital data from MGS have greatly added to our
understanding of the chemistry and structure of the Martian crust and mantle.
MGS geophysicists have detected signs that the northern third of Mars may have
been shaped by plate tectonics early in its history. The North Pole manifests a remarkably
thin and uniform crust but this region was sunk to create lowlands, while the thicker crust
of the southern two-thirds of the planet has bobbed up to create highlands. This is much in
contrast to the Earth. The relatively thin crust of the northern lowlands is expected in a
plate tectonics regime resembling the Earth’s. The other possibility may be a giant plume
rising from the deep interior beneath the North Pole and sinking in the south.
Because MGS observations revealed the faintly preserved remains of a huge impact
crater on the northern lowlands, the crust there must have formed in the first half-billion
years of the Solar System when such giant impacts were common and the planet managed
to power up its own magnetic field. MGS observation also confirms that the crustal
formation in the north ended about the time the Martian magnetic field died.
The solar tidal deformation of Mars is observed by Mars Global Surveyor (MGS)
along with the weak magnetic field of Mars of today, and the discovery by the spacecraft of
the presence of a strong magnetic field in the past suggests a liquid Martian core.
Multianvil HP studies indicate that a minimum temperature of 1,400 K at the CMB is
required to sustain a liquid outer core.
Petro-Tectonic Features of Terrestrial Planets 221

However, Bertka and Fei determined the liquids’ temperature in the Fe – FeS3
system at high pressure, which defines the minimum temperature for an entirely liquid
Martian core, with an inferred core temperature to be greater than 1,800 K at the
CMP pressure.

2.7. The crust and cratons (“keels”)

2.7.1. Continental lithosphere

The buoyancy of Archean cratons makes them highly resistant to mixing through
convection process. There arises a clear dichotomy between partial melts (e.g., mafic and
silicic crust) and mantle residues (e.g., peridotites), as discussed below with reference to
Osmium ratios.
The Re/Os ratios in crustal rocks are seen to be much higher than in the mantle
rocks. The Re abundances in crustal rocks are 1– 10 times higher than mantle abundances.
Highly elevated 187Os/188Os ratios are observed in continental crust and sediments. The
silicon crustal rocks and sediments, which are mostly reworked, show widely variable
contents of Re and Os, but mafic rocks retain Re and Os abundances similar to MORB
(e.g., Shirey and Walker, 1998).
The continental crust is characterized by high ratios of Rb/Sr and Nd/Sm.

2.7.2. Subcontinental mantle

Samples of subcontinental mantle are obtained as mantle xenoliths in basalt and

kimberlite magmas as also by the tectonically exposed massifs, called “alpine peridotites”.
Two types of mantle xenoliths are common: the peridotites (olivine-dominant) and
eclogite (pyroxene-dominant). These two types show distinctive Re/Os and 187Os/188Os
ratios. Melt fractionation leads to high Re/Os ratios in mantle melts with complementary
proportion in peridotites. The Re –Os systematics of mantle samples such as peridotites
and eclogites provide good age constraints, which date the earliest melting event as 3.2 b.y.
old (see Shirey and Walker, 1998).
Earth’s continental crust, typically 40 km thick and presently covering about 2/5th
of the surface, accounts for only about 0.4% of the Earth’s volume The continental crust,
rich in granite and associated rocks, differs substantially from the denser, thinner oceanic
crust composed predominantly of basaltic lavas. It contains perhaps 30% of the total
terrestrial budget of the heat-producing radioactive elements (K, U and Th and other
elements such as Zr, B, Rb and Cs), whose sizes and valencies exclude them from being the
common minerals in the Earth’s mantle.
Most terrestrial planets generate basaltic crust and it can be supposed that an early
basaltic crust existed in the Earth; Venus may be an analogue of it. The crusts of other
planets and satellites are mostly basaltic or icy. However, in the refractory Ca– Al-rich
Moon, the crust consists mainly of anorthosite with basalt. The common and familiar
granite is perhaps unique to the Earth.
222 Chapter 2

Cratons formed earlier than 2.5 b.y. ago are Archean in age. They form the oldest
parts of Earth’s tectonic plates. These form the nuclei of land masses and seem to remain
unmodified by later tectonism. Their tectonic longevity is presumed to have been derived
from their ”keels” which go deep (. 200 km) into the Earth (Nyblade, 2001). The craton
keels are depleted in basaltic constituents (Al2O3, FeO and CaO) and volatile molecules
(H2O and CO2) compared with “fertile” mantle, generating basaltic volcanism along
MOR. This depletion makes the keels strong enough to resist wholesale destruction by
tectonic forces but, strangely, cratonic keels are neutrally buoyant and do not cause any
significant perturbations in Earth’s gravity field. The results of the study by Lee et al.
(2001) of the isopycnic (equal density) condition in the lithospheric mantle indicate that
more continental crust may have formed before 2.5 b.y. ago than is indicated by the
present-day distribution of cratons.
The Earth experienced an extensive episodic growth in its crust in the late Archean
or early Proterozoic, with only minor additions since then. Younger sediments of
Proterozoic age contained a significant population of zircons derived from late-Archean
rocks (3.0 – 2.5 b.y.; Stevenson and Patchett, 1990). This suggests that there was rapid
continental growth starting about 3.0 b.y. ago but, before that time, there were only
scattered small areas of continental crust. Early Archean continental crust was probably
dominated by sodium-rich granites (tonalites) rather than the K-granites of the present
upper continental crust. The post-Archean crust is typically depleted in Eu (normalized
REE patterns).
Heat flow from the mantle into cratons, which are the old, stable, buoyant parts of
continents, is approximately constant regardless of the craton’s age and heat flow through
cratons more than 2.5 b.y. ago was similar to that at present, whereas the heat flow out of
Earth was much higher because of its steady cooling. Buoyancy alone is insufficient to
stabilize 200 –300 km thick cratons for billions of years. A viscosity , 1,000 times the
normal upper-mantle viscosity of 1020 –1021 Pa s is also necessary.

2.7.3. Plate tectonics, magmatism and hotspots

In the Solar System, plate tectonics is a unique style of crustal motion in the Earth
alone, while in the Moon and Mars crustal dynamism is manifested by the processes of
melting, magmatic migration and differentiation.
In the Earth, a new generation of terrestrial crust is created by extrusion of basaltic
rocks taking place mostly at submerged “mid-ocean” and “back-arc” ridges. Oceans cover
almost two-thirds of the crustal Earth. This oceanic crust is presently composed of basalt,
overlain by a thin veneer of recent sediments. The latter are 1– 2 km thick, depending on
age, and are typically younger than 2 £ 108 years.
At the MOR, the basaltic magma chambers form plutons along with lavas and
feeder dykes, which move away from the source to the subduction zone to be transported
down to the Earth’s interior. Occasionally, slices of old oceanic crust are pushed above the
sea level to from “ophiolites”, as are seen in Oman, Newfoundland and Papua
New Guinea. A study of such ophiolites helps construct the model for the generation of
a new crust.
Petro-Tectonic Features of Terrestrial Planets 223

In plate-collisional zones, heat is generated by dissipation of shearing energy and,

added to this, the mantle heat at depths causes melting of the subducted plate. This results
in a wide spectrum of basalts through andesite and dacite and finally to rhyolite. The rocks
types associated with extruded basalts are alkaline in character and are considered as
products of fractional crystallization of basaltic parent magmas.
The other magmatism related to plate tectonics is manifested as a hotspot, as seen at
Hawaii, La Reunion or the Galapagos islands. The hotspot loci present a constant angular
separation from each other, bearing a frame of reference of plate motion. The migration of
a hotspot from the depths of the mantle to the surface is independent of plate motion. The
ultimate rooting of hotspot sources has been suggested to lie at the transition zone
(,670 km depth) between the upper and lower mantle or at a still deeper region at the
CMB (2,900 km depth).
The hotspot activity is now seen to be the most common form of expression of deep
magmatic activity of many other Solar System bodies. Evidently, to unravel the nature of
the combine thermal and material transport in the inner planets, an understanding of this
hotspot activity is important. The hotspots on the Earth are the manifestations of the
recycling and partial melting of subducted plates (after a residence time in the mantle in the
order of 1 Ga).
In the mantle, the convective velocities are of the order of centimetres per year.
Hence, the “transit” across the mantle takes ,100 m.y. and one “overturn” takes ,400–
500 m.y.
Mantle convection is the driving mechanism for plate tectonics and the related
processes such as continental drift, earthquakes, volcanoes and orogenies. The
neighbouring planets, Mars and Venus, currently do not show plate tectonics, although
both might have experienced it in the distant past. Slow cooling of the plates over a period
of 4.5 b.y. and some radiogenic heating contribute to the energy for motion of the plates.
Mantle dynamics is largely affected by subduction mantle rheology (non-
Newtonian) and plume dynamics. Erupted magmas indicate several chemically
distinct reservoirs, surviving mixing by convection (“whole-mantle”). In the Earth,
convection reaches the surface, where the oceanic plates act as the upper thermal boundary
layer participating in the convective motion. These plates, generated at the MOR, re-enter
the mantle in the form of “slabs” at subduction zones in less than 200 m.y.
Subducting slabs can be traced to the base of the transition zone by seismology. A
current model holds that slab material may pile up for some time in the transition zone until
it either descends into the lower mantle as an avalanche (Tackley et al., 1993) or is swept
aside by large-scale flow (Allegre and Turcotte, 1986).

2.8. The mantle

2.8.1. Geochemistry

The upper mantle consists of several chemically stratified layers, which include the
layers of peridotite, pyroxenite and garnetite produced during the crystallization of the
magma ocean.
224 Chapter 2

A knowledge of the composition of the uppermost mantle has been gathered

from basalts, which are products of partial melting from peridotites caught up in
ophiolitic complexes and massifs and from xenoliths brought up from as deep as 200 or
250 km by basalts or kimberlites. The partial melting of the uppermost mantle
to generate basalts left a residual peridotite depleted in low-temperature minerals
(“pyrolite model”).
A lherzolite peridotite is composed mainly of olivine and orthopyroxene (enstatite)
with some clinopyroxene (diopsidic). The aluminous phases, which are seen to be
occurring with increasing depth, are: plagioclase, spinel or pyrope garnet (majorite-
bearing). These make up the respective lherzolites, such as plagiocalse –lherzolite, spinel –
lherzolite and garnet – lherzolite. The major phase changes in upper-mantle mineralogy are
coincident with the 410 and 660 km seismic discontinuities given the mantle geotherm.
The lithosphere thermal boundary layer marks the transition from an adiabatic thermal
regime in the convecting mantle to a conductive thermal regime in the lithosphere. This
thermal boundary layer is shallow beneath the MOR, leading to melting by adiabatic
upwelling. In contrast, the lithospheric geotherm is steeper beneath ancient continents,
leading to a deep lithospheric thermal boundary layer. Another important rock type in the
mantle is metasomite. These rocks have enriched chemistries in large-ion lithophile (LIL,
typical of the crust) elements (K, Na, Sr and Ba) and in high-field strength elements (HFS,
e.g., Ti, Zr, Nb and P). These have unusual mineralogies of ferri-phlogopite (mica),
K-richterite (amphibole) and exotic alkali titanites in substrates of previously depleted
harzburgite.
Based on the phase study of the systems MgO –FeO – SiO2 and CaSiO3 – MgSiO3 –
Al2O3, the mineralogy and volume fraction of the Earth’s mantle is depicted in Fig. 2.6.

2.8.1.1. Mantle end members

Five mantle end members that can be suggested on the basic of Hauri and Hart
(1997) are as follows.
(a) Two end members show signatures of Sr- and Nd-isotope depletion: (i) DMM,
depleted MORB mantle and (ii) FOZO, depleted mantle with high 3He/4He.
(b) Three end members show isotopic signatures of enrichment: (i) EM1 and (ii) EM2 —
enriched in (U þ Th)/Pb. Intermediate to these end members can be explained
by mixing. The depleted (DMM and FOZO) components have depleted 187Os/188Os
ratios, while the enriched end members (EM1, EM2, HIMU) have elevated ratios of
the same.
For the mantle the depleted components (DMM,FOZO) are composed of peridotite,
while the enriched components (EM1, EM2, HIMU) contain probably recycled oceanic
crust to a significant extent (. 10%). Presence of sedimentary component in a minor way
may contribute to the variability of Sr, Nd, and Pb.
All enriched hotspots show elevated 187Os/188Os, suggesting recycled oceanic crust
to be a major contributor to most hotspots.
From many evidences it is suggested that primitive undifferentiated reservoir
(such as the BSE) is not likely to exist in the mantle (Van Keken et al., 2002). The entire
reservoir has been depleted to form continental and oceanic crusts. This is supported by
Petro-Tectonic Features of Terrestrial Planets 225

wide spread observation of depleted mantle components of both mid-oceanic ridges and
hotspots.

2.8.2. Petro-tectonics

Plate tectonics is driven by solid-state mantle convection, which carries deep-

mantle material upwards until it begins to melt at about 30– 100 km depth. On melting,
a mantle’s incompatible trace elements, mainly U, Th and K, go to the melt. The
melt may extrude through the oceanic or continental crust, thereby bringing the
chemical message of the mantle to the geochemists. Consequent to the melting,
the remaining part of the mantle becomes depleted of these incompatible elements. The
66,000-km-long MOR system offers valuable samples of depleted rocks, mostly basaltic
in composition.
Oceanic and crustal materials buried deep inside the mantle may resurface as
”mantle plumes”, which mainly create volcanic islands. The OIB are less depleted or even
enriched in the incompatible elements, in contrast to the depleted MORBs. The former
appears to come from an upper continental layer and the latter from a deeper layer.
Generally, MORB, OIB and continental crust are enriched in incompatible elements
(relative to the primitive mantle). OIB are much more enriched in incompatible elements
than MORB. The chemistry of OIB lies between that of the continental crust and the
depleted MORB source. This is because of the back-mixing of various types of continental
material into the mantle. However, there are many cases which cannot be explained
this way.
The processes participating in the inhomogenization of the mantle are manifested in
the following:
Mid-oceanic ridge samples the uppermost mantle and migrates laterally.
Occasionally, a ridge intersects a plume, when an oceanic crust of anomalous chemistry
is produced, e.g., Iceland.
Volcanic seamounts, littering over the oceanic crust, perhaps represent the
products of the local melting anomalies in the upper mantle, e.g., Cameroon Line
Volcanoes.
Continental flood basalts and oceanic plateaus are possibly the mushroom-like
head of a plume. A line of young volcanoes may be seen to connect to an old flood basalt
province, e.g., the Deccan Traps of India seem to lie in line with the currently active
hotspot of Reunion Island.
Seafloor spreading is caused by extrusion of basaltic partial melt from the mantle
(leaving a refractory residue).
Subduction at the plate margins causes injection of crustal material into the mantle,
where a rehomogenization takes place by convection.
Distinct contrasts in the isotopic composition and trace element content of intra-
continental rift magmas with those of MOR are suggestive of heterogeneity in the chemical
composition of the mantle beneath the continents and oceans. For example, the
intra-continental rifts, such as the one extending from Ethiopia to Zambia in eastern
226 Chapter 2

Africa, produced highly alkaline lavas and carbonate magmas characteristically enriched
in lanthanides, Nb, Zn and Hf.

2.8.3. Xenoliths

Mantle xenoliths carried up by basaltic or kimberlite magmas offer clues to

decipher the crystallization process in the mantle. The xenoliths fall into two classes:
cumulates and residues. The cumulates are produced by fractional crystallization by
basaltic magmas (Chen et al., 1992) while the residues are products of partial melting and
separation of basaltic magmas. The chemistry of the residues suggests the parent basaltic
magma to be lherzolite in composition. The xenoliths may be regarded as residues of
partial fusion or crystallized eutectic melts.

2.8.4. Deep-mantle flow and Wilson cycle: American Cordillera

The large-scale Cordillera structures seen in the North and South American plate
are indicative that Andes cordillera in the south and Laramide Rockies were formed in a
similar manner and their respective plates were driven westward by deep-mantle flow. The
Atlantic spreading was also similarly driven.
Marginal basins are opened up at convergent margins, e.g., in the western Pacific.
Russo and Silver (1996) suggest the existence of a causal link between the opening and
closing of ocean basins and the formation of calderas and marginal basins. Cordillera
formation occurs at the leading continental margin during the spreading phase of the
Wilson cycle, whereas marginal basins form during the closing phase.

2.8.5. Diversification of rock types

2.8.5.1. Petrogeny’s residua system

There are diverse primary basaltic magma compositions produced in space and time
in the Earth. The diverse compositional trends are best evaluated in terms of plate
tectonics. Certain magmatic differentiations are associated with plate-boundary regimes
(i.e., spreading ocean ridge or compressional island and continental volcanic arcs).
Basaltic magmas represent the high-temperature end members of a spectrum of liquid
composition, lowest-temperature end members are equivalent to granite or syenite and
H2O-rich silicate liquids to pegmatites. On a different sequence of crystal separation, a
residual liquid is left over. This, known as “Petrogeny’s residua system”, i.e., the Na2O –
K2O–Al2O – SiO2 system, crystallizing felspars, quartz and felspathoids. The end product
thus becomes granitic or pegmatite. Granites broadly represent the lowest melting
temperature products.
In the upper continental crust, granite is the most common rock. The question
arises: Is the genesis of granite always by fractional crystallization alone or by other
processes? Generally, the evidence such as high temperature (# 8008C) in the country
rocks around granite plutons, magmatic texture and the mineralogy of the granite suggest it
Petro-Tectonic Features of Terrestrial Planets 227

to be of igneous origin. But in regions affected by ultra-metamorphism manifested by

palingenesis, the process of granitization of crustal sedimentary and igneous rocks seems
most tangible. In a regional metamorphism, wherever the temperature of the continental
crust is raised to the condition of melting, the mixture of the melting temperature products
forms granitic melt, which migrates upwards to shallower crustal levels.

2.8.5.2. Effusive rocks

In volcanism with a high discharge rate, the rock types involved are andesite, dacite
and rhyolite. The eruption products often manifest the inverse stratigraphy of the sub-
surface chemical zonation of the tapped magmatic chamber, such that the bottom (earliest)
part of the deposit is equivalent to the top (first erupted) of the magma chamber. The
chemical zonation in the magma chamber occurs through the process of fractionation and
Soret diffusion of elements in the long-lived thermal gradients in the liquid present. The
magma chemistry is also modified by partial assimilation of the wall-rock.
Highly fluid fissure-fed basaltic lavas, with high rates of emission, flow out to
extensive flat plateaux, such as the Deccan Traps and Columbia River Plateau. The
eruption behaviour of volcanoes has recently been studied in great detail during the mild
eruption of Mount St Helens (1980), which ejected about 1 km3 of material.
The viscosity of magma is a function of SiO2 content and temperature. Gas-poor
basaltic magmas of low viscosity (,3 £ 102 P) erupt along an elongated rift or flow down
low slopes of high elevation, as seen in Mauna Loa on Hawaii. Relatively high viscous
(.105 P) rocks are SiO2-rich, such as rhyolite.

Basalt sources. A consensus has been reached regarding the generation of basaltic magma
from partial melting of peridotitic upper mantle. The conclusion is based on the evidence
offered by: (1) the results of high-pressure studies reaching beyond 25 GPa, and
corresponding to a depth of about 1,000 km and 2,0008C, (2) the mantle xenoliths, (3) the
slice of upper mantle rocks enclosed in some faulted crusta and (4) the seismic wave
velocity versus time curve of the mantle. The mantle peridotite occurs as a solid crystalline
mass predominantly of olivine and orthopyroxene, with subordinate amounts of
clinopyroxene and garnet in the pressure range above 1.5 GPa.
Beneath the continents, the upper mantle shows heterogeneity with localized
entrapment of basaltic melts, which crystallize as eclogite (sodic clinopyroxene þ garnet).
Locally, a reaction with hydrous supercritical fluids with the host results in the formation
of amphibole and phlogopite.

Andesite sources. Andesite makes the major volume of the rock types of the volcanic arcs
related to the subduction zones (viz. the Andes of Peru and Chile). A melting of basalt and
sediment layers of the subducted lithosphere possibly results in andesite magmas.
However, it is more likely that andesite is derived at low pressure (,1 GPa) by fractional
crystallization from basaltic magma, formed by partial melting of the peridotite overlying
the subducted lithosphere. Release of supercritical fluids such as H2O and CO2 and
elements such as K, Cs, Rb, Ba, Sr and Pb and other trace elements from the subducted
228 Chapter 2

sediments may bring about a dramatic lowering of the melting temperature of peridotite
and its fractionation.
However, magmas more siliceous than andesite (i.e., SiO2 . 55 – 60 wt% and
MgO , 5 wt%) cannot be generated by partial melting of peridotite but such high SiO2
rocks are the products of fractional crystallization of basaltic melts. Such evidences are
seen in island arcs, ridges and hotspots. In such cases, the solidus of crustal material has
been exceeded, resulting in siliceous melts to generate rocks varying from granite
(70 – 76 wt% SiO2) to andesite.

2.8.5.3. Calc-alkaline magmatism: LIL enrichment and “Pb paradox”

The generation of calc-alkaline magmas in subduction zones is thought to be the
most important mechanism for the growth of continental crust since the Proterozoic eon.
Fluid transport from the subducted slab into the zone of melting also contributes to the
trace-element and radiogenic-isotope characteristics of calc-alkaline magmas (e.g.,
Arculus, 1994; Miller et al., 1994).
Through experimental work on the partitioning of trace elements between fluids,
silicate melts and minerals, Keppler (1996) suggests that the agent responsible for the
transport of trace element in the subduction zone may be an alkali-chloride-rich aqueous
fluid, which can generate the trace elements- and isotope-enrichment pattern typical for
calc-alkaline magmas, including the enrichment of LIL elements, Pb and U, and the
characteristic depletion in Nb and Ta.

Chloride fluids: hard acid/base relation. Calc-alkali magmas can be products of

metasomatism involving alkali-chloride-rich fluid and the trace element composition of
the continental crust is largely a result of the complexing properties of chloride in
supercritical aqueous solutions. Because the composition of the fluid released in the
subducted slab depends on the composition of the seawater, the oceanic crust has been
affected by the chemical evolution of seawater (Keppler, 1996).
Using the concept of “hard” and “soft” bases and acids, it is possible to say which
elements are preferentially mobilized by chloridic fluids. Chloride, as a moderately hard
base, should preferentially react with moderately hard acids, such as Rbþ, Ba2þ, Sr2þ and
Pb2þ, while extremely hard acids, such as Nb5þ and Ta5þ, are not expected to react with
chloride and thus not be transported by the fluid.

Lead paradox. One of the most intriguing features of calc-alkaline magmatism and
perhaps the entire continental crustal composition is the “lead paradox” (e.g., Miller et al.,
1994). Compared with U and Th, Pb behaves less incompatibly during the formation of
oceanic crust but the relative compatibilities are reversed in the continental crust above the
subduction zone.
This may be explained by considering that, in equilibrium with silicate melts or
minerals such as clinopyroxene, Pb partitions much more strongly than U or Th into a
chloride-rich fluid. Metasomatism by such a fluid will therefore transport much more Pb
than U or Th from the subducted slab or the mantle wedge into the zone of melting.
Petro-Tectonic Features of Terrestrial Planets 229

The aspect of Pb concentration in K-hollandite structure in subducted lithologies has been

discussed in Section 11.6.2.3.

Radioactive disequilibria: fluid transport. Sr partitions preferentially into the hydrous

chloride fluid in equilibrium with crystalline phases such as clinopyroxene and other
Mg-silicates. This mechanism of partitioning causes contamination of Sr isotopes in the
zone of melting by Sr from the subducted oceanic slab, thus causing some relative
enrichment of radiogenic Sr-isotope composition of many calc-alkaline magmas (e.g.,
Arculus, 1994).
Such magmas also show radioactive disequilibria (Green, 1994) with strong
enrichment of 226Ra over 230Th. These disequilibria may also represent a fluid transport.

2.9. Earth’s rheology and dynamism

2.9.1. Lithospheric rheology and dynamism

The kinematics of surface motions can be described by the rigid-body motions of a

small number of plates, which show relative motions across narrow (tens of kilometres
wide) regions of deformation and intense seismicity at their boundaries. The continents
overriding some of the plates behave as continuously deforming solids rather than rigid
masses. To evaluate the continuous deformation, a knowledge of the spatial derivatives of
stress is required, the quantity of which is, however, more difficult to acquire than knowing
the velocities in the crustal region.
The governing equation for the deformation of a continuous lithosphere is the
stress-balance equation, which states that gradients of stress are balanced by the force of
gravity per unit volume:

dsji =dxj ¼ 2rgi

where sji is the ijth component of the stress tensor, xj is the jth coordinate direction, r is the
density and gi is the ith component of acceleration due to gravity (Note the left handed
coordinate system). However, stress or strain at depth within the lithosphere cannot be
measured, hence the vertical gradients of stress cannot be determined using the above
equation. However, evidences are gathering for rheological layering in the lithosphere and
such layering may play an important role in deformation of the upper crust.
Gravity anomalies reveal that, on length scales of .100 – 200 km, density contrasts
within the lithosphere are isostatically compensated. Thus, the weight per unit area of any
column of rock is supported by the vertical traction, szz ; on its base
ðz
szz ðzÞ ¼ 2g rðz0 Þdz0 ð2-21Þ
0

where z is the depth, z0 a variable of integration and g the acceleration due to gravity.
230 Chapter 2

England and Jackson (1989) developed a relation in terms of deviatoric stress as:

Lðdtxx =dx 2 dtzz =dx þ dtxy =dy Þ ¼ dG=dx

ð2-22Þ
Lðdtyy =dy 2 dtzz =dy þ dtxy =dx Þ ¼ dG=dy

where L is the thickness of the lithosphere, tij is the ijth component of the deviatoric stress
tensor, averaged vertically through the lithosphere and
ðL ðL ðZ
G ¼ 2 szz ðzÞdz ¼ g rðz0 Þdz0 ð2-23Þ
0 0 0

The stresses in equation (2-22) can be related to deformation by specifying a

relation between the vertically averaged deviatoric stresses and the vertically averaged
strain rates in the lithosphere. Differences in G are equal to differences in the gravitational
potential energy per unit area between columns (Molnar et al., 1993). The difference in DG
in gravitational potential energy per unit area between any two isostatically balanced
columns of lithosphere can be simplified as

DG ¼ ðgrc =2Þ½1 2 ðrc =rm ÞðS22 2 S21 Þ ð2-24Þ

where rc and rm are the densities of crust and mantle and S1 and S2 the thickness of crust in
the two columns.
The potential energy calculated from strain rates and the potential energy calculated
from topography are completely independent quantities. Agreement between these two in
an area would provide support for the hypothesis that equation (2-22) describes the active
deformation there.
Weakening of the plates occurs at the boundaries above the downwellings because
of the negative buoyancy causing a concentration of stress. In contrast, the weakness at the
spreading centres arises only because the plates are thin and the yield strength is low to
allow evolving plate tectonics to occur. The transformational boundaries (strike-slip faults)
are associated with toroidal motion. [Note: Any 3D solenoidal velocity field can be divided
into a polonoidal component, associated with divergence in a horizonatal plane, and a
toroidal component, associated with vertical velocity. The torodial component is
associated with the strike-slip (shear) component at plate boundaries.] A low-viscosity
region underneath the plates lubricates spreading centres, making them more localized and
less episodic.
Any fault (brittle failure) on the plate can reach the maximum depth of ,15 km,
below which the deformation is in the form of a distribution of microcracks. Shear zones
can be formed through recrystallization by deformation under high (10 – 100 MPa) stress
and lithospheric cooling, aided by viscous dissipation. These zones can form in only
,50,000 years by rapid release of elastic energy. Peridotites from typical rift-zone
environment or xenoliths from OIB show stress magnitudes of ,1 –10 MPa, whereas
peridotites from collision zones (low-temperature, high-pressure rates) show much higher
stresses (,50– 300 MPa; Jin et al., 1998).
Petro-Tectonic Features of Terrestrial Planets 231

2.9.2. Mantle rheology

The mantle is solid except for some shallow (,100 km depth) regions where a small
percentage of partial melting may take place. Over a long time scale, mantle rocks deform
like a fluid with viscosity ,1021 Pa s by the movement of vacancies and dislocations
through the crystal lattice. The deformation mechanism in mantle rocks ranges from brittle
failure to viscous creep. Strain weakening may also occur.
If the mode of flow is simple shear with large strain, the recrystallization becomes
stress controlled. Dislocation creep may be a major deformation mechanism. Diffusion
creep may contribute to plastic deformation in the upper mantle. The seismic anisotropy in
the upper mantle may offer strong evidence for dislocation creep. This has been further
discussed in Chapter 15.
Rocks exhibit different deformation mechanisms, ranging from brittle failure to
viscous creep, depending on the P, T differential stress and past history of deformation
(e.g., Ranalli, 1995). Plate tectonics is not affected by viscous creep with T-dependent
viscosity. Instead, under a rigid lithosphere, convection takes place.
High-resolution seismic tomography (e.g., Zhou et al., 1996) and numerical
modelling (e.g., Honda et al., 1993) suggest that the interactions of convection currents with
the transition zone are complicated, causing the geometry of the subducted slabs to be
variable. The complications arise from the changes in rheological properties and phase
transformations. Except for the wadsleyite phase (Dupas-Bruzek et al., 1998), no significant
experimental studies have been performed so far on the effects of phase transformation on
the rheological properties. However, the results of Karato et al. (1998) strongly support the
idea that a large reduction in grain size in the olivine-to-spinel transition in cold slabs leads to
a significant rheological weakening (Riedel and Karato, 1997).

2.9.2.1. Decompression and magma fragmentation

Samples of magma from the 1980 eruption of Mount St Helens that rapidly
depressed from initial pressures of up to 12 MPa (at temperatures in the range 750 –8258C)
fragmented to form pyroclastic products that are, in many respects, similar to those formed
in real eruptions. This may suggest that even relatively cool magma bodies can be very
hazardous when subjected to rapid unloading events (Alidibirov and Dingwell, 1996).
Magma rheology, together with the properties of the tensile pulse, will dictate the
mechanical nature of the magma response (brittle vs. ductile) (Webb and Dingwell, 1990).
The effective volume strain rate implied by decompression rate of 40 MPa ms21 in the first
0.15 ms is 101.6 s21 (for bulk modulus of magma, K ¼ 10 GPa). As the magma approaches
viscoelastic behaviour during decompression, the contribution of viscous (non-brittle)
dissipation of energy increases (Dingwell and Webb, 1989).

2.9.3. Seismic tomography: Iceland hotspot and Nazca plate

A seismic network records a number of distant earthquakes from a variety of

directions The differences in arrival times of compressional and shear waves can then be
inverted to recover the 3D seismic velocity structure beneath the network. This technique
232 Chapter 2

is known as seismic tomography. Compressional and shear-wave arrival times will be

inverted to determine the 3D structure of the crust and the upper mantle beneath the
network, which has an aperture of 2,000 km along its long-axis direction.
The details of upper-mantle structure, which is not resolvable by the global network
of seismic observatories, have now been better imaged by the deployment of portable
broadband seismic experiments (Carnegie Institution Yearbook 96/97, pp. 55– 61).
The tomographic results may represent the average isotropic heterogeneity, even if
the individual seismic observations are affected by the anisotropic structure. For example,
velocity anomalies in the upper portion of the oceanic mantle are commonly interpreted in
terms of the progressive cooling with/without localized reheating of a mechanical and
thermal boundary layer consisting of rigid oceanic lithosphere and an underlying, less
viscous, asthenosphere. The seismic data around Hawaii reveal a regional anomaly in
elastic anisotropy, which produces variations of seismic velocities, arising perhaps from
thermal effects.
Beneath central Iceland, the tomographic images depict a low-velocity anomaly
having the shape of a vertical cylinder of radius , 150 km and extends vertically
from ,100 – . 400 km depth. The anomaly depicts the contours for shear velocity 3%
less than normal and outlines a cylindrical zone of hot upwelling material extending
from more than 400 km depth to the uppermost mantle (i.e., ,100 km depth). A
portable network in Iceland reveals evidence for an upwelling plume to more than
700 km depth. An upward-travelling compressional wave converts part of their energy to
shear waves.
Tomographic images and geodynamic modelling suggest that subducting slabs
descend to the lowermost mantle. These paleoslabs cause thermal anomalies and high
overall seismic velocities. The contrast in material properties between what was the former
oceanic crust and oceanic mantle-lithosphere is primarily responsible for the velocity
anisotropy at the D00 zone.
The upper-mantle structure across the entire South American continent has been
imaged by spatial resolution. The eastward dipping, high-velocity anomaly at the western
margin of the continent traces the subduction of the Nazca plate beneath South America
deep into the lower mantle. High velocities extending to at least 300 km depth at the
eastern end of the profile correspond to the lithospheric roots of the Sao Francisco craton,
one of the most ancient continental segments.

2.9.3.1. Anomalous low-velocity zone

There is, however, a strange observation of an anomalous low-velocity zone
directly below the Paraná basin, the site of huge flood-basalt eruptions ,130 m.y. ago and
alkalic volcanism 80– 90 m.y. ago. The question arises of how this anomaly could persist
so long after the end of magmatism and a presumed mantle upwelling. The sufficiently
high vertical extent of the anomaly suggests that the mantle material moving with the
South American crust must have a thickness of several hundreds of kilometres. [Note:
tectonic plates are usually not so thick].
Petro-Tectonic Features of Terrestrial Planets 233

2.10. Convergent plate boundaries

New continental crust is extracted from the mantle by magmatic processes, whereas
old crust is recycled into the mantle at subduction zones The rates of these are shown
below:
Addition of mantle material to crust: ,1.6 km3 yr21
Amount of sediment subduction: 0.5 –0.7 km3 yr21
Amount of loss of continental crust to mantle: 1.6 km3 yr21
At the convergent plate boundaries, where subduction occurs, three tectonic units
may be identified as:
(1) The subducting slab: consisting of oceanic crust of basalt, dolerite and gabbro,
hydrated and pristine oceanic upper mantle and the overlying pelagic sediments.
(2) The mantle wedge: above the subducting slab lies the wedge-shaped peridotitic mass.
(3) The crust above the wedge: often develops uplifted mountain chains, island arcs, etc.
During its descent through the mantle, the subducting slab undergoes transform-
ations through metamorphic reactions to melting (Fig. 2.15). The peridotitic mantle wedge
gets metasomatized by the ingress of fluids or melts from the subducting slab. This
transformation causes the formation of minerals uncommon in pristine (unaffected)
mantle. The melting causes near-surface volcanic and hydrothermal activity and often
addition of the crust (Fig. 2.15).
The crust in island arcs is dominated by intermediate granitoids, along with mafics
like gabbro at depths. Often, the island arcs are typified by active volcanism that are
explosive due to the release of H2O arising from dehydration of the hydrated minerals in
the subducting slab.
Oceanic lithosphere generated at MOR is recycled into the mantle. Hydrothermal
alteration close to the MOR gives the crust a high initial H2O content and low slab
temperatures stabilize hydrous phases to greater depths than in surrounding hotter mantle.
Stabilities of these hydrous phases play a major role in controlling subduction zone
processes such as dehydration, melting and magmatism.
Buoyancy that is inherent in the very nature of the slab by virtue of lithology —
basalt underlain by depleted peridotite — becomes increasingly important as pressure
increases. A schematic model for the subduction of continental crust into the mantle and
exhumation of ultra-high pressure (UHP) metamorphic rocks by reverse buoyancy
tectonics is shown in Fig. 2.16 (Science, 285, 1999).

2.10.1. Subducting slabs

Subducting slabs at the convergent plate boundaries can be traced to the base of the
transition zone by seismology. A current model holds that slab material may pile up for
some time in the transition zone until it either descends into the lower mantle as an
avalanche (Tackley et al., 1993) or is swept aside by large-scale flow (Allégre and
Turcotte, 1986).
234 Chapter 2

Figure 2.16. A schematic model for the subduction of continental crust into the mantle and exhumation of the
UHP metamorphic rocks by reverse buoyancy tectonics in the temperature zones of 600–8008C (from Science,
285, 1999) (see also Fig. 11.4).

During subsidence of a slab in the upper mantle, a temperature increase of 18C

corresponds to pressure increment of 1/4 kbar. This is an order of magnitude greater than at
the Moho discontinuity and is 20 times greater than near the surface.
Detailed laboratory investigations of phase equilibria, volatile solubility in melts
and silicate solubility in fluids, partitioning of trace and major elements between minerals,
melts and fluids have helped construct the models of phase transformations in the
subduction zone.
In subduction zones, the generation of magmas occurs when hydrous fluids
released from the subducting slab infiltrate the mantle wedge and the temperature
exceeds the water-saturated solidus. The onset of melting would depend on the relative
rates of water release from the subducted slab and water incorporation in olivine and
pyroxene.
Upon release from the slab, the fluid would contain significant amounts of
halogens, particularly chlorine (derived from any material coming in contact with
seawater). The fluid-causing metasomatism of the source region of magmas in
subduction zones might, therefore, be a highly concentrated salt brine rather than a dilute
aqueous solution.
Water-rich basaltic and volcanic rocks plus the ocean-floor sediments are the
principal reservoir of H2O in the subducted oceanic lithosphere. In the basaltic component,
pargasitic amphibole may be the dominant H2O carrier (e.g., Davies and Stevenson, 1992)
whereas, in the pelitic component, phengitic white mica is commonly presumed to be the
supporting repository of H2O (e.g., Domanik and Holloway, 1996). This has been further
discussed in Chapter 13.
At pressures of 5– 10 GPa, 500– 1,000 ppm by weight of H2O can be incorporated
into the olivine structures. The pyroxene structure can contain more. The high equilibrium
solubility of H2O in olivine has important geochemical consequences. Any estimate of
water partitioning between upper mantle and transition zone must obviously consider
pyroxene as well.
Petro-Tectonic Features of Terrestrial Planets 235

2.10.1.1. Slab tomography: volatiles and partial melting

Tomography evidence suggests that some slabs penetrate into the lower mantle.
However, the lower velocities are suggestive of either higher temperatures (e.g., Anderson
et al., 1992) or the presence of melt (Zeilhuis and Nolet, 1994).
Some of the seismically slowest regions in the Earth are seen to occur above the
subducting slabs. These most likely reflect partial melting, caused by the volatiles released
from the slabs (e.g., Nolet et al., 1994). During subduction, the slab experiences high
temperature and partial melting of the slab occurs in the transition zone owing to the
presence of volatiles (Gasparik, 1993). The underlying mantle, however, is cooled by
the slab and may attain higher velocities than the subducting slab. Such velocity inversion
can obscure discovery of the continuation of the slab.
Tomographic images of subduction zones indicate that the slabs maintain their
rigidity and integrity in the upper mantle, encounter resistance at the 670 km discontinuity
and could stall there and get softened and deformed, bent and thickened (e.g., Fukao et al.,
1992). This scenario is consistent with an incipient melting by the presence of volatiles
during thermal equilibrium of the slab with the surrounding mantle.
A recent viscosity model predicts a low-viscosity layer in the transition zone (e.g.,
King and Masters, 1992). The partially melted slab may contribute to this low viscosity and
this viscous mass may be largely recycled by upwelling under the spreading centres. The
speed of upwelling may be a few meters per year.

Doubly-seismic slab: hydrous phases. The subducting slab descends as a cooler tongue
into the mantle and shows itself as seismically faster than the mantle it dips into. Often the
subduction zone itself is marked by a double seismic zone (e.g., Abers, 1996). The
interface between the slab and the overlying mantle wedge constitutes the upper zone. This
possibly includes metamorphosed oceanic crust, a hydrated metabasalt system (e.g.,
Peacock, 1993) and sediments. Below this zone, , 20 –40 km deep, lies the second layer
reflecting the compositional heterogeneity within the slab.
Perhaps ,20% of the slab volume is made up of oceanic crust and sediments. The
seismic data seem to be consistent with mineralogies involving hydrous phases such as
lawsonite, glaucophane, tremolite and zoisite (discussed in detail in Sections 13.3
and 13.4). Helffrich (1996) modelled a low-velocity layer near the subduction zone to a
depth of , 65 km as a lawsonite blueschist, which transform to eclogite assemblages at
greater depths.

2.10.1.2. Deflections of seismic discontinuities: NW Pacific subduction

Short-period seismic data are less coherent than long-period data. However, these
results coupled with other studies indicate that, in the vicinity of the subducting slabs, the
410 km discontinuity is elevated by about 15 km while the 670 km discontinuity is
depressed by 20 –50 km compared with the global averages of discontinuity depths. This is
consistent with the effect of a cold subducting slab.
The slab broadens or turns near 670 km, thus spreading the cold-slab temperatures
into the adjacent mantle. This model receives support from the observation of velocity
inversions for northwest Pacific subduction zones, which image horizontal extension of the
236 Chapter 2

slab just before 670 km. This apparent deflection could be a result of the resistance to slab
penetration through the 670 km phase discontinuity. However, two subducting slabs, the
Japanese and Farallon plates, are known to reach the CMB (e.g., van der Hilst et al., 1997).
High-frequency seismic energy can be reflected only by relatively sharp interfaces,
whereas low-frequency arrivals can be generated even by a fairly gradual change in
properties. Additional reflectors (anomalies) have been noted at various levels, the
shallower ones have been seen to occur only locally at 220 and 520 km depths.

2.10.1.3. Deep-focus earthquakes: fossil slab at transition zone

Old, strong and dense slabs may descend to the bottom of the transition zone where
the mineral assemblages are reconstituted at near-equilibrium conditions. Below the
critical temperature, Tc, the slab may remain strong enough. Stresses resulting from the
downward thermal buoyancy forces and the viscous forces resisting slab descent
presumably cause deep faulting (of unspecified character) at temperatures below Tc. Deep
earthquakes occur by transformational faulting in metastable peridotite where the slab
stresses satisfy the criterion for such failure. This causes a release of seismic energy and
latent heat anchored at the 670 km discontinuity could lose support owing to incipient
melting caused by volatiles. This may lead to a sudden movement of the slab and trigger
deep-focus earthquakes (Gasparik, 1993). Such a mechanism is consistent for the Bolivian
earthquake of 1994, which was apparently caused by a sudden movement along a pre-
existing hydrated fault and was triggered by the lubricating effect of the released volatiles
(Silver et al., 1995).
In the northern part of Tonga trench subduction occurs at a rate of , 250 mm yr21,
sufficient to produce a slab extending to the base of the seismogenic zone (700 km) in just
3 m.y. — much too fast for thermal assimilation (Green, 2001). Chen and Brudzinski
(2001) show that a very large remnant (fossil) slab appears to be “floating” beneath Fiji. On
this, an “outboard” earthquake occurs. The slab is neutrally buoyant (i.e., of the same
density as its surroundings). They conclude that the fossil slab originated in the fossil
Vitiaz trench, which runs west – northwest from the current northern extreme of the Tonga
subduction zone. The fossil subducted slab exhibits earthquake distributions similar to
those in the currently active slab.
The subducting slabs are the coldest parts of the mantle and the cold material
preferentially supports larger stresses and fails by fracture rather than flow.
The cold thermal structure of the slab may cause the peridotite to bypass the
equilibrium olivine ! spinel boundary because of sluggish reaction rates. A sluggish
reaction such as olivine ! spinel cannot produce transformational faulting and hence
cannot produce the earthquakes. Therefore, at greater depths when the transition zone
minerals convert at near-equilibrium conditions to the lower-mantle assemblage of
magnesiowüstite (Mg, Fe), perovskite and Ca-perovskite, no earthquake can occur.
Earthquakes cannot occur in the portions of the slab that have transformed to spinel not in
the lower mantle. The depth of the deep focus earthquake is, thereby, restricted to 670 km.
Deep earthquakes occur within subducting slabs only of oceanic lithosphere.
Petro-Tectonic Features of Terrestrial Planets 237

2.10.2. Subducting mafic, ultramafic rocks and sediments

Subduction of oceanic lithosphere continuously transports rocks that are produced

near the Earth’s surface, particularly MORB and associated sediments, back into the deep
mantle. These introduced heterogeneities are stirred and distorted by convection in the hot
mantle. Eventually, they become buoyant and rise as plumes that melt to produce the OIB.
The time taken for this recycling process is thought to be typically about a billion years.
Subducted basaltic crust may become buoyant at the mantle’s 660 km discontinuity
and may remain so down to a depth of 800 km where it may transform to a garnetite layer
(Irifune and Ringwood, 1993; Ringwood, 1994).
Hirose et al. (1999) investigated the MORB basalts to pressures of 64 GPa
(ø1,500 km depth) and found these to be transformed to a perovskitic lithology at
,720 km depth, where the P –T boundary of phase transition is positive (in contrast to
the negative slope for peridotite). It is presumed that this perovskitic mass would
gravitationally sink into the deep mantle but the melting data of Hirose et al. (1999)
suggest that the former basaltic crust would be partially molten at the base of the lower
mantle, where the temperature exceeds 4,000 K.
In meta-sediments, chlorite þ kyanite and staurolite þ quartz are stable only to
pressures of about 2.0 GPa (, 60 km depth). At higher pressures, hydrous assemblages
comprise talc þ chloritoid þ phengite. Between 4 and 5 GPa, talc þ kyanite break down
to Mg –Al-pumpellyite or OH-topaz. The storage of H2O in sediments to more than
200 km depth is ensured by phengite, lawsonite and Mg – Al-pumpellyite under
temperatures as high as 9008C (see Fig. 2.17; Mysen et al., 1998).

2.10.2.1. Subduction of oceanic lithosphere: upper to lower mantle

The subduction of oceanic lithosphere into the Earth’s deep interior is thought to
drive convection and create chemical heterogeneity in the mantle Because of differences in
chemistry, density and melting temperature between the basaltic crust and the underlying
peridotite, the oceanic lithosphere may not subduct uniformly.
The major alteration product of ultramafic oceanic lithosphere is serpentine
(antigorite). The high-pressure stability limit of antigorite is marked first by the formation
of phase A at temperatures below , 6008C at P , 6 GPa.
At T . 6008C, antigorite dehydrates to produce olivine, enstatite and fluid:

antigorite þ phaseA ¼ enstatite þ fluid

When this temperature is exceeded and the check point is overcome, a series of
dense, hydrous magnesium silicates are formed. As the pressure is increased above the
stability of antigorite (6 GPa/, 6008C), phase A first appears. Phase A possibly transforms
to phase D or to phase E depending on temperature as the pressure is increased further
(Burnley and Navrotsky, 1996; Pawley and Wood, 1996). At even higher pressure, phase
D, with or without phase B, may be stable (e.g., Ohtani et al., 1995). For further details, see
Section 13.2.
238 Chapter 2

Figure 2.17. Phase relations for H2O-saturated mid-ocean ridge basalt. Experimental data compiled by Schmidt
and Poli (1998) [Source: Mysen et al., 1998 q1998 Mineralogical Society of America].

Intensely serpentinized peridotites in a variety of tectonic settings are reported in

the vicinity of both fast (e.g., East Pacific Rise) and slow (e.g., Southwest Indian Ridge)
spreading ridges, as well as along passive continental margins (e.g., Galicia margin). The
oceanic Mohorovicic discontinuity (Moho) may represent the limit between serpentinized
and non-serpentinized peridotites, i.e., the serpentinization front (e.g., Coulton et al., 1995;
Muller et al., 1997).

670 km deep slab samples(?): Malaita pipe in Papua. In an “Earth-shattering discovery”,

fresh, seemingly unaltered minerals from as deep as 670 km have been reported from the
island of Malaita in the Solomon Islands (east of Papua New Guinea).
In an exposed top of a volcanic pipe (pushed up by Pacific– Australia plate collision
, 23 m.y. ago), a rock is reported to contain micron-size diamond, majorite – garnet and
even silicate perovskites, which are suggested to have been produced from oceanic basalt
subducted to lower depths of the transition zone, i.e., 670 km.
The Malaita pipes blasted through mantle laden with old slabs of tectonic plate that
had sunk, just as slabs now sink into deep-sea trenches off Japan and South America. The
slabs can pile up in the transition zone before falling deeper into the lower-mantle. Seismic
tomography has revealed the presence of such slabs.
Petro-Tectonic Features of Terrestrial Planets 239

2.10.2.2. Mid-oceanic ridge basalt

Mafic rock production at ridges. Mid-oceanic ridge and rift are areas where Earth is
chemically differentiating. Today MOR volcanism creates a mafic crust 7 km thick and
leaves a mantle residue as thick as 70 km.
If the current production rate of oceanic crust is integrated over 4.5 b.y. a minimum
of 2.6 £ 1026 g of mafic material was produced at ridge and reintroduced into the mantle at
subduction zones, equivalent to 6.5% of the mantle mass. The complementary mass of
depleted residue is 10 times the amount (Hauri, 2003).
Oceanic ridges may contain samples of abyssal peridotites. These samples represent
the uppermost part of depleted residue left after MORB generation.
Trace and major elemental and isotopic evidence indicates that MORB is produced
by mixing of melts derived from enriched and depleted components. The enriched
component is possibly related to garnet pyroxenite (; metabasalt) inclusions in the
depleted peridotite MORB source region (e.g., Hirschmann and Stolper, 1996). This
association suggests “marble-cake” mantle hypothesis, originally proposed for pyroxenite
veining peridotite massifs (Allegre and Turcotte, 1986).
Hirose et al. (1999) have studied the phase relations and melting temperatures of
MORB at pressures up to 64 GPa (<1,500 km depth). They found that at , 720 km depth,
when the basaltic crust is transformed to a perovskitite body, it is no longer buoyant.
This transition boundary has a positive P– T slope, in contrast to the negative slope of the
transition boundary in peridotite. Thus, the perovskititic mass of the original basaltic
lithosphere would sink deep into the mantle.
The melting data obtained by Hirose et al. (1999) also suggest that, at the base of the
lower mantle, the former basaltic crust would be partially molten if the temperature there
exceeds 4,000 K. The phase relations in MORB composition up to 27 GPa are presented in
Fig. 2.18 (compiled by Hirose et al., 1999, see caption).
The phase relations for H2O-saturated MORB with the experimental data compiled
by Schmidt and Poli (1998) were shown earlier in Fig. 2.17.
The phase assemblage in metabasalt at blueschist conditions (10 –50 km depth) is
lawsonite þ glaucophane þ chlorite þ albite (or jadeite) ^ phengite (Fig. 2.17). With
increasing temperature, lawsonite reacts to produce epidote or zoisite, while chlorite
decomposes to form garnet. At temperatures above 6008C and 1.5 – 2.2 GPa, an amphibole-
bearing eclogite results. At pressures above , 2.4 GPa and temperatures below 6608C
(corresponding to about 70 km depth), amphibole breaks down to chloritoid and
blueschist transforms to lawsonite-eclogites (Fig. 2.17). Amphibole-bearing eclogite
is transformed to zoisite-bearing eclogite. At pressures near 3.0 GPa, zoisite reaches
its maximum stability in metabasalt, resulting in an almost dry eclogite at con-
ditions above 3.0 GPa, 7008C (corresponding to a depth ,100 km in the subducting
metabasaltic slab).
At lower temperature, lawsonite persists. Lawsonite reaches its maximum
temperature stability at 8308C and 8.4 GPa. At pressures above the zoisite stability,
lawsonite decomposes through a continuous reaction of the type:

lawsonite þ diopside þ garnet1 ¼ garnet2 þ coesite=stishovite þ H2 O ð2-25Þ

240 Chapter 2

Figure 2.18. Phase relations in MORB composition under pressure of up to 27 GPa. Solid lines represent solidus
and liquidus temperatures. Between 25.5 and 26.5 GPa lithology changes from garnetite to perovskite. The half-
filled symbols represent mixed assemblages (from Hirose et al., 1999).

where garnet2 is richer in grossular and pyrope components than garnet1 (Schmidt and Poli,
1998). In the coesite stability field, this reaction has a positive DV and a positive dP=dT
slope which steepens with increasing pressure. When stishovite replaces coesite (above
,8 GPa), DV of reaction (2-25) becomes negative, which results in a negative dP=dT slope
for the reaction. The maximum pressure stability of lawsonite in subduction zones is,
therefore, 8– 9 GPa. In a potassium-bearing metabasalt, phengite forms ubiquitously to
pressures of 10 –11 GPa (Domanik and Holloway, 1996; Schmidt, 1996; Ono, 1998).
Other minor hydrous phases in metabasalt are paragonite, talc and staurolite.
Paragonite forms at 1.4 ^ 0.2 GPa (at 500– 6508C) and decomposes near 2.2 GPa and
500 –7008C. The conditions of talc and staurolite occurrence are relatively restricted,
although talc might be an important high-pressure phase in Mg-gabbros (Mysen et al.,
1998, p. 110).
Because MORB has high Si, Al, Fe and Na content, the minerals developed at
mantle pressures are substantially lighter in nature compared with those obtained from
transformed mantle peridotite, viz. perovskite, CaSiO3-perovskite and Mg-wüstite
(Kesson et al., 1998). The high Al2O3 content of MORB also results in high majorite –
perovskite transition pressure than seen in peridotite transition.
The zero-pressure density of basaltic crust with perovskitic lithology is
4.23 g cm23. Hirose et al. (1999) calculated the zero-pressure densities at 24 and
26 GPa as 3.87 and 3.92 cm23, respectively. These values are consistent with those
obtained much earlier by Irifune and Ringwood (1993).
At 24 GPa (/2,023 K), the mineral assemblage that was observed was composed of
majorite þ stishovite þ CaSiO3-perovskite. Along with this, an aluminous phase with
Ca –ferrite structure is noted. This observation is consistent with the earlier experiments
by Irifune and Ringwood (1993). At 26 GPa (/. 2,473 K), a new Al – Ca phase is
found (Fig. 2.18). This is akin to the CAS phase described by Irifune et al. (1994).
Petro-Tectonic Features of Terrestrial Planets 241

A majorite – perovskite transformation is seen to occur at the pressure of 26 GPa

(< 720 km depth) and 2,000 K (Hirose et al., 1999).
Under the 660 km discontinuity, if the slabs of basaltic lithosphere accumulate to
form a megalith of $ 60 km thickness, its transformation to a denser perovskite lithology
would cause it to penetrate deep inside the mantle. This transition boundary has a positive
P –T slope, whereas the transition boundary in the underlying harzburgite has a negative
P –T slope (Irifune and Ringwood, 1987).
However, no major phase transformations have been reported at higher pressures up
to 100 GPa. This suggests that these phases remain stable in the deep mantle, excepting
majorite which is fully transformed to perovskite at P .27 GPa. At this pressure
(/, 4,000 K), the partial melt generated should have a compositional enrichment in MgO,
FeO (with depletion in SiO2 þ Al2O3). Thus, the solid residue in MORB composition at
lower-mantle pressure would become denser because of its higher iron content.
At a depth of 1,500 km (<64 GPa), the melting temperature of basalt is about
250 K lower than that of mantle peridotite (Zerr et al., 1998). Extrapolation to 135 GPa
yields a melting temperature of MORB of about 4,000 K at the CMB.
Thus, if in the temperature locally reaches 4,000 K in the D00 region, which may be a
graveyard for subducted lithosphere, the crustal material of basaltic body would partially
melt. This melt at the base of the mantle can account for the recent seismic observations
of the seismic anisotropy (Kendall and Silver, 1996) and anomalously slow P-wave
velocities (Williams and Garnero, 1996; Revenaugh and Mayer, 1997). However, under
such a scenario, the temperature of the outer core must be higher than the 4,000 K required
for melting MORB perovskite (Hirose et al., 1999). The temperature difference over the
thermal boundary between core and mantle may reach 1,500 K and hot mantle
plumes, including partially molten slab materials, are likely to arise from this depth
(Hirose et al., 1999).

2.10.3. Mantle wedge

The mantle wedge above the subducting slab is characterized by low velocity and
high attenuation (i e., low Q value), which is in sharp contrast to the high velocity and low
attenuation in the interior of the slab. This indicates that the mantle wedge is hot and the
slab is cool.
Frictional heat may arise from the shear flow in the boundary zone between the
slab and the mantle wedge. The frictional heat is equivalent to the mechanical work done
by the slab motion and is measured from the shear stress multiplied by the slab velocity per
unit time and unit area of the boundary surface. In Newton flow with a constant velocity,
the heat production is proportional to the velocity squared but mantle material is non-
Newtonian and strongly temperature-dependent.
Primary basaltic magma is generated at , 100 km in depth, where the dehydration
of hydrous minerals in the subducting slab supplies volatile components to the mantle
wedge. The extruded basaltic magmas should have experienced temperatures as high as
1,4008C in equilibrium with the mantle peridotite. The ascending mantle diapir absorbs
heat from the surrounding mantle, which is hot due to convection induced by the slab drag
242 Chapter 2

(Tatsumi, 1986). The ascending flow results in arc volcanism. The viscosity of the mantle
wedge is as low as 1019 Pa s.
Mantle xenoliths often offer a glimpse of the upper-mantle wedge above subducting
oceanic plates. Ultramafic xenoliths are commonly spinel peridotites (olivine þ
orthopyroxene þ clinopyroxene þ spinel ^ pergasitic amphibole) reflecting equilibrium
pressures in the 1 –2 GPa range and at temperatures near 1,000 ^ 2008C (Takahashi,
1986). Garnet peridotite xenoliths (olivine þ orthopyroxene þ clinopyroxene þ garnet)
are common in kimberlites. Overgrowth of olivine by orthopyroxene in the mantle wedge
offers evidence of the influx of silica-rich fluids from the subducting slab. Similarly,
influx of LIL elements and light to middle REE cause a relative depletion of high-field-
strength elements (Nb and Ta, in particular; Ionov and Hofmann, 1995) (see also
Section 2.8.1).

2.10.3.1. Arc magmatism: alkali and H2O activity

Incompatible elements such as K and Na are seen to be systematically enriched
toward the backarc. Enrichment of incompatible elements is accompanied by a decrease in
silica content and the chemistry of the magma becomes more alkaline. Experimental
petrology has shown that basaltic melt becomes more alkaline in equilibrium with mantle
minerals as partial melting takes place at a higher pressure. Thus, an alkaline basaltic
magma can be produced relatively deep in the mantle.
In the backarc region, the reaction of the magma with the surrounding mantle
minerals would also make the magma more alkaline. It is seen that alkaline volcanoes in
Korea and Northern China are distributed farther from the Japan Trench. However, no
Benioff-Wadati zone is observed beneath these volcanoes.
A wide variability in magma composition is a major characteristic of arc volcanism
and is due to the differentiation process in the crust.
The melting-phase relations of arc magmas appear consistent with high H2O
activity (Kushiro, 1990), which is also reflected in higher oxygen fugacities in the island
arc upper-mantle conditions (Arculus, 1994). The ultramafic xenoliths in arc lavas are , 2
orders of magnitude higher in values of oxygen fugacity compared with those of cratonic
kimberlites and abyssal peridotites. The influx of H2O into the mantle wedge is reflected by
the occurrence of sodic amphiboles (paragasite, seen in xenoliths).
The H2O contents determined in eruptive magma (glass) or in the glass trapped in
phenocrysts (e.g., Sobolev and Chaussidon, 1996) offer some clues to the H2O budget in
the slab– mantle wedge complex. High H2O content has been observed in glass inclusions
in olivine from arc magmas (Harris et al., 1984). Analysis of glass inclusions in spinel
harzburgite from Philippines indicates H2O contents of the source magma to be , 5 wt%
(Schiano et al., 1995). Gaetani et al. (1993) concluded that arc magmas may entrain
# 6 wt% H2O dissolved during crystallization. Through partial melting studies relating
to Japanese arc, Kushiro (1990) concluded that melt probably contained , 1 wt%
H2O, whereas magnesian andesite melt (from peridotite partial melting) might contain
,8 wt% H2O.
Petro-Tectonic Features of Terrestrial Planets 243

2.10.4. Hotspots and mantle plumes: OIB versus MORB

Oceanic hotspots are generally accepted to be the manifestations of plumes of hot,

upwelling mantle material, but the nature of such flows remains enigmatic (see, for further
discussion, Section 2.13.2). Hawaii is the Earth’s largest active intra-plate volcanic centre
and the Hawaiian mantle plume occurring here represents an ancient slab of subducted
ocean floor. The original isotopic composition of the subducted ocean floor has survived
with little change despite a billion years of mantle convection (Halliday, 1999).
The magma generating from melting plumes under places as Hawaii is known as
OIB and is chemically and isotopically distinct from the MORB. Mostly, OIB is produced
by melting slabs of old MORB that have been recycled by subduction. Paradoxically, most
erupted OIB are isotopically more heterogeneous than MORB (Allègre et al., 1995),
although they are chemically (major element) less variable.
To explain the heterogeneity, Putirka (1999) opines that the upper portion of the
slab is metamorphosed basaltic rocks with high levels of sodium, calcium and aluminium.
This is very distinct from the more Mg-rich and Al-depleted peridotite, which makes up the
bulk of the mantle. He models a partitioning of trace elements between peridotite and melt
as a function of temperature and pressure (depth). The Na/Ti ratio is shown to be
particularly sensitive to pressure during melting.

2.10.4.1. Iceland mantle plume

Iceland is one of the most thoroughly investigated hotspots. Researchers have
debated whether these plumes rise from the bottom of Earth’s lower mantle, 2,900 km
down, or are rooted only a few hundred kilometres down in the upper mantle (Science, 23
May, 1997, p. 1198). Seismic waves slow down as they pass through hot body. By plotting
wave speeds from distant earthquakes, seismologists had already created CT-scan-like
images of a shallow plume below Iceland. Shen and his colleagues, through a study of
1,500 pairs of P- and S-waves, were convinced of a plume arising from the lower mantle
and having a narrow column of rocks ,1258C hotter than its surroundings. A zone of
partially molten rock has been identified at the base of the mantle below Iceland (see Kerr,
1998; Science, 31 Jan, 1997, p. 614). Wolfe et al. (1997) undertook a regional broadband
seismic experiment to determine the 3D velocity structure of the upper mantle beneath
Iceland using the relative travel times of body waves from teleseismic earthquakes.
Inversion solutions of these data show a cylindrical zone of low P- and S-wave
velocities that extends from 100 km to at least 400 km depth beneath central Iceland. The
radius of the low-velocity anomaly is , 2% for P-waves and 4% for S-waves, indicating
that Iceland is underlain by a hot, narrow plume of upwelling mantle. Mantle anisotropy
could account for some of the differences in P- and S-wave models. In particular, a narrow
zone of upwelling flow beneath central Iceland would tend to induce vertical alignment of
the crystallographic a-axes of mantle olivine crystals, which would produce a vertically
oriented fast direction of anisotropy for P-waves but not for S-waves (Kendall, 1994). The
results suggest that the mantle beneath the eastern Iceland is shallower than 100 km and it
displays the highest average wave speeds in the region.
244 Chapter 2

The low-velocity anomaly in the upper mantle beneath Iceland can be interpreted as
the locus of the active plume and its interaction with the spreading plate boundary.

2.10.4.2. Plumes and underplating

Mantle plume and the associated magmatic underplating work as an important
mechanism for driving regional surface uplift and denudation of large portions of the
continents. Such uplift occurs rapidly because substantial volumes of basalt melts are
added to the crust over geologically short periods of time (1 – 10 Myr) and can shed large
amounts of elastic sediments being shed into surrounding basins (White and Lovell, 1997).
Episodic magmatic underplating of the continental shelf of northwestern Europe
was related to the activity of the Iceland plume. This suggests that individual pulses of
sedimentation provide a potentially sensitive measure of plume activity and so may be
used to resolve time-dependent fluctuations in mantle-plume activity predicted by
theoretical studies of mantle convection. Around the British Isles, the upliftment and
denudation can be related to Paleogene magmatic underplating. The magma was generated
by adiabatic decompression of asthenosphere with a potential temperature of 1,450 –
1,5008C, requiring the existence of mantle plume.

2.10.4.3. Megaplumes
Two huge “megaplumes” under Africa and the Pacific are observed in seismic
tomographic models. Inversions of normal mode-splitting functions for 3D mantle-density
structure indicate that these are anomalously dense, in great contrast to the density
anomaly of hot thermal upwellings. Combined lithospheric and crustal recycling can
explain the high 3He/4He reservoir and the 1– 2 b.y. old reservoirs. The missing heat-
producing elements (e.g., U and Th, which produce 4He) are lost to the basaltic oceanic
lithosphere generated through partial melting of mantle. [Note: Any high value of 3He/4He
would indicate a frozen value of the primitive oceanic lithosphere.] The high 3He/4He
reservoir is contained in two megaplumes. From a uniformitarianist view, this reservoir
would be recycled oceanic lithosphere.

2.11. Upper mantle

A highly homogeneous mantle received widespread support with the discovery of

plate tectonics, which is consequential to a vigorously convecting interior. Cumulative
evidence shows that the models of a largely homogeneous Earth’s mantle, such as pyrolite
or piclogite (see Table 2.2), do not stand testing with the experimental and seismic
evidence, while this points to a suggestion for a layered mantle. A two-layered mantle
suggests independent convection regimes in the upper and lower mantle, separated by a
largely impermeable 670 km discontinuity.
Liu (1994) showed that in the transition from (Mg,Fe)2SiO4 spinel to (Mg,Fe)SiO3-
perovskite and magnesiowüstite, the Clapeyron slope was negative at lower temperatures
but becomes positive at higher temperatures. Consequently, whereas the downwelling flow
was still impeded, hot plumes ascended across the positive boundary with ease.
Petro-Tectonic Features of Terrestrial Planets 245

The olivine-rich composition of the primitive mantle is derived from the

composition of mantle xenolith. This composition, however, deviates from those of
meteorites and the Sun, which have an Mg/Si ratio close to unity (e.g., Anders and
Grevesse, 1989). Large-scale differentiation producing an olivine-rich upper layer requires
the presence of a complementary silica-rich layer at greater depths. Liu (1979) suggested
that the 670 km discontinuity is a chemical boundary between an olivine-rich upper mantle
and a more silica-rich lower mantle.
The partitioning of siderophile elements between silicate melt and sulfur-bearing
molten iron suggests that the extraction of iron from the mantle during core formation
occurred at 28 –30 GPa, and thus at depths not exceeding much more than 670 km. The
400 km discontinuity is usually explained as the phase boundary between the a- and b-
phase of olivine but Gasparik (1989) opined that transformation from pyroxene to garnet
occurring in compositions close to the enstatite –jadeite join at 13 –13.5 GPa is univariant
and hence can produce a sharp discontinuity at 400 km depth. In contrast, the a- to b-
olivine transformation is divariant and would occur over a depth interval 10 –20 km
(Katsura and Ito, 1989). On this basis, Gasparik (1997) opined that the univariant
transformation from pyroxene to garnet is the most likely cause for the 400 km
discontinuity.
Within the mantle, large cells of convection are induced by instabilities and driven
by temperature gradient. In a mantle-convection cell, the strain is very heterogeneous and
changes as deformation proceeds. In the whole lower mantle, convection can take place
owing to the ubiquitous presence of volatiles whereas, in the upper mantle, convection is
limited to the volatile-bearing regions, which include oceanic lithosphere, asthenosphere,
the transition zone between 520 and 670 km depth and the narrow conduits associated with
spreading centres, subduction zones and hotspots. The presence of volatiles is essential for
the recycling of the subducted oceanic lithosphere.
187
Os/188Os values. Fertile peridotites and chondritetic meteorites have estimated
BSE values at around 0.128 (e.g., Luck and Allegre, 1992). Abyssal peridotites show lower
values than this. This occurs due to long-term depletion of rhenium and consequent
depletion of 187Os/188Os.
Due to the removal of Rb and Sm the upper mantle shows depletion in Sr and Nd
isotopes. The mass of Rb and Sm removed from the upper mantle is similar to the mass of
these elements in the continental crust (e.g., Hofmann, 1988).
The 187Os/188Os ratios (like Re/Os ratios) show nearly perfect separation between
peridotites and eclogites. The pyroxene-dominated mafic layer in alpine peridotites
contains very high 187Os/188Os ratios, with ages as old as 1.3 b.y. (e.g., Reisberg et al.,
1991). These layers typically constitute 5% of the mass of any given peridotite body
(Hauri, 2002).
Re –Os system. The Re– Os isotope system is uniquely related to the partial melting
phenomenon than any other radiogenic isotopic pair. In general the Re depletion age of the
upper mantle is regarded as 2 b.y. Based on this the Re abundances of the upper mantle is
estimated at 0.17 ppb (for the BSE it is 0.26 ppb; Hauri and Hart, 1997).
Of the total mass of the mantle if 50% of the mass is regarded as depleted then the
mass of Re depleted from the mantle is estimated at 1.7 £ 1017 g, compared with only
246 Chapter 2

1 £ 1016 g of Re estimated to reside in the continental crust (Esser, 1991). The

Re/Os ratios for residual peridotites are distinct from those for mantle-derived magmas.
The 187Os/188Os ratio of seawater is 10 times higher than the mantle value. The
Fe- and Mn-oxyhydroxides scavenge Os from seawater; hence Os contents of MORB are
very low, but Re/Os ratios of MORB are high. This is the reason why 187Os/188Os ratios
are always higher than those of seafloor peridotites. The 187Os/188Os ratios are low in
peridotites while high in mafic rocks, with almost no average between the two types.
Re –Os ages of pacific and Atlantic MORB are about 0.557 m.y., while Indian
Ocean MORB shows a value of 1.37 m.y. The older age of the latter correlates with
a slower mean-spreading rate compared with the former two.
The 187Os/188Os ratio of the upper mantle beneath MOR is best preserved by
abyssal peridotite and MORB magnetic sulfides which are free from Os of sea water origin
(Hauri, 2003). In 187Os/188Os ratios between seafloor peridotites and hotspot basalts there
is a marked difference. Like MORBs nearly all hotspot basalts have 187Os/188Os
ratios higher than abyssal peridotites and BSE. Os isotopic composition of OIB reflects
long-lived elevated Re/Os ratios in the mantle, which are higher than in BSE.
Ancient subcontinental peridotites exhibit uniquely depleted Os isotope ratios. In
basalts from oceanic hotspots the 187Os/188Os ratios are uniformly elevated and those from
abyssal peridotites are depleted. The relationship between 187Os/188Os and other isotopes
(Sr, Nd, Pb, O) suggests that most of the mafic components in the convecting mantle
originate from ancient subducted oceanic crust.

2.11.1. Upper-mantle anisotropy

Anisotropy in the upper mantle is notably observed beneath Hawaii where strong
VSH . VSV (VSH.SV, velocity of SH- and SV-waves). The anisotropy persists down to the
deep (, 200 – 300 km) upper mantle (Montagner and Guillot, 2000) and the anisotropy in
the mantle wedge where strong trench-parallel polarization of the fast S-waves occurs is
observed close to the trenches (e.g., Smith et al., 2001).
In old oceanic mantle (. 80 m.y.), anisotropy appears to be different from that in
young oceanic mantle and SV . SH anisotropy. It is suggested that a change in
mechanism of seismic anisotropy with age arises possibly from kinematic to dynamic
(stress-controlled) mechanism due to decrease in temperatures.
Stress-controlled mechanism may cause polarization anisotropy of surface waves.
The seismic anisotropy seen under the European continent in which the P-wave velocity
direction is seen to be significantly inclined to the horizontal plane, attributed to the stress-
controlled linear preferred orientation (LPO) near the possible rift zones or due to the
presence of fossil subducting slabs. If stress is homogeneously distributed, grains with hard
orientation will dominate the LPO, but if strain is homogeneous, grains with soft
orientation will dominate.
In the deep upper mantle, the tectonic processes involve large strains and dynamic
recrystallization should play an important role in the microstructural development,
including LPO (e.g., Krarato, 1984). When dynamic crystallization occurs, temperature
Petro-Tectonic Features of Terrestrial Planets 247

controls the mechanism of seismic anisotropy. However, the mechanisms of LPO and the
seismic anisotropy in the deep upper mantle are not as well understood as in the uppermost
oceanic mantle.
Strong LPO develops during upwelling and the preferred orientation stabilizes
during spreading and attenuates during subduction. Thus, the processes that occur within
crystals on an atomic to microscopic scale transform to large dimensions to explain such
properties as anisotropy of the lower mantle (consisting of oxides) and the inner core of the
Earth (consisting of iron alloys).
Christensen (1984) reviewed the LPO of naturally deformed ultramafic rocks in
ophiolites (rocks of the uppermost oceanic mantle). The LPOs of olivine in most of the
ophiolites show orthorhombic axial symmetry, the axial symmetry axis corresponding to
the olivine [100] axis maxima and sub-parallel to paleospreading directions. This result is
consistent with the control of LPO by the dislocation glide, the olivine slip system being
[100] (010) or [100] {0kl}, which is consistent with the experimental results (Carter and
Avé Lallemant, 1970).
The seismic anisotropy in the uppermost oceanic mantle (possibly extending to the
lower lithosphere) is consistent with the LPO of olivine and pyroxenes of ophiolites
(e.g., Nicholas and Christensen, 1987). The LPO in ophiolites and the seismic anisotropy
in the uppermost mantle may be due to the absence of significant dynamic recrystallization
arising due to the relatively low strains from rapid cooling. Kinematically controlled
ophiolites are seen in ophiolites. However, near oceanic ridges where dynamic
recrystallization occurs, the seismic anisotropy may be attributed to being kinematically
controlled by very high temperatures, possibly exceeding the solidus.

2.11.2. Mantle minerals versus discontinuities

The two major Mg-silicate phases that dominate the mantle are Mg2SiO4 and
MgSiO3; their P –T domains are important for geophysics. The phase diagram of
the MgSiO 3 system shows stability fields of single phases like pyroxenes
(protoenstatite, orthoenstatite and clinoenstatite), garnet, ilmenite and perovskite. The
two-mineral domains appear for b-phase þ stishovite, and g-spinel þ stishovite. At
high temperature (.1,800 K), pyroxene transforms to garnet, then to ilmenite and
to perovskite.
Earth’s mantle is a multi-component system; its volume percentage of
phases present as a function of depth is schematically presented in Fig. 2.19 (after the
style of Ringwood, 1989), where the near-vertical boundaries will generate abrupt
change.
The density discontinuity at 400 km arises from olivine to wadsleyite (b-olivine)
transition and the 670 km transition demarcates the spinel (g-olivine) to silicate
perovskite þ magnesiowüstite phase change. Above 400 km, all silicon are [iv]Si while
below 670, all silicon are [vi]Si.
The 670 km seismic discontinuity, dividing the transition zone from the lower
mantle, coincides with the perovskite phase-transition boundary. Perovskite of
248 Chapter 2

Figure 2.19. Relative proportions of minerals in the mantle. Hatched regions indicate divariant phase transitions
in olivine polymorphs. Cross-hatched regions denote uncertainty in phase proposition of mw (magnesiowüstite)
and cpv (Ca-silicate pv). Dotted lines indicate approximate pressures of oxide transitions [Source: Bina, 1998;
Mineralogical Society of America].

composition , (Mg0.9Fe0.1)SiO3 seems to be the dominant lower-mantle mineral (Liu,

1979). It is thought to be the Earth’s most abundant mineral. It is orthorhombic, in which Si
occupies near-regular octahedral coordination, while Mg is in a larger site with eight
nearest-neighbour O atoms.
Silicate perovskite plus magnesiowüstite (Mg,Fe)O account for the relatively high
seismic velocities in the lower mantle (670 – 2,900 km), in which the velocities increase
smoothly with depth (see Fig. 2.3).
For most of the complexities of seismic velocity –density between 400 and 670 km,
the mineralogical models with octahedral silicon work satisfactorily. The pressure –
temperature field of MgSiO3 (Fei et al., 1990) is presented in Fig. 2.20. In the pressure
range from ,10 to 30 GPa in the transition zone, the high-pressure silicates contain both
IV
Si and VISi.
At high pressures, stishovite is the stable form of SiO2 and has a rutile structure with
exclusively [vi]Si. At still higher P, SiO2 attains CaCl2 structure with eight-coordination for
Si (Kingma, 1994) and ultimately it amorphizes (Sharma and Sikka, 1996).
Petro-Tectonic Features of Terrestrial Planets 249

Figure 2.20. Phase diagram for MgSiO3. viSi-bearing phases: perovskite (PV), garnet (GT), and ilmenite (IL), and
pyroxene (PX), spinel (SP), stishovite (ST), b-Mg2SiO4 (b) and liquid (L) (from Fei et al., 1990, q1990 American
Geophysical Union).

Consequently, the depth-wise scenario in the upper-mantle mineralogy may be

depicted as below (see also Figs. 2.12 and 2.19).

2.11.2.1. Upper mantle

350 km: pyroxene and aluminous garnet enter into solid solution, forming
majorite with garnet structure.
400 km: olivine transforms to b-phase.
Transition zone: b-phase changes to g-spinel, garnet to ilmenite structure and CaSiO3
molecules get into calcic perovskite.
650 km: Disproportionation of (Mg,Fe)SiO3 occurs to form perovskite and
magnesiowüstite. This may be the reason for seismic discontinuity. It
is seismically marked as the thin subducted oceanic lithosphere
trapped at the interface between the upper and lower mantle
(Ringwood and Irifune, 1988).

2.11.2.2. 400, 520 and 670 km discontinuities

The 400 km discontinuity making the a-olivine to wadsleyite phase is
also considered as the boundary between the pyroxene and garnet layer. The
boundary is univariant in Na-rich pyroxene compositions and therefore can produce a
sharp discontinuity. The original bottom of the magma ocean could correspond to the
520 km discontinuity. Following the solidification of the magma ocean, the komatiitic
and basaltic volcanism was triggered by volatile-rich mantle plumes originating in the
deeper parts of the transition zone or the lower mantle. The elevated temperature necessary
250 Chapter 2

for the origin of komatiite apparently reflects residual temperature from the magma
ocean stage.
At 520 km depth pressure, if the mantle temperature is assumed to be , 1,2508C,
the phase boundary for the transformation clinopyroxene þ majorite garnet to majorite
garnet þ calcium silicate perovskite (cpx þ Gt to Gt þ Ca Pv) should occur as shown in
Fig. 2.7. On either side of the phase boundary, majorite garnet makes up most of the mid-
transition zone “non-olivine” mineralogy. The Ks and G of CaPv are much higher than
those of cpx. At this depth pressure, the phase equilibria studies show that CaPv makes up
only , 5 –10% of the total mineralogy and the seismic velocity jump can scarcely exceed
1 –3% (which can also be brought about by b-phase ! g-spinel transition). However, the
intermittent nature of 520 km could indicate that it represents isolated lenses or layers of
“fast” material rich in garnet ^ CaPv.
The discontinuity between the transition zone and the lower mantle is marked at
around 670 km (some postulate it at 660 km). This abrupt discontinuity is explained as an
isochemical phase transformation of g-spinel to (Mg,Fe)SiO 3-perovskite þ
magnesiowüstite.
The size of the velocity jump at 670 km discontinuity is twice as large as that
occurring at 410 km. Thus, PREM gives DVp and DVs of 4.8 and 6.6%, respectively, for
670 km and only 2.6 and 3.6% for 410 km. The compelling evidence for a
convection barrier at 670 km is the negative dT=dP slope of the g-spinel to
(Mg,Fe)SiO3-perovskite þ Mg-wüstite transformation and the effect of the subducted
slab on the 670 km topography.

2.11.2.3. Ca-phases in mantle discontinuities

The Ca content of pyrolite is close to the maximum Ca solubility in garnet. The
exsolution of CaSiO3-perovskite from garnet, occurring at pressures higher than 17 GPa,
has been expected to produce the high-velocity gradients observed in the seismic velocity
profiles of the transition zone between 500 and 670 km depths. The maximum Ca content
of garnet is found at 17 GPa, corresponding to 15 mol% of CaSiO3. If pyrolite contains
40% garnet, the total amount of CaSiO3-perovskite formed by exsolution from garnet
between 17 and 21 GPa is 3 mol%.
An immiscibility gap may exist to split garnets to Na-rich and Na-poor and Ca-rich
and Ca-poor varieties. Such an immiscibility could produce a sharp discontinuity in an
eclogite mantle at 400 km depth.
A sharp 520 km discontinuity corresponding to the breakdown of the diopside
omphacite to garnet and CaSiO3-perovskite can only be present in a Ca-rich mantle.

2.11.3. Mantle melting and extraction

Mantle depletion refers to melt extraction of basaltic (pyroxene þ felspar ^

olivine) components (Si, Fe and Ca) from geochemically fertile mantle lherzolite (garnet
or spinel þ olivine þ Ca– Mg – pyroxene). The residue that remains in the mantle is low in
Si and high in Mg/Fe ratios and the rocks produced are dunite (olivine only) and
harzburgite (olivine þ Mg – pyroxene).
Petro-Tectonic Features of Terrestrial Planets 251

Melt extraction in the early Earth also included olivine- and pyroxene-rich
komatiite (volcanic peridotite). Basaltic composition that is trapped in the mantle
crystallizes garnet and Ca– Na pyroxene (omphacite). This bimineralic rock is known as
ecologite. Subducted ocean-floor basalts are transformed to amphibolite and, at higher
pressures, to eclogite.
Above the subducting plate, an ascending flow of mantle is driven by its internal
buoyancy associated with higher temperature conditions and partial melting. The
subducting slab supplies volatile components that facilitate partial melting in the flow.
The ascending mass requires incoming heat to preserve the convection. The flow that has
passed the top of the mantle moves toward the backarc. The outgoing flow is gradually
cooled so that the magma in it is partly solidified with more incompatible elements.

2.11.3.1. Deep-mantle melting: melt sinking

At the very high pressures as are prevalent in the Earth’s mantle and core, the
density of a silicate melt may approach and even exceed that of the corresponding
crystalline phases with which it is in equilibrium. At a specific depth, the condition of equal
buoyancy could be reached: melts below that depth would sink, above it they would rise.
For stipulating the base of the lower mantle (, 2,900 km depth) experiencing a
pressure of 136 GPa, Stixrude and Bukowinski (1990) used a purely phenomenological
criterion. The Helmholz free energy, which is a function of the volume and temperature, is
parameterized to fit the available experimental data (principally phase equilibria and EOS).
They predicted that the melting temperature of MgSiO3 perovskite is essentially
independent of pressure above , 50 GPa and has a weak maximum. Seismological
studies nonetheless show no evidence for such widespread melting and again, throughout
the lower mantle, the density contrast between solid and liquid may be negligible.
Stixrude and Bukowinski’s (1990) calculated density changes for the liquid are
consistent with those determined for molten diopside (CaMgSi2O6) up to 38 GPa using
shock-wave techniques. However, there have been no direct EOS measurements for
MgSiO3 liquid at high pressure.
Experiments on metasilicate glasses at mantle pressures indicate that tetrahedral
silicon can be preserved to relatively high pressures (. 35 GPa) with the compression
achieved mostly by distortion and repacking of SiO4 tetrahedra. Evidently, the higher
degree of freedom available in the liquid and glass permit higher densities relative to the
crystal, without necessarily increasing silicon coordination.
The melting interval for perovskite at the pressures of D00 is estimated to be 4,500 –
5,000 K at the CMB (Knittle and Jeanloz, 1989). However, perovskite is unstable in the
D00 layer and dissociates to form (Mg,Fe)O and SiO2.

2.11.3.2. Depletion and mixing: non-Newtonian high-viscosity blobs

Partial melting of the mantle of MOR and other magmatic environments causes
differentiation. Through subduction, these heterogeneous differentiated products can
re-enter the mantle, where convection tends to homogenize these heterogeneities. Different
locations of the mantle lead to different magmatic products with varying concentrations of
incompatible trace elements. MORB upwelled from shallow mantle show depletion in
252 Chapter 2

incompatible elements and are fairly homogeneous in composition, whereas OIB

occurring in Hawaiian chain islands are less depleted and are thought to represent
upwelling plumes from a deep, hot boundary layer. Their compositions seem to be
products of mixing between depleted MORB mantle (DMM) and three other reservoirs.
These reservoirs are radiometrically dated to be 1 –2 b.y. old — equivalent to several
“overturns” of mantle convection (see Section 1.24). Thus, OIB are not primordial and are
possibly products of the mixing of mantle with recycled oceanic or continental crust. Some
OIB display an anomalously high ratio of 3He/4He. 3He is primordial, dating back to the
Earth’s formation, whereas 4He is continuously formed from radioactive decay of U and
Th. Thus, high 3He/4He indicates a primitive reservoir which remained isolated since the
early Earth.
Many geochemists suppose the upper-mantle composition to be DMM whereas the
lower mantle is chemically primitive. However, high-resolution global models of 3D
mantle velocity obtained by seismic tomography indicate subducting oceanic plates plunge
down to the lower mantle. In many locations, the slabs appear to flatten out temporarily
above the 660 km discontinuity before descending further into the lower mantle.
Mixing efficiency is reduced by lateral viscosity variations caused by non-
Newtonian rheology of high-viscosity blobs. These blobs of viscosities 10– 100 times the
viscosity of mantle may float to the top or sink to the bottom to form a layer.

2.11.4. Peridotite mineralogy at depths

At 400 km, peridotite composition predominates. Peridotite composition in the

Earth’s mantle shows a series of transformations: a ! b, b ! g, g ! pvs þ MW and
cpx ! maj-gt.
Here, olivine (a-phase) transforms to modified spinel (b-phase). In the transition
zone (400 –670 km), majorite – garnet becomes the dominant constituent in place of
pyroxenes (Liu, 1977).
Among naturally occurring peridotites, garnet lherzolite xenoliths in kimberlite
pipes offer the deepest samples coming from the inside of the Earth. The equilibrium
pressure and temperature of the garnet lherzolite xenoliths have been much investigated by
experimental petrologists and a number of geothermometers and geobarometers have been
proposed (Gasparik, 1984; Bertrand et al., 1986).
At low pressure, the Mg# (¼ Mg/Mg þ Fe) value of garnet is much lower than seen
in other silicates but this value increases with pressure and becomes higher than that of co-
existing spinel (g-phase) at 20 GPa (Akaogi and Akimoto, 1977; Takahashi and Ito, 1987).
The major phase transformation in the mantle occurs thus: (i) a-phase to b-phase
then b-phase to g-phase in orthosilicate composition (Akimoto, 1972); (ii) pyroxene to
garnet (Akaogi and Akimoto, 1977); (iii) g-spinel to MgSiO3-perovskite plus
magnesiowüstite (Ito and Yamada, 1982).
In addition to these phases, Takahashi and Ito (1987), in their model geotherm with
maximum pressure 26 GPa (, 720 km) and a maximum temperature (, 1,6008C), found
two accessory phases in the natural peridotite composition: (1) Ca –P, a Ca-rich phase
Petro-Tectonic Features of Terrestrial Planets 253

which is probably an unquenchable, high-pressure polymorph of diopsidic pyroxene, and

(2) Al –P, an aluminous phase with unknown crystal structure.

2.11.4.1. Mantle silicate framework

Silicates dominate throughout the Earth’s mantle (composed essentially of olivine,
clinopyroxene, garnet and perovskite) to a depth of , 2,900 km. In low-pressure silicates,
silicon is essentially tetrahedral. The silicon coordination number acts as the major crystal-
chemical difference between the Earth’s crust and lower mantle. Above the depth
of , 250 km, Si in essentially in 4-fold coordination while below 670 km it is entirely
six-coordinated. Between 250 and 670 km, Si occurs both as [iv]Si and [vi]Si.

Olivine (a-(Mg,Fe)2SiO4). Olivine has isolated Si tetrahedra sharing three edges with
(Mg,Fe)O6 octahedra. The octahedra share some edges with other octahedra. At 400 km
discontinuity, Mg-rich olivine transforms to b-(Mg,Fe)SiO4, which has an Si2O7 group and
(Mg,Fe)O6 octahedra. The bulk modulus of this phase is 165 GPa. At higher pressure, g-
(Mg,Fe)SiO4 forms with a face-centred cubic (fcc) structure. The bulk modulus of this
form is 183 GPa (Weidner et al., 1984).

Clinopyroxene (Mg,Fe,Ca)2Si2O6. Large cations like Ca are only accommodated at an

irregular dodecahedral 8-fold coordination site at lower pressures in a clinopyroxene
structure, in which each tetrahedron shares two corners with other tetrahedra to make a
chain. At higher pressure, Na gets into this 8-fold site to form high-pressure jadeite.

Garnet. In the cubic pyrope-almandine series, (Mg,Fe)3SiO12, the 8-fold dodecahedral

site, is occupied by large cations at moderate pressures. But at higher pressures, Si starts
entering octahedral sites, leading to a tetrahedral form of garnet.

Perovskite. In silicate perovskite, ASiO3, when A is relatively large (e.g., Ca), the cubic
form is stable whereas for small A-site cations (such as Mg), an orthorhombic structure
results.
In addition to the above observations, it must be noted here that any structure with
an octahedrally coordinated tetravalent cation must be of high pressure.

Other dense structures. At the Earth’s lower mantle (.25 GPa), all silicates are observed
to transform to one of the seven dense structures in which all Si are six-coordinated. These
seven structures — rutile, perovskite, ilmenite, hollandite, Ca-ferrite, pyrochlore and
K2NiF4 — are well known topologies.

Octahedral silicon in mantle silicates. At the top of the Earth’s lower mantle at pressures
between 8 and 30 GPa, the common silicates undergo transition from four- to six-
coordinated silicon-bearing silicates. Above 27 GPa, many silicates transform to
perovskite structure, in which SiO6 octahedra form a 3D corner-linked network,
while larger R cations fill positions with oxygen coordination of eight or greater.
254 Chapter 2

The quintessential characteristics of the structural elements of some dominant phases of

mantle are discussed below.
Dense high-pressure phases commonly incorporate shared edges between
octahedra. In stishovite, hollandite and pyroxene, a combination of edge and corner
sharing is observed but in phase B (Section 13.2.4), each Si octahedron shares all edges
with adjacent Mg octahedra.
In garnet, wadeite and pyrochlore, the Si octahedra form part of a corner-linked
framework but additional cations in eight or greater coordination share edges and faces
with the octahedra. Ilmenite shows an unusual face-sharing between Mg and Si octahedra
as well as corner- and edge-sharing.
All SiO6 groups display rigid-body vibrational motion. The dense high-pressure
silicates with higher oxygen coordination display the longest Si –O bond distances. The
mean Si – O bond lengths in these compounds average more than 1.78 Å, while octahedral
volume average almost 7.7 Å3. The compositions and calculated densities of silicates with
SiO6 octahedra are presented in Table 2.6.

Systematics of silicon coordination in the mantle. The other silicates having both
octahedral and tetrahedral silicon are garnet (majorite), pyroxene, wadeite and anhydrous
B-phase. Structures like rutile, hollandite and Ca-ferrite are formed from edge-sharing
chains of silicon octahedra.

TABLE 2.6
Compositions and calculated densities of silicates with SiO6 octahedra

Composition Mineral name Structure type r0 calca (g/cm1)

High-pressure phases with SiO6 groups only

SiO2 Stishovite Rutile 4.29
CaSiO3 – Cubic perovskite 4.25
MgSiO3 – Ortho perovskite 4.10
MgSiO3 – Ilmenite 3.81
ZnSiO3 – Ilmenite 5.25
KAlSi3O8 – Hollandite 3.91
BaAl2Si2O8 – Hollandite 5.3
CaAl2Si2O8 – Hollandite 3.9
NaAlSiO4 – Calcium ferrite 3.91
Sc2Si2O7 – Pyrochlore 4.28
In2Si2O7 – Pyrochlore 6.34
Ca2SiO4 – K2NiF4 3.56
High-pressure phases with SiO6 1 SiO4 groups
MgSiO3 Majorite Garnet 3.22
MnSiO3 – Garnet 4.32
Na(Mg0.5Si0.5)Si2O6 – Pyroxene 3.28
K2Si5O24 – Wadeite 3.09
Mg14Si5O24 – Anhydrous phase B 3.44
Mg12Si4O19(OH)2 – Phase B 3.37
a
r0 calc ¼ density calculated from unit-cell parameters at room pressure and temperature.
Petro-Tectonic Features of Terrestrial Planets 255

TABLE 2.7
[vi]
Predicted and observed high-pressure Si compounds based on the substitution 2[vi]Al ! [vi](Mg þ Si)

Structure type Known [vi]Al phase Possible [vi]Si phase

Corundum/ilmenite Al2O3 MgSiO3a

Pyroxene NaAlSi2O6 Na(Mg0.5Si0.5)Si2O6a
Garnet Mg3Al2Si3O12 Mg3(MgSi)Si3O12a
Pseudobrookite Al2TiO5 (MgSi)TiO5
Gibbsite Al(OH)3 (MgSi)(OH)6
Diaspore AlO(OH) (MgSi)O2(OH)2
Kyanite Al2SiO5 (MgSi)SiO5
Staurolite Fe4Al18Si8O46(OH)2 Be4Al14(Mg2Si2)Si8O46(OH)2b
Clinozoisite Ca2Al3Si3O12(OH) Ca2Al(MgSi)Si3O12(OH)
Lawsoniet CaAl2Si2O7(OH)2H2O Ca(MgSi)Si2O7(OH)2H2O
Crodierite Mg2Si5Al4O18 Mg2Si5(Mg2Si2)O18
a
Known high-pressure phase.
b
In staurolite Fe2þ is four-coordinated, a configuration unlikely in a high-pressure phase. Thus, Be is substituted
for Fe.

From the few systematic relations observed among the structures, the following
criteria for [vi]Si phases can be admitted.
1. Three structure types — rutile, hollandite, Ca-ferrite — are formed from edge-sharing
chains of silicon octahedra.
2. All the seven high-pressure [vi]Si structures without tetrahedral Si are isomorphs of
room-pressure oxides with trivalent or tetravalent transition metals (Ti, Mn or Fe) in
octahedral coordination.
3. Three high-pressure magnesium silicates (ilmenite, garnet and pyroxene) are derived
from room-pressure [vi]Al structures by the substitution 2[vi]Al ! [vi](Mg þ Si). Based
on this substitution, a number of high-pressure phases have been predicted to occur in
addition to the observed ones (see Table 2.7).
Sergei Stishov (see Stishov and Popova, 1961) showed the transition occurs from a
relatively open quartz framework of corner-sharing tetrahedra to the dense rutile-type
structure of stishovite with edge-sharing chains of silicate octahedra. The corresponding
increase in density, from 2.65 to 4.41 g cm23 in 0– 8 GPa (Ross et al., 1990), has
profound implications in seismic discontinuity. Stishovite has a simple TiO2 structure with
edge-linked chains of SiO6 octahedra that run parallel to the c-axis and the octahedra are
corner-linked to four adjacent chains. Two symmetrically distinct atoms — Si at ð0; 0; 0Þ
and O at ðx; x; 0Þ with x approximately 0.3 — define the structure in space group P42 =mnm:
Stishovite, synthesized above 10 GPa, is the stable form of free SiO2 throughout
most of the Earth’s volume. The presence of stishovite grains in sediments near the KT
boundary layer supported the hypothesis that a large impact, rather than volcanism, led to
the mass extinction , 65 Ma ago.
Silicon, when coupled with other electronegative cations such as P or C, may
become an octahedrally coordinated network former. In these unusual materials, the
polyhedra tend to be linked by corner sharing. Such octahedral corner sharing may also
occur in a variety of organic molecular crystals with viSi (see Flynn and Boer, 1969).
256 Chapter 2

The third group of viSi silicates possesses open framework structures with corner
sharing between silicon octahedra and other polyhedra of electronegative cations,
notably ivP.

2.12. Lower mantle

2.12.1. Phases

The lower mantle is dominated by (Mg,Fe)-perovskite and oxide phases such as

magnesiowüstite (MgO), CaO, and SiO2. CaSiO3-perovskite serves as the “invisible”
phase because its density is similar to the lower mantle. Using the first-principles plane-
wave pseudopotential method Karki and her co-workers (viz. Karki and Crain, 1998)
investigated the high-pressure (1 –140 GPa at 0 K) structure and elasticity of the major
silicates (MgSiO3 and CaSiO3-perovskite) and oxides (MgO, CaO and SiO2), which are
considered to be the major constituents of the lower mantle. They calculated
the longitudinal and shear-wave velocities of these silicates and oxides and their
assemblages. These data were compared with the seismologically derived wave velocities
for this region of the Earth. The maximum FeO content in the perovskitic lower mantle
along the geotherm is shown in Figs. 2.21 and 2.22.

Figure 2.21. FeO solubility in perovskite as a function of temperature at 26 GPa. The solid squares stand for
single-phase perovskite, whereas the solid triangles present the three-phase assemblage (perovskite þ
magnesiowüstite þ stishovite). The compositions of perovskite in three-phase assemblage were determined by
X-ray diffraction (open circles) and with electron microprobe (open squares). Data in Ito et al. (1984) and Ito and
Takahashi (1989) are shown for comparison (horizontal bars). The solid line is the best fit
XFe ¼ 20.0394 þ 9.14 £ 1025T(8C) (Fei et al., 1996, q 1996 American Geophysical Union).
Petro-Tectonic Features of Terrestrial Planets 257

Figure 2.22. The maximum FeO content in a perovskititic lower mantle along the geotherm (Stacey, 1992). Error
bar range represents the composition limits at 2008C above and below the geotherm (Fei et al., 1996, q 1996
American Geophysical Union).

The calculated EOS of CaSiO3 in the cubic perovskite structure (Karki and
Crain, 1998) agrees well with the EOS determined at lower-mantle pressure (e.g., Wang
et al., 1996).
MgO is found to be stable in the NaCl structure (B1-phase) up to 450 GPa and then
transforms to the B2-phase (CsCl structure). The predicted B1 –B2-phase transition
pressure is comparable with the all-electron linearized augmented plane wave (LAPW)
prediction of 510 GPa (Mehl et al.,1998). SiO2 is predicted to transform from stishovite to
the orthorhombic CaCl2-type structure at 47 GPa. The CaCl2 structure is then predicted to
transform to the columbite structure (Pbcn) at 98 GPa and finally to the pyrite structure at
226 GPa (Karki et al., 1997). The elastic constants of these systems vary strongly with
pressure, especially near the vicinity of structural changes.
For MgSiO3 perovskite, the calculated zero-pressure derivative of the shear
modulus G0 ¼ 1:65 ^ 0:05 compares favourably with the value of 1.8 ^ 0.4 from recent
ultrasonic interferometric data up to 8 GPa (Sinelnikov et al., 1998).
The CMB exhibits the largest contrast in both seismic wave speeds and density
in Earth’s interior. A sharp discontinuity in wave speed is noted with substantial
variations in topography located several hundred kilometres (, 250 km) above the CMB.
This implies the existence of a chemical or phase boundary. This may be accompanied by
strong lateral and radial variations in temperature and composition. The strongest
anomalies at CMB extend some 500 –1,000 km into the mantle. Thermal differences may
be the principal cause of anomalies, aided by compositional heterogeneity.
Partial melt in the lower mantle could be denser than its co-existing solids (Ridgen
et al., 1989), particularly if iron partitions preferentially into the liquid. If the ULVZ is
produced by downward descent of melt from the overlying mantle, it would in
consequence be enriched in incompatible (and possibly volatile) elements (Knittle,
1998, AGU). Chemically differentiated oceanic crustal material might have penetrated to
258 Chapter 2

D00 providing a distinctive hotspot geochemical behaviour and possibly generating

laminated features that cause seismic anisotropy of the lowermost mantle (Kendall and
Silver, 1996).
Material of (Mg,Fe)2SiO4 composition appears to undergo eutectic-style melting at
CMB pressures and temperatures near 4,300 K (Holland and Ahrens, 1997).

2.12.2. Solidus in the lower mantle

The melting temperature of (Mg,Fe)SiO3-perovskite increases rapidly with

pressure (Shen and Lazor, 1995). However, the melting curve of magnesiowüstite,
(Mg,Fe)O, the second most abundant lower-mantle mineral, has a much smaller slope
(Zerr and Boehler, 1993). This difference in the melting behaviour of the end-members
complicates the predictions of eutectic composition in the lower mantle.
The solidus of a pyrolite-like composition, approximating that of the lower mantle,
was measured by Zerr et al. (1998) up to 59 GPa using CO2 laser heating in a DAC. The
solidus temperatures are seen to be ,700 K lower than the melting temperature of
the three major components: Mg-wüstite, Mg – Si – perovskite and Ca – Si – perovskite.
The melting curve of magnesiowüstite is seen to follow a Lindemann melting law
(see Section 5.12.3.1). They obtained a solidus temperature at the CMB (135 GPa) of about
4,300 K. Indeed, olivine is seen to break down into a perovskite –magnesiowüstite
assemblage at a shock melting temperature of 4,300 ^ 270 K at 130 GPa (Holland and
Ahrens, 1997). From the melting temperature measurements on the Fe –O –S system
(4,000 ^ 200 K), the solidus temperature of about 4,300 K is deduced. This occurs above
the outer core at the CMB.

2.12.3. Fe, Si enrichment in the lower mantle

An accurate measurement of the thermoelastic properties of silicate perovskite can

help assess whether the lower mantle is relatively enriched in Fe and Si with respect to the
upper mantle
Depending on the atomic ratio of Mg/(Mg þ Fe), the (Mg,Fe)SiO3 pyroxene
transforms to (Mg,Fe)SiO3-perovskite or to a three-phase assemblage of (Mg,Fe)SiO3-
perovskite, (Mg,Fe)O Mg-wüstite and SiO2 – stishovite. Under pressure (26 GPa/1,6008C),
however, a solubility of FeO in (Mg,Fe)SiO3-perovskite to a maximum of 11 mol% has
been reported (e.g., Ito and Takahashi, 1989). Perovskite co-existing with magnesiowüstite
and stishovite yielded the maximum solubility of FeO in perovskite; this increases with
increasing temperature (Fig. 2.21). The best fit of the experimental data to a linear function
yields (Fei, 1996):

XFe ¼ 20:0394 þ 9:14 £ 1025 Tð8CÞ:

Figure 2.21 also shows that (Mg0.8Fe0.2)SiO3 pyroxene would transform to the
three-phase assemblage at any temperature below melting. However, a large temperature
Petro-Tectonic Features of Terrestrial Planets 259

gradient in the experimental set up, driving the chemical diffusion, may produce
inhomogeneity in the sample.
If the lower mantle is perovskitic, where the (Mg,Fe)/Si ratio is 1.0, for a given
geotherm, the upper bound of the FeO content is defined by the maximum FeO solubility in
perovskite. Fig. 2.22 illustrates the maximum Fe/(Fe þ Mg) ratio along the geotherm of
Stacey (1992) at the top of the lower mantle. If the temperature at the top of the lower
mantle is less than 2,000 K, the maximum FeO content in the lower mantle is , 12%.
Because Fe strongly partitions into Mg-wüstite in a system of co-existing
perovskite and magnesiowüstite, the maximum Fe/(Fe þ Mg) ratio in a perovskite –Mg-
wüstite lower mantle (where (Mg þ Fe)/Si . 1.0) could be much higher than the
maximum solubility in perovskite. For stishovite to exist in a perovskititic lower mantle,
either the (Mg,Fe)/Si ratio is less than 1.0 or the Fe/(Fe þ Mg) ratio in perovskite is greater
than the maximum solubility of FeO in it.
Oxygen fugacities of peridotites and mantle-derived basalt are too high to allow
stabilization of iron metal in the mantle.

2.12.3.1. Effects of Fe
Fe composition affects the bulk modulus ðKT Þ of FexO (McCammon and Liu, 1984)
The KT values are estimated to range between , 150 and 154 GPa for x ¼ 0:90 – 0:95: The
maximum FeO content in the perovskitic lower mantle along the geotherm (Stacey, 1992)
is shown in Fig. 2.22.
The Neel temperature (TN) of FexO decreases linearly as xincreases from 0.90 to
0.95 but, from 0.95 to 0.98, there is a large discontinuity and TN increases with x
(McCammon, 1992). McCammon (1993) concluded that pressure reduces the Fe3þ in
FexO wüstite equilibrated
P with Fe.
The Fe3þ= Fe ratio decreases with increasing P pressure to at least 18 GPa. The g–
1-phase transition in Fe would favour a larger Fe3þ= Fe ratio at higher pressures and
lower temperatures but the transition boundary crosses the Earth’s geotherm only in the
deep mantle. P
In (Fe,Mg)O in equilibrium with metallic Fe, the maximum Fe3þ= Fe is seen to be
0.05 (McCammon,
P 1993). The electrical conductivity of (Fe,Mg)O increases with the
Fe3þ= Fe ratio and observation suggests that the lower-mantle conductivity is dominated
by (Fe,Mg)O (Wood and Nell, 1991; Li and Jeanloz, 1991). For further discussion on
conductivity, see Section 15.7.

Fe partitioning between perovskite and periclase. In the lower mantle Fe partitioning

may cause two distinct layers. The upper layer would consist of a phase mixture of Mg –
Si – perovskite and ferro-periclase (Mg, Fe)O with about equal iron partitioning between
these two phases, while the lower layer would consist of almost iron-free perovskite and
iron-rich ferro-periclase. At this pressure region iron in ferro-periclase undergoes
transformation from a high spin (HS) to low spin (LS) state. At such high pressure the
change in electronic properties of iron would make it behave such as to make perovskite
completely depleted in iron, while periclase will host the “drained-out” iron causing its
enrichment by several fold (Badro et al., 2003). This stratification may have profound
260 Chapter 2

implications on the transport properties of the lower mantle. The boundary is supposed to
be located between 1,700 and 2,300 km depths. Between the two phases the free enthalpy
change due to Mg –Fe substitution reaction can be evaluated from the calculation of
thermodynamic equilibrium constant as:
LnðKÞ ¼ lnðFe=MgÞmw 2 lnðFe=MgÞpv ¼ 2ðDG=NkB TÞ

As ferro-periclase undergoes spin transition, it is enriched in Fe, which results in its LS

transition point shift to higher pressures at lower depths. LS ferro-periclase is a much better
radiative thermal conductor than HS ferro-periclase (Sherman, 1991) because it is
transported to thermal radiation in the near IR. Dominant iron-depleted perovskite could
create an electrically, thermally, and rheologically insulating lid above the CMB.

2.12.3.2. P- and S-velocities and shear modulus

On the seismic velocities, the effect of Fe may be large compared with the other
secondary elements such as Ca, Al, Na and others but Fe has only a negligible effect on the
shear modulus. The effect of Fe on shear modulus in orthopyroxene is calculated from:
GðXFe Þ ¼ Gð0Þð1 2 0:31XFe Þ
The inclusion of Fe may reduce the G value of perovskite in a manner similar to the
reduction of G in magnesiowüstite (e.g, Wang and Weidner, 1996).
For Mg, Fe– silicate perovskite and magnesiowüstite, the effect of iron is the same
and is calculated from
GðXFe Þ ¼ Gð0Þð1 2 0:59XFe Þ
where in all cases XFe ¼ Fe/(Fe þ Mg).

2.13. Core –mantle boundary

The region near the CMB must have played a significant role in both the core and
mantle dynamic systems throughout their subsequent 4.5 Gyr of evolution. Heat must be
flowing from the core into the mantle in order to sustain the geodynamo, which is the core
magneto-hydrodynamic flow regime producing the Earth’s magnetic field.
The CMB exhibits the largest contrast in both seismic wave speeds and density in
Earth’s interior. A sharp discontinuity in wave speed is noted with substantial variations
in topography located several hundred kilometres (, 250 km) above the CMB
(see Fig. 2.23). This implies the existence of a chemical or phase boundary. This may be
accompanied by strong lateral and radial variations in temperature and composition. The
strongest anomalies at CMB extend some 500–1,000 km into the mantle. Thermal
differences may be the principal causes of anomalies, aided by compositional heterogeneity.
The CMB is possibly the major discontinuity in the Earth, with a density contrast of
4.49 cm23 between the core and the mantle (compared with 2.4 g cm23 between the crust
and the atmosphere) and a viscosity contrast of the order of 1021 P. It is the seat of
Petro-Tectonic Features of Terrestrial Planets 261

Figure 2.23. The fossil slab above the CMB may get complexly folded to generate discontinuities in the seismic
wave velocities.

energetic exchanges and coupling between core and mantle that may be of extreme
geodynamic importance. Brown and McQueen (1986) estimated an ICB temperature of
5,800 (^ 500) K on the basis of Hugoniot temperatures determined by thermodynamic
integration. The CMB temperature was noted as 4,000 (^ 500) K. Ahrens et al. (1998) also
estimated the core-side temperature at CMB to be 3,930 (^ 630) K. It is possible that the
increased T within D00 could drive the deformation mechanism of the deep mantle from a
super-elastic regime (Karato et al., 1995) into a regime of power-law creep. The cold core of
fossil slab of basaltic composition may be folded up to generate complicated seismic
response (Fig.2.23).
The CMB also shows lateral inhomogeneity in velocity, which is explained as
uneven distribution of dregs of dense materials. Correlation between the surface
topography and the geoid shows that the amplitude of the topography, dynamically
maintained by convection at the CMB, should be about 3 km (Hager et al., 1985).
The core is not in equilibrium with the mantle and, therefore, the CMB must be
chemically active and heterogeneous.

2.13.1. Minerals at CMB

The potential mineral constituents of the CMB region are listed in Table 2.8 below
with their structure, stability range and the techniques employed for determining these.
Siderophile (“metal-loving”) elements are supposed to have segregated with iron
metal to the core of the Earth. Equilibrium between metallic and silicate liquid at high
temperature (3,000 – 4,000 K) was thought to be responsible for the siderophile element
partition in the Earth’s mantle (Murthy, 1991) but the metal – silicate partition coefficients
would either increase or decrease with temperature, depending on the valence state and
thermodynamic properties such as enthalpy and entropy of formation of the oxide species
(Jones et al., 1992).
262 Chapter 2

TABLE 2.8
Minerals at CMB

Material Structure Maximum observed stability range Technique

(Mg,Fe)SiO3 Orthorhombic perovskite 127– 135 GPa at .2,000 K Laser-heated diamond cella
CaSiO3 Cubic perovskite 135 GPa T unknown Laser-heated diamond cellb
SiO2 CaCl2-structured 124 GPa at .1,273 K Laser-heated diamond cellc
MgO NaCl-structured to 227 GPa at 300 K Diamond celld
Mg0.6Fe0.4O NaCl-structured to 200 GPa at 23,700 K Shock wavee
FeO heated NiAs-structured 100 GPa at 900 K Externally Diamond cellf
FeSi B20-structured 49 GPa at 1,500 K Laser-heated diamond cellg

(a) Knittle and Jeanloz (1987), Kesson et al. (1998); (b) Kesson et al. (1998), Mai et al. (1989); (c) (Calcutta)
Tsuchida and Yagi (1989); (d) Duffy et al. (1995); (e) based on no observations of volumetric discontinuities on
shock compression: Vassiliou nad Ahrens (1982), temperature estimate from Svendsen and Ahrens (1987); (f) Fei
and Mao (1994); (g) Knittle and Williams (1995).

2.13.2. Hotspots and CMB

Oceanic islands commonly form linear chains with progressively increasing age
away from the active region, reflecting nearly stationary “hotspots” beneath the moving
lithospheric plate (discussed earlier in Section 2.7.1). The roots of the Iceland and
Hawaiian hotspots have been identified at the CMB by the seismologists (e.g., Helmberger
et al., 1998; Russell et al., 1998). These hotspots are long-lived, erupt large volumes of
magma and have distinctive geochemical signatures, in particular a high ratio of 3He/4He,
which is interpreted as 3He emanating from an undegassed region of the mantle.
Seismologists have constructed images of a mantle plume conduit in the upper
mantle (Wolfe et al., 1997) and transition zone (410 –660 km depth) (Shen et al., 1998). An
ULVZ, detected at the CMB beneath Iceland, reflects a hot, partially molten source of the
Iceland mantle plume (see Sections 2.9.1 and 2.10.4.1). Such a zone is defined as a
5 –40 km thick layer at the base of the mantle with velocity reductions of 10% for P-waves
and 30% for S-waves. This is observed from the delay and amplification of the core phase
known as SKPdS with respect to the core-phase SKS. These velocity reductions are
consistent with the hypothesis of partial melt. At the CMB below southeast of Hawaii,
observe that the anisotropy of ScS phases indicates a change from horizontal to vertical
fabric, which they attribute to plume upwelling (Russell et al., 1998)
Chemically, the hotspot volcanic lavas are distinct from those from oceanic ridges.
This difference arises because the hotspot magmas are contributed by the melting of the
subducted crustal plates at the convergent plate boundaries. The crust shows much
higher abundances of samarium and its daughter neodymium. Consequently, the hotspot
basalts contributed by molten slabs show higher contents of Sm and Nd compared
with MORB.
Following the isotopic evidence in Section 1.23.1, it is possible that hotspot
volcanoes reflect the direct transfer of heat from core to surface (Brandon et al., 1998).
The ultimate cause of mid-plate “hotspot” volcanoes like Hawaii is narrow plumes of hot
Petro-Tectonic Features of Terrestrial Planets 263

mantle rising from the boundary between core and mantle but the Hawaiian volcano seems
to contain a small component of material derived directly from the Earth’s metallic core.

2.13.3. Anisotropic structures at CMB

The discovery in the early 1980s of a deep-mantle shear-wave reflector several

hundred kilometres above the CMB introduced the analysis of reflections and refractions.
Core-grazing and core-reflected S-waves can also be analyzed for any shear-wave splitting
associated with anisotropic structure near the base of the mantle. PKP phases and their
scattered precursors also provide critical information about small-scale structure in D00 .
With an ambient temperature averaging around 3,000 K, the boundary layer at the
base of the mantle should have a temperature contrast of 1,000 ^ 500 K. This hot thermal
boundary layer has to be a likely source of boundary-layer instabilities and the thermal
plumes are presumed to ascend from the CMB to produce hot spots (like at Hawaii and
Iceland), transporting ,10– 15% of the surface heat flux (Sleep, 1990). Downwelling of
subducted slabs of oceanic lithosphere to this zone results in unique hotspot chemistry
(Woodhead et al., 1993). Plume-like upwellings rising from the hot CMB thermal
boundary layer generate hot-spot volcanism at the surface and subduction-related cold
downwellings produce locally seismologically fast (and geochemically anomalous) zones
at the CMB. The interactions between hot upwellings, cold downwellings and the CMB
are, however, very complex. Improvised seismic tomography reveals deep-mantle elastic
velocity heterogeneity with large scale (. 2,000 km) patterns, which are coherent for high-
and low-velocity regions at all depths (Masters et al., 1996).
The mechanism of coupling the core and the mantle influences the core flow and is
responsible for the observed irregularities in the rotation of the planet (Loper and Lay, 1995).
A continuing chemical reaction across the CMB may cause electromagnetic coupling.
Studies have revealed the presence of intermittent stratification of the D00 region in many
areas, with 1.5– 3% depths of 150 – 300 km above the CMB (Lay, 1995). In localized
patches, the discontinuity has either 100 –300 km wavelength topography of about 50 km
height, corresponding to acute lateral gradients in velocity structures noted from the
fluctuations in amplitudes and travel times of the reflected phases (Lay et al., 1997).

2.13.3.1. Seismic anisotropy in D 00 layer

Seismic tomography depicts that the base of the lower mantle has the largest lateral
variations in velocities This is the definitive evidence for a D00 discontinuity lying 250–
300 km above the CMB.
Seismologically determined anisotropism in the lowermost mantle is usually
explained in terms of partial melting, chemical inhomogeneities and/or solid – solid
mineral phase transitions (Williams and Garnero, 1996).
The D00 shell at the base of the Earth’s mantle is thought to be a thermal and
compositional boundary layer where vigorous dynamical processes are taking place (Loper
and Lay, 1995). An important property of D00 is its seismic anisotropy, manifested by
different velocities of shear waves, polarized vertically or horizontally on diffraction from
the CMB (Lay and Young, 1991). These lateral variations of anisotropy can be mistaken
264 Chapter 2

for lateral velocity variations. Notwithstanding this lateral variation in the degree of
anisotropy, there are large regions which appear to be isotropic.
If the lowermost mantle is considered as finely stratified and homogeneous, it will
behave to long waves as transversely isotropic. In such a medium, the diffracted
horizontally polarized S-waves (SHdiff) and the vertically polarized S-waves (SVdiff) will
propagate independently, but the former is faster than the latter (Vinnik et al., 1989).
For normal distribution of random variations of S-wave velocity in the stack of thin
horizontal layers with mean m and standard deviation s, the SH/SV velocity ratio for
horizontal propagation is expressed as:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
SH=SV ¼ 1 þ 4ðs=mÞ2

For S-wave anisotropy of 15%, s is thus evaluated as 28% of m.

The separation in SH and SV travel times (Dt) for rays which turn above D00 can be
explained by upper-mantle anisotropy. Those travel times for rays which turn within D00
show greater values of Dt, thereby constraining most of the lower-mantle anisotropy in the
D00 region. The anisotropic region extends from the CMB to the discontinuity 250 km
above the CMB.
Signals from the D00 layer exhibit significant shear-wave splitting, implying that the
anisotropy is vertically distributed throughout most of the D00 region, which, however,
show variability in thickness. Birefringence splits the S-waves into orthogonal
polarizations that travel with different velocities owing to organized mineralogical or
petrographical fabrics in the D00 region, contributing to the discontinuity structure (e.g.,
Lay et al., 1998, AGU). The shear-wave splitting observed beneath the circum-Pacific
regions is such that horizontally polarized SH-waves arrive ahead of vertically polarized
SV-waves. This observation is consistent with transmission through a D00 layer, in which
this effect could be produced by hexagonally symmetric crystals with the symmetry axis
oriented vertically or by stacks of horizontal thin layers with strong velocity contrasts
(e.g., Kendall and Silver, 1996).
Kendall and Silver (1998) noted that regions below the Caribbean and the
N.Pacific / Alaska show clear evidence for D00 (or CMB) anisotropy. The constraints in
the region are: (1) S / ScSH-phases arrives earlier than S /ScSV-phases, (2) SKS does not
appear to be sensetive to the D" anisotropy.
Fig. 2.24 shows a clear separation of >5 sec. between the radial and transverse
component are observed of the shear phases which turn within the CMB (D00 ) layer.
S-wave velocity can be lowered significantly by partial melting and/or by
accumulation of crystalline iron-alloy, the product of chemical reactions between iron of
the outer core and perovskite (Jeanloz, 1993).
Both downwellings (e.g., subduction of slabs) and upwellings (e.g., plumes) would
give rise to strong shear flows near the CMB that may produce anisotropy in D00 , in great
contrast to the isotropic character of the overlying lower mantle. In this region, structural
laminations, possibly contributing to shear anisotropy, are caused by (a) the downloading
of subducted oceanic crust (possibly partly molten), (b) core –mantle reaction products
swept up by mantle flow or (c) oriented inclusions of partial melt (e.g., Lay et al., 1998).
Petro-Tectonic Features of Terrestrial Planets 265

Figure 2.24. A clear seperation of >5 sec. between the radial and transverse component are observed of the
shear phases which turn within the CMB layer (from Kendall and Silver, 1998).

Oriented melt inclusions are particularly efficient in generating the anisotropy in the
lowermost mantle.
Relatively weak variations of P-wave velocity would support partial melting. If
the layered structure in D00 is generated by convective mixing and shearing, then the
strong anisotropy in D00 should be accompanied by strong wave scattering. One
anomalous region of D00 , when projected onto the surface of the Earth, closes near to
Polynesia, which is seen to be a region of unusual thermal agitation. Here, a large-scale
thermochemical plume may be present at the top of the lower mantle (Vinnik
et al., 1997). The anisotropy in D00 could somehow be related to the same plume
(Vinnik et al., 1998).
Summarizing the observations above, two hypotheses may be constructed to
explain the anisotropy of the layered D00 region:
(a) In the reaction-zone hypothesis (Knittle and Jeanloz, 1991), D00 is thought to contain
silicate and iron-alloy products of a reaction between the mantle and core. As iron
alloys have shear velocities 30 –40% lower than silicates (Jeanloz, 1990), this model
can explain the anisotropy assuming a 5% iron-alloy concentration segregates as
elongated inclusions throughout the 250 km thickness of the D00 layer. Introduction of
iron into D00 can reduce the shear velocities.
(b) The other hypothesis is derived from seismological data suggesting that slabs do
descend to the bottom part of the lower mantle and may even be folded to rest
horizontally on the CMB (Weber, 1994). The harzburgitic and basaltic slab (now
transformed to high-pressure phases) with thermal anomaly will give rise to an
anisotropy. The low melting Fe-, Al- and Ca-bearing phases in the basaltic fraction
of the slab will significantly reduce the shear modulus and lower the seismic velocity.
266 Chapter 2

2.13.3.2. Anisotropy caused by paleo-slabs

There seem a number of reasons to suspect that the cause of the anisotropy is
associated with subducting slabs descending to the CMB region. Tomographic images and
geodynamic modelling suggest that slabs descend to the lowermost mantle and remain
there as paleo-slabs. The high seismic velocities arise possibly from the thermal anomalies
in the slab, what was formerly oceanic crust and oceanic-mantle lithosphere (Kendall and
Silver, 1998). The contrasting velocities between the former-basalt and former-peridotite
(oceanic-mantle lithosphere) give rise to an extrinsic anisotropy in the D00 layer, which is
evidenced in the central Pacific, the Caribbean and North Pacific/Alaskan regions.
The lower-mantle tomography shows evidence for vertical, slab-like features
extending down to the CMB (named “Caribbean anomaly”), ascribed as subduction of the
former Farallon plate (Grand, 1994). If CMB is indeed the graveyard of subduction slabs,
then the slabs are seen to play major roles in the two zones — the crust and the CMB.
If the slabs can thus influence the thermal, compositional and seismological
properties of the CMB, then they might also play a significant role in the mantle’s two major
thermal-boundary layers. The subducted slabs, which are chemically stratified, may
maintain a thermally cold signature to depths of the CMB. However, it is still not clear how
the thermally cold core (with perhaps a basaltic top) of subducted slabs, with initial thickness
near 150 km, might be placed virtually horizontally at , 250 km above the CMB.
At elevated pressure, the collapse of the spin moments of some metal ions bear
significant implications in the elastic behaviour and the EOS of the lower-mantle materials,
particularly in the D00 region.

2.13.3.3. Carribbean and Pacific evidence

At the base of the mantle, such as below the SW Pacific Ocean, the vertically
polarized waves are seen to be slower than horizontally polarized waves. This anisotropy
may be the result of the mixing and shearing of strongly heterogeneous material in the
boundary layer. The diffracted P-wave velocities, Pdiff, beneath the SW Pacific is
anomalously slow (Garnero and Helmberger, 1995; Vinnik et al., 1998). This may be due
to a partial melting (Williams and Garnero, 1996). Again, the low horizontal P-wave
velocity in a fine-layered horizontally stratified medium becomes feasible if the variations
of the S-wave velocity between the layers are much stronger than those of the P-wave
velocity. This may, however, also arise due to LPO (Karato, 1998).
The D00 zone anisotropism is also evidenced in the region below the Caribbean and
Central America. This D00 anisotropy has been observed from anomalous high-amplitude
diffracted SV phases (shear waves polarized in the source– receiver plane and diffracted
along the CMB). This evidence of shear-wave splitting is diagnostic of seismic anisotropy.
For the anisotropy, two possible mineralogical causes are invoked: (1) LPO and (2) shape-
preferred orientation (SPO) of constituent minerals. Some argue that the anisotropy is
associated with the fragmented subducted material in the D00 zone.
Interestingly, the D00 beneath the Caribbean is transversely isotropic and the rays
with near-vertical paths through D00 , such as SKS, show no splitting (Meade et al., 1995).
There is a discontinuous increase in SH-wave velocity (VSH) at the top of D00 in the
Caribbean region. The velocity anisotropy, however, is not due to the LPO of the
Petro-Tectonic Features of Terrestrial Planets 267

lower-mantle minerals, especially of the prevalent perovskites, which fail to develop

strain-induced LPO. Some models give rise to long-wavelength transverse isotropy with a
vertical symmetry axis and exhibit VSH . VSV for horizontally propagating waves (Sayers,
1992).
Wave forms displayed at many Canadian stations and North Pacific/Alaska exhibit
significant amounts of SKS splitting, indicating receiver-side upper-mantle anisotropy (see
Fig. 2.24). From SKS splitting, wide ranges of azimuth estimates are made of the upper-
mantle anisotropy. The Canadian National Seismograph Network (CNSN) has published
some measurements on the upper-mantle anisotropy. The primary seismic constraints in
the Alaska and Carribbean regions are: (i) S/ScSH-phases arrive earlier than S/ScSV-
phases, (ii) SKS does not appear to be sensitive to D00 anisotropy and (iii) these are regions
of high seismic velocities and a D00 discontinuity. An example of clear S/ScS splitting is
seen beneath Argentina (Kendall and Silver, 1998). The anisotropy may arise from LPO of
constituent minerals or from SPO of inclusions, which are horizontally placed tabular
bodies. The anisotropy and the thermal anomaly may contribute to the high aggregate
shear velocities.

2.14. Reaction between mantle and liquid-core

Oxides and molten iron are seen to react at high pressure and temperature,
suggesting FeO is extracted from the oxide. This mechanism provides light elements (e.g.,
oxygen) to be mixed in the outer core.
This reaction mechanism may also cause in the D00 zone an enrichment of FeO
phase, which is possibly anti-B8 or B8 – anti-B8 polytypes (Wentzcovitch et al., 1998).
Such phases would show high metallic conductivity with strong anisotropic character. This
region would thus control the magnetic field to be propagated from the core.
Structural transitions such as transition to anti-B8 or B8 in D00 could also change
phase relations and contribute to the heterogeneity in this region (Jeanloz and
Romanowicz, 1997).
At high pressures (20 – 30 GPa), chemical reactions between silicates and iron
alloys occur in the form:

ðMgx Fe12x ÞSiO3 þ 3½ð1 2 xÞ 2 sFe ¼ xMgSiO3 þ sSiO2

þ ½ð1 2 xÞ 2 sFeSi þ ½3ð1 2 xÞ 2 2sFeO ð2-3Þ

where 1-x is the amount of Fe in reacting mantle material and s is the amount of silicate
generated by the reaction (Knittle and Jeanloz, 1991). If FeSi and FeO are segregated from
their complementary MgSiO3 and SiO2, a dense, highly conductive iron-alloy layer could
be generated (Manga and Jeanloz, 1996). The reaction is greatly enhanced when it occurs
between a partially molten layer at the base of the mantle and the outer core. Any upward
flow within such a layer would transfer iron from the outer core to the lowermost mantle,
causing a reduction in the stratification.
268 Chapter 2

On the basis of exploratory experiments, Knittle and Jeanloz (1991) suggested a

reaction between the liquid iron of the Earth’s outer core and crystalline silicates (Mg, Fe,
SiO3) of the mantle:
Mg0:9 Fe0:1 SiO3 þ 0:15Fe ¼ 0:9MgSiO3 þ 0:2FeO þ 0:05FeSi þ 0:05SiO2 ð2-4Þ
ðpvÞ ð1-phaseÞ ðpvÞ ðhppÞ ð1Þ ðstÞ

The 1 iron is presumed to be stable at CMB (Brown and McQueen, 1982). The free energy
of the above reaction is relatively insensitive to the thermal expansion model.
The FeO at CMB presumably occurs in a high-pressure phase (Jeanloz and Ahrens,
1980). For this high-pressure phase of FeO, the heat capacity, Cp, bears the following
relation, derived from the Dulong– Petit value

Cp ¼ 3nR þ a2 VTKT

Here, n is the number of atoms per formula unit, a the thermal expansion coefficient, V the
molar volume and KT is the bulk modulus. The product aK is assumed to be constant and is
independent of temperature (Birch, 1952).
At CMB condition when SiO2 reacts with 1-iron, the following are produced:
SiO2 þ 3Fe ¼ 2FeO þ FeSi
ðstÞ ð1Þ ðhppÞ

These reactions also indicate that iron will react with lower-mantle silicates to form
Fe –Si alloy in order to minimize the Gibbs free energy in such a reaction as:
MgSiO3 þ 3Fe ¼ MgO þ FeSi þ FeO
ðpvÞ ð1Þ ðmwÞ ðhppÞ

However, the free energy of formation in this reaction has a very positive value, and
hence this reaction is not likely to take place in terrestrial planets at very high temperatures
(. 4,500 K) and pressures (. 300 GPa).
A modification of the reaction by Song and Ahrens (1994) leads to:
Mg0:9 Fe0:1 SiO3 þ 0:3Fe ¼ 0:9MgSiO3 þ 0:3FeO þ 0:1FeSi ð2-5Þ
ðpvÞ ð1Þ ðpvÞ ðhppÞ

In terms of Gibbs free energy, the above reaction occurs at higher pressures than the
reaction originally proposed by Knittle and Jeanloz (1991).
Amongst other possible reactions, the reaction
Mg2 SiO4 þ ð3 2 3xÞFe ¼ xMgSiO3 þ ð2 2 xÞMgO þ ð1 2 xÞFeSi þ ð2 2 2xÞFeO
ðgÞ ð1Þ ðpvÞ
ð2-6Þ
can proceed at much lower pressure and temperature than prevalent at present CMB. But
perhaps this could have occurred in the earlier, smaller proto-Earth or in the present Venus.
If the high-pressure phase of FeO can co-exist stably with perovskite and stishovite,
the reaction between iron and silicates can occur at much lower temperatures and pressures.
This could have occurred during the very early period of Earth’s accretion stage and possibly
Petro-Tectonic Features of Terrestrial Planets 269

still occurs in the centre of the present Mars and Venus. This may well explain the
compositional difference in iron content between the Martian mantle and the Earth’s mantle.
It seems that core iron does not tend to react with pure iron-free polymorphs of
Mg2SiO4 and MgSiO3. Possibly, iron is drained via reaction from the lowermost mantle
into the core. This would have been a mechanism for core formation and core – mantle
segregation.
The thermodynamic calculations (Song and Ahrens, 1994) revealed that the
reaction (2.4) can occur at 130 GPa (< 135 GPa at CMB) and at as low a temperature as
900 K. The basis of the thermodynamics employed is that, as pressure increases along the
isotherm, the difference in free energies between the reactants and the products increases
and the products become more and more stable relative to the reactants. Also, as
temperature increases, the transition pressure decreases. The calculation by Song and
Ahren (1994) shows that the transition pressure for the post-spinel phase at 1,600 K is
23.7 GPa, which is closely in agreement with 23.1 GPa, determined by experiment (Ito and
Takahashi, 1989).

2.14.1. Ultra-low-velocity zone

In this zone, the compressional and shear-wave seismic velocity are reduced by
almost 10 –20 and 10– 50%, respectively. A relation between ULVZ and volcanic hot
spots has been proposed (Williams and Garnero, 1996).
The ULVZ is believed to be , 5 –50 km thick on the mantle side and its bottom
manifests signatures of infiltration of liquid outer core into it, which may have induced
partial melting of the lowermost mantle. Ultra-low velocities at the base of the mantle
suggest higher than average temperatures (or the presence of fluxing components).
A newly discovered laterally varying ultra-low-velocity layer in the lowermost
section (5 – 40 km) of the lower mantle (e.g., Garnero and Helmberger, 1995) exhibits a
velocity drop (9 – 10% or more of P and S velocities). This may arise from partial melting
along with attendant chemical reactions, heat transport and boundary-layer dynamics.
Extensive regions of an ULVZ are found beneath the central Pacific Ocean, Alaska,
Iceland and Africa. In the central Pacific, the ULVZ shows significant lateral variations in
thickness and/or depth (e.g., Garnero and Helmberger, 1996). Regions like those beneath
Alaska and the Caribbean have relatively high shear-wave velocities in D00 , with a 2 –3%
SH velocity increase near 200 –280 km above the CMB.
Recent seismic studies (Vidale and Hedin, 1998) have suggested the existence of
ULVZ beneath Iceland and the western Pacific region. Short- and long-period precursors
of the PKP phase have been used to study an ULVZ near the CMB beneath the western
Pacific. Synthetic seismic methods computed from a hybrid method handles seismic-wave
propagation through two-dimensional complex structures. Some phases arriving before the
PKP phases in synthetic seismograms of 1D models (such as PREM; Dziewonski and
Anderson, 1981) are essentially caused by diffraction of the rays into the fluid-core shadow
zone, but they manifest very long-period character.
Long-period precursors are explained by Gaussian-shaped ULVS of 60 –80 km
height with P velocity drops of at least 7% over 100 –300 km. Short-period precursors
270 Chapter 2

suggest the presence of smaller-scale anomalies accompanying these larger Gaussian-

shaped structures. These fine structures may be areas of partial melt caused by vigorous
convection or the instability of a thermal boundary layer at the mantle base or both
(Wen and Helmberger, 1998).
In ULVZ, the shear-wave velocity may decrease up to 3 times the decrease in
P-wave velocity (Revenaugh and Meyer, 1997). The acute velocity heterogeneity of
the ULVZ is a likely source of strong scattering of short-period phases that have long been
associated with short-wavelength heterogeneity in D00 (Vidale and Hedlin, 1998).
The spatial correlation of the ULVZ with hotspots indicates that deep-seated partial
melt may enhance the ascent of a plume to the surface. The zone from which plumes
extract material is likely to be both highly inviscid and chemically different from the
overlying mantle. Plumes appear to have a causal relationship with continental extension
and breakup at the surface of the Earth (e.g., Storey, 1995).

2.15. The Earth’s magnetism and orbital obliquity

The Earth’s main geomagnetic field stems from electric currents in the liquid outer
core. These currents derive from convective flow, with a typical overturn time of about 500
years. The magnetic field of the solid inner core can change only by diffusion of the field,
which has a time scale of , 3,000 years. This inner core thus stabilizes the geomagnetic
field. The influence of the inner core delays full reversal of the field, during which time the
original polarity may establish itself in the outer core.
There are periodic variations in the eccentricity of the Earth’s orbit (100,000-year
period) and in its obliquity (41,000-year period). The Earth’s obliquity influences
geomagnetic field intensity, with a period around 41,000 years. For further discussion on
the properties of the Earth’s core, readers are advised to see Section 14.2.8.

2.15.1. Mantle plume and geomagnetic reversals

The mantle-plume activity controls magnetic-reversal frequency. For the last

150 m.y., plume activity correlates inversely with magnetic-reversal frequency. A pulse in
plume activity during mid-Cretaceous time caused the nearly synchronous magnetic
superchron, a period of normal polarity lasting 41 m.y., when the Earth’s magnetic field
did not reverse. Thus, the events just above the CMB control processes in the outer core
where the Earth’s magnetic field is generated.
As the plumes rise from the unstable base of the thermal-boundary layer, an area of
abrupt temperature increase called D00 (above CMB), the thermal boundary layer is thinned
to fuel the plumes. This causes the vertical temperature gradient across the boundary layer
to increase, allowing more heat to be conducted out of the outer core. In turn, more
vigorous convection occurs inside the outer core in response to the abnormal heat loss. An
increase in convective activity would produce more instabilities in the outer-core fluid,
which would lead to more magnetic reversals (Larson, 1993).
Petro-Tectonic Features of Terrestrial Planets 271

2.16. Mars

Mariner-9 and Viking Orbiter images reveal that the Southern Hemisphere has
preserved the effects of the early ($ 3.7 b.y.) heavy bombardment of Mars and other bodies
of the inner Solar System. In contrast, the southern hemisphere is flat and less impacted,
implying younger in age (Solomon, 2002). This “crustal dichotomy” has also been
analyzed later, since 1997, by Mars Global Surveyor (MGS) and Mars Orbiter Laser
Altimeter Experiment (Smith et al., 2001). The preservation of the dichotomy implies that
the lower crust cooled rapidly after early crustal formation, perhaps as a result of deep
hydrothermal circulation (Permentier and Zuber, 2002).

2.16.1. Crust and mantle

The radius of Mars is about half that of Earth and so Mars probably heated up and
cooled off more quickly. Thus, the geological activities such as tectonism, volcanism and
escape of volatiles occurred earlier in Mars than in the Earth.
Currently, Mars is a single-plate planet with a thick, rigid outer shell but perhaps in
its early history the planet displayed thinner and even possibly mobile plates (Sleep, 1994).
The evolution of the crust and mantle of Mars has been better understood with the
aid of orbital global geophysical measurements of topography and magnetics, Earth-based
and orbital spectra, orbital and lander (Viking and Pathfinder landers) images of the surface
and geochemical measurements of Martian meteorites, such as shergottites, nakhlaites and
chassignites (SNC) (McSween, 1994).
The surface is composed of igneous rocks covered by oxidized weathering
products. The red colour of the cover is due to the pressure of Fe3þ-bearing minerals.
Organic compounds, contributed by meteorites, are commonly destroyed by the presence
of peroxide oxidants.
Based on the moment of inertia of Mars, geophysical arguments have been
developed. These are supported by cosmochemical evidence and the model for the
composition and structure of its mantle and core. Using the mass and radius of Mars and
the moment of inertia factor as constraints, the mantle density and core size can be
calculated for a given core composition. However, because of the uncertain distribution of
the non-hydrostatic mass about the Tharsis plateau region, the moment of inertia factor of
Mars remains uncertain.
Calculations show the Martian mantle to be denser than the Earth’s, possibly
because of the increased abundance of iron in the mantle of Mars.
For interpretation of the seismic discontinuities in the Martian interior, a profile of it
may be obtained by the proposed Intermarsnet mission, which would attempt to ascertain
the structure and seismic profile of Mars.
The bulk composition of Mars is chondritic (carbonaceous chondrite representing
primitive Solar System composition).
Certain of the basaltic achondritic meteorites obtained from Antarctica, the SNC
are believed to be the products of ejecta from Martian surface by meteorite impacts within
the past 1 –20 Myr. The trapped gases in these isotopically match the Martian atmosphere
272 Chapter 2

(Bogard et al., 1984) and, geochemically, the meteorites are essentially picritic and basaltic
lava. The meteorite ALH 84001, as evidence for past biological activity, has a
crystallization age of ,4.5 Gyr and probably represents an ancient Martian crust.
Most models of the bulk composition of Mars are derived from cosmochemical
argument and are constrained by geophysical information, including the planet’s mean
density and moment of inertia (Morgan and Anders, 1979), the latter constraining two
of the three variables: mantle density, core size and core density. Martian composition
is constrained by the chemistry of SNC meteorites. The elemental ratios in SNC and
the chondritic abundance predict an SNC parent-body mantle composition. Based on
these, Bertka and Fei (1997) derived the Martian bulk (MB) composition, as discussed
below.

2.16.2. SNC and LHB: ALH 84001

The Moon was differentiated and cooled more than 4.45 Gyr ago but extensive
bombardment continued until 3.8 Gyr ago. Large, multi-ringed impact basins formed on it
during the closure of the period, i.e., during 4.0 –3.8 b.y. ago. Dark and flat area basins
formed after this period reflects the end of the late heavy bombardment (LHB). Lunar
highlands were formed during the cataclysm. Projectiles from the Moon should have struck
the neighbouring planetary bodies like the Earth, Mars, Venus and Mercury.
A chance break up of a large comet or asteroid, perhaps by planetary tidal
disruption during a close encounter (as happened to comet Shoemaker-Levy 9 in 1992),
could also have caused LHB.
On the basis of textural and isotope data, the Antarctic meteorite ALH 84001 is
considered to belong to SNC classes of meteorites. Ash et al. (1996) report that it is
4.0 Gyr old. Ar – Ar isotope dating puts the creation of ALH 84001 at the same time
as the LHB. Ar – Ar ages for meteorites (Bogard, 1995), especially for basaltic
meteorites that are probably derived from the asteroid Vesta, and also for ordinary
chondrites, indicates that a LHB also occurred in the asteroid belt during the period
3.5 –4.1 Gyr ago.

2.16.3. Martian mantle composition

The upper Martian mantle is similar to Earth’s, composed primarily of olivine. At

depth, a transition zone appears consisting of a more densely packed spinel structure,
followed downward to the lower mantle by a still denser perovskite structure of the Mg – Fe
silicate [(Mg,Fe)SiO3). In the Earth, the olivine –spinel transition occurs at 400 km depth
while this occurs at , 1,000 km depth in Mars. In Mars, the CMB is presumed to exist at
1,300 –1,700 km upwards from the centre.
Until recently, only limited geophysical data were available to determine the
composition or structure of the Martian interior. Models for the composition were derived
form the Martian meteorites (SNC). Bertka and Fei (1999, Yb. 98– 99) determined the
high-pressure mineralogy of a one-meteorite-based model and used this information to
model the structure of the Martian interior (Fig. 2.25). They compared the available data
Petro-Tectonic Features of Terrestrial Planets 273

Figure 2.25. Model mineralogy (in weight percent) of the MB, Dreibus and Wänke (1985). Martian mantle
composition as a function of pressure (Fei et al., 1995; Bertka and Fei, 1997, q1997 American Geophysical Union).

with data on the mass distributions of the planet’s interior obtained from the Mars
Pathfinder mission. They observed that the notion that the terrestrial planets have
refractory element bulk compositions equivalent to a primitive C1 meteorite is not
tenable. They confirmed, however, that the mantle of Mars is richer in iron than the
Earth’s mantle.
Bertka has been conducting melting experiments and density determinations of
model core compositions. The types of melts that would be produced from partial melting
of an iron-rich hydrous source region are under investigation by Bertka (yb. 98/99).
The temperature profile of the Martian mantle is shown in Fig. 2.26, along with the
location of the CMB (Fei et al., 1995) and the silicate– perovskite transition boundary (Ito
and Takahashi, 1989) (see caption to Fig. 2.26).
MB used in experiment by Bertka and Fie (1997) has an iron-rich chondritic
composition similar to what Dreibus and Wänke (1985) estimated for it. Compared with
terrestrial spinel lherzolite (KLB) (Takahashi, 1986), it is more Fe- and Na-rich but it
shows a chondritic, not peridotitic, Mg/Si ratio (Table 2.9).
274 Chapter 2

Figure 2.26. Temperature profiles of the Martian mantle showing the locations of experimental data and the
location of the core– mantle boundary (Fei et al., 1995) and the silicate-perovskite transition boundary (Ito and
Takahashi, 1989) (dashed line). Solid circles denote experiments performed with the Dreibus and Wanke (1985)
mantle composition (referred to as MB); open circles denote experiments performed by Kamaya et al. (1993) with
the Morgan and Anders (1979) mantle composition. Areotherms shown are liquid core (line a), solid core (line b)
(Longhi et al., 1992), and liquid core estimated on the basis of new melting data in the system Fe–FeS (Fei and
Bertka, 1995) (line c) (from Bertka and Fei, 1997, q 1997 American Geophysical Union).

MB of Mars has chondritic ratios for Mg/Si and Al/Ca and magnesium number
Mg#75 (Mg# ¼ atomic[Mg/(Mg þ Fe2þ) £ 100]), with 0.5 wt% Na2O (Table.2.9). It
should be noted, however, that the Mg/Si ratio as seen in samples of Earth’s upper mantle
(and also in KLB) resulted from differentiation which did not operate on Mars.

TABLE 2.9
Bulk Mantle Compositions

Earth Mars
Takahashi Morgan and Anders Dreibus and Wänke Bertka and Fei
(1986) KLB (1979) (1985) (1997) MB

SiO2 44.48 42.1 44.40 43.68

TiO2 0.16 – 0.14 –
Al2O3 3.59 6.5 3.02 3.13
FeO 8.10 16.0 17.90 18.71
MnO 0.12 – 0.46 –
MgO 39.22 30.2 30.20 31.50
CaO 3.44 5.3 2.45 2.49
Na2O 0.30 – 0.50 0.50
P2O5 0.03 – 0.16 –
Cr2O3 0.31 – 0.76 –
Mg# 89.6 77.1 75.0 75.0
Petro-Tectonic Features of Terrestrial Planets 275

At 17 GPa pressure in the transition zone of MB mantle, the appearance of , 2 wt%

of magnesiowüstite is attributed to more Mg-rich compositions, as displayed by
pressure increase (Bertka and Fei, 1996). At transition zone pressure (14 – 24 GPa),
magnesiowüstite is found at temperatures , 508C below the solidus. However, for the
melting interval of KLB-1 (Zhang and Herzbery, 1994) and Allende meteorite, Agee et al.
(1995) reported the presence of magnesiowüstite in their experiment (at 14 GPa) at a
temperature , 1508C below the solidus of the Allende meteorite.
The large abundance of CaO and Al2O3 in the M –A (Morgan and Anders)
(Table 2.9) mantle composition may be responsible for the large modal abundance of
majorite and magnesiowüstite. The experimental work of Bertka and Fei (1997) at
. 23 GPa indicates that the MB Martian mantle assemblage (along a high-temperature
areotherm), consists of Mg – Fe silicate perovskite, magnesiowüstite and majorite
(Fig. 2.25). Significantly, CaSiO3-perovskite is indicated as absent.

2.16.3.1. Mantle geochemistry: SNC

The overall bulk composition of Mars is assumed to possess chrondritic (C1 type)
Fe/Si ratio and C1 abundance of all refractory elements (Longhi et al., 1992). From the
measured FeO/MnO ratio of the SNC meteorites, the iron content of the Martian mantle is
calculated. The FeO/MnO ratio seen in SNC meteorites is stipulated to be the same as in
the Martian mantle. The total MnO in the mantle should nearly account for the total MnO
of Mars. This is because MnO would not be present in the core and the crust is very much
thinner compared with the Martian mantle.
The distribution of Fe between the core and mantle of Mars is calculated from the
measured ratio of FeO/MnO in the SNC meteorite. This is interpreted as the
(FeOmantle)/(MnOplanet) ratio, whereas the FeO/MnO ratio in C1 is interpreted as
equivalent to (FeOplanet)/(MnOplanet).
As for FeO, the distribution of NiO and CoO between mantle and core is calculated.
Derived from a SNC meteorite model, the Martian core is considered sulfur-rich and all
sulfur is assumed to be concentrated entirely at the core. The core chemistry is essentially
Fe with S (14.2 wt%), Ni (7.6 wt%) and Co (0.4 wt%). The weak magnetic field of Mars
suggests that its core is either entirely solid or entirely liquid (Longhi et al., 1992) and there
is no thermal-boundary layer at the CMB. Presumably, the core formation was
contemporaneous with Mars accretion.
The Morgan and Anders (1979) cosmochemical model (see Table 2.9) scales CaO
and Al2O3 abundances to uranium abundance but this estimate of CaO and Al2O3 may be
on the high side because it was based on a thermal history model, which did not receive
support from later models dealing with Martian core formation (Schubert et al., 1992;
Bertka and Fei, 1997). These workers also calculated the volatile content of the Martian
mantle and found that to be depleted compared with that in the Earth.
However, the K/U ratio of the SNC meteorites is observed to be higher than that
measured from the Martian orbiter (Longhi et al., 1992). The Na2O concentration
predicted from the SNC meteorites stands as higher than that noted by Morgan and Anders
(1979). Moreover, most workers, on the basis of geophysical and geochemical
276 Chapter 2

Figure 2.27. Projection of MB and KLB bulk mantle compositions, and the MB 24 GPa mineral phases into the
system MgO–FeO–SiO2 mol%. Filled circle denotes MB bulk mantle, open circle denotes KLB bulk mantle
composition, open square denotes MB 24 GPa phases. Oxide mix was the staring material (Bertka and Fei, 1996b)
and solid circles denote MB phases (glass as starting material) (from Bertka and Fei, 1997, q 1997 American
Geophysical Union).

assumptions, predict a mantle composition of Mars that is more iron-rich than the Earth’s
mantle.
Bertka and Fei (1997) performed experiments with model MB composition
(Table 2.9, last column) and the results obtained (Fig. 2.27) suggest that the Martian upper
mantle consists of olivine, garnet, clinopyroxene and orthopyroxene up to the depth
pressure of 9 GPa. Above 9 GPa, orthopyroxene is absent and olivine, garnet (or majorite)
and clinopyroxene co-exist up to 13.5 GPa, where the appearance of g-spinel (ring-
woodite) marks the beginning of the Martian transition zone. The g-spinel co-exists with
olivine, clinopyroxene and majorite. At 14 GPa, g-spinel and olivine are replaced by b-
phase (wadsleyite). At 15 GPa, both g-spinel and b-phase are present. By 17 GPa, the
modal abundance of g-spinel increases at the expense of b-phase. Above 17 GPa, g-spinel
co-exists with majorite. The beginning of the lower mantle is marked by the transition of g-
spinel to Mg – Fe silicate perovskite plus magnesiowüstite. The results of Bertka and Fei
(1997) indicate that, given a high-temperature aerotherm, the Martian lower mantle
consists of Mg – Fe silicate perovskite, magnesiowüistite and majorite.

2.16.3.2. Mantle-phase stability: MB versus KLB

At depths, spinel to Mg –Fe silicate perovskite phase transition should mark the
transition to lower mantle. In Martian mantle, the intersection of the aerotherm with the
phase transition of spinel to Mg –Fe silicate perovskite should also correspond to the upper
boundary of the lower mantle. In the iron-rich Martian lower mantle, temperature should
play a significant role.
Bertka and Fei (1996) reported the stable lower-mantle assemblage to be Mg – Fe
silicate perovskite, majorite, magnesiowüstite, CaSiO3-perovskite and stishovite.
This is plotted (Fig. 2.27) in a MgO – FeO – SiO2 ternary as a function of temperatures
Petro-Tectonic Features of Terrestrial Planets 277

(1,200 and 1,7508C). In Fig. 2.28, the MB composition as well as a terrestrial spinel
lherzolite, KLB, representing Earth’s mantle composition, are plotted. In the plot, the
terrestrial KLB composition falls in the stability field of perovskite plus magnesiowüstite,
both outside the low- (1,2008C) and high-temperature (1,7508C) triangles. The MB
composition, however, falls within the triangle defined by the co-existing Mg – Fe silicate
perovskite, magnesiowüstite and stishovite.
In the Martian interior, if the temperature profile is low (i.e., ,1,7008C),
then stishovite may be stable in the Martian lower mantle, even though it is absent in the
Earth’s lower mantle. In the Earth’s mantle, CaSiO3-perovskite appears at 18 GPa
(Takahashi and Ito, 1987) but, in more Fe-rich composition (as in MB composition),
CaSiO3-perovskite forms at high pressure (e.g., 23.5 GPa). In the MB transition zone and
also in the lower mantle, majorite is the principal Ca-bearing phase. The CaO content
would increases with depth.
The Martian mantle structure is very much dependent on the thermal profile of the
Martian interior. With time since its origin Mars, is losing its internal heat and, because of
its high Fe-content (with respect to Earth’s), the Martian lower mantle is more sensitive to
temperature. From its early history of high interior temperature, Mars has progressed
to yield to a much thinner lower mantle, rather to a nearly vanishing stability field of
Mg-Fe silicate perovskite. However, the presence of a Martian lower mantle played
significantly in the evolution of plume volcanism and Martian mantle convection (Harder
and Christensen, 1996).

2.16.3.3. Fe-rich Martian mantle: density increase

As stated earlier, the composition of the Martian mantle is predicted to be iron-rich
compared with the Earth’s mantle (Dreibus and Wanke, 1985). Bertka and Fei (1996)
determined the mineralogy of Fe-rich Martian mantle composition up to 24 GPa. They
observed that the Martian transition zone is marked by the appearance of a small modal
abundance of g-spinel co-existing with olivine, Mg-wüstite, clinopyroxene and majorite at
3.5 GPa and , 1,7008C. The transformation of olivine to its high-pressure polymorphs
g-spinel and b-phase will produce a small density increase of , 2.4%. The appearance of
CaSiO3-perovskite at 21 GPa has only a minor effect on density because the modal
abundance of this phase is small. If the Martian mantle contains a lower mantle, both Mg-
wüstite and stishovite may be present in the lower mantle along with perovskite and
majorite. The lower-mantle transition will be marked by a density increase of about
7.5 wt%.
Bertka and Fei (1996) carried out experiments up to 24 GPa pressure on
anhydrous mullite components such as CaO – MgO –FeO –Al2O3 – SiO2 –Na2O system,
presenting a mantle composition derived from SNC meteorites. Magnesiowüstite’s
presence in the transition zone is possibly caused by the shift of olivine, b-phase,
g-spinel and clinopyroxene or majorite towards more Mg-rich composition at
higher pressure.
With high Fe-content in Mars, an assemblage of stishovite þ Mg-wüstite þ
majorite þ perovskite seems consistent. Mg-wüstite may be present in the Martian
278 Chapter 2

Figure 2.28. The terrestrial lherzolite, KLB, composition falls in the stability field of PVS+Mw, while MB
composition falls within the triangle defined by the coexisting Mg-Fe pvs, Mg-Wu and stishovite (see text) (from
Bertka and Fei,1997).
Petro-Tectonic Features of Terrestrial Planets 279

transition zone and Mg-wüstite plus stishovite in the Martian lower mantle. In contrast, in
the Earth’s mantle, Mg-wüstite is restricted to the lower mantle, where stishovite has to be
absent (Ito and Stixrude, 1993).

2.16.3.4. Olivine –spinel phase transition

The phase relations calculated for the system Mg2SiO4 –Fe2SiO4 at 1,7008C (Katsura
and Ito, 1989; Fei et al., 1991; Bertka and Fei, 1997) are presented in Fig. 2.27. Bertka and
Fei (1997) stated that, in the Martian mantle, which is Fe-rich, the transition zone is marked
by the appearance of a small modal abundance of g-spinel. In contrast, in the Earth’s mantle
(Mg# , 90), olivine first transforms over a small pressure interval (, ,0.5 GPa) to b-
phase, which is stable to , 4 GPa when the transformation to g-spinel begins. In the Martian
mantle, the transformation to b-phase begins after the initial appearance of g-spinel. b-
phase is then stable over a small pressure interval (, 1.5 GPa) before the next transformation
to g-spinel begins. The iron abundance in the Martian mantle results in the transition zone
beginning at 0.5 – 1 GPa less pressure in the Martian mantle compared with the Earth’s and
with the initial appearance of g-spinel rather than b-phase.
In general, between the two mantles of the Earth and Mars, both the variation in
olivine– spinel phase transitions and the difference in pressure – depth relations (because of
the smaller size of Mars) will result in a Martian transition zone covering a larger depth
range. It is also marked by smoother discontinuities in density or seismic profiles than
those observed in the Earth’s transition zone (Bertka and Fei, 1997).

2.16.3.5. Mantle-flow: viscosity

Hemisphere-scale mantle flow with long-wavelength pattern of heat loss, valid for
the Moon, does not show promise for a planet with a core (iron) as large as Mars.
Hemisphere-scale convection is possible if, like the Earth’s, the mantle of Mars is layered in
viscosity such that the upper mantle is at least 100 times less viscous than the lower mantle.
Such a layer could correspond to a Martian asthenosphere, as occurs on Earth beneath the
lithosphere. A long-wavelength convection pattern cools the core and mantle more
efficiently than shorter-wavelength convection and could have contributed to the demise of
the core dynamo in the early existence of Mars. This also involved rapid cooling of the crust.

2.16.3.6. Magmatic water in Mars

The surface morphology of Mars offers scope for speculation of an ancient ocean
but the chemical analyses of Martian meteorites of igneous origin offer evidence for low
magmatic water content. However, a study on pyroxene present in Shergotty meteorite —
a basaltic achondritic mass ejected 175 m.y. ago (Cretaceous period) from Mars —
revealed that water contents of pre-eruptive magma on Mars could have been up to 1.8%
(McSwee et al., 2001). But the presence of enriched soluble trace elements in the inner
cores of pyroxene (formed at depths) corresponding to the outer rims (formed near the
surface) implies that water could be present at depth but was largely lost in pyroxene as it
was carried upwards by magma ascent. The ascending magmas could possibly have
280 Chapter 2

delivered significant quantities of water to the Martian surface in late recent times to
produce the erosional features observed in the surface morphology.

2.16.4. Magnetism

On the Earth, the continental magnetic anomalies are caused primarily by the
magnetic induction by the Earth’s dipole field, whereas the oceanic crust displays magnetic
stripes dominated by remanence. Presently, Mars lacks an internally generated dipole
magnetic field and crustal anomalies most likely represent remanent magnetism from an
internal field in the distant past (Acuna et al., 1999). Stevenson (Nature, 412, pp. 214 –219)
presents evidence for a vigorous core dynamo that operated during Noachian (. 3.8 b.y.),
before the end of heavy bombardment.
The minerals responsible for remanent magnetism are primarily magnetite (Fe3O4),
titanomagnetite (Fe2O3 – FeTiO3), hematite (Fe2O3), magnetite (g-Fe2O3), pyrrhotite
(Fe7S8), etc. Indeed, the minerals have been detected by MGS orbital spectroscopy, lander
surface analysis and from Martian meteorites. Aqueous alteration might also have
contributed to magnetization (e.g., McSween et al., 2001). Magnetization of
Martian meteorite ALH 84001 may also be related to an early Martian dynamo. However,
the existence of localized magnetic anomalies in the northern lowlands and part of Tharsis
of younger ages (, 3 b.y.) also invokes evidence for a late turn-on of the Martian dynamo.

2.16.4.1. Core formation and magnetism

Tungsten isotope study reveals the Martian core to have formed within 30 Myr of
the formation of the Solar System. Tungsten anomalies correlate with neodymium
anomalies and date the period of differentiation of the mantle leading to generation of
the crust. However, the crust formation was synchronous with core formation and Mars has
not experienced any significant large-scale connective mixing of the mantle or impact
re-distribution of the crust since the time of core formation (Zuber, 2001).
Detection of magnetized rocks in ancient parts of the crust on Mars indicates
that the planet once had a magnetic field. The absence of magnetization in younger
rocks suggests that the internal dynamo ceased early in Mars’ history (Acuna et al., 1999).
Interpretation of tungsten isotopes suggest that core formation of Mars as well as of
Earth was effectively completed 60 m.y. after the start of accretion (Halliday and Lee,
1999). For Mars, we do not know if the core is solid or liquid. The melting relations could
place a constraint on the state of the core for a given composition and the core temperature.
The eutectic melting T is used to evaluate the efficiency of Fe-rich core segregation in the
early core-formation processes.

2.17. Venus

The crust of Venus’ surface is dominated by vast volcanic plains, several volcanic
rises and major plateaux. Most of the crust appears to be between 1 and 500 m.y. old.
The plateaux seem to record early extension in a hot crust; later deformation and volcanism
Petro-Tectonic Features of Terrestrial Planets 281

are consistent with cooling and thickening of the crust in the plateaux with time. Phillips
and Hansen (1998) suggest that these and other features can be related to the initial
formation of crustal plateaux above hot mantle plumes and, as the planet cooled with time,
a change in style of convection in the mantle took place.
There is no evidence of a magnetic field on Venus at the present time and its surface
temperature is too high to preserve a record of a former magnetic field (Russell, 1993).

2.17.1. Gabbro ! eclogite transition in Venus

Because basalt transforms to eclogite at high pressure and because eclogite is likely
to be denser than mantle material, the thickness of the crust of Venus may be limited by the
gabbro– garnet granulite– eclogite phase transitions (e.g., Turcotte, 1989). The bulk
composition and radioactive heat production of the crust on Venus, where measured, are
similar to those of terrestrial basalts (e.g., Surkov et al., 1987).
Maxwell Montes, standing up to 7 km above the adjacent highland plateaux,
constitute the highest mountain belt on Venus. Because the thickness of the crust is likely
to be limited by the gabbro– garnet granulite – eclogite phase transitions, this relief is
difficult to reconcile with the assumption of thermodynamic equilibrium and a standard
Airy isostatic model.
The micromechanisms in Venus governing the gabbro – eclogite transition,
involving chemical and phase changes, are not well understood.

2.18. Mercury

The three flybies of Mercury by Mariner 10 (1974 and 1975) discovered the planet’s
interval magnetic field, measured the UV signatures of H, He and O in Mercury’s
atmosphere, the time-dependent nature of Mercury’s magnetosphere and characteristics of
its surface materials.
The questions pertaining to Mercury relate to the origin of an anomalously high
metal/silicates ratio and its relation to the planetary accretion process, its geological
evolution and cooling history, the mechanism for its magnetic-field generation, the state of
its core and the processes controlling the volatile species in Mercury’s poler deposits,
exosphere and magnetosphere.
The discovery by a Mariner-10 flyby of Mercury’s intrinsic magnetic field suggests
the presence of a molten core. Harder and Schuberl (2001) opined an outer core with radius
of , 1,800 km and a thickness of , 500 km can be present if a small percentage of S is
mixed with iron in the core to reduce the melting temperature.
Observation that the metal/silicate ratio of Mercury is higher than any other planet
or satellite allows further thoughts on the early accretionary process of the protosolar
nebula.
After launch by a Delta 2925H-9.5, two flybies of Venus, two flybies of Mercury
and one orbit insertion were accomplished on the third Mercury encounter. These flybies
will explore the regions unexplored earlier by Mariner 10 and data will be gathered on
Mercury’s exosphere, magnetosphere and surface composition. The Mercury Messenger
282 Chapter 2

mission will undertake global mapping and characterization of the exosphere, magneto-
sphere, surface and interior (see S.C. Solomon et al., 2001).

2.19. Galilean satellites

Io, the smallest of the Galilean satellites, is noted for its active volcanism
This volcanism provides clues to the crustal-surface deformation and its relation to internal
dynamics. Large mass movement occurs at Euboea Montes in the southern hemisphere of
Io. This large slumping block of material probably represents a landslide related to uplift
and compression within the interior of Io beneath this area (see Schenk and Bulmer, 1998).

2.20. Interplanetary flights of planetary materials

The isotopic ratios of gas trapped in the unusual SNC meteorites indicate their
provenance, which is that they are actual pieces from Mars (Bogaard and Johnson, 1983).
Meteorites that originated on the Moon were identified and one might be even from
Mercury (Palme, 2002).
Such interplanetary transport, lifted by giant cratering impacts, might have “seeded”
life on one planet from the other, perhaps from the Earth to Mars or Mars to the Earth.
TheMoon may be a repository for ancient meteorites from other planets, including
Venus, Mars and the Earth itself. However, its surface has been tranquil for the past
3.8 b.y. compared with the pervasive geological and geochemical evolution of planets such
as the Earth, Venus and Mars.
Armstrong et al. (2002) propose that , 20,000 kg of ancient terrestrial rocks may
have been implanted in a 10 £ 10 km2 region on the Moon: , 200 kg are from Mars and
1 –30 kg from Venus. However, most of these lie well under “mega-regolith” hundreds of
metres deep, formed by ejecta blasted around impact basins several hundred kilometres in
diameter. Armstrong et al. (2002) estimate that , 7 ppm of lunar surface materials are of
terrestrial origin. Hence, it is possible that the samples brought back by lunar missions
might contain a few grams of terrestrial material. However, searching such samples from
Earth or Venus ejected out in catastrophic explosions, landing on the Moon and buried by
saltation and “gardening” is like trying to find a needle in a haystack.
During the short 50 m.y. period between 3.9 and 3.85 b.y. ago, several dozen
impact basins were formed on the Moon. This period, known as LHB, witnessed
bombardment by asteroids of the Earth and other terrestrial planets and Jupiter’s moons
were possibly also affected. This was the period when on the Earth 100-plus basins were
formed by bombardment just when life was struggling to gain a foothold on our planet —
and possibly on others as well (Levison et al., 2001).
 283

Chapter 3
Structural Types of Major Phases: AB, AB2, A2B3, ABX3, ABX4, AB2X4
and A2B2X7

In this chapter, the structural elements of the compounds having formula types AB, AB2,
A2B3, ABX3, ABX4, AB2X4 and A2B2X7 are discussed along with their type examples of
mineral phases and their respective special characteristics and properties under pressure.
Some of the examples, however, appear again in later sections when they are discussed in
the context of the consideration of the mineral groups. Nevertheless, some greater detail on
the structural and other property manifestations under high pressure of such phases as TiO2
and Fe2O3 are considered more appropriate here than elsewhere.

3.1. AB structures

AB compounds show five major structural types: (i) NaCl or rocksalt (B1)
(Fig. 3.1a), (ii) CsCl (B2) (Fig. 3.2), (iii) ZnS or zinc blende (B3) (Fig. 3.3a), (iv) ZnS or
wurtzite (B4) (Fig. 3.3b) and (v) NiAs (B8) (Fig. 3.1b).
Cation radii values quantitatively account for the phase transitions observed in the
series between “ionic” B1 (NaCl, six-coordinate), B2 (CsCl, eight-coordinate), four-
coordinated B3 (zinc blende) and B4 (wurtzite) crystal structures. In the ideal structure, the
anions occupy tetrahedral holes in a cubic cation lattice. In the zinc blende structure, the
occupancy is strictly charge-ordered and r 22 ¼ r þþ : For MO stoichiometry with face
centred cubic (fcc) lattice, in which the number of ions and holes equalize and the octahedral
holes are occupied, the resulting symmetric structure does not allow any dipole induction.
The NaCl (rock salt or B1) structure is conceived as composed of two
interpenetrating fcc arrays of Naþ and Cl2 ions to give 6-fold coordination around each
ion by the other (6 : 6). Per cell, there are four NaCl formula units. This structure can also be
viewed as of cubic close packing of Cl ions (rCl2 ¼ 1:81 Å) with Na ions (rNaþ ¼ 0:95 Å)
occupying the octahedral interstices or, in other words, as cation-centred NaCl6 octahedra
extending in three dimensions sharing all their edges. NaCl structure in many oxides
contains good amounts of vacancies at both cation and anion sites. The total bandwidth of the
valence energy bands in B1 and B2 NaCl are shown in Fig. 3.4. CsCl structure repeats a
pattern of Cl (or Cs) at the cubic origin and Cs (or Cl) at the cube centre (Fig. 3.2). Each atom
is coordinated with eight atoms of the other kind, which is known as cubic coordination. In
p
this ionic solid, when ions touch along the cubic diagonal, a 3 ¼ 2rCsþ þ 2rCl2 :
284 Chapter 3

Figure 3.1. Crystal structure of (a) NaCl (B1) and (b) NiAs (B8).

To account for why there are eight Cl2 ions coordinated to each Csþ ion in CsCl but
only six Cl2 ions to each Naþ ion in NaCl, Linus Pauling, twice a Nobel Laureate,
formulated rules to predict the structural coordination of ions in ionic materials in terms of
ionic size. The coordination number NC ; depending on the ionic ratio of cation to anion
radii (i.e., rc =ra ), determines the stability of structures. Up to an rc =ra ratio of 0.155, a
linear structure is predicted, whereas a triangular planar ðNC ¼ 3Þ array is stable at
0:225 . rc =ra . 0:155: Beyond these ratio values, three-dimensional tetrahedrons
ðNC ¼ 4Þ; octahedrons ðNC ¼ 6Þ; cubes ðNC ¼ 8Þ and close-packed face-centred cubes
or cuboctahedrons ðNC ¼ 12Þ become progressively more stable.
Values of r in nm are rðNaþ Þ ¼ 0:098; rðCsþ Þ ¼ 0:165 and rðCl2 Þ ¼ 0:181:
For NaCl, rðNaþ Þ=rðCl2 Þ ¼ 0:098=0:181 ¼ 0:544: Therefore, NC ¼ 6: For CsCl,
rðCsþ Þ=rðCl2 Þ ¼ 0:912: Therefore, NC ¼ 8: This explains the difference in structure of
NaCl and CsCl.

Figure 3.2. Crystal structure of caesium chloride (B2). Two interpenetrating simple cubic lattices (for Csþ and for
Cl2) generate this structure.
Structural Types of Major Phases 285

Figure 3.3. A comparison of the (a) sphalerite (zinc blende, B3) and (b) wurtzite (B4) structure.

The CsCl (or B2) structure consists of interpenetrating primitive cubic arrays of
Csþ and Cl2 ions with 8 : 8 coordination (Fig. 3.2).
The ZnS (zinc blende or B3) structure is much like the diamond structure. If one
half of the carbon atoms in the diamond structure forming the fcc array are replaced by zinc
and other half by sulphur, a zinc blende structure with 4 : 4 coordination is obtained. The
derivative of this structure is manifested by chalcopyrite, CuFeS2, with ordering of
monovalent and trivalent cations at the zinc sites. Cu3SbS4, famatinite, is obtained by the
substitution of Cuþ and Sb5þ ions in place of 2Fe3þ in chalcopyrite.

Figure 3.4. Total width (eV) of the valence energy bands in B1 and B2 NaCl (Bukowinski and Aidun, 1985,
q 1985 American Geophysical Union).
286 Chapter 3

The NiAs (or B8) structure is hexagonal with Ni octahedrally coordinated by As

atoms, which again are surrounded by six Ni atoms at the corners of a trigonal prism. In
this structure of hexagonal close packing of As, the octahedral sites are occupied by metal
atoms. The NiAs octahedra share common faces along the c-axis bringing metal atoms
close together. The NiAs structure has a Madelung constant (1.693) smaller than that of
NaCl (1.748). In it, the Ni– As bond is fairly covalent and there is Ni –Ni bonding.
Compounds with NiAs-type structure are seen in the sulphides, selenides, tellurides,
phosphides, arsenides, antimonides and bismuthides of transition metals.
HgS structure consists of spiral chains along the c-axis with each Hg atom having a
linear coordination. In such a structure, the d10 system causes formation of strongly
directional linear sp hybrid orbitals arising out of the close proximity of ns, np and
(n 2 1)d orbital energies.
CuO crystallizes in a distorted PtS structure because of the Jahn – Teller (JT) effect
on Cu ion. In PdS etc., showing tetragonal structure, the metal has a square planar (dsp2)
coordination and the anion has a tetrahedral coordination.

3.1.1. NaCl (B1): alkali halides

NaCl transforms from B1 to B2 structure at , 27 GPa and 300 K and , 21 GPa and
1,000 K (e.g., Li and Jeanloz, 1987). The theoretical and experimental zero-pressure
parameters for NaCl (B1) and NaCl (B2) are shown in Tables 3.1 and 3.2, respectively.
The effect of pressure on the effective ionic radii and second nearest-neighbour distances in
NaCl is shown Fig. 3.5. In the B2 phase, the Cl radius is slightly larger than half the Cl –Cl
distance, implying that the Cl ion is deformed and its radius is not well defined. Thermal
conductivity of NaCl in the B2 structure was estimated to be at least three times that of
CsCl (Slack and Ross, 1985).
The electronic structure of the lowest exciton states of alkali halides has not been
clearly understood so far. There is currently no quantitatively successful first-principles
computational approach for excitons.
The alkali halides can be studied using three-photon spectroscopy, the PIB model,
the charge-transfer model and so on. The principles of these are discussed below.

TABLE 3.1
Theoretical and experimental zero-pressure parameters for NaCl (B1)

A (Å) Density (g/cm3) K0 (GPa) K 00 K0 K 000

Theory 5.645 ^ 0.005 2.158 ^ 0.006 23.7 ^ 1.5 4.83 ^ 0.46 26.1 ^ 2
Experimental 5.64 ^ 0.001 2.163 ^ 0.001 23.3 ^ 23.78a 5.02 ^ 6.0a 24.77 to 223.84b
23.4 2 26.4c 3.9 2 4.92c
23.8 ^ 7.5d 4.0 ^ 3.9d
a
From ultrasonic measurements, summarized by Birch (1978).
b
Spetzler et al. (1972, 1996) and Birch (1978).
c
From static measurements, summarized by Birch (1978).
d
Diamond anvil (Sato-Sorensen, 1983).
Structural Types of Major Phases 287

TABLE 3.2
Theoretical and experimental zero-pressure parameters for NaCl (B2)

A (Å) Density (g/cm3) K0 (GPa) K 00 K0 K 000

Theory 3.458 ^ 0.005 2.358 ^ 0.01 26.6 ^ 0.7 5.20 ^ 0.2 29.0 ^ 2
Experimental 2.947a 121 ^ 23a 4.0a,b

Experimental zero pressure parameters obtained by Sato-Sorensen (1983) from fit to data above 20 GPa.
a
Sato-Sorensen (1983).
b 0
K 0 ¼ 4:0 assumed in fit.

3.1.1.1. Exciton in alkali halides

The studies of the exciton in alkali halides are based mostly on the “charge-
transfer” model and the “excitation” model (Dexter, 1957). In the empirical “HP formula”
(Hilsch and Pohl, 1928), it is assumed that the primary excitation process for exciton
formation involves the transfer of an electron from one of the halogen ions to a
neighbouring alkali ion as:

am e 2
Eex ¼ EA 2 EI þ ð3-1Þ
a
where EA is the electron affinity of the halogen atom, EI the ionization energy of the alkali
atom, am the Madelung constant and a the nearest-neighbour distance. This formula gives
the best agreement with experiment both for the zero-pressure lowest exciton energy and
its pressure shift.

Figure 3.5. Effect of pressure on the effective ionic radii and second nearest neighbour distances in NaCl (see text)
(Bukowinski and Aidun, 1985, q 1985 American Geophysical Union).
288 Chapter 3

(a) Three-photon spectroscopy. In probing the pressure dependence of the electronic

energy (< ultraviolet) levels in solids, three-photon spectroscopy has proved to be more
advantageous compared with one-photon spectroscopy (Lipp and Daniels, 1991).
Using the three-photon spectroscopy, it is seen that the energy of the lowest exciton
peak at ambient pressure exceeds the transmission limit of all existing standard pressure
cell window materials such as diamond and sapphire. Three-photon absorption can be
detected via luminescence of the self-trapped exciton. There are two luminescence bands
of the self-trapped exciton in both NaCl and KBr, as discussed below.

(i) NaCl spectra: polariton branch. The two peaks in the NaCl spectra (Zhang et al., 1994)
represent the two lowest exciton states separated by spin – orbit splitting of the valence
band and the exchange interaction of the electron and hole. The spectra of NaCl (unlike
other alkali halides) have only one resolvable resonance at 8.01 eV on the lowest upper
polariton branch. The polariton energy is seen to be independent of wave vector k (at least
between k ¼ 0 and 3kmax).

(ii) Blue shift of polariton energies. Throughout the pressure range of 0 – 6 kbar, the three-
photon excitation spectra show two peaks due to the resonances at wave vectors k and 3k
on the lowest upper polariton branch (Zhang et al., 1994). The energies of both polariton
peaks ½1ðkÞ; 1ð3kÞ increase with increasing pressure. This blue shift is linear.
The pressure shifts of the polariton energies for 1(k) and 1(3k) have been found
to be:
d1ðkÞ d1ð3kÞ
¼ ð15:3 ^ 0:4ÞmeV=kbar; ¼ ð14:5 ^ 0:4ÞmeV=kbar
dP dP
Within experimental error, there is no difference between the pressure shift of the
longitudinal exciton and the transverse exciton.
Formation of an exciton can induce electronic polarization in the surrounding
medium. Employing polarization correction, the lowest exciton excitation energy for NaCl
is observed to be 7.95 eV and for KBr it is 6.72 eV (see Fig. 3.6). With increasing pressure,
the exciton energies show linear blue shifts. The calculated blue shifts in both NaCl
(10.5 meV/kbar) and KBr (11.6 meV/kbar) appear to be lower than the experimental
values. The contributions towards the blue shift of the exciton energy induced by pressure
are from the changes in the Madelung term and the overlap term. The self-energy terms are
seen essentially to be pressure independent (in the pressure range 0 – 6 kbar).
The band-gap energy differs from the lowest exciton energy only by the exciton
binding energy. Preliminary experimental results indicate that the exciton binding energy
does not change significantly with pressure (Lipp and Daniels, 1993). Therefore, the
exciton is a good probe of band-gap variations with pressure.

(b) Charge-transfer model. The simple charge-transfer model assumes that the lowest
exciton states of the crystals arise from the configuration in which an electron is transferred
from a negative ion to the nearest-neighbour positive ion, creating two neutral atoms as
defects.
Structural Types of Major Phases 289

Figure 3.6. Three-photon excitation spectra of KBr at pressures of 99 bar (A) and 4,946 bar (B); T ¼ 15 K (Zhang
et al., 1994, q 1994 American Physical Society).

(c) The Hugoniots. The theoretical and experimental shock-wave Hugoniots for the B1,
B2 and mixed phase fields of NaCl are shown in Fig. 3.7 (from Bukowinski and Aidun,
1985) using Fritz et al. (1971).

3.1.1.2. NaCl structure at lower mantle

The nature and behaviour of minerals with rock salt structure have drawn much
attention from geophysicists because minerals with rock salt structure such as (Mg, Fe)O
and also MgO and FeO are believed to make up the majority of the Earth’s lower mantle.
However, there is uncertainty concerning the high-pressure behaviour of the 3d transition
metal oxides such as CaO, NiO, FeO and MnO, and the resulting effect on element
partitioning in both the Earth’s lower mantle and core and the implications for core
formation.
Of the Mn compounds, MnO, MnS and MnSe crystallize in the rock salt (B1)
structure at ambient conditions, while MnTe crystallizes in the NiAs (B8) structure. MnSe
transforms to the B8 structure at 298 K and 9 GPa.

MgO and ZnO: octupole polarization. The lowest energy isomorph of MgO is NaCl
(B1) structure, whereas that of ZnO is four-coordinated wurtzite (B4) structure.
Indeed, Zn2þ shows the strongest CFSE at a tetrahedral site amongst the transition
elements. The application of pressure (9 –9.5 GPa) on four-coordinated ZnO results in a
phase transition to six-coordinated B1 structure. For a post-transition metal ion like
290 Chapter 3

Figure 3.7. NaCl Hugoniot. Theoretical and experimental shock wave Hugoniots for NaCl. The B1, B2, and
mixed phase fields are the ones suggested by Ahrens et al. (1982). Approximate error due to uncertainties
in isothermal data of Sato-Sorensen (1983) is shown for corresponding Hugoniot (Bukowinski and Aidun, 1985,
q 1985 American Geophysical Union).

Zn2þ, the octupole polarizability, V; might be large because of octupole allowed d ! p

transitions.
Electronic structure calculations show that octupole polarizability of Zn2þ
(, 29 a.u.) is , 50 times greater than that of Mg 2þ, suggesting that cation
octupole polarization could be responsible for the different coordination preferences of
the two ions.
Structural Types of Major Phases 291

3.1.2. CsCl (B2) structure

3.1.2.1. Cs-halide (B2), CsI: metallization

As discussed in an earlier section, the crystal structure of CsCl (Fig. 3.2) appears as
interpenetrating simple cubic lattices, one for Csþ and the other for Cl2; CsI has a similar
B2 structure.
In the field of high-pressure research, caesium iodide, CsI, is the equivalent of the
molecular geneticists’ Drosophila or Escherichia coli in offering considerable scope for
investigation of widely diverse problems. Studies of CsI’s transformations and
transmutations under pressure (such as phase transitions, EOS, optical spectra, soft
modes, disproportionation and metallization) by static compression, shock waves and
theory have served as testing grounds, both for new ideas about compressed matter and for
new techniques to study it (Hemley, 1998).
CsI is a prototype for the study of the dielectric-to-metal transition. Because it has
the lowest energy gap (, 6 eV) among the alkali halides and because of its high
compressibility, it is a favourable material for the study of metallization under pressure.
It is also isoelectronic to solid xenon.
At ambient pressure, CsI has a B2 structure. It undergoes a continuous distortion
starting from the B2 phase to the hcp phase at 15 GPa and close packing of the ions up to at
least 200 GPa (Mao et al., 1990). The intermediate phase between 15 and 200 GPa has
orthorhombic symmetry. Optical experiments confirmed the metallization of CsI at
, 100 GPa (Williams and Jeanloz, 1986). Extrapolation of the pressure dependence of
the optical absorption edge gives an estimate for the onset of metallization near 100 GPa. An
increase of the infrared reflectivity, indicating metallization, is found around this pressure.
Eremets et al. (1998) from Osaka University pressed CsI into a metallic state
(Williams and Jeanloz, 1987) and directly measured the temperature and pressure
dependence of the electrical resistance of the sample. They were the first to provide direct
evidence for metallization at 115 GPa. Upon further increase in pressure, a characteristic
drop in resistance was found near 180 GPa at 2 K. Moreover, application of a magnetic
field caused the resistance to reappear — a convincing sign of superconductivity. The
superconductivity changed under the influence of the magnetic field to a lower critical
temperature and disappeared above 0.3 T. The critical temperature is seen to decrease with
increasing pressure.

3.1.2.2. Convergence with rare gas: solid Xe

CsI, an ionic salt having a simple cubic structure, contains closed-shell ions Csþ and
2
I . Two atoms in xenon gas, Xe2, are isoelectronic with CsI and condense to fcc crystals at
low temperature or moderate pressure (Fig. 3.8a). The atoms are neutral and are bound by
weak van der Waals forces, arising from mutual polarization of electrons on adjacent atoms.
At high pressure (.50 GPa), the EOS of CsI and Xe2 show striking convergence
(Fig. 3.8b). The transformations culminate in metallic band-overlap states. For CsI, the
high-pressure diffraction pattern indicates an hcp structure, with the Cs and I atoms
(which are indistinguishable in the high-density state) arrayed on the lattice points of the
292 Chapter 3

Figure 3.8. Pressure and structure transition in CsI and Xe (after Jeanloz, R. (1989) Ann. Rev. Phys. Chem.,
40, 237).

hcp structure. Between 100 and 150 GPa, both cross the divide between insulator and
metals. Hence, despite the difference in their birth forms — one a compressible ionic crystal,
the other a weakly bonded van der Waals solid — they transform to hcp structure with
identical densities at such pressures.
Structural Types of Major Phases 293

3.1.3. NiAs (B8) structure

The NiAs (B8) structure is represented by MmXn, where M is a metal ion and X is a
B group element. X ions are in approximately close-packed layers which in turn are
stacked in the ABAB… sequence and ðm , nÞ M ions are in octahedral coordination. This
is in contrast to the NaCl (B1) structure in which the layers of the X ions are stacked in the
ABCABC… sequence, where none of the octahedra shares faces.
Ideal NiAs structure has hexagonal symmetry. Each X ion is surrounded by six M
ions at the corners of a trigonal prism. The axial ratio ðc=aÞ varies from 1.22 to 1.97 for
different compounds. For hexagonally close-packed spheres, the ideal axial ratio ðc=aÞ is
1.633. A structure having a lesser value for this ratio is stabilized by cation – cation bonding
through overlapping of d-orbitals (along the octahedra) and those in the bi-pyramidal
interstices. Such compounds manifest metallic conductivity.
There are two crystallographic modifications of Fe-substituted NiAs (B8) structure,
depending on whether Fe is in the Ni (normal B8) or in the As (inverse B8) sites. In normal
B8 structure, the Fe atoms form chains along the c (the hexagonal) axis, the magnetic
collapse is continuous and the transition is continuous because of the short Fe– Fe distance.
In inverse B8 structure, the transition is sudden and occurs at extreme pressure
(.500 GPa).

3.1.3.1. Chemical bonding

In the NiAs structure, the axial ratio ðc=aÞ becomes close to 1.63, when it assumes
hcp structure. The structure is electrostatically unstable with respect to NaCl (B1) or CsCl
(B2) structures but it is stabilized by appreciable covalent character of its bonding. Non-
transition-metal compounds can have this structure under high P, T conditions. When the
chemical bonds are appreciably covalent, the transition-metal ions tend to have six-
coordination rather than four; hence, the NiAs structure occurs in preference to sphalerite
or wurtzite structure. Thus, the NiAs structure stabilizes when the transition-metal ion is
octahedrally coordinated and the bonding is dominantly covalent.
In an ideally close-packed hexagonal array ðc=a ¼ 1:633Þ; an ion occupies the
same space as in a cubic close-packed array but cation –cation distance along the c-axis of
the hexagonal close-packed array is considerably shorter than the cation –cation distance
in the cubic-packed array. Zemann (1958) computed the Madelung constant for the
NiAs structure of various axial ratios. The maximum of these constant occurs for the NaCl
structure. Thus, if electrostatic forces are the only interaction among the ions, the NaCl
structure is more stable than the NiAs structure.
Anions having p3 and sp3d2 orbitals favour the NaCl structure rather than the NiAs
structure but some hybrid orbitals such as spd4 and pd5 have a trigonal-prismatic
configuration. With a central X ion having sp3 hybrid orbitals, the distorted tetragonal
orbitals “resonate” at the six positions by “pivoting around the central ion” (Pauling,
1961). Such ions which form quadrivalent bonds prefer trigonal prismatic sites.
This resonating quadricovalent bond of Pauling can explain the magnetic properties of
the NiAs structure.
294 Chapter 3

In transition-metal compounds, two competing tendencies operate. Electrostatic

forces among the ions tend to arrange the ions in the NaCl structure in order to lower the
Coulomb energy of the system. The covalent bonds, on the other hand, tend to arrange the
ions in the NiAs structure in order to lower the distortion energy (in the covalent bonds).
The covalency of the bonds is consequently determined by the choices between these
two structures.

3.1.3.2. Hexagonal close packing and c=a ratio

The hcp metals are different from fcc and bcc metals in having an anisotropy
parameter represented by the c=a ratio. The model of hard-sphere packing specifies the
ideal values of the c=a ratio to be 1.633. Most of the hcp metals have axial ratios close to
this value but there are some metals whose axial ratios are very different. The deviation
from the ideal c=a value can be explained by the gain in the band-structure energy through
lattice distortion.
With pressure, the variation of the axial ratio is continuous but there is a clear
changepin the slope of the volume dependence of the p axial ratios at a common value of
c=a ¼ 3ð¼ 1:732Þ: The hcp structure with c=a ¼ 3 has special symmetry both in real
and reciprocal spaces. A number p of hcp reciprocal lattice vectors are degenerate and have
the same magnitude at c=a ¼ 3: The origin for the anomaly could be the topological
transition of the Fermi surface of such materials. Further theoretical and experimental
studies of this anomaly are highly desirable.
The most distorted hcp metals (namely Zn and Cd) manifest decrease in the axial
ratios with pressure and approach the value of 1.59 in the pressure range 100 –200 GPa.
The smallest axial ratio for hcp structure under pressure has been found in Ba(II), for which
c=a ¼ 1:50 (at 12.6 GPa). Ba(II) transforms to Ba(IV) exactly when the axial ratio
becomes 1.50. The hcp structure with c=a ¼ 1:50 attains high symmetry and also a number
of reciprocal lattice vectors become degenerate. This remarkable decrease in the axial ratio
with pressure can be related to the pressure-induced s – d electronic transition preceding
phase II. The increased d-character of the valence electrons in phase II adopts directional
bonding and may favour the distorted and anisotropic structure.

3.2. AB2 structure

The structural types of AB2 compounds may be described as: (a) fluorite, (b) rutile,
(c) pyrite and (d) cristobalite (see Fig. 3.9a – d). The most common dense form of AB2
compounds (where A is Si, Ti, Ge, etc.) is rutile structure (C4) (Fig. 3.9b), in which the
oxygen sublattice (B) can be seen as largely distorted face-centred cubic and in which only
one of the two octahedral sites may be filled by silicon. This structure is stable for ionic
compounds with rc =ra , 1:73: Quadrivalent metal oxides crystallize in this structure. It
consists of an infinite array of MO6 octahedra linked through the opposite edge along the c-
axis. For rutile-type oxides, the linear axial compressibilities of a ðba Þ and c ðbc Þ bulk
moduli (K) and their derivatives with respect to pressure ðdK=dPÞ are presented in
Table 3.3.
Structural Types of Major Phases 295

Figure 3.9. Structure of AB2 compounds: (A) (a) fluorite, (b) rutile, (c) pyrite. (B) The structure of cristobalite (c)
shows the same coordination found in all silicates; the Si is in tetrahedral coordination, and O is in 2-fold
coordination.

Rutile is related to ABX3 perovskite-type structure (where AB is MgSi, CaTi,

SrZr, etc.) in which an absence of A cation would be balanced by a modification of the
octahedral links, thus reducing the size of the vacant polyhedra.
Transition in TiO2 depends on the nature of compression. Compression in different
directions, [100] and [110], leads to CaF2-type structure (or a twinned type), whereas [001]
compression leads to the CaCl2-type structure. Calculations employing first-principles-
derived pair potential (Matsui and Tsuneyuki, 1987) showed that the AO2 system at the
transition fluctuates between two equivalent configurations, having a . b or b . a unit
cells, respectively, of the CaCl2 structure.
The predicted stability of Pa3-type silica was confirmed by calculations and the
transition pressure from the rutile or CaCl2-phase was predicted to be above 100 GPa
(Cohen, 1992).
The CaF2 fluorite (C1) structure consists of a cubic close-packed array of cations in
which all the tetrahedral sites are occupied by anions (Fig. 3.9a). This is akin to CsCl
structure, in which the alternate cubes are occupied by cations. This structure is generally
seen in cases where the radius ratio exceeds 1.732. Fluorite structure often shows
non-stoichiometry, accommodating the positions of anions and cations (4 : 8 coordi-
nations) such that an antifluorite structure may be generated. Examples of C1 structure are
296 Chapter 3

TABLE 3.3
Comparison of the linear axial compressibilities of a (ba) and c (bc) bulk moduli (K) and their derivatives with
respect to pressure ðdK=dPÞ for rutile-type oxides

Compound ba bc Ka dK=dP Method of study References

(£ 103 GPa21) (£ 103 GPa21) (GPa)

SiO2 1.32 0.60 306 – Brillouin Weidner et al. (1982)

GeO2 1.52 0.59 258 7b X-ray Hazen and Finger (1981)
1.68 0.62 259b 6.2 Ultrasonic Wang and Simmons (1973)
SnO2 1.73 0.78 218 7b X-ray Hazen and Finger (1981)
1.77 0.76 212b 5.5 Ultrasonic Chang and Graham (1975)
TiO2 1.80 0.90 216 7c X-ray Hazen and Finger (1981)
1.94 0.87 214 6.8 Ultrasonic Manghnani (1969)
a
Unless otherwise stated, isothermal bulk moduli and their derivatives with respect to pressure are presented.
b
Assumed value used in the calculation of KT from Brich–Murnaghan equation of state.
c
Isotropic adiabatic bulk modulus, derived from the Voigt–Reuss–Hill approximation.

shown by many oxides, sulphides, selenides, etc. Simulations for SiO2 in fluorite structure
(Matsui and Kawamura, 1987) have revealed that a new modified fluorite having space
group Pa3 could be formed.
The FeS2 pyrite (C2) structure consists of S22 2 molecular ions (Fig. 3.9c). The
structure is closely related to NaCl, from which it may be derived by replacing Na by Fe
and Cl by S22 22
2 , with the centre of the S2 ion occupying the Cl ion position. Each Fe ion is
octahedrally coordinated but each S is tetrahedrally surrounded by one S and three Fe.

3.2.1. SiO2 polymorphs

Silica of AB2 structure exhibits an exotic range of crystalline isomorphs. At room

pressure, polymorphs are composed of corner-linked SiO4 tetrahedra. This is illustrated
with the structure of cristobalite in Fig. 3.9d. With higher pressure, it transforms to six-
coordinated rutile structure (stishovite) and then to a CaF2 or a-PbO2 structure. The details
of these are discussed further in Section 12.1.
Due to its bonding characteristics, silica (SiO2) at different temperatures and
pressures yields to a variety of structures and physical properties. These are described by
various spatial arrangements of SiO4 tetrahedra. The transformations among the SiO2
polymorphs have been studied thoroughly over the years. The three polymorphs that are
most stable at low pressures are quartz, cristobalite and tridymite. The 4-fold coordinated
silicon (IVSi) results from strong sp3 bonding and a large number of packing sequences is
allowed by soft, deformable Si – O –Si linkage joining the rather rigid SiO4 units.
When SiO4 tetrahedra share all the corners, they make silica (SiO2). Commonly,
SiO2 occurs in three structural forms, the stability temperatures of which are:
a-quartz (C8): ,1,143 K
Tridymite (C10): 1,143 – 1,743 K
Cristobalite (C9): .1,743 K
Structural Types of Major Phases 297

Cristobalite structure is essentially a diamond structure in which oxygen atoms are

inserted between each pair of silicon atoms. This results in Si with tetrahedral coordination
and oxygen in a linear coordination (4 : 2). Tridymite also shows a 4 : 2 coordination.
Crystobalite (C10) is related to wurtzite (B4), just as cristobalite (C9) is related to the zinc
blende (B3) structure.
Occurrences of common HP silica polymorphs are shown below:
Coesite ! eclogites from UHP metamorphic terrains
Coesite þ stishovite ! impact craters
Free silica (stishovite) ! D0 zone

3.2.1.1. Si-coordination
Fourfold silicon (IVSi) is the fundamental building block of such technologically
important materials as amorphous silica optical waveguides, quartz crystal oscillators and
siliceous molecular sieves. It is one of the most common structural motifs found in
minerals that make up the Earth’s crust.
At higher pressures, silica increases its coordination number as in the high-density
phase, stishovite, in which 6-fold silicon VISi forms a network of SiO6 octahedra, a
configuration that gives rise to a more ionic Si –O bond (vide Section 12.1.2.1).
Four- and six-coordinated silica polyhedra have long been considered fundamental
building blocks of minerals, glasses and melts relevant to the mineralogy and geochemistry
of the Earth’s interior. The IVSi ! VISi transformations are observed at pressures in excess
of 8 GPa (.7,000 K) by the formation of stishovite. The low-pressure, low-temperature
polymorph a-quartz undergoes a crystalline-to-crystalline transformation at 21 GPa prior
to pressure-induced amorphization (e.g., Kingma et al., 1993). Similar transformations
have been observed for cristobalite and for the isomorphic forms of quartz, such as GeO2
and AlPO4 (e.g., Gillet et al., 1995).
At still high pressures around 50 GPa, stishovite transforms to a CaCl2 structure
type which also contains VISi (Kingma et al., 1995). Increases in silica coordination have
also been inferred from spectroscopic and diffraction measurements of statically
compressed silica glass or quartz amorphized at room temperature (e.g., Meade et al.,
1992; Williams et al., 1993).

3.2.1.2. Polarization and chirality

The lowest-energy structures in the absence of polarization are ideal for cristobalite
and tridymite. The former has an fcc oxide lattice whereas tridymite is hexagonally
packed. These structures have linear Si –O – Si triplets. These two structures are less stable
than the other form of cristobalite (real) and a- or b-quartz. The latter ones attain greater
stability because of the bending of the Si – O –Si bond. Polarization stabilizes the bent Si –
O – Si bonds which are formed in triplets. The polarization energy of the four-coordinated
structure is greater than that of the six-coordinated one. The high symmetries of the oxide
sites in the ideal cristobalite, tridymite and stishovite structures preclude such polarization
effects. In quartz, the SiO4 tetrahedra are arranged in hexagonal spirals. Depending on the
direction of spiral chirality, quartz exhibits dextro or laevo (D or L ) optical activity.
298 Chapter 3

Looking down on the SiO4 plane (ab-plane) of a quartz crystal, one can visualize
the alternating up and down spirals with clockwise- or anti-clockwise-induced dipoles
responsible for the piezoelectric properties.

3.2.2. TiO2

Rutile (TiO2), a minor constituent of igneous and metamorphic rocks, is

isostructural with stishovite (SiO2). High P experiments on TiO2 revealed the existence
of several high-pressure polymorphs (e.g., Haines, 1996). A a-PbO2-structured polymorph
is reported from different petrological settings: (i) as an inclusion in garnet from a
diamondiferous quartzo-feldspathic rock, Saxogian Erzgebirge (Hwang et al., 2000) and
(ii) as a phase in shocked gneisses from the Ries crater, Germany.
Rutile (TiO2) is reported to transform to an orthorhombic a-PbO2 structure in a
DAC at , 13.3 GPa (Liu, 1975). The apparent zero-pressure volume for the high-pressure
phase is approximately 70– 72% of that for rutile.
The lattice parameter of 4.455 Å for the cubic phase of TiO2 would imply a Ti –O
distance of 1.929 Å in 8-fold coordination of the structure is of fluorite type. This is less
than the normal octahedral Ti – O distance (.1.95 Å; Brown and Shannon, 1973).
DAC experiments on TiO2 in pressures .55 GPa revealed the following four dense
polymorphs:
No. Phase type Crystal Stability References
system (space group)
1 a-PbO2 phase, TiO2 II Orthorhombic (Pbcn) ,14 GPa, 300 K Olsen et al. (1999)
2 Baddeleyite (ZrO2) Monoclinic .14 GPa, 300 K Gerward
structured MI (P21/c) and Olsen (1997)
3 OI Orthorhombic (Pbma) .28 GPa Goresy et al. (2001)
structured(Pbca)
4 Cotunnite (PbCl2), Orthorhombic (Pbma) .55 GPa Dubrovinsky et al. (2001)
OII 9-coordinated Ti Orthorhombic

The MI- and OI-TiO2 phases synthesized in experiments above 14 and 28 GPa,
respectively, are metastable and invert instantaneously upon laser irradiation at ambient
pressure to the a-PbO2-structured phase (Dubrovinsky et al., 2001).
The baddeleyite (ZrO2)-structured TiO2 phase from the Ries impact crater,
Germany is 11% denser than rutile. This is the reported natural occurrence of an ultradense
TiO2 phase with Ti cations in seven-coordinated oxygen polyhedra. The XRD lines and
their Miller indices are presented in Table 3.4. In Ries, the high-pressure polymorphs
formed in the impact event were preserved during the post-shock decompression period. It
is presumed that the minor presence of FeO (0.14%) and Nb2O5 (0.20%) may be
responsible for preventing the back-transformation. The presence of residual stress could
also stabilize baddeleyite-structured TiO2, as it did for the diaplectic SiO2 glass (included
in garnets).
However, post-shock heating (.1,0008C) would back-transform these high-
pressure phases to rutile structure.
Structural Types of Major Phases 299

TABLE 3.4
Indexed peaks of the X-ray diffraction pattern and Miller indices collected from the natural baddeleyite-structured
TiO2 in the assemblage from Ries (Goresey et al., 2001)

dobs a (Å) lobs b (%) h k l dcalca (Å) lcalcb (%)

4.548 11 1 0 0 4.547 12
3.486 12 0 1 1 3.484 12
3.357 14 1 1 0 3.359 11
3.247 3 1 1 0
2.929 100 1 1 21 2.931 100
2.626 91 1 1 1 2.625 70
2.494 24 0 2 0 2.493 19
2.437 42 0 0 2 2.435 37
2.276 14 2 0 0 2.273 15
2.188 3 1 2 0 2.186 2
2.078 19 1 1 22 2.092 7
2 1 0 2.069
1 2 21 2.054 10
2.017 40 2 1 21 2.017 40
1 0 2
1.873 10 1 1 2 1.870 11
1.814 3 2 0 22 1.812 4
1.810 9 2 1 1 1.811 11
1.742 40 0 2 2 1.742 42
1.686 42 1 2 22 1.693 15
2 2 0 1.680 29
1.647 8 2 2 21 1.649 6
1.548 31 1 3 0 1.561 15
2 0 2 1.544 13
0 1 3 1.544 9
2 2 1 1.533 5
1 1 23 1.534 7
1.510 12 1 3 21 1.511 11
1.463 12 1 3 1 1.463 14
1.451 11 3 1 21 1.452 13

Rutile reflections of broad reflections dobs and dcalc are the observed and calculated d values, respectively. Iobs and
Icalc are the observed and calculated intensities, respectively.
a
a ¼ 4.606(2) Å, b ¼ 4.986(3) Å, c ¼ 4.933(3) Å, b ¼ 99.17(6)0.
b
Intensities are calculated with lattice parameters given above and coordinates of atoms for the ZrO2-structured
TiO2 polymorph.

The EOS (Birch – Murnaghan) for synthetic baddeleyite-type TiO2 (Dubrovinsky

et al., 2001) are KT ¼ 304 ^ 6GPa; K 0 ¼ 3:9 ^ 2 and V0 ¼ 16:93 ^ 3cm3 =mol:
It may be noted here that baddeleyite-structured TiO2 in slabs near to the translation
zone will accommodate appreciable amounts of ZrO2 and HfO2 in solid solution, causing
appreciable partitioning of Zr and Hf.
For the cubic phase of TiO2, an isentropic zero-pressure bulk modulus of
5.75 ^ 0.30 Mbar and a first pressure derivative greater than 8 were found (Liu, 1975). In
the rutile – cubic transition in TiO2, there is a volume change (18 or 29%). This is too large
300 Chapter 3

to be observed in SiO2 (; 2– 3 Mbar under static condition) under shock-compression

study by Trunin et al. (1970) up to 6.5 Mbar.

3.2.2.1. Cotunnite type: hardest polymorph

The hardness of ionic and covalent materials is related to their elastic properties and
increases with bulk modulus and shear modulus (see Section 5.6.2).
Anatase or rutile transforms at or over 12 GPa to baddeleyite structure. A new
polymorph of TiO2 with nine-coordination to oxygen in the cotunnite (PbCl2) structure is
synthesized at P . 60 GPa and T . 1,000 K. This is one of the least compressible and
hardest polycrystalline materials (see Table 3.5).
Cotunnite-type TiO2 phase could be compressed at ambient temperature to at least
80 GPa. Fitting the P – V data collected at ambient temperature to a third-order Birch–
Murnaghan EOS gave values of: K300¼ 431 þ 10 GPa, K 0 ¼ 1:35 ^ 0:1 GPa and
V0 ¼ 15:82 ^ 3 cm3/mol.
The polycrystalline high-pressure cotunnite-structured TiO2 is harder than
stishovite and much harder than alumina (Dubrovinsky et al., 2001).

3.2.2.2. Crystallographic shear (cs) planes

The crystallographic shear (cs) planes of TiO2 structure with varying composition
are known to be as follows:
Parent structure cs planes Series formula Approx. comp. range
TiO2 (rutile) 
{121} TinO2n21 ð4 , n , 10Þ TiO1.75 –TiO1.90

{132} Ti8O2n21 (16 , n ,, 37Þ TiO1.90 –TiO1.9375

3.2.3. Post-stishovite phase

For the stishovite (rutile) ! a-PbO2 transformation, there is an intermediate stage

through the CaCl2 structure (Haines and Leger, 1997). The transformation is second order
and involves minor distortions to the tetragonal rutile structure through octahedral rotation.

TABLE 3.5
Hardness of polycrystalline materials

Material Bulk modulus Knoop hardness References

(GPa) (GPa)

B4C 200 30(30) Srikanth et al. (1991) and Sung and Sung (1996)
SiC 248 29(29) Sung and Sung (1996)
Al2O3 252 20(19) Leger (1996)
SiO2, stishovite 291 32(33) Leger (1996)
WC 421 30(30) Teter (1998)
Cubic BN 369 (32) Sung and Sung (1996)
Cotunnite-type TiO2 431 38
Sintered diamond 444 (50) Sung and Sung (1996)
Structural Types of Major Phases 301

In general, the a-PbO2 phase of many dioxide systems (e.g., SiO2, GeO2, TiO2, SnO2 and
PbO2) is about 2% denser than rutile phase.

3.2.3.1. Stishovite(TiO2) ! a-PbO2 structural transformation

Stishovite transforms to orthorhombic CaCl2 type form near 50 GPa (Cohen, 1992;
Kingma et al., 1995). Molecular dynamics and ab initio calculations by Belonoshko et al.
(1996) indicate the possibility of SiO2 with Pb2n symmetry at P . 120 GPa. Later Karki
et al. (1997) based on first-principles calculations reported a transition of CaCl2 structure to
Pnc2 (pyrite-type) structure at a pressure of 226 GPa. However, Pnc2 and a-PbO2
structures should have the same total energies and diffraction lines.
The high-pressure SiO2 forms reported by different workers are noted as follows:
(i) Fe2N-type structure (Liu et al., 1978)
(ii) a-ZrO2/baddeleyite (e.g., El Goresy et al., 2000)
(iii) 3 £ zig-zag chain form (Haines et al., 2001)
(iv) undetermined stishovite-like form (Yamakata and Yagi, 1997)
In the octahedral chains of stishovite, the Si – Si distance is 2.668 Å while in a-PbO2
structure the Si – Si distance is increased to 2.722 Å because of the rippled structure.
However, the distance between the centres of unoccupied octahedral voids becomes much
shorter and, consequently, the translation period becomes shorter. This causes 5%
shrinkage of the unit cell, leading to an increased density of a-PbO2. The change in
arrangement of the Si4þ ions during stishovite (TiO2) ! a-PbO2 transformation allows for
more efficient packing, leading to an increase in density accompanied by an increase in
the nearest-neighbour Si – Si distance and a lowering of the electrostatic repulsion
(Dera et al., 2002).

3.2.3.2. Baddeleyite-type structure

The TiO2 system provides an analogue to the ZrO2 (baddeleyite) system. McQueen
et al. (1967) have reported that TiO2 undergoes a phase transition from 33 GPa terminating
at 100 GPa with , 20% volume reduction. The transition begins at 12.2, 17.0 or 33.7 GPa
along the [100], [110] or [001] axis, respectively. While Syno et al. (1987) report that this
transition terminates at , 70 GPa and another transition takes place around 100 GPa.
The a-PbO2 phase of TiO2 may be a metastable from of high-pressure phase,
having a flourite structure. Using laser heating, two different high-pressure phases have
been identified: (i) a hexagonal phase with volume 10.5%, less than that of rutile at 25 GPa
after heating to , 1,0008C (Liu, 1978) and (ii) an orthorhombic phase at 20 GPa
(, 1,0008C).
A high-pressure phase of TiO2 was studied by shock-wave experiments by Siato
et al. (1990). At 20 GPa (7708C), a baddeleyite (ZrO2) structure was observed. The
coordination of Ti increased from six to seven across the rutile to baddeleyite transition
with volume reduction of 9%. Thus, the structural changes with the phase transition are
noted in the sequence: rutile ! a-PbO2 ! baddeleyite.

Extraterrestrial occurrence: Mars. The PbO2/baddeleyite-like structure of SiO2,

occurring as an accessory phase, has been reported from the Martian meteorite, Shergotty
302 Chapter 3

(Sharp et al., 1999; El Goresy et al., 2000). However, the electron diffraction results (Sharp
et al., 1999) indicate the presence of a new orthorhombic SiO2 phase. Dera et al. (2000)
studied the small HP phase of SiO2, obtained from the Shergotty meteorite, employing the
XRD patterns and Rietveld refinement procedure. They identified the SiO2 to have a-PbO2
structure with the crystallographic parameters as follows:


a ¼ 4:097 A; 
b ¼ 5:0462 A; 
c ¼ 4:4946 A

The r is 4.295 g/cm3, which is 0.23% greater than stishovite (r — 4.285 g/cm3).

3.3. A2B3 structure

The crystal structure of A2B3 compounds is categorized based on whether the A – B

bond is predominantly covalent or ionic. Ionic A2B3 compounds are typified by
sesquioxides of d-transition metals and commonly show Al2O3 corundum structure.
Corundum structure is seen in a-Fe2O3 and such others, viz. Ti2O3, Cr2O3, etc. In these, the
cations have octahedral coordination and oxygen shows nearly tetrahedral coordination
(6 : 4). The unit cell volume vs. ionic radius systematics for the corundum structure
sesquioxide is presented in Fig. 3.10.

Figure 3.10. Unit cell volume vs. ionic radius systematics for the corundum structure sesquioxide (from Goto
et al., 1982).
Structural Types of Major Phases 303

The ilmenite structure is derived from the corundum structure by replacing the
cations in alternate layers with Fe and Ti.

3.3.1. Fe2O3

The corundum structure of a-Fe2O3 ðS:G: ¼ R3cÞ  consists of a unit of two

molecules, where four Fe3þ ions are occupying c sites with point-group symmetry 3 and six
O32 ions e sites with symmetry 2. These correspond to a primitive rhombohedral unit cell.
A magnetic spin – flip transition in hematite, known as Morin transition (Morin,
1950), has been shown to be quite sensitive to the effects of crystallite size (Nininger et al.,
1978), mechanical grinding and static high pressure (e.g., Syono et al., 1984).
Hematite (a-Fe2O3) is an anti-ferromagnetic material exhibiting a weak ferromag-
netism above the Morin temperature, TM (, 260 K), due to spin canting. The symmetry of its
 isomorphous with a-Al2O3 (corundum).
crystal structure is rhombohedral, D63d – R3c;
The anti-ferromagnetic easy axis is parallel to the crystal c hexagonal axis
above/below TM. TM is known to decrease systematically with decreasing particle size,
primarily due to lattice expansion associated with small particles (Nininger and Schrocer,
1978). However, TM is known to increase with static high pressure.
a-Fe2O3 undergoes a high-pressure transition near 50 GPa — from a high spin at
ambient pressure to low spin in high-pressure form (e.g., Venturini et al., 1985). The
transformation is almost complete at 60 GPa (e.g., Olsen et al., 1991). The transformation
is reversible and the structure reverts on release of pressure to that of a-Fe2O3. The
compression and decompression do not produce any pronounced hysteresis in the P – V
relation.
At about 50 GPa, hematite (rhombohedral, anti-ferromagnetic insulator) transforms
to a phase with low-spin iron. The free energy of HP phase is necessarily less than that of
the low-pressure phase.
Hematite (a-Fe2O3) is seen to undergo a structural phase transformation at
, 50 GPa, accompanying drastic change in electronic properties. From shock-wave
(Hugoniot) experiments, it is known that this transition involves a density increase of 14%. A
structural phase transformation of Fe2O3 has been observed with a transition pressure of
, 55 GPa; the high-pressure form is orthorhombic. The lattice parameters (for Z ¼ 4) at
60 GPa are

a ¼ 4:59ð3Þ A; 
b ¼ 4:97ð3Þ A; 
c ¼ 6:68ð5Þ A
At pressure between 55 and 63 GPa, Fe2O3 has been observed to transform from anti-
ferromagnetic to paramagnetic state (Mao et al., 1977). The electrical resistivity is seen to
start decreasing at pressures at , 50 GPa (Endo and Ito, 1982). At about 60 GPa, the X-ray
diffraction d-values begin to reduce suddenly and, at 85 GPa, the diffraction lines
corroborate the corundum structure (Yagi and Akimoto, 1982). The estimated bulk moduli
and density of Fe2O3 obtained by different workers are shown in Table 3.6.
In Fe2O3, the ferric iron occurs in octahedral coordination. Its high-pressure
perovskite structure possibly leads to the disproportionation: 2Fe3þ ! Fe2þ þ Fe4þ.
If iron remains in the high-spin Fe3þ state, this produces a tolerance factor of 0.83
304 Chapter 3

TABLE 3.6
Estimated bulk moduli and density for Fe2O3

K0 (GPa) Method

(a) Low-pressure phase

Goto et al. (1975) 193 Shock wave
Liebermann and Schreiber (1968) 202.7a Ultrasonic
Wilburn et al. (1978) 199 High-pressure X-ray
Olsen et al. (1991) 230(5) High-pressure X-ray (synchrotron source)

K0 (GPa) r0 (g/cm3) Method

(b) High-pressure phase

Goto et al. (1975) 277 6.22 Shock wave

K0 is set equal to 4.
a
K0 ¼ 4:53 determined from the same data.

(increasing from 0.75). Mössbauer study indicated high-pressure transition at , 55 GPa

with two different types of iron above the transition. This is represented by a quadrupole
doublet and a magnetic sextet in the Mössbauer spectrum (Syono et al., 1984). The
presence of a hyperfine field is consistent with a high-spin configuration ðS ¼ 2Þ and
the magnitude of the observed field is comparable with the values expected for Fe2þ. The
relatively large size of the high-spin Fe2þ cation implies that it occupies the A site while
the B site could be occupied by Fe4þ. Some, however, ruled out the presence of Fe4þ in
high-pressure Fe2O3. In SrFeO3 structure, “Fe4þ” results from rapid electron transfer
between Fe2þ and Fe5þ ions on adjacent sites.

3.3.1.1. Structural and spin transition

Shock-wave experiments on hematite (a-Fe2O3) have shown that it undergoes phase
transition between 0.4 and 0.8 Mbar (Syono et al., 1977). This transition may be induced by
a high spin –low spin transition in Fe3þ ions, possibly through corundum – corundum
transformation with a concomitant contraction of the radii of Fe3þ ion (Syono et al., 1971).
By 57Fe Mössbauer resonance spectroscopy, it has been proved that Fe2O3 is anti-
ferromagnetic below 0.55 Mbar but, at 0.63 Mbar, it becomes paramagnetic (Mao et al.,
1977). The shock-wave experiments also revealed that, between 44 and 52 GPa, this
structural change also accompanies a decrease in electrical resistance by several orders of
magnitude (Kondo et al., 1980). Through Mössbauer experiments (Suzuki et al., 1985), it is
seen that, in Fe2O3 under high pressure (.50 GPa), the valence state of iron transforms from
Fe3þ (high spin) to a combination of different valence states. A typical Mössbauer spectrum
obtained from the high-pressure phase at 69 GPa of hematite is shown in Fig. 3.11
(Kurimoto et al., 1986). The six-line magnetic spectrum overlies the paramagnetic doublet.
The behaviour of a metastable (intermediate) HP phase in the stability domain of
the low-pressure phase shows that the electronic transition is pre-empted by crystal-
lographic transition. However, in the HP phase, electronic transition occurs as a result of a
Mott transition.
Structural Types of Major Phases 305

Figure 3.11. 57Fe Mössbauer spectrum of the high pressure phase of hematite at 69 GPa. The spectrum shows two
components: one is magnetically split, the other shows a paramagnetic doublet (Kurimoto et al., 1986).

3.3.1.2. X-ray emission spectra

The Fe Kb X-ray emission spectra of Fe2O3 show that, under pressure near 72 GPa,
it undergoes high-spin to low-spin transition with a structural change. In Fe2O3, iron is in
the þ 3 state (3d5 configuration) and the total 3d magnetic moment is not equal to zero in
the low-spin state. The peak observed (Badro et al., 1999) is due to emission from the
valence band and the final state has a valence band and core-hole. The shape of the peak
indicates a valence-band electronic density of states projected by point-group symmetry
Oh. (Note: The emission spectrum in the region of the valence band reveals information on
the width of the band.) This part of the spectrum provides a direct measure of the
symmetry-projected anion p bandwidth. Several high P – T transformations in Fe2O3 have
been observed with the laser-heating in situ XRD technique (Mao et al., 1999).

Orthorhombic oxide perovskite. The high-pressure (, 60 GPa) form of Fe2O3 attains

orthorhombic perovskite structure (Pbnm). The orthorhombic unit cell contains four
formula units of Fe2O3. The unit parameters of the perovskite structure are related to the
lattice constant, ac, of the cubic unit cell by the relations:
p
a < b < ac 2 and c < 2ac
p
Thus, the c=a and c=b ratios should be close to 2 (¼ 1.41) in orthorhombic
distortion.
The crystal structure is perhaps a GdFeO3-type perovskite with space group Pbnm.
The volume per formula unit in the orthorhombic phase is 10% smaller than in a-Fe2O3.
The volume change is due to a combination of a valence change and a HS ! LS transition.
Williamson et al. (1986) investigated a-Fe2O3 in shock loading at peak pressures
from 8 to 27 GPa. Their studies employing Mössbauer and magnetic measurements revealed
that large fractions of the Fe sites do not exhibit TM transition. Small particles are seen to
have reduced hysteresis compared with bulk samples (Nininger and Schroeer, 1978). Their
306 Chapter 3

a-Fe2O3 powder had a “zero-field” magnetization M0 of 0.274 emu/g at 300 K. The zero-
field magnetization above the Morin transition can be split into an isotropic moment and an
anisotropic “intrinsic” moment in the (111) plane, attributed to the Dzyaloshinsky – Moriya
spin-canting mechanism (Moriya, 1960).
Above TM, the magnetic response of hematite is determined by an in-plane
anisotropy, which is very sensitive to stress, and the large residual strain could produce a
“hardening”. Under extreme conditions, a small amount (< 1%) of magnetite (Fe3O4) has
been formed by the shock. These observations by Williamson et al. (1986) can be due to
reduced crystallite size or large defect density associated with the residual strain. Large
residual strain leads to an apparent decrease in the “intrinsic” net magnetization above the
Morin transition, presumably due to strain-induced “magnetic hardening”.

3.3.2. Al2O3

Corundum (Al2O3) is rhombohedral (space group R3c),  and has 10 atoms per
primitive cell. The AlO6 octahedra are quite regular; the Al– O distances at zero pressure
are 1.86 and 1.97 Å and Al lies on the 3-fold symmetry axis. When doped with Cr3þ,
corundum can be used as a pressure calibrant in static high-pressure diamond-anvil
experiments (e.g., Mao et al., 1986). Cr3þ-doped corundum (called “ruby”) is also
commonly used as the window material in dynamic high-pressure shock-wave
experiments (e.g., Yoo et al., 1992).
Corundum seems to be stable over a wide pressure range from 0 to over 175 GPa
(Jephcoat et al., 1988). Cynn et al. (1990), using the potential induced breathing (PIB)
model, predicted a high-pressure phase transition of corundum. A phase transition is
predicted at high pressure (90 GPa) from the corundum structure to the Rh2O3(II) structure
using first-principles calculations with the linearized augmented plane-wave (LAPW)
method, wherein no assumptions about ionicity, bonding or the form of the charge density
are invoked (Marton and Cohen, 1994). These LAPW calculations make no
approximations other than the local density approximation (LDA), which has been
shown to be very accurate for other ionic oxides such as MgO (Mehl and Cohen, 1988),
SiO2 (Cohen, 1992) and MgSiO3 (Stixrude and Cohen, 1993).
Energy calculated with PIB predicts a transition to the Rh2O3(II) (Pbna) structure at
a pressure of 90 or 62 GPa, depending on whether the Thomas– Fermi (TF) or Kohn –
Sham (KS) form of the kinetic energy functional is used (Cynn et al., 1990). KS kinetic
energy is known to be more accurate in the PIB ionic model for determination of phase
transitions than the TF kinetic energy (Isaak et al., 1990). The PIB model also shows an
elastic instability in corundum at high pressures (Cohen, 1987).
Later, Yoo et al. (1992) observed an increase in the thermal emission at pressures of
200 GPa, which indicates a phase transition. The calculations show that, in the lower
mantle, free Al2O3 is unlikely to be in the form of corundum. Moreover, Al2O3 is probably
not sufficiently abundant to give an observable seismological signal.
A Raman spectroscopic study at P , 20 GPa was performed by Xu et al. (1995) on
corundum (Al2O3) and Kieffer’s model was applied for evaluating the high-pressure
thermodynamic properties.
Structural Types of Major Phases 307

3.3.2.1. Quadrupole polarizability

The corundum structure is observed to have higher stability over the less dense
bixbyite structure. In bixbyite, the coordination is almost tetrahedral while in corundum
the tetrahedra are twisted towards a planar D4h geometry. In the corundum structure, there
is an appreciable field gradient at the oxygen site which can induce a quadrupole; this is
observed from nuclear quadrupole resonance (NQR) studies of the field gradients at 18O
and Al nuclei. The quadrupole polarizability (c) of O2 has been estimated as 5– 7 a.u.
(Hafner and Raymond, 1968). The c value of corundum measured as 6 a.u. predicts its
stability over bixbyite by a fair margin. In MgO, the ab initio quadrupolar polarizability
shows a value of 26 a.u.

3.4. ABX3

3.4.1. Perovskite –ilmenite

The perovskites of the general formula ABX3 may be regarded as derived from BX3
structure. The BX3 framework consists of corner-shared BX6 octahedra. The large A cation
occupies the body-centred 12-coordinate position. The close-packed AX3 layers are
stacked one above the other with B cations occupying octahedral holes surrounded
by oxygen.
In an ideal cubic perovskite structure (Fig. 3.12; Zao et al., 1995), the atoms just
touch one another, the B –X distance isp
equal to a=2 and the A – X distance is 2ða=2Þ where a is the cube unit cell length. In a
perovskite structure, the following relation between radii of ions holds:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðRA þ RX Þ ¼ 2ðRB þ RX Þ

Even when the relation is not exactly obeyed in the ABX3 compounds, the
perovskite structure may still be retained. For this structure, the ratio relation is known as
the tolerance factor, t.

ðRA þ RX Þ
t ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð3-2Þ
2ðRB þ RX Þ

For ideal perovskite structure, t is unity but the usual range of “t” values for a perovskite
structure is , 0:75 , t # 1:0; allowing the structure to distort to tetragonal or
orthorhombic symmetry.
A small size of A ion causes a tilting of BX6 octahedra to optimize A – X bonding.
Many ferroelectric oxides show distorted perovskite structure at ordinary temperatures and
become cubic at high temperature. The perovskite structure can tolerate vacancies (or
excess) at A or X sites showing non-stoichiometric compositions such as A12xBX3 and
ABX32x, viz. CaFeO2.5 (brownmillerite) or LaMnO32x (^ La vacancies), respectively.
308 Chapter 3

Figure 3.12. The octahedral framework of an ideal cubic Pm3m perovskite. The centrosymmetrically distorted
orthorhombic Pbnm perovskite is derived through the combination of the octahedral rotation F about [001] and
the octahedral tilting f about [110]. It can also be conceived as the result of tilting F about the 3-fold [111] axes of
the octahedra (Zhao et al., 1994, q 1994 Mineralogical society of America).

3.4.2. Ilmenite structure: stability

The ilmenite (FeTiO3, R3)  and corundum (a-Al2O3, R3c)  structures appear as likely
candidates for high-pressure silicate structure in which all cations assume octahedral
coordination. In high-pressure FeTiO3 perovskite phase, Fe occupies the A site and
remains in the Fe2þ state (Leinenweber et al., 1995). For FeTiO3 perovskite, the tolerance
factor (t) is 0.82, which is similar for most distorted RE orthoferrites (t ¼ 0:82 in LuFeO3).
Magnesio-silicate ilmenites are unique in structure — each silicon octahedron
shares a face with adjacent magnesium octahedron. No other known silicate structure dis-
plays such face sharing between a silicon polyhedron and another tetrahedron or octahedron.
The stability field of silicate ilmenite is restricted. MgSiO3 ilmenite is formed above
20 GPa, but only above 25 GPa it is transformed to perovskite. A presence of only 10%
iron completely eliminates the ilmenite field for most divalent cations, except for Zn2þ,
which forms a stable ilmenite, ZnSiO3 (Liu, 1977).
Germanates and silicates of Fe, Mg, Co, Zn and also titanites of these elements
occur in ilmenite as well as in the perovskite structure under pressure. These structures can
also contain large cations such as Ba, Sr, Ca and Cd but all other transition elements are
destabilized in 8- and 12-fold coordination sites relative to octahedral sites because of the
ligand field effects (causing less CFSE, etc.). This explains why we do not expect
transition-metal perovskites.
Structural Types of Major Phases 309

Pressure-induced disordering in ilmenite may cause breakdown of its structure to a

slightly more dense corundum structure. Corundum and perovskite phases seem to have
higher entropies than ilmenite leading to negative dP=dT slopes of transitions.

3.4.2.1. Polymorphism of FeTiO3: LiNbO3 structure

The energetics of the ilmenite to FeTiO3 “lithium niobate” structural transformation
were investigated by Mehta et al. (1994) through transposed T drop calorimetry. A DAC
compression measurement indicated that the volume change for ilmenite to “lithium
niobate” transition (at 2 0.34 cm3/mol) is nearly P-independent. Thermodynamic cal-
culations show that the ilmenite to “lithium niobate” phase boundary is metastable with
respect to the ilmenite – perovskite phase boundary and the stable phase at high P is
probably the perovskite phase.

3.4.3. Ilmenite solution in olivine: Alpe Arami massif

A rock (garnet lherzolite) from the Alpe Arami massif, Switzerland is seen to
contain forsteritic olivine with 20-mm rods of ilmenite (FeTiO3). As the rock was supposed
to have undergone ultra-high pressure metamorphism (, 14 GPa), titanium was dissolved
in olivine. This later became exsolved because of reduction in pressure during the upward
movement of the rock (Bozhilov et al., 1999). For further discussion on such exsolution
features, see Section 11.6.3.5.
The oldest generation olivine is seen to contain up to 1% by volume of rod-shaped
precipitates of FeTiO3 parallel to [010] of olivine and up to , 0.2% by volume of tabular
chromite FeCr2O4 precipitates aligned parallel to (100) (Dobrzinetskaya et al., 1996).
These precipitates of highly charged cations (Ti4þ, Cr3þ) are older than the
prevalent dislocation microstructure (Green et al., 1997) and therefore pre-date Alpine
deformation. The older generation olivine exhibits a unique lattice preferred orientation
(LPO). This observation led Dobrzhinetskaya et al. (1996) to infer an extreme depth of
origin (.300 km) of the host peridotite (garnet lherzolite).
Experiments were conducted on nominally anhydrous ilmenite-saturated lherzolite
compositions at pressures from 5 to 12 GPa and temperatures from 1,400 to
1,700 K. Concentrations of TiO2 in olivine were recorded ranging from ,0.1 wt% at
1,400 K and 5 GPa to .1.0 wt% at 1,700 K and 12 GPa. TiO2 concentrations of 0.6 wt%
or more were measured for temperature .1,650 K at 10 GPa and temperature $1,500 K at
12 GPa (Dobrzhinetskaya et al., 1996).

3.5. ABX4

In this structural group, the most intriguing compound to interest high-pressure

physicists is AlPO4, the structural features of which are discussed below. A discussion on
AlPO4 is further continued in Section 14.14.1.
310 Chapter 3

3.5.1. Berlinite/scheelite structure: AWO4

ABX4 compounds of scheelite-type structure contain eight -coordinated Anþ cation

(n ¼ 1, 2, 3) and Bm (m ¼ 5, 6, 7) cations. Scheelite-type tungstates and molybdates of high-
pressure form have been studied up to 6.0 GPa by Hazen et al. (1985). The stable structure of
ABO4-type compounds is controlled, in large measure, by the cation radius ratios, A/B.
Compounds of the type AWO4 and AMoO4, for example, occur in the scheelite form if the
radius of A is greater than 0.90 Å but in the wolframite structure if the radius of A is less than
0.90 Å. The W or Mo tetrahedra do not show any significant change with P and T, which,
however, causes change in the effective radius of the eight-coordinated cation.

3.5.2. Berlinite and crystobalite: AlPO4

The hexagonal unit cell of a-berlinite (see also Section 14.14.1) contains three
molecular units of AlPO4 (Fig. 3.13; Christie and Chelikowski, 1998). The L and D forms
have P3121 and P3221 symmetry, respectively (Wyckoff, 1964) and are defined by the
lattice constants ðc; aÞ and eight internal parameters (u1 ; u2 ; x1 ; y1 ; z1 ; x2 ; y2 and z2 ).
The parameters u1 and u2 define the Al and P atom positions in the unit cell, respectively.

Figure 3.13. Structure of berlinite (a-AlPO4) (from Christie and Chelikowski, 1998, q 1998 Springer-Verlag,
Heidelberg).
Structural Types of Major Phases 311

The set of parameters (x; y and z) define O atom positions for two non-equivalent sites
(for details, see Thong and Schwarzenbach, 1979).
A Murnaghan EOS is used to determine the equilibrium energy, the equilibrium
volume and the bulk modulus. The volume vs. pressure curve represents the calculated
pressure using the semi-empirical Murnaghan (1974) expression:
"  0 #
K0 V0 K 0
P¼ 0 21 ð3-3Þ
K0 V

with the theoretical K0 value calculated from assumed K 00 ¼ 4:

The behaviour of c and a is linear with volume. The variation of the aluminium
parameter, u1 ; and phosphorus parameter, u2 ; shows a dependence with volume ðV=V0 Þ
(Fig. 3.14). The aluminium parameter, u1 ; shows an excellent agreement with
experimental values of Christie and Chelikowsky (1998). The volume and the internal
structural parameters as a function of pressure show good agreement with experimental

Figure 3.14. Theoretical (solid circles) aluminium and phosphorus structural parameters, u1 and u2 ; respectively,
versus the molecular volume for a-berlinite and comparison to experimental (open squares). The volume has been
normalized to the ambient experimental volume, 77.2 Å3 per molecular unit (Christie and Chelikowsky, 1998).
312 Chapter 3

results. This is also consistent with ab initio pseudopotential work performed on

isostructural a-quartz (Chelikowsky et al., 1991). Under pressure, berlinite manifests the
following features:
(1) At 12 GPa, disordering of oxygen sublattices takes place.
(2) The P – V curve shows first-order phase transition at 30 GPa.
(3) PO4 tetrahedra distort continuously.
(4) It does not amorphize at 30 GPa. A new phosphate phase (Cmcm) appears at , 30 GPa:

12 – 40 GPa
a-berlinite ! Cmcm phase
K0 ¼ 36 GPa very incompressible

This hindrance to amorphization is a consequence of kinetic preference and steric

hindrance plays a vital role in bringing about this frustration. Thus, AlPO4 cannot be a true
memory glass. For further discussion on berlinite, see Section 14.14.1.
Both SiO2 and AlPO4 have high-temperature cristobalite phases, which are
metastable at low temperature. These undergo a first-order but reversible a – b phase
transition at , 500 K. Despite their crystallographic differences, the mechanisms of the a–
b phase transitions in the cristobalite phases of SiO2 and AlPO4 are very similar.
The b ! a transition in AlPO4 cristobalite is from cubic ðF 43mÞ to orthorhombic
ðC2221 Þ; whereas that in SiO2 cristobalite is from cubic ðFd 3mÞ  to tetragonal (P4321
or P41 21 2).
Hatch et al. (1994) explained that these crystallographic differences stem from the
fact that there are two distinct cation positions in AlPO4 cristobalite, as opposed to one in
SiO2 cristobalite, and the ordered (Al, P) distribution is retained through the phase
transition. As a result, there are significant differences in their crystal structures and
domain configurations resulting from the phase transition and Landau free-energy
expressions. They performed symmetry analysis of the “improper ferroelastic” transition

from F 43m ! C2221 in AlPO4 cristobalite on the basis of Landau formalism and the
projection operator methods.
The b ! a transition in AlPO4 cristobalite results in tetrahedral configurations of 12
a-phase domains, in two classes: (1) transformation twins from a loss of the 3-fold axis and
(2) anti-phase domains from the loss of the translation vector 1/2[101] and 1/2[011]
(F ! C).
In contrast to a-SiO2 cristobalite, the a-AlPO4 cristobalite ðC2221 Þ does not have
chiral elements ð43 ; 41 Þ and, hence, enantiomorphous domains are absent. These
transformation domains are essentially macroscopic and static in the a-phase, and
microscopic and dynamic in the b-phase. The order parameter, Q, coupled with the strain
components, initiates the structural fluctuations causing the domain configurations to
interchange dynamically in the b-phase.
An analysis of the MAS NMR data (29Si, 17O and 27Al) on the a –b transitions in
SiO2 and AlPO4 cristobalites (Spearing et al., 1992; Phillips et al., 1993) essentially
confirms the dynamical model proposed earlier for SiO2 cristobalite (Hatch and Ghose,
1991) and yields a detailed picture of the transition dynamics. The NMR data on the
Structural Types of Major Phases 313

b-phase above the transition temperature, Tc, cannot be explained by a softening of

the tetrahedral rotational and translational modes alone, but require the onset of an
order –disorder mechanism resulting in a dynamic averaging due to rapidly changing
domain configurations considerably below Tc (Hatch et al., 1994). (Note: Berlinite as a
“memory glass” has been discussed in Section 14.14.1.)

3.5.2.1. GaPO4 and AlAsO4

GaPO4 and AlAsO4 show quartz structures. These were investigated by Sowa and
Asbahs (1996) near the pressure at which the quartz-type structures become unstable, using
single-crystal X-ray precession technique and polarized light microscopy.
AlAsO4 transformed into a new high-P phase before amorphization took place. On
release of P, the compound showed a “memory effect”. Non-reversible amorphization of
GaPO4 occurred under non-hydrostatic conditions.

3.6. A2BX4 structure

3.6.1. Tetragonal structure: K2NiO4

The body-centred tetragonal structure formed by four body-centred unit cells are
shown in Fig. 3.15 (Stixrude andp Cohen, 1995). The bcc structure corresponds to c=a ¼ 1
and the fcc structure to c=a ¼ 2:
Some complex oxides crystallize in tetragonal K2NiF4 structure (Fig. 3.16a and b),
which is akin to perovskite structure. K2NiF4 structure consists of perovskite slabs of one
unit cell thick, in which one slab is stacked over the other along the c-direction.

Figure 3.15. The body-centred tetragonal structure. Body-centred unit cells (four)are shown in thin lines. The face-
centred unit cell is shown in bold lines. The shaded atoms can be regarded as body-centred or face centred. The bcc
p
structure corresponds to c=a ¼ 1 and the fcc structure to c=a ¼ 2 (Stixrude and Cohen, 1995, q 1995 AAAS).
314 Chapter 3

Figure 3.16. (a) View of K2NiF6 octahedral structure showing one sheet of corner sharing NiF6 octahedra in z ¼ 0
level and a single octahedron at the z ¼ 21=2 level. The F anions are the larger shaded circles. The separation
between Ni2þ sheets along the c-axis is seen. (b) The spin directions on the Ni2þ are indicated by arrows (after
Freeman, 1995).
Structural Types of Major Phases 315

The adjacent slabs are displaced relative to one another by 12 12 12 such that the c-axis of the
tetragonal structure is roughly equal to three times the cell-edge of the cubic perovskite.
K2NiF4 has space group I4=mmm; with two formula units per unit cell. To this
tetragonal family belong the compounds like K2MnF4, RbMnF4, K2CoF4, Rb2FeF4 and
Ca2MnO4. In K2NiF4, the three-dimensional anti-ferromagnetic ordering sets in at
TN ¼ 97:1 K. Below TN ; the crystal is likely to be twinned even without any applied stress.
The K2NiF4 structure consists of corner-sharing sheets of NiF4 octahedra with
Ni2þ –Ni2þ separations of 4 Å. The anti-ferromagnetic arrangement of Ni2þ spins is
indicated by arrows in Fig. 3.17. Of particular interest is the potential coupling between
ferroelasticity and anti-ferromagnetism. Planes normal to a contain ferromagnetically
coupled Ni2þ ions, whereas those normal to b are anti-ferromagnetically coupled.

3.6.2. Spinel structure

The spinel MgAl2O4 has eight formula units (Mg8B16O32) in a unit cell. The
structure consists of a cubic close-packed array of oxide ions in which Mg2þ ions occupy
1/8th of the A (tetrahedral) sites and Al3þ ions occupy half of the B (octahedral) sites. This
distribution of cations, as in MgAl2O4 (the common spinel), is called the “normal” spinel
structure. At the other end is the “inverse” spinel structure in which the other half of these
occur in the B (octahedral) site. A common example is Fe3þ 2þ 3þ
A (FeB , FeB )2O4, magnetite,
which is a ferrimagnet.

Figure 3.17. Ac magnetic lattice of K2NiF4 below TN. Spins of Ni2þ ions are indicated by arrows. The K and F
ions are omitted for clarity. In this magnetic cell a < a1 þ a2 , and b < a1 2 a2 :
316 Chapter 3

In between these two extremes, lie the intermediate spinels with randomization of
cation distribution over all the 24 cation sites. The 2 : 3 spinels such as A2þB3þO4
(A ¼ M2þ: Mg, Mn, Fe, Co, Ni, Cu and B ¼ M3þ: V, Cr, Mn, Fe, Co, Rh, etc.) are well
known. Further discussion on spinels is presented in Chapter 9.

3.7. A2B2X7

3.7.1. Pyrochlore structure

Ternary oxides, such as those of A2B2O7 composition, commonly crystallize in

pyrochlorite structure. Oxide pyrochlore of formula A2B2O7 and space group Fd3m is fcc
in structure. This is a fluorite structure in which 1/8 of anions are removed so as to create an
almost 6-fold coordination around B ions, while A cations remain in the cubic coordination
as in fluorite.
The anions have two types of coordination: six anions have (2A þ 2B) tetrahedral
coordination while the seventh anion has (A þ 3B) tetrahedral neighbours — examples:
Ln2B2O7, Cd2Nb2O7, Pb2B2O72x, TlNbO3, TlTaO3, etc. Griggs (1983) systematized such
intermetallic compounds into valence-, electron- and size-factor compounds.

3.7.1.1. Frustration and magnetic “spin ice”

Pyrochlore lattice is composed of corner-linked tetrahedra. Because of the
structure, two out of every four spins cannot align anti-parallel. When tetrahedra are
linked to form a pyrochlore lattice, it becomes every bit as frustrated as any possible anti-
ferromagnet.
Within the pyrochlore lattice, anti-ferromagnetic interactions can become
unfrustrated and produce a conventionally ordered magnet, whereas ferromagnetic
interactions can become strongly frustrated. This anisotropy becomes strong in
pyrochlores where the magnetic atom is a rare earth, such as holmium (Ho) or dysprosium
(Dy). Here, the anisotropy constrains each spin to point either directly into or away from
the centre of each tetrahedron.
This frustrated pyrochlore offers a magnetic analogue of “spin ice”, wherein the
hydrogen atoms order in the ice (H2O) lattice (see Section 1.11.4.2). In this analogy, the
spins represent hydrogen positions around a central oxygen atom, so the “two in, two out”
rule is satisfied in the sense of two hydrogen close to and two further away from each
oxygen. The “ice rule” ensures that the structure consists entirely of H2O molecules
(see Section 1.11.3).
Spin-ice compounds are best represented by magnetic pyrochlore compounds, as
stated earlier, with Ho or Dy, such as Ho2Ti2O7 or Dy2Ti2O7. They possess residual
entropy at low temperatures. The work of Harris (1999) reveals that the two analogous
systems, Dy2Ti2O7 and ice, have the same entropy, even though the ordering dynamics of
the spins and hydrogen atoms are vastly different (Pauling’s “finite entropy”). The spin ice
presents a real threat to the Third Law of Thermodynamics. Possibly the disordered state of
water ice is the real ground state at low temperature.
Structural Types of Major Phases 317

Figure 3.18. Temperature dependence of magnetic moment measured with an applied magnetic field of 0.5 T. The
temperature dependence of resistivity at 0.2 and 5 T is also shown (right-hand axis) (from Shimakawa et al., 1996,
q 1996 Nature Publishing Group).

An answer to the question that reaches into the heart of what causes glassiness in the
first place lies in finding the origin of glassy behaviour in chemically ordered magnets.

3.7.1.2. Tl2Mn2O7: GMR

In Tl2Mn2O7, pyrochlore manganese ions form MnO6 octahedra and occupy the
16c site in the structure (fcc). The octahedral crystal field around Mn in MnO6 causes 3d
orbitals of manganese to split to t2g and eg states. In the MnO6 octahedra, the Mn –O –Mn
bond angle is , 1308, in contrast to the 1808 bond angle in the simple MnO6 network in
perovskite structure.
For Tl2Mn2O7, Shimakawa et al. (1996) determined the temperature dependence of
the magnetic moment in an applied magnetic field of 0.5 T. The temperature dependence
of the resistivity at 0, 2 and 5 T are shown in Fig. 3.18. The sample shows ferromagnetic

Figure 3.19. Field-induced changes in resistivity at 120, 135 and 150 K. Magnetoresistance (MR) ratio Dr=r 0 T at
135 K is also shown (right-hand scale) (from Shimakawa et al., 1996, q 1996 Nature).
318 Chapter 3

behaviour at low temperature. The susceptibility above 200 K obeys the Curie – Weiss
Law, x ¼ C=ðT 2 uP Þ: The paramagnetic Curie temperature (u P) obtained from the
Curie – Weiss plot is 164 K.
The giant magnetoresistance (GMR) of Tl2Mn2O7 with pyrochlore structure was
investigated by Shimakawas et al. (1996). The GMR properties of RE-Mn-perovskites are
discussed later in Section 15.4. The resistivity with zero applied field (r0T) shows a broad
maximum slightly above Tc and decreases markedly below this temperature. The sharp
decrease in resistivity in accordance with the development of spontaneous magnetization
strongly suggests that the electron transfer is closely related to local spin alignment. This is
similar to the case with hole-doped perovskite manganese oxides.
Negative GMR effects are observed at around Tc. There is no difference in
magnetoresistance (MR) effects for longitudinal and transverse configurations. At around
Tc, the resistivity decreases dramatically when a magnetic field is applied. Figure 3.19
shows field-induced changes in resistivity at 120, 135 and 150 K (Shimakawa et al., 1996).
Magnetoresistance with field-induced magnetization (above Tc) and the sharp
decrease in resistivity with the development of spontaneous magnetization (below Tc) are
both governed by the same magnetic-ordering-related electron-transfer mechanism.
 Section C

Basics for Pressures Studies

This page is intentionally left blank
 321

Chapter 4
Principles of Techniques

4.1. Introduction

The physical world encompasses a wide range of domains of pressure fields and we
are generally aware of the pressures in terms of atmospheric pressure, called bar. The
different pressure units involved in different phenomena are shown in Table 4.1.
The conversion for units of stress commonly used is shown in Table 4.2. In the early part of
the twentieth century, Bridgman established the principles of high-pressure techniques
with a two-stage piston-cylinder and opposed anvil devices (P.W. Bridgman, Collected
Experimental Papers, Harvard University, 1964). He was awarded the Nobel Prize in
Physics for this pioneering work. In 1955, diamond was synthesized from graphite. Later,
Drickamer and his co-workers developed the anvil devices to pressures of 50 GPa for
measurements of optical spectra, electrical resistance, X-ray diffraction (XRD) and
Mössbauer effect (Drickamer, 1966).
In 1973, Barnett et al. developed the diamond-anvil devices with metal gaskets
which are capable of generating hydrostatic pressure of 10.3 GPa at room temperature
(Barnett et al., 1973). During the last 30 years, technical development has taken place in
high-pressure researches with diamond-anvil devices.
High-pressure – temperature profiles of some major planets are shown in Fig. 4.1.
The most importantly versatile high-pressure equipment is the diamond-anvil cell (DAC)
(Fig. 4.2).
Materials may exhibit remarkable behaviour when they are subjected to extreme
pressures up to and beyond 300 GPa (3 Mbar) at temperatures ranging from millikelvin
to above 5,000 K. The P – T range for static high-pressure experiments along with
estimated temperature profiles for the Earth (geotherm) and other planets are shown
in Fig. 4.3a (from Mao and Hemley, 1998). Results of such studies bear
implications which span from fundamental chemistry and physics of Earth and
planetary science to materials science and high technology. The recent, important high-
pressure studies include dense silicates, hydrous phases (Chapter 13), simple oxides and
iron (Chapter 14).
New XRD and spectroscopic studies have uncovered novel phenomena in
Fe-bearing oxides and silicates, including electronic-spin transition and magnetic
collapse. These studies also involve measurements of the elasticity, texture and flow
properties of iron — which help in interpreting the seismological responses from the
322 Chapter 4

TABLE 4.1
The known pressures units and phenomena (Wentorf and Devries, 1987)

Pressure units
Pa bar atm. psi

Cylinder, gasoline engine 106 10 9.8692 145

Hydraulic jack, compressed gases 107 100 98.692 1,450
Ocean depths 108 1,000 986.92 1.45 £ 104
Metal forming 5 £ 108 5,000 4,934.6 7.25 £ 104
Diamond synthesis, depths of moon 5 £ 109 50,000 49,346 7.25 £ 105
Center of Earth 3.64 £ 1011 3.64 £ 106 3.59 £ 106 5.28 £ 107
Center of Jupiter ,1013 ,108 ,108 ,109
White dwarf star, degenerate matter ,108 ,1013 ,1013 ,1014

Note: A volume change of 4 ml g mol21 at 10 GPa means a free-energy change of 10 kcal g mol21, which is
significant with the energies of chemical reactions, 20– 50 kcal g mol21.

Earth’s core. SiO2 shows a major high-pressure transition at 50 GPa, which

collaborates with the geophysical signature of free silica in the lower mantle.
Light-scattering techniques help track the pressure dependence of the stability of
quartz.
Amongst the high-pressure experiments, most important have been those involving
high-intensity synchrotron radiation, X-ray fluorescence spectroscopy and new high-
pressure inelastic X-ray scattering. By 1990, the high-pressure researches reached
1.8 Mbar and detected evidence of a transition to metallic hydrogen in the molecular form.
SQUID. To measure static susceptibility, external coils are used with a
superconducting quantum interference device (SQUID) detector to measure the small
changes in magnetization.

4.1.1. Insulator –metal transition

It is now known that hydrostatic pressure would dramatically change the energy
surfaces of an electron in a solid. That is why an insulator behaves as a conductor when the
originally empty conduction band is forced to overlap the full valence band. Under
pressure, discontinuous changes in the topology of the Fermi surface of a metal may occur
when an energy band is pushed above or below the Fermi energy level.
For computational purposes, the electrons are divided into two groups: core
electrons and band electrons. The first are treated as highly bound atomic-like states, the
latter are properly treated as eigen states of the irreducible representations of the group of
the point K in the Brillouin zone. At core pressure, all electrons outside the L-shell ðn ¼ 2Þ
are treated as band electrons.
The insulator or semiconductor is characteristic of the energy-band gap whereas the
metal is typical of the partially occupied conduction band.
 Principles of Techniques
TABLE 4.2
Conversion table for units of stress commonly used in Earth sciences

Bar kbar Dynes cm22 Atmosphere kg cm22 N m22 Pascal (Pa) Gigapascal Pounds in.
(GPa) s22 (psi)

Bar 1.0 10 – 3 106 0.9869 1.0197 105 105 1024 14.503

kbar 103 1.0 109 0.9869 £ 103 1.0197 £ 103 108 108 10 – 1 14.503 £ 103
Dynes cm22 1026 10 – 9 1.0 0.9869 £ 1026 1.0197 £ 1026 1021 1021 10210 14,503 £ 1026
Atmosphere 1.1033 1.0133 £ 1023 1.0133 £ 106 1.0 1.0333 1.0133 £ 105 1.0133 £ 105 1.0133 £ 1024 14.695
kg cm22 0.9807 0.9807 £ 1023 0.9807 £ 106 0.9678 1.0 0.9807 £ 105 0.9807 £ 105 0.9807 £ 1024 14.223
Newton m22 1025 1028 10 0.9869 £ 1025 1.0197 £ 1025 1.0 1.0 1029 14.503 £ 1025
Pascal (Pa) 1025 1028 10 0.9869 £ 1025 1.0197 £ 1025 1.0 1.0 1029 14.503 £ 1025
Gigapascal 104 10 1010 0.9869 £ 104 1.0197 £ 104 109 109 1.0 14.503 £ 104
(GPa)
Pounds in. 6.895 £ 1022 6.895 £ 1025 6.895 £ 104 6.805 £ 1022 7.031 £ 1022 6.895 £ 103 6.895 £ 103 6.859 £ 1026 1.0
s22 (psi)

To use this table start in the left-hand column and read along to the column for which a conversion is required. Thus, 1 atm ¼ 1.013 £ 105 N m22 whereas
1 N m22 ¼ 1.02 £ 105 kg cm22. Sources: Baumeister and Marks, 1967; The Symbols Committee of the Royal Society (1971).

323
324 Chapter 4

Figure 4.1. Advances in P – T of static laboratory experiments are compared with respect to the pressure–
temperature profiles of several major planets (Duffy and Hemley, 1995, q 1995 American Geophysical
Union).

Figure 4.2. Schematic view of the diamond anvil cell and the different methodology that can be adopted using
DAC.
Principles of Techniques 325

Figure 4.3. (a) Differential stress resolution and pressure capabilities of several apparatus used in
experimental deformation studies. Temperature at which the strength of olivine (at a strain rate of
1026 s21) equals experimental accuracy of stress shown on the right (Kirby, 1985). (b) The zones of the Earth
corresponding to the depths and pressures (Kronenberg and Kirby, 1991, q 1991 American Geophysical
Union).

The insulator –metal transitions in tetrahedrally bonded crystalline and amorphous

semiconductors have been studied by measurements of the optical gap, electrical
resistance, XRD and Raman scattering as a function of pressure and temperature.
The transition pressures of amorphous phases are much lower than those of the crystalline
state because the amorphous states are high-energy metastable phases which contain
considerable strain energies.
As a prototype of insulator –metal transition in diatomic molecular solids, the
behaviour of iodine has been studied. Iodine under pressure shows two distinct
transitions: (i) metallization to molecular metallic phase at , 16 GPa and (ii) dissociation
to monatomic metallic phase at 21 GPa. At 21 GPa, iodine dissociates to the monatomic
metallic phase with rc ¼ 2:899 Å.
The cohabitation (i.e., co-existence) of metal and insulator in rare-earth
manganate perovskite is discussed in later sections. Application of a small magnetic
field to some poorly conducting material restores some ferromagnetic alignment, causing
the resistance to drop substantially. This phenomenon has been dubbed as colossal
326 Chapter 4

magnetoresistance (CMR) because it is much larger than that observed in standard

magnetic materials (Ramirez, 1997). CMR arises from competition between a metal and
an insulator.

4.1.1.1. Mott insulators

Mott proposed a transition for a crystalline array of atoms. If the interatomic
distance ðrÞ is larger than the critical value ðrc Þ; the material is a Mott insulator with
interatomic Coulomb repulsion ðUÞ: If r . rc ; the material is a metal with hopping integral
ðtij Þ from j atom to i atom.
A Mott insulator is a material in which the conductivity vanishes as temperature
tends to zero, even though band theory would predict it to be metallic, e.g., NiO, LaTiO3
and V2O3. However, the high-Tc cuprates are the only Mott insulators known to become
superconducting when the electron concentration is changed from one per cell.
A Mott insulator is fundamentally different from a conventional (band) insulator. In
the latter system, conductivity is blocked by the Pauli exclusion principle. Virtual charge
fluctuations in a Mott insulator generate a “super-exchange” interaction, which favours
anti-parallel alignment of neighbouring spins. In many materials, this leads to long-range
anti-ferromagnetic ordering. For detailed discussion on high-temperature superconduc-
tivity, the reader is advised to read Orenstein and Millis (2000).

4.1.2. High-pressure techniques

The main HP devices are piston-cylinder, multi-anvil apparatus and the DAC. The
pressure in multi-anvil apparatus can range from 5 to 27 GPa (; 750 km) and is suitable
for analysis of large samples to provide accurate phase-equilibrium data. A solid-medium

Figure 4.4. Example of a high-pressure UV– visible–near IR optical set up (from Hemley et al., 1998, q 1998
Mineralogical Society of America). (a) A double beam instrument for reflectivity measurements. All reflecting
microscope optics are used to remove chromatic aberrations and a grating spectrometer. (b) Modification of
the double beam system for absorption measurements. One of the reflecting objectives is replaced by an
achromatic lens. The system is capable of measurements from 0.5–5 eV; similar systems are in use in a
number of high-pressure laboratories (from Syassen and Sonneschein 1982). (c) Schematic of the optical setup
at the U2B IR beamline. The focused and collimated synchrotron beam in the IR region is directed either to an
IR microscope or to a custom designed, long-working distance microscope. The latter system consists of
symmetrical (Spectra-Tech/Nicolet Irus) reflecting objectives (Cassagrain type) to focus the IR beam onto the
specimen for both absorption and reflection measurements on small samples, either free-standard or in high
pressure cells, cryostats, or furnaces. At the same time, this integrated system allows full spectral
characterization of the same sample with UV, visible, and laser Raman and fluorescence spectroscopy (from
Hemley et al., 1998, q 1998 Mineralogical Society of America). (d) Schematic diagram of a confocal set-up
installed on a micro Raman spectrometer (Dilor XY). Two optically conjugated diaphragms, D1 and D2, are
placed in the optical train. D1 has a fixed diameter and allows the laser beam to be focused on the sample to
2 mm-sized spot. D2 is a variable aperture confocal hole allowing the collection of the Raman back-scattered
light, matched to the diameter of the scattering cylinder within the sample. Diamond anvil cells or heating
stages can be inserted at the laser focal point. The heating wire technique is schematically outlined (Gillet,
1996, q 1996 Springer-Verlag).
Principles of Techniques 327
328 Chapter 4

Fig. 4.4 (continued )

Principles of Techniques 329

HP deformation apparatus such as the Griggs apparatus can be used to , 3 GPa, whereas a
gas-medium high-pressure deformation apparatus such as the Paterson apparatus can be
used only to , 0.5 GPa.
Experimental techniques for high-pressure studies have experienced breakthroughs
that would have been deemed incomprehensible only a few years ago. The development in
second- and third-generation synchrotron sources has extended the capabilities for
studying materials over a wide P – T range.

4.1.2.1. Synchrotron source

Synchrotron sources provide intensely bright radiation which can be gainfully
employed in high-pressure infrared spectroscopic studies. The intensity at IR
wavelengths helps enormously in probing the microscopic samples when the
synchrotron source is coupled with a Fourier-transform infrared (FT-IR) interferometer
and special microscope for high-pressure cells. A gain in sensitivity is obtained up to
three orders of magnitude with respect to FT-IR measurements with a conventional
source. However, IR laser methods (non-linear optical techniques and diode lasers)
allow high brightness but a pulsed time structure provides important advantages for the
synchrotron method.
Hemley et al. (1998) have described a synchrotron infrared facility for high-
pressure spectroscopy and microspectroscopy at the National Synchrotron Light Source
(NS LS). Located at beam line U2B on the VUV ring of the NS LS, the facility utilizes a
commercial FT-IR together with custom-built microscope optics designed for a variety
of DAC experiments, including low- and high-temperature studies. The system contains
an integrated laser optical/grating spectrometer for concurrent optical experiments
(Fig. 4.4c).
They also observe that, for the U2B beam line, the use of the synchrotron source
becomes advantageous for samples smaller than 50 mm in the mid- and near-IR range.
Samples down to 10 and 25 mm diameter can be studied in transmission and reflectivity,
respectively, with synchrotron radiation at 600 – 1,200 cm21 (Note: Frequencies are
expressed in units of cm21, where n (cm21) ¼ w (rad s21)).
A new-generation synchrotron radiation facility operated at 8 GeV (Spring-80 and a
beam line equipped with 1,500-ton multi-anvil apparatus (SPEED-1500) are available in
Japan for studies in high-pressure mineral physics (Utsumi et al., 1997).

4.1.2.2. Synchrotron radiation

Of the extraordinary properties of synchrotron radiation, the following merit special
mention (Mao and Hemley, 1998):
(a) Its energy range extends continuously from the far infrared (1024 eV) to hard X-ray
(105 eV).
(b) Highly parallel radiation originates from a very small source (0.01 –0.1 mm) and
produces a very small focal spot (submicron in X-ray region) with exceptional
brightness or spatial resolution.
(c) The time structure is well defined for 10 –100 ps per pulse.
330 Chapter 4

(d) Radiation is linearly polarized in the plane of the electron orbit and elliptically
polarized above and below the plane.
Molecular structures are studied well by X-ray spectroscopy — bright X-rays
are produced from large accelerators. GRL uses synchrotron photon beam sources, avail-
able from second-generation Nations Synchrotron Light Sources (NSLS) and third
generation Advanced Photon Source (APS) at Argonne National Laboratory. GRL has
devised and employed a fully dedicated high-pressure synchrotron infrared spectroscopy
from NSLS.

4.1.2.3. Multi-anvil and DAC

The four types of multi-anvil apparatus used by different workers employing
synchrotron source for XRD studies are presented in Table 4.3.
Continued refinements in the DAC have increased the ability to study materials
under multi-megabar range (. 200 GPa). Such effort has also led to collateral development
of the long piston-cylinder megabar cell design (Mao and Bell, 1978), bevelled anvils
(Mao and Bell, 1979) and also extension of the range of optical (Liu and Vohra, 1996) and
X-ray (Ruoff et al., 1992) studies and the advent of new classes of megabar devices (Mao
and Hemley, 1996). Such studies have been facilitated by the concomitant development of
capability for microanalysis under high pressure. Hemley et al. (1997) have reported the
development and application of techniques that permit imaging and measurement of
stress – strain distributions and deformation of materials under multi-megabar pressures. In
highly stressed states, the behaviour of materials can differ considerably from those near
ambient conditions. For example, materials which are otherwise strong and hard can
deform plastically and develop texture at megabar pressures (. 100 GPa) (Mao and
Hemley, 1991).
However, estimates of the yield strength of materials under these conditions
have been inferred from indirect measurements and extrapolation from lower pressure
data (e.g., Chai and Brown, 1996), or from higher pressure experiments using limited
stress – strain geometries (e.g., Duffy et al., 1995).
For megabar pressure studies, it is important to determine directly the stress – strain
distributions of all load-bearing components of the high-pressure devices, including elastic
deformation of the anvils and elastic and plastic deformation of the gaskets. Direct
determinations of the elastic deformation of diamond of the DAC beyond 200 GPa could
not be attained but this has been inferred from optical absorption studies (Mao and
Hemley, 1991) or modelled theoretically.
The strain distribution in gaskets and the elastic deformation of the diamonds were
determined by Hemley et al. (1997) using direct imaging of the topography of the
diamond-anvil surface under in situ pressure around 300 GPa. It was observed that
diamond can accommodate remarkably large strains localized over small areas in the anvil
tip. Further development of ultra-high-pressure techniques requires relieving stress
concentrations associated with large elastic deformation.
The differential stress resolution and pressure capabilities of several items of
apparatus used in deformation studies are shown in Fig. 4.3a,b along with the depths in the
Earth and the corresponding pressures of the PREM model. Laser-heated DAC
 Principles of Techniques
TABLE 4.3
Types of multi-anvil apparatus used with synchrotron X-ray diffraction

Type of apparatus Anvil geometry Cell geometry Anvil material Direction of Reference(s)
P, T range diffraction vector

DIA Six anvil square Cube WC: 15 GPa, 2,0008C Vertical Shimomura
Tips SDC: 20 GPa, 1,2008C et al. (1985)
MA-8 or 6/8 Eight second-stage Octahedron WC: 30 GPa, 2,0008C Inclined 35.38T-Cup: from Vaughan et al. (1998),
cubes compressed ram force direction (T-Cup) or Clark (1996)
by six subparallel to load
first-stage wedge direction (Walker module)
6/8 in DIA MA8 second Octahedron WC: 30 GPa, 2,0008C Horizontal, with Shimomura
stage assembly SDC: 40 GPa, 2,0008C first stages notched et al. (1992)
compressed in
a large DIA
Paris-Edinburgh Opposed anvils Sphere WC: 15 GPa, 1,5008C Through anvil gaps Khvostantsev (1984),
with SDC: 30 GPa, 1,2008C Besson et al. (1992)
semi-spherical
recess in
each anvil

331
332 Chapter 4

experiments, shock-compression experiments and theoretical calculations are producing

accurate data on phase diagrams, EOS and melting. However, the temperature and pressure
regime of multi-anvil devices is limited to the top of the lower mantle. Again, the observed
lateral and depth variations in seismic velocities may have thermal, structural, dynamical
and/or chemical causes.

4.1.2.4. Measurement techniques

The four major measurement techniques for high-pressure studies are given below:
Diffraction X-ray Powder
Single crystal
Amorphous
Neutron Crystalline
Amorphous
Spectroscopy X-ray EXAFS, XANES
Optical Absorption
Reflectivity
Luminescence
Multiphoton
Raman Coherent and anti-Stokes Raman
Spontaneous Raman
Hyper-Raman
Infrared acoustic Brillouin scattering
Acoustic emission
Ultrasonic interferometry
Impulsive stimulated scattering
Inelastic neutron scattering
Mössbauer effect
Electron paramagnetic resonance (EPR)
Nuclear magnetic resonance (NMR)
Nuclear quadrupole resonance (NQR)
Transport Electrical Resistivity
Super-conductivity
Thermoelectric power
Magnetic thermal conductivity Susceptibility
Macroscopic Deformation and Strength
Macroscopic length

In these, the in situ methods include synchrotron X-ray diffraction and

spectroscopy, synchrotron infrared spectroscopy and optical techniques (e.g., Raman
and Brillouin scattering). The deviation from hydrostatic conditions can be accounted for
by estimating pressures from subsets {200, 420} and {222, 220} of NaCl diffraction lines
(and by assigning the corresponding difference to a pressure uncertainty of the
measurements).
Principles of Techniques 333

4.2. Diamond-anvil cell

Simultaneous high P – T measurements are essential for surmising the state of the
phases within the Earth and planetary interiors. By DAC technology, static pressures in
excess of 500 GPa and temperatures from 3 to 7,000 K have been achieved (Weathers and
Basset, 1987). A schematic view of the DAC for XRD and spectroscopic measurements
was shown earlier in Fig. 4.2.
Diamond anvils at zero pressure have an absorption threshold of 5 eV (for type IIa
diamonds). In diamonds of type I (with nitrogen impurity), the effective threshold is below
4 eV. Under pressure, the band gap of diamond increases while the absorption threshold
(possibly the gap) decreases in energy. Piezoelectric crystals become polarized when
subjected to a mechanical stress (compressive or tensile).
DAC is transparent in the IR, visible and near UV (to , 5 eV), and also to X-rays
with energies greater than 7 keV. However, a nitrogen impurity causing a pale yellow
colour may produce absorption bands in the infrared. High-quality synthetic diamonds
have an extremely low level (50 ppb) of nitrogen impurity and other defects as compared
with most natural diamonds. The gem-quality diamond is synthesized by combination of
chemical-vapour deposition (CVD) and high-pressure, high-temperature (HPHT)
technique. Diamond not only possesses very high strength, but also very high thermal
conductivity, about five times that of copper (see also Section 1.15.6.1).
Fluorescence and Raman spectroscopic measurements have been carried out on
synthetic isotopically pure 12C diamond to 320 GPa and synthetic isotopically mixed
diamond to 370 GPa. These studies indicate that synthetic diamonds stay yellow
fluorescent to pressures as high as 300 GPa with blue-green laser excitation, and beyond
370 GPa with red excitation. Synthetic diamond anvils provide ideal windows for
spectroscopic studies at ultra-high pressures.
The small volume at the highest pressures in the DAC and the long path lengths of
the X-ray beams through the multi-anvil devices need extremely bright X-ray sources for
unambiguous results. Third-generation synchrotron sources such as ESRF, APS and
Spring-8 with modern undulator insertion devices have arrived as solutions to the problem.
The sample should be embedded in a pressure medium with low thermal conductivity that
does not absorb laser radiation. The preferred orientation developed in the crystallites in
the anvil set-up has to be accounted.
Moissanite anvil cell. High-pressure studies using DAC suffers imitation induced
by the small anvil size and interference by the diamond spectra. Other gem anvils, such as
sapphire and cubic zirconia, have not been able to reach much beyond 26 GPa. Moissanite
anvils can be made two to three orders of magnitude larger than conventional diamonds
(Xu et al., 2002).

4.2.1. Properties of the gaskets

A gasket serves to encapsulate the sample, support the tip of anvils, and build a
gradient from ambient to peak pressure. The strength of the support depends on the
tensile strength and initial thickness of the gasket. Because a gasket provides both
334 Chapter 4

sample containment and anvil support at high pressures, the gasket material must exhibit
ductility as well as high strength under loading. Gasket materials are chosen primarily
on the basis of their internal friction or extrusion resistance under pressure, and also for
their desirable compressibilities, thermal stabilities, chemical inertness and ease of
fabrication. Between two parallel culets, the thin, flat portion of the gasket sustains a
large pressure gradient within the gasket and reduces the pressure gradient in the
sample.
The gasket acts as a supporting ring to prevent failure of the anvils due to the
elevated stress at the edge of the anvil faces. Yet most of the properties of relevant
materials have not been measured at high loads. Rheological tests on materials at mantle
pressures have been conducted since only recently. The problem commonly noticed in
DAC experiments is that the gasketing material slowly creeps away from the high-pressure
centre of the assembly. Rubie and his co-workers (1995) have attempted to measure
deformation in various kinds of 10/5, 14/8 and 18/11 (octahedron edge length/truncation
edge length in mm) multi-anvil assemblies.
In the piston-cylinder case, the piston may carry the heating current and a
thermocouple or other sensing wires can be led out through the gaskets. The pressure
inside the heated chamber may vary as a result of density changes produced by thermal
expansion or phase changes resulting from the heating. For example, pyrophyllite, a layer-
lattice aluminosilicate, may transform to a denser assembly of coesite and kyanite, thereby
reducing local pressure. Its thickness, t; is related to the shear strength, s; and pressure
gradient, DP=Dr (Sung et al., 1977):
t ¼ s=ðDP=DrÞ
In XRD studies, the gasket thickness ðtÞ can be calculated using Beer’s law:
t ¼ ð1=mÞlnðl0 =lÞ
where m is the average (or effective) extinction coefficient and l0 and l are the intensities of
X-rays incident to and transmitted through the gasket. Hemley and Mao (1997) determined
l0 from measurements (at the centre of the culet) at the maximum load, where the gasket
thickness is minimum (, 3 mm), and the effective m was determined from measurements at
the culet edge before diamond deformation (where the thickness was 45 mm). Calibrations
at intermediate loads and measurement of plastic deformation of gaskets recovered at zero
pressure indicate no measurable effects of pressure on the effective extinction coefficient at
these energies.
Beryllium gasket in a DAC allows both the incident and transmitted/scattered
photons to pass through the low absorbing Be gasket instead of the diamond. The usable X-
ray can, in principle, be extended to include the spectral region from 3 to 10 keV, making
possible a wide variety of X-ray spectroscopic studies of the electronic structure of
materials under pressure.

4.2.2. Pressure medium and calibration

Most pressure-transmitting medium, including mixtures of alcohol, alcohol and

water and rare gases such as Ar and Ne, become solid or stiff around 15 GPa (e.g., Zhang
Principles of Techniques 335

and Ahsbahs, 1998). This induces deviatoric stresses and reduces the diffraction quality of
crystal. Above , 8 GPa, the pistons may not be strictly incompressible and their
deformation may be significant.
High-pressure studies are often hampered by the difficulty of calibrating pressure.
In some calibrants, the pressure vs. property relation is employed, such as in P – V EOS
(determined by XRD). However, the most commonly employed technique comprises
spectroscopy, such as ruby fluorescence, which has been calibrated to 180 GPa against
EOS of simple metals (Bell et al., 1986).
In a DAC, T is measured from radiation spectra and P from shifts in the ruby
fluorescence line (e.g., Chudinovskikh and Boehler, 2001). Intense X-rays (from
synchrotion sources) allow P to be calibrated from the EOS of standards such as of gold
or platinum. One class of EOS yields pressures that place the g ! pv þ mv transition at
660 km, whereas the other yields pressures , 2 GPa lower, placing the transition at
600 km. The controversial 2 GPa (; 60 km) apparent shift of the g ! pv þ mv transition
appears to signal disagreements about high-pressure EOS of standard reference materials
(see Bina, 2001).
For laser-heating experiments, the pressure medium should have (i) low shear
strength to minimize deviatoric stresses in the sample, (ii) low thermal conductivity,
(iii) high chemical inertness, (iv) low absorption of the laser light and (v) low emissivity.
The ideal pressure media are the noble gases.
For pressure calculation (or, rather, calibration) an internal standard is used. In
pressure calibration, changes in the calibrant are usually in the form of changes in phase,
crystal structure or electrical resistance. Pressure-induced phase transitions are observed as
discontinuous changes in electrical conductivity. H2O has been used as a pressure calibrant
in HDAC (see Section 4.3). NaCl (m ¼ 1:5 GPa) and Al (m ¼ 2:6 GPa) are less commonly
used as pressure standards. For NaCl, the commonly used XRD lines are (200), (220) and
(222).
Pressures derived from lattice parameters of NaCl are generally lower than those of
Au; the difference arises from their respective elasticity mismatch (Wang et al., 1998). In
NaCl, the B1 – B2 (CsCl structure) transition occurs at , 29 GPA. The B1 – B2 phase
transformation in NaCl is sluggish and these two phases continue to exist to , 32 GPa.
Alkali halides (e.g., KBr, CsI, CsCl, etc.) showing low thermal conductivity at high
pressures are also well suited, albeit in restricted conditions. At high pressure, the
calibration is made using gold only.
The highest recorded pressure achieved by a DAC is 400– 500 GPa (Jayaraman,
1986). The pressure is applied by rotating the driving screw. The typical anvil-
face diameters are 750 mm for studies up to 20 GPa, allowing sample cavity diameters of
400 mm and heights of 50 – 100 mm. The sample size is typically 200 £ 200 £ 40 mm.
All gaseous materials solidify at high pressure. Commonly, a methanol/ethanol
(4 : 1) mixture is used. This allows measurement of pressures from hydrostatic to 10 GPa
at room temperature. Gases of low shear strength such as Ar or He are better to use. At
room temperature, Ar solidifies at 12 GPa and He above 10 GPa. At high temperature, ruby
pressure calibration is unsuitable because of the broadening of the ruby R lines, as well as
the decrease in intensity (Wunder and Schoen, 1981). The EOS measurement may also
336 Chapter 4

underestimate the pressure at high temperature when the electronic thermal pressure
becomes significant.
The pressure calibrations involving XRD (Liu and Vohra, 1996), including the
secondary ruby scale, are carried out for the axial geometry.
Unit-cell parameters may be measured to a few parts in 104 by routine single-crystal
XRD procedures but pressure measurements are limited to no better than ,^1% below
10 GPa. Resolving small differences in the compressibilities of two similar specimens is
thus impossible using conventional methods, in which two or more crystals are studied in
separate experiments.
To improve the accuracy of primary calibration for anvil devices, a pair of variables
has recently been chosen. They are the density ðrÞ measured with XRD and the acoustic
velocity ðVw Þ measured with ultrasonic or Brillouin scattering on the same sample under
the same compression (Mao and Hemley, 1996b). Pressure is derived from:
ð
P ¼ Vw2 dr

The resultant P – r relation serves as a primary pressure standard.

4.2.2.1. Quasi-hydrostatic stress

Above 12 GPa and RT, all pressure media solidify and therefore exert some degree of
non-hydrostatic stress on samples. Purely hydrostatic conditions can be achieved if the
pressure-transmitting medium remains liquid or gaseous under pressure. This is the case for
the 4 : 1 ethanol/methanol mixture up to 12 GPa (Piermarini et al., 1973; Fujishiro et al.,
1981). Above these pressures, vitrification of the mixtures occurs and quasi-hydrostatic
conditions prevail. Similar quasi-hydrostatic conditions are obtained with soft solids like
KBr, NaCl or CsI. Rare gases such as helium, argon and neon also provide quasi-hydrostatic
conditions.
The deviation from hydrostatic conditions can be accounted for by estimating
pressures from subsets {200, 420} and {222, 220} of NaCl diffraction lines and by
assigning the corresponding difference to a pressure uncertainty of the measurements.
Meng et al. (1993) have demonstrated by XRD that neon becomes significantly non-
hydrostatic above 15 GPa and that deviatoric stresses of the order of 0.15 GPa develop in the
compressed sample. The degree of hydrostaticity can be inferred from broadening of the
line-width of ruby fluorescence lines used for measuring pressure (Piermarini et al., 1973) or
by measuring pressure difference within the gasket hole. For measuring small deviatoric
stress, gold-cell volume change is effectively used (see Section 5.3.1.1 and Fig. 5.3).

4.2.2.2. Shear stress (s)

Estimates of the shear stress ðsÞ acting on the sample parallel to the diamond culet
can be obtained through the following relation (Meade and Jeanloz, 1988):
s ¼ 1=hðdP=drÞ
where h is the thickness of the gasket and ðdP=drÞ the pressure gradient across the gasket
hole.
Principles of Techniques 337

TABLE 4.4
High-pressure high-temperature reference phase transition (Wentorf and Devries, 1987)

Phase pair Transition aid End points (GPa, 8C)

Sillimanite–kyanite Water (0.8, 700)–(3.0, 1,700)

Quartz–coesite Water (2.7, 650)–(4.5, 2,100)
NaCl, liquid–solid – (2.5, 1,250)–(7.0, 1,600)
Graphite–diamond Mn, Ni (4.5, 1,200)–(10, 3,100)
Co, Fe
Coesite–stishovite Water (8.5, 450)–(10, 1,850)
ZnO (NaCl– wustite) Water shear (9.5, 200)–(11.5, 600)
aFe– 1Fe Shear (10, 500)–(11.3, 25)
ZnSiO3: clinopyroxene–ilmenite Water (11, 1,000)

For instance, for a run performed on KBr, the typical values at 15 GPa are
h ¼ 60 mm and dP=dr ¼ 1 GPa/100 mm. This leads to s ¼ 0:3 GPa, which agrees well
with that obtained by XRD (Meng et al., 1993). (Note: Without shear stresses being
developed, epilayers and multiple quantum layers or super-lattices cannot be generated.)

4.2.3. Reference-phase transitions

The high-pressure, high-temperature phase transitions of a few reference materials

are presented in Table 4.4. The numbers in parentheses are P and T coordinates that mark
the ends of a straight boundary line between two phases across the transition. The high P; T
conditions are shown in the right in parentheses.
For pressure calibration at room temperature, phase transitions in metals (Bi, Ba
and Pb) or semiconductors (ZnTe, ZnS, GaAs and GaP) are commonly used but such
calibration is of a limited use in high-temperature experiments. However, some important
phase transitions at corresponding temperatures and pressures may be noted for practical
purposes.
Phase transition Temperature (8C) Pressure (GPa) Source

a ! b olivine (Mg2SiO4) 1,400 14.3 Katsura and Ito (1989)

1,600 15.2 Katsura and Ito (1989)
Enstatite (MgSiO3) ! bol þ stishovite 1,400 16.5 Sawamoto (1987)
Majorite garnet ! MgSiO3 pvs 2,000 21.5 Gasparik (1996)
b-olivine ! perovskite þ periclase 2,000 22.4 Gasparik (1990)
(Mg2SiO4 ! MgSiO3 þ MgO)
1,800 22 Gasparik (1993)

The efficiency of pressure generation is seen to decrease at high temperatures such

as between 1,800 and 2,0008C but, at lower pressures, the pressure calibration is
independent of temperature (Gasparik, 1998).

4.2.3.1. CaO – MgO – SiO2 system

In the system CaO – MgO –SiO2, the subsolidus phase boundaries and melting
curves produce a grid of sufficient density and coverage at pressures up to 25 GPa and
338 Chapter 4

2,6008C temperature. The grid based on phase relations in the system CaO – MgO – SiO2
appears to be particularly suitable for calibration.
Experiments on the system are carried out using split-sphere apparatus, USSA-2000
(Gasparik, 1998). Calibration of the temperature distribution in the sample is performed at
the nominal temperatures of 1,400, 1,600 and 1,7008C using the compositions of two
co-existing pyroxenes on the join Mg2Si2O6 –CaMgSi2O6 (Gasparik, 1989).

4.2.4. Temperature control

4.2.4.1. Cryogenic methods

A variety of lower pressure devices (max. P , 7 GPa), including piston-cylinders,
have been used to liquid helium temperatures. DAC experiments into the megabar range
(. 100 GPa) have been performed with cryostats at liquid helium temperatures. There
have also been a few studies at lower pressure (, 1 GPa) to temperatures below 0.1 K
(Hemley and Mao, 1997).

4.2.4.2. Laser heating

High temperatures can be achieved by focussing a beam of an infrared high-power
laser on a spot (of , 30 mm across). The IR radiation of the laser beam is not absorbed by
the diamonds but is readily absorbed by most silicates and oxides containing small
amounts of iron. By modulating the laser power, temperatures from 300 to 5,000 K can be
achieved and the temperature obtained can be measured by pyrometry.
Temperatures greater than 3,000 K in a DAC can be achieved by heating the sample
by far-IR CO2 lasers and Nd-YAG and Nd-YLF lasers operating at 10.6 mm (e.g., Boehler,
1996).
YAG laser heating without thermally insulating media causes temperature gradients
in silicates ranging from 100 to 1,0008 per micrometer depending on the peak temperature
(Heinz and Jeanloz, 1987). For simultaneous high P – T diffraction studies, laser heating in
combination with synchrotron X-ray methods is used (Fiquet et al., 1996).
However, difficulties in experimentation at high temperatures and high
pressures have led to the importance of theoretical approaches which are expected to pro-
vide not only good calculated geometries, but also consistent sets of thermochemical data.
Semiconductors are studied as a function of temperature, which acts to change the
population of particles (such as electrons), or excitations (such as lattice vibrations) in
various energy levels.

4.2.5. Ruby (Al2O3:Cr31) calibration

For pressure calibration, ruby fluorescence scale is conveniently used. Ruby scale
has been calibrated up to 180 GPa.
The ruby chip in association with the sample is excited by a laser to the twin
fluorescence lines R1 and R2. The separation of the doublets (R1 and R2) does not change
Principles of Techniques 339

with pressure up to 10 GPa. Nevertheless, the relative intensity of the fluorescence can be
used to estimate the temperature in the range 5– 300 K. At elevated temperature, the band
maxima show broadening, intensification and shift to longer wavelengths. This is most
pronounced for the 18,000 cm21 band (especially in E==c position). Above 5008C, ruby
changes its colour to green.
Under pressure, intensification and blue shift of spin-allowed (4A2 – 4T2) transitions
are seen to occur (Goto et al., 1979). The shifts in unpolarized bands are noted as in the
following:

Transition Band position Pressures Blue shifted position

21
4 4
A2 ! T2 18,020 cm 46 GPa 19,880 cm21
21
4 4
A2 ! T1 24,690 cm 32 GPa 25,910 cm21

However, the ruby R1 fluorescence line, representing a peak of the spin-forbidden

4
A2 ! 2E(G) transition doublet, occurring at 14,405 cm21 (694.2 nm) under ambient
pressure (1.013 £ 105 Pa), shows the red-shift with pressure as:

Transition Band position Pressures Red-shifted position

21
4 2
A2 ! E(G) 14,405 cm 112 GPa 13,600 cm21

This shift in the R1 line is used as a pressure gauge in experiments using a DAC
(Mao et al., 1978). The pressure ðPÞ is calculated from the equation:
( 5 )
Dl
PðGPaÞ ¼ 0:380 £ 8 1þ 21 ð4-1Þ
694:2

where Dl is the pressure-induced wavelength shift of the ruby fluorescence line at

694.2 nm at atmospheric pressure. This relation holds remarkably well until 172 GPa. The
relation has a slope of 0.36 nm GPa21 and the precision of measurements is typically
0.05 GPa. Under quasi-hydrostatic condition, the relation becomes (Mao et al., 1986) the
following (applicable up to 80 GPa):
( 7:665 )
Dl
PðGPaÞ ¼ 0:248 £ 4 1þ 21 ð4-2Þ
694:2

By using ruby (Al2O3:Cr3þ), another calibration is obtained from the above relation
(Piermarini et al., 1975) as:

P ¼ 2:74DlðkbarÞ ð4-3Þ

where P is the pressure and Dl is the wavelength shift in Å from l0 ¼ 6,942 Å.
340 Chapter 4

4.2.6. Diamond window

The diamond window in a high-pressure cell has for many years been known to be
especially suitable for optical, infrared, gamma-ray (Mössbauer) and other in situ
spectroscopy measurements. The hydrostatic-pressure environment offered by this
window has resulted in a great improvement in spectral resolution. The spectroscopic
technique that has been possible at high pressure with high resolution involves Brillouin
scattering (the photon– phonon interaction process), which yields elastic properties of a
sample (Bassett et al., 1982).
The diamond-window high-pressure cell may contain a gas in its sample chamber
which is pressurized to the liquid (or dense fluid) state and even to such a higher pressure
state that is rendered to a weak, single or polycrystalline solid. Hydrostatic or quasi-
hydrostatic pressure can be generated above 1 Mbar, even at low temperature. As a result
of this development, not only can samples be studied under various degrees of shear stress,
but also single crystals can be preserved for structure determinations even to very high
pressure (Hazen and Finger, 1981).

4.3. Hydrothermal diamond-anvil cell

Bassett et al. (1993, 1996) developed a Hydrothermal diamond-anvil cell (HDAC)

to make accurate measurements on samples under pressures and at temperatures between
2190 and þ1,2008C. In such a system where water is used as the pressure medium, its EOS
can be used to determine the pressure from the measured temperature. When the EOS is
well known and the density of the fluid is accurately measured (by homogenization
temperature or freezing temperature), the accuracy of such pressure measurements is
significantly improved compared with those obtained by other pressure-measuring
methods (with DAC).
The techniques which employ HDAC include optical microscopy, XRD, infrared
and Raman spectroscopy and luminescence. By employing this cell, Bassett et al. (1996)
developed three pressure – temperature calibrations from: (a) H2O EOS, (b) rapidly
reversible second-order phase transitions and (c) shift of the Raman spectrum of diamond.
The EOS of H2O determined by using HDAC agrees well (Shen et al., 1992; Bassett et al.,
1996) with the P – T boundary of the a – b quartz transition, determined earlier by Mirwald
and Massonne (1980) for several different isochores.

4.3.1. Pressure calibration

In the HDAC, three new pressure calibrants have been characterized using H2O as
the pressure medium: BaTiO3, Pb3(PO4)2 and PbTiO3. Their applicable temperature
ranges are , 120, , 180 and , 4908C, respectively. The common features of these
calibrants are (Bassett et al., 1996) that (i) they are all based on ferroelastic phase
transitions that are rapid, (ii) the transition temperature decreases as pressure increases and
(iii) they are inert in most pressure media in a wide P – T range. The tetragonal – cubic
Principles of Techniques 341

phase transition for BaTiO3 and the monoclinic – trigonal transition for Pb3(PO4)2 can both
be detected by the disappearance and reappearance of the twin boundaries during heating
and cooling, respectively.
Chou et al. (1995) used HDAC to study the eutectic melting of the assemblage
Ca(OH)2 þ CaCO3 with excess H2O and found evidence for “portlandite II” phase.

4.4. Diffraction and spectroscopic techniques in pressure studies

In optical experiments, the problem of chromatic aberrations (l-dependence on

focussing) is avoided by using reflection optics (e.g., mirror objectives). A versatile UV –
visible – near IR absorption and reflecting system, using a conventional continuum light
source, has been designated for high-pressure applications (Fig. 4.4a – d). The scope of HP
studies includes changes in band gaps, excitations, crystal-field spectra, impurity bands,
and defects.
For transparent minerals (wide band-gap insulators), the measurement in the
visible range gives the real part of the dielectric function (i.e., off-resonance).
Absorption and reflectivity data help us understand the changes in the electronic
structure of minerals close to insulator – metal transitions at high pressures, including
the appearance of the associated inter- and intra-band excitations (examples: diamond
and MgO under pressure show a decrease in refractive index ðnÞ; and an increase in
band gap). However, many other materials show a decrease in band gap with pressure.
Brillouin scattering provides an independent measurement of the refractive index.
High-pressure metal –insulator transition also shows conductivity changes in the
optical range; e.g., at the pressure-induced transition of graphite to an insulating state,
graphite becomes transparent (e.g., Utsumi and Yagi, 1991).
HP and HT in X-ray diffractometry make one of the most useful methods to
determine an EOS. In order to derive the EOS from shock-compression data or
from data of acoustic velocities obtained by the ultrasonic or Brillouin scattering
method, it is necessary to use the high-temperature thermal properties (e.g., thermal
expansivity) at ambient pressure. The principles of the high-pressure spectroscopic
technique are discussed in the following section, followed by those of diffraction
methods.
From refractive indices measured by optical interferometry (Hemley et al., 1991),
one can determine the high-pressure dielectric properties. Changes in crystal symmetry by
pressure are revealed by distortion of interfacial angles or birefringence under crossed
polarization.

4.4.1. Optical spectroscopy

Optical spectroscopy is one of the most informative techniques for ultra-

high-pressure studies. This method is indispensable for studies of low-Z containing
materials for their disordered phases, phase transitions, pressure calibration and lattice
342 Chapter 4

dynamics. The set up for a high-pressure UV – visible –near IR optical studies is presented
in Fig. 4.4a and b.
Pressure effects on optical absorption spectra of transition elements in minerals
reveal the change of crystal-field stabilization (or orbital hybridization) energy, oxidation
state, site occupancy, d- or f-electron spin state and change-transfer processes. These in
turn affect radiative heat transfer, electrical conductivity and oxidation processes in the
Earth’s deep interior (Mao and Bell, 1972). Especially selected diamonds at low pressure
may be transparent from the near UV (5.5 eV) to 4 mm and from 5.5 to 10 mm in the
infrared. Common diamonds (Type I) reach only 3.5 eV in the UV and show strong
absorption in the 7– 10 mm region.
Reflectivity measurements in the visible and infrared have been extensively used as
a diagnostic techniques in recent studies of metallization of condensed permanent gases
(Reichlin et al., 1989).
The energy of the synchrotron radiation beam shown in Fig. 4.4c extends
continuously from the far infrared (1024 eV) to hard X-ray (105 eV). A very small source
(0.01 – 0.1 mm) gives out highly parallel radiation and the time structure is very well
defined by 10– 100 ps per pulse. The availability of synchrotron sources and energy-
dispersive X-ray analysis has reduced data collection times and increased the general
capability of techniques by permitting very fine collimation (, 4 mm) of the incoming
beam (Ruoff et al., 1990). For micro-Raman studies, the set up of the spectrometer is
shown in Fig. 4.4d (refer caption for details).

4.4.1.1. Crystal-field under pressure: theory

The application of CFT in the partially filled d-shell of transition elements in
minerals has found importance in the geochemical distribution of these elements through
the evolution of the host rocks (see Section 1.27). As stated elsewhere, the orbital
hybridization of the central ion and its neighbour (e.g., oxygen atoms) is considered as the
explanation of the spectral changes. Spectral measurements help substantially in
answering the problems such as site occupancy, coordination number in dense glasses
and radiative heat transfer (e.g., Keppler et al., 1994).
In crystal-field theory, the energy level of the d-orbital (eg and t2g orbital) of a cation
is effected by the location of the oxygen ligands (Burns, 1985). The interaction among the
cationic orbitals and ligands determines the stability of the transition elements in a
crystalline site. The t2g orbitals have the greatest electron density in a direction between the
x; y and z axes. Under pressure, the spin state changes from high spin to low spin, as
illustrated in Fig. 4.5 with a d6 ion.
The eg – t2g splitting in a hydrated silicate is due to OH-diagonal interaction and not
shifts in the diagonal elements of the Hamiltonian matrix. The eg –t2g splitting can be
considered as arising from a ligand-field effect, but it is not at all related to changes in
electrostatic field at the atomic site. The splitting is due to d –d interactions between
next-nearest-neighbour cations and due to p – d and s– d interactions with neighbouring
oxygen ions (throughout the Brillouin zone). The d– d interactions lead to splitting
approximately as 1=r 5 and the p– d and s– d interactions give splittings that would vary
approximately as 1=r 7 (Harrison, 1980). The metal – metal d– d splitting dominates.
Principles of Techniques 343

Figure 4.5. High and low spin states of d6 ion.

ol=melt
Figure 4.6. ›lnDt =›P of divalent Ni, Co. Fe, Mn and Mg (from Taura et al., 1998).
344 Chapter 4

A strong pressure dependence is seen in the partition coefficients of Ni, Co and Fe

distributed between melt and olivine crystals. The effect is greater in these elements
compared with those of other divalent cations showing no crystal-field effects (e.g., Mg,
Mn and Ca) (see Fig. 4.6). The figure of Taura et al. (1998) shows a comparison of
ol=melt
pressure dependence, ðdlnDi =dPÞ; of partition coefficients among Mg, Mn, Fe, Co and
Ni calculated from Fig. 4.6. The degree of pressure dependence shows the order:
Ni . Co . Fe.
The splitting of 3d orbitals of Fe2þ in octahedral sites (i.e., FeO6 octahedron) in
Fe2SiO4(fayalite) imparts a crystal-field stabilization enthalpy (CFSE $ 0) to the structure
which amounts to (see Fig. 4.7a –d):
 2 
CFSE ¼ 2 5 D þ ðd þ sÞ=3 ð4-4Þ

where D (or 10Dq) is the energy difference between the bari-centres of eg orbitals (d2z and
dx2 2y2 ) and t2g orbitals (dxy ; dyz and dzx ), d is the maximum splitting of t2g orbital energy
levels and s is the energy splitting between the two lowest t2g levels. Because there are two
Fe2þ ions per formula unit of Fe2SiO4, the factor 2 arises. In general, D increases with

Figure 4.7. Schematic diagram showing the crystal field splitting d-orbitals of Fe2þ in FeO6 octahedron (b) the B
octahedron of spinel; (c) The M1 (Ci) octahedron of olivine; and (d) the M2 (C5) octahedron of olivine. The
symmetry of each octahedron is also shown.
Principles of Techniques 345

decreasing volume of the FeO6 octahedra. To a first approximation:

D / d25
Fe – O

Both d and s are functions of the distortion of FeO6 octahedra. The term 2ðd þ sÞ=3;
designated as j; is the extra stabilization enthalpy gained by Fe2þ (per Fe2SiO4 formula
unit) as a result of distortion of FeO6 octahedra.
The crystal-field splitting of the 3d orbitals not only imparts a CFSE which
contributes to the lowering of the lattice energy, but it also reduces the electronic
configurational entropy, Se ; by removing orbital degeneracy. However, the lowering of
electronic configurational entropy due to the crystal-field splitting, SCFS, tends to offset the
enthalpy effects by lowering the free energy and this effect can be especially significant at
high temperatures. The crystal-field splitting brings about an overall change of free energy,
GCFS, which equals ðCFSE þ TSCFS Þ; signifying that CFS is temperature dependent. From
the values of CFSE and SCFS, the contribution of the crystal-field splitting to the free
energy (GCFS) can be calculated. For example, the difference of GCFS between spinel and
olivine measures the crystal-field stabilization on the olivine ! spinel transition as:
DGCFS ¼ DCFSE 2 TDSCFS
Increasing pressure tends to increase the DGCFS (i.e., make it less negative) due to
reduction of DCFSE.
The overall DG may be separated into two components:
DG ¼ DG0 þ DGCFS
where DG0 is the overall free energy change for the olivine ! spinel transition under the
hypothetical condition of zero crystal-field splitting. DGCFS may be approximated as <
DV: DP; where DV is the volume change of the transition and DP is the over-pressure.
When
DG0 ¼ 0
DP ¼ DPCFS
and
DPCFS ¼ Pt 2 PCFS
where Pt is the transition pressure and DGCFS < DV: DPCFS or, DPCFS < DGCFS =DV:
The relation can be illustrated with the example of fayalite, Fe2SiO4.
Burns and Sung (1978) showed that for an experimentally determined phase-
equilibrium boundary for the olivine ! spinel transition in Fe2SiO4 (revised earlier by
Akimoto, 1977) the equation of the univariant curve is:
PðkbarÞ ¼ 34:6 þ 0:025Tð8CÞ
The experimental results reveal that the crystal-field stabilization is less sensitive to
temperature for the models A (DPCFS < 9 GPa) and B (DPCFS < 5 GPa), but is temperature
dependent for the model C (DPCFS is 2– 7.5 GPa). It is also seen that contributions from
346 Chapter 4

crystal-field stabilization in all three models of Fe2SiO4 spinel lower considerably the
observed transition pressure ðPt Þ from the hypothetical crystal-field-free values.

4.4.1.2. Transition-metal compounds: ionic model

In a cubic field, the splitting parameter D ð¼ 10DqÞ and Racah parameter B measure
the strength of the Coulomb and the exchange interaction between d-electrons. The point-
charge model with the simultaneous use of Slater 3d-wave function gives a reasonable
value of D for the first transition-metal ions in cubic crystals. A quantum-mechanical
evaluation of D was first developed by Tanabe and Sugano (1954) with an ionic model. In
that treatment, the non-orthogonality between the d-orbital and ligand orbitals is taken into
account and the electrostatic interactions involve exchange interactions. Also, this ionic
model is restricted to the Heitler –London framework. In the ionic model, the value of
Racah parameter B may be deviated from that of a free ion because of the orthogonality
between the d and ligand orbitals. Covalency or electron delocalization plays an important
role in determining the value of D as well as in giving the transferred hyperfine interaction.
The parameters D and B can be determined from the optical spectra of transition-
metal compounds in the visible region. Pressure experiments show that the pressure
dependence of D upon distance R between the metal ion and the ligand closely
approximates that obtained from the point-charge model, such as:
dðlnDÞ=dðlnRÞ ø 25
With pressure, the value of parameter B decreases. This is interpreted as due to an
increase in covalency with decreasing R. Application of a simple MO theory, in which the
Coulomb and exchange integrals of ligand orbitals are neglected, can explain the reduction
of B due to the increase in covalency.
For the quantum-mechanical treatments of a many-electron system with many
nuclei, the discrete variational ðDVÞ 2 Xa calculation method can be employed. This
method performs numerical integration of many-centre integrals.
The R dependence of B can be derived from the calculated splitting DE2 of the
d-levels caused by the spin polarization. In terms of Racah parameters B and C; the
splitting can be expressed as:
DE2 ¼ md ð7B=2 þ 7C=5Þ
where md is the total number (difference) of spin-up electrons minus that of spin-down.

Energy levels
Distorted sites. In the tetragonally distorted (along the highest symmetry axis), the t2g and
eg orbital groups split, obeying the baricentre rule. The split levels in t2g and eg orbitals are
usually labelled d and a; respectively.
In a site compressed along the tetrad axis, the dxy and dx2 2y2 orbitals are more stable
(lower in energy by lying below the baricentre of t2g and eg levels). Conversely, in a similar
site elongated along that axis, the dxz and dyz orbitals are stabilized by 1/3d and the dzz
orbitals by 1/2a relative to the t2g and eg baricentres, respectively. The corresponding
destabilizations for the dxy and dx2 2y2 orbitals are 2/3d and 1/2a, respectively.
Principles of Techniques 347

The stabilization energies in tetragonally distorted octahedral sites in oxides are

given by Burns (1970, p. 24), which shows that in regular octahedral sites the most stable
ions are with d3, d8 and low-spin d6 configurations. The other ions attain stability by
distortion of the octahedral site (through JT mechanism).

Distortion and JT (JT) ions. Jahn and Teller (1937) state that, if the lowest energy level is
degenerate, the symmetry of the molecules will spontaneously distort to remove the
degeneracy and make one level more stable (i.e., lower in energy) than the other.
Consequently, excited states possessing degenerate electronic configurations will be
distorted. Therefore, the position of a transition-metal ion in a crystal structure is
determined by several factors, one of which is the force of the JT distortion.

Octahedral co-ordination site. Ions in which one or three electrons occupy eg orbitals
such as d4, d9 and low-spin d7 ions are susceptible to JT effect distortion in octahedral sites.
Because of this effect, compounds of Cr2þ, Mn3þ, Ni3þ and Cu2þ are usually distorted
from the type structures shown by other transition-metal compounds. These cations
illustrate that the concept of ionic radius is not a rigorous atomic property. Cu2þ ions occur
in sites distorted from regular octahedral symmetry in such natural compounds as CuO
(tenorite), Cu2(OH)2CO3 (malachite) and dioptase (Cu6Si6O18·6H2O). Likewise, Mn3þ
ions (d4 ions) occur in distorted coordination sites in compounds like g-MnOOH
(manganite).
JT distortion is small in the t2g group compared with that in the eg level. Generally,
the value of energy separation d (in t2g group) is of the order of 10 – 100 cm21, while the
value of a (separation in eg levels) goes as high as 1,000 – 4,000 cm21, when D0 extends
between 10,000 and 20,000 cm21.

Tetragonal coordination site. Electronic configurations predict that no JT distortion of a

tetragonal site is necessary for such low-pain d4 ions. However, ions having d3, d4, d8 and
d9 configurations (e.g., Cr2þ, Mn3þ, Ni3þ and Cu2þ) are more stable in a distorted
tetragonal site. JT distortion of a co-ordination site can be ascertained by such techniques
as X-ray, electron and neutron diffraction, and paramagnetic and nuclear magnetic
resonance measurements.

Jahn – Teller effects. Unusual valencies get stabilized by low-symmetry environments,

particularly when the cation itself is prone to JT effects (e.g., d1 and d9 ions).
JT effects are responsible for axial distortion in d9, d7 (LS) and d4 (HS) ions. Cu2þ
d ion has the configuration (t2g)6(eg)3. One of the eg orbitals contains two electrons and the
9

other has one. These two eg orbitals (viz., dx2 2y2 and dz2 ) are degenerate. The doubly
occupied orbital will experience greater repulsion from the ligands, causing lengthening of
the metal – ligand bonds in the direction of this orbital. Thus, if dz2 orbital is doubly
occupied the two metal – ligand bonds. This lengthening of the M– O bond along the z-axis
will lead to a lowering of the energy of the dz2 orbital. See Fig. 15.5 for the change
occurring in energy levels due to lattice distortion and JT effect. The M – O bonds along z
are longer than the other four.
348 Chapter 4

JT distortion. Diverse behaviour is expected when the polyhedra involve transition

elements, showing JT effect, which is associated with the degeneracy of the electronic
ground states of d9 and high-spin d4 transition-metal cations in holosymmetric (commonly
octahedral) ligand fields. The electronic distortion due to the JT effect, effecting
topologically induced distortion, is found to contribute significantly to the compressional
behaviour of the polyhedra. The influence of pressure on the JT effect can be determined
quantitatively for a coordination, which is essentially free of any type of topologically
induced distortion. The increase of the JT distortion with pressure opposes the
compressional changes induced by normal bond shortening with pressure.

Dynamic and cooperative JT effect (CJTE). If there is a coupling between the electronic
states and vibrational modes, dynamic JT effect occurs. Through the CJTE, the phase
transition is driven by interaction between the electronic states of one of the ions in the
crystals and the phonons. Measurements of heat capacities, elastic constants and other
properties offer clues for the CJTE phenomenon.

4.4.1.3. Pressure on crystal-field parameters

Burns (1987) extracted polyhedral moduli from high-pressure electronic absorption
spectra. The basis for this result is the relation between the octahedral crystal-field
parameter D0 ¼ 10Dq and the mean central-ion ligand distance R of the octahedra as
theoretically derived from the point-charge model of the crystal-field theory:
D0;3d N ¼ 10Dq3d N ¼ 3=5ðZL e2 Þ=Rs kr5 l ¼ constant=R5 ðe:g:; Lever; 1984Þ
In the evaluation of high-pressure spectra, the supposition is that the effective
charge of the ligands ðZ; e2 Þ and the mean value of the fourth power of the radial distance
of the 3d-orbitals from the nucleus, kr4l, are constant (Langer et al., 1997).
The mean linear polyhedral compressibility bi is:
bi ¼ 1=R0 ðDR=DPÞ
and the polyhedral modulus is:
Kpoly ¼ 1=3bi
For evaluating the effect of pressure on the crystal-field splitting D; the first-order
Birch – Murnaghan EOS can be employed as:
" #
3 D 7=5 D 5=3
r ¼ K0 2 ð4-5Þ
2 D0 D0

where K is the bulk modulus of the octahedra. The subscript zero represents the values at
zero pressure and room temperature.

Pressure effects on absorption spectra. High-pressure absorption spectra have measured

some oriented minerals in a DAC using gaskets containing liquid (often methanol) under
80 kbar and polarized light. However, measurements at higher pressures up to 300 kbar
Principles of Techniques 349

have been made using normal (unpolarized) light on minerals such as Mg –olivine
(forsterite), Fe– olivines (fayalites), spinels g-FeSiO4, pyroxenes, spinel, g-NiSiO4, ruby,
piemontite, Fe2þ in periclase, almandine garnet, etc.
Under pressure, there is a tendency for a blue shift of the absorption band position
and also a small increase in intensity. The intense bands in olivine (10,500 – 10,800 Å)
and pyroxene (9,000, 18,500Å) show shifts to higher energies of about 10 – 15 cm21 per
kbar pressure.
The charge-transfer bands, however, do not show pressure effects to a signifi-
cant level.

Pressure effects on covalency. The effect of pressure on the covalencies of Cr3þ –O,
Fe3þ –O and Fe2þ – O bonds in pyroxenes and garnets (natural and synthetic) have
been investigated up to 200 kbar by Abu-eid and Burns (1976). Pressure-induced
increase in covalency is small up to 200 kbar pressure and it manifests as a slight
decrease in the B Racah parameter. But this increase in covalency can hardly have an
appreciable effect on the ionicity of the M –O bond in upper-mantle minerals, where
ionic bonding dominates down to the transition zone. Before any significant change in
the ionic character of the M– O bond at high-pressure depths, structural changes take
place such as phase transitions or disproportionation to oxide components. In the
lower mantle, the chemical bond in the oxides is most likely to be of metallic
character. This metallization is indicated by the increase in the electrical conductivities
of (Mg,Fe)2SiO4 polymorphs at pressure above 100 kbar (Mao, 1973) and the shifts of
their absorption edges to lower energies. At still higher pressures, a disproportionation
of the polymorphs transforms to oxide components.
The magnitude of energy changes in B with pressure corresponds to the shortened
transition-metal ion – ligand distances. Up to 200 kbar pressure, a shortening in the average
Cr3þ –O bond distance by 0.7 –1.0% has been observed by experiment, along with a
decrease in the energy value of B by 1.5– 2% (Abu-eid and Burns, 1976).

4.4.1.4. Racah parameters and band shifts

Racah parameter, B, which is a measure of the inter-electronic repulsion and
exchange interactions, provides a qualitative indication of bond covalency. From
optical spectra, the value of B, derived for a transition-metal ion coordinated to
ligands, is always less than the value obtained from the free gaseous state. The
decreasing Racah B parameters, manifested in the nephelauxitic series, represent the
increasing covalency in ligand – metal bonding. Such a decrease in shown in Fig. 4.8
for B vs. Cr3þ in Al2O3.
Under pressure with decreasing M– O bond distance, the Racah parameter and the
nephelauxitic ratio ðB=B0 Þ decrease (Schmetzer, 1982). Such decreases, though small,
have been observed in a host of minerals: e.g., in uvarovite (Cr – garnet) under pressure
over 20 GPa, in ruby (Cr – corundum) under pressure over 10 GPa (Drickamer and Frank,
1973) and in Mn2þ – oxygen bonded minerals (Fig. 4.9). These relatively small decreases
in Racah B parameters of Cr3þ- and Mn2þ-bearing minerals suggest that, for these
350 Chapter 4

Figure 4.8. Racah parameter B versus pressure for Cr3þ in Al2O3 (Source: Drickamer and Frank, 1973).

compounds, the covalent bonding characters do not change significantly over the pressure
range prevalent in the Earth’s mantle.
Increased pressure on g-phase (ringwoodite) induces a blue shift of the Fe2þ CF
band (11,430 cm21) towards the visible region (Mao and Bell, 1972). Octahedral Cr3þ and
Ni2þ CF bands also show blue shifts (Yagi and Mao, 1977).
However, above 20 GPa, a strong pressure-induced red-shift of O ! Fe absorption
edge obscures the Fe2þ crystal-field band. Such a pressure-induced effect working
similarly in olivine, garnet, magnesio-wüstite, etc., may affect the radiative heat transfer
and electrical conductivity in the Earth’s interior, as discussed elsewhere.
In garnets, however, increased temperature induces small red-shifts and
intensification of spin-forbidden CF peaks located in the visible region, arising from
Fe3þ ion (in andradite) and Mn3þ ion (in spessartine).

4.4.1.5. Examples: Cr3þ-bearing minerals

Langer et al. (1997) studied the high-pressure electronic absorption spectra at room
temperature and at pressures 1024 , P(GPa) , 8 were measured in the spectral range
380 , l(nm) , 780 (26,218 . (cm21) . 12,820) on analysed single-crystal slabs
(, 20 mm thick) of Cr3þ-bearing spinel (I), kyanite (II), corundum (III), pyrope (IV) and
uvarovite (V) using DAC cell technique in combination with single-beam spectrometry.
Ligand-field theoretical evaluation of the spectra yielded the following results:
(i) The octahedral crystal-field parameter, 10DqCr3þ [6]
, linearly shifts on increasing pressure
to higher energies with slopes (d10DqCr3þ [6]
/dP) ðr $ 0:92Þ:
(I) spinel: 103.1 (cm21/GPa)
(II) kyanite: 99.5 (cm21/GPa)
(III) corundum: 104.0 (cm21/GPa)
Principles of Techniques 351

Figure 4.9. The variation of B value with Cr–O distance in a suite of chromium minerals. The B value for
each mineral was computed using the energies of v1 and v2 transitions tabulated by Burns (1975).
The abbreviations stand as follows: epid: epidote; alex: alexandrite; trem: tremolite; tour: tourmaline; Cr-diop:
Cr-diopside; Fo: forsterite; per: periclase; uv: uvarovite; rub: ruby; sp: Spinel and esk: eskolaite (Abu-eid and
Burns, 1976).

(IV) pyrope: 111.7 (cm21/GPa)

(V) uvarovite: 110.3 (cm21/GPa)
(ii) The Racah parameter BCr3þ [6]
, reflecting the covalency of the Cr –O bonds, does not
significantly change with pressure up to 8 GPa but a decrease of BCr3þ [6]
, reflects an
increase in covalency.
The octahedral compression moduli of K of Cr3þ [6] from pressure slopes of
10DqCr3þ [6]
, are:
for
(I) spinel: 312 þ 48 GPa
(II) kyanite: 297 þ 70 GPa
(III) corundum: 298 þ 44 GPa
(IV) pyrope: 275 þ 35 GPa
(V) uvarovite: 257 þ 32 GPa
352 Chapter 4

4.4.1.6. Intervalence charge transfer

In edge- or face-shared coordination polyhedra, M – M intervalence CT is
polarization dependent. Under high pressure, the shortened M –M bond lengths enhance
the transition probability and the intensity increases. However, with increasing
temperature, the intensity of intervalence charge transfer (IVCT) decreases. In general,
energies of homonuclear IVCT transitions (e.g., Fe2þ ! Fe3þ, Ti3þ ! Ti4þ) are relatively
insensitive to P changes, whereas heterogeneous IVCT transition (e.g., Fe2þ ! Ti4) shows
red-shifts to longer wavelengths at high pressures. Evidently, the pressure-induced
variations of CF bands are distinct from the IVCT band.

4.4.1.7. O ! M charge transfer

In a coordination polyhedron, the high-energy UV light induces electron
transfer between oxygens and the cations within the polyhedron. This UV absorption
band extends into the visible region. The energies of O ! M CT in transition metals show
the order:

Cr3þ . Ti3þ . Fe2þ . Fe3þ . Ti4þ ð4-6Þ

Under higher P and T; the absorption band in the UV region shows red-shift with increased
absorption in the visible region.
The O ! M CT for olivines and spinels are discussed below.

Olivines. In the upper mantle at a sufficiently high pressure and temperature

(14 GPa/, 1,0008C), the absorption edge of olivine, (Mg, Fe)2SiO4, moves well into the
infrared. This red-shift by pressure causes an opacity in fayalite above 15 GPa and a rapid
increase in its electrical conductivity. Forsteritic olivine manifests this increase in
electrical conductivity only at high pressures beyond 30 GPa (Mao, 1973).
The Fe2þ (M1) band at 8,000 cm21 (1,200 nm) also shows blue shifts. Some (Abu-
Eid, 1976; Hazen, 1977) interpret this shift as showing that the M1 site in olivine becomes
less distorted (lesser M– O length) at high pressure.
At elevated pressures, all Fe2þ CF bands of olivine intensify and O ! Fe CT
absorption edge shows red-shift, as does the weak, spin-forbidden Fe2þ peaks near
16,260 cm21 (615 nm) and 22,075 cm21 (453 nm).
Under pressure, the UV absorption edge (due to charge-transfer) of iron-bearing
olivine is red-shifted and at sufficiently high pressure (14 GPa) and temperature
(¼ 1,0008C at upper-mantle condition) it moves well into the infrared. In fayalitic olivine
at 15 GPa pressure, this red-shift of the absorption edge causes an optical opacity and high
increase in conductivity. In Fa-poor olivine (Fa18), this property is attained only at a higher
pressure of 30 GPa (Mao, 1973) and beyond.

Spinels. The pressure-induced blue shifts of Fe2þ CF band (11,430 cm21) to the visible
region and red-shift of oxygen ! Fe absorption edge are observed in oxide and silicate
spinels with iron. The strong pressure-induced visible region absorption of Fe2þ in
g-Fe2SiO4 and magnesio-wüstite at 30 GPa suggests that these phases would effectively
Principles of Techniques 353

block the black-body radiation at wavelengths shorter than 1,500 nm (Mao, 1973). Thus, in
the lower mantle, Fe2þ in magnesio-wüstite would effectively block the black-body
radiation in the red and near-infrared regions. This effect, therefore, should influence the
radiative heat transfer, electrical conductivity and other properties of geophysical
significance in the Earth’s interior. More of this aspect has been discussed elsewhere
(Chapter 15).

4.4.1.8. M– M bonding and Cr dimerization

Evidence is observed for the changes of subtle stereochemical features with
pressure, such as the increasing planarity of the fourfold square-planar oxygen
configuration. The electronic absorption bands arise from orbital overlap. The diamagnetic
response would suggest electron pairing from orbital overlap due to metal – metal bonding.
An increase in orbital overlap occurs due to reduction in interatomic distances under
pressure, while temperature enhances the probability for electronic transfer between the
two Cr atoms. All these offer evidence for a weak metal – metal bonding interaction and a
silicate-bridged chromium dimer. Multiple metal – metal bonding and dimerization might
also stabilize other transition-metal species in silicate melt species in silicate melt
structures at high P – T and thus play an important role for geochemical fractionation of
these elements during the evolution of the Earth (Miletich et al., 1999).

4.4.1.9. Crystal-field effect on transition pressure

Crystal-field effects due to unpaired electrons in 3d-transition elements may
influence (lower) the pressure of transition. For example, in the P21 =c ! C2=c transition
in clinopyroxene, the crystal-field effect of Fe2þ or Cr2þ stabilizes HP C2=c
clinopyroxenes with respect to P21 =c structure (at lower pressures). The average cation
size also contributes in delineating the transition. This lowering of transition pressure has
been known more remarkably in the transformation of high-pressure olivine to spinel in
Fe2SiO4, Co2SiO4 and Ni2SiO4 (3 –7 GPa), in great contrast to that for Zn2SiO4 and
Mg2SiO4 (13 and 12 GPa) (Syono et al., 1971). This behaviour is due to the excess crystal-
field stabilization energy (CFSE) for Fe2þ, Co2þ and Ni2þ in the octahedral sites in spinel
and those in olivine (Burns and Sung, 1978). This argument should be valid for other
systems as well. For Fe2þ and Cr2þ containing clinopyroxenes, there will be an additional
contribution to the free energy ðDGCFS Þ associated with the P21 =c to C2=c transition due to
the difference in CFSE of these ions in the octahedral sites of these phases.
The energy contribution DGCFS consists of an enthalpy term DHCFS and an entropy
term TDSCFS (e.g., Burns, 1993). The sign on the enthalpy is determined by the contribution
of CFSE at the octahedral sites, which in the HP C2=c polymorph are significantly smaller
than those in the P21 =c structure ðD / R25 Þ and a positive term arising from the fact that
the M2 sites in C2=c structure are less distorted than those in the P21 =c structure. The
decrease in distortion will, however, also give rise to a counter-balancing positive DSCFS
term arising from an increase in electronic entropy.
In the case of clinopyroxene, a stabilization of the C2=c form by Fe2þ and Cr2þ was
observed by Arlt et al. (1998), who concluded that the volume decrease of the octahedral
site is the dominant factor and leads to a negative sign of DGCFS :
354 Chapter 4

4.4.1.10. Exchange-coupled pair bands

In a host of minerals containing Fe2þ and Fe3þ cations in adjacent polyhedra, there
occur situations where E-vector runs parallel to Fe2þ – Fe3þ pairs. The electronic d– d
transitions of ions and in exchange-coupled pair (ECP) such as Fe2þ – Fe3þ and Fe3þ – Fe2þ
can generate a host of such properties, e.g., pleochroism, magnetism and interesting
electrical behaviour. These electronic d –d transitions of ions in ECP may from two types
of bands. One arises from spin-forbidden transitions, e.g., of Mn2þ –Fe3þ pairs, and the
other originates from spin-allowed transitions of Fe2þ –Fe3þ pairs. However, the latter
transition is distinct from IVCT as there is no electron transfer between the pair nuclei on
the ECP excitation. The ECP bands, in comparison with the corresponding d –d transitions
in isolated ions, are more intense and strongly polarized. As for the band energies, ECP
bands behave as in the usual d –d transition. Both depend on the changes in the M –O
distance and polyhedral forms,- affected by P, T changes (e.g., Burns, 1994). In a centro-
symmetric crystal-field of 3dN-ions, they increase in intensity on raising temperature.
In general, the influence of pressure on the intensity of the ECP bands is much stronger
than the influence of pressure on the intensities of the Fe2þ –Fe3þ IVCT band.
In the absorption spectra, the high-energy band arising from Fe2þ – Fe3þ IVCT is
usually designated as band I and the low-energy band arising from the spin-allowed d– d
transitions (5T2g ! 5Eg transition) of Fe2þ is designated as band II. The other lower energy
band due to second Fe2þ d– d transition is marked as band III.
Bands II and III of Fe2þ d –d transition bands in the spectra of vivianite, phlogopite,
biotite, elbaite and schörl occur when the E-vector is parallel to the possible Fe2þ ! Fe3þ
pairs. The response of these bands to pressure and temperature differs from that of the
isolated Fe2þ d– d transitions band. The energies of IVCT bands in spectra of vivianite and
biotite show opposite pressure-induced shifts, which may be caused by difference in
compressibility of neighbouring Fe2þ –Fe3þ-centred octahedra in silicate (biotite) and
phosphate (vivianite) structures.
Taran et al. (1996) studied the ECP such as Fe2þ –Fe3þ in single crystals of
vivianite, phlogopite, biotite, elbaite and schörl in the pressure range of 1024 # P
(GPa) # 8 and temperature range of 79 # T (K) # 597. The two Fe2þ d – d transition
(manifesting ECP effects) are seen to occur at 11,000 –14,000 cm21 (II) and 8,400 –
9,150 cm21 (III), depending on the host environment. The band energies shift to higher
values with pressure, while the opposite is true for higher temperatures.
In single-crystal spectra of vivianite, the unusually high intensity of both Fe2þ
bands at 12,100 cm21 (band II) and 8,400 cm21 (band III) was suggested to be due to an
exchange interaction of Fe2þ with neighbouring Fe3þ ion. Under high pressure, band II
shows a sharp increase in intensity and shifts by 110 cm21/GPa to higher energy.
On the other hand, the Fe2þ – Fe3þ – IVCT band (band I) at 16,000 cm21 is hardly
influenced by pressure and shifts to lower energy by , 60 cm21/GPa. This behaviour is
typical of IVCT (Abu-Eid, 1974; Mao, 1974). At high pressure, a shoulder near
10,000 cm21 is observed, which is regarded as band III, the low-energy component of the
d –d transition of Fe2þ occurs at 8,400 cm21 at low pressure. This implies a high energy
shift amounting , 320 cm21/GPa. The low symmetry splitting of the 5Eg excited state of
Principles of Techniques 355

Fe2þ would then decrease on pressure. This phenomenon is observed in many Fe2þ bearing
minerals (Abu-Eid, 1974; Langer, 1990).
Under P and T, the ECP bands II and III differ greatly from ordinary d –d bands.
The latter do not change in intensity under pressure (e.g., Burns, 1994). The bands II and
III of elbaite (turmaline) in p-polarized spectra are attributed to 5T2g – 5Eg electronic
transition of isolated Fe2þ ions (Mattson and Rossman, 1987). The intensification of these
bands on heating is typical of 3dn ions in centro-symmetric sites. An increase of ECP-band
intensity in s-polarized spectra of tourmaline (elbaite) on cooling is due to thermal
contraction of the structure resulting in an increasing overlap of the metal –metal orbitals.
Alternatively, it can be suggested that the intensification of the ECP bands on cooling is
due to a decrease in population of excited vibronic states of the electronic ground state
(Taran et al., 1996).

4.4.1.11. CFSE and elastic property change: Fe2þ

Weidner et al. (1982) have noted that the elastic properties of minerals are
particularly sensitive to the replacement of Mg2þ by 3d transition-element ions. Cations
which show no crystal-field stabilization energy at octahedral or tetrahedral sites, such as
3d5 or 3d10 ions (e.g., Mn2þ or Fe3þ, etc.) replacing Mg2þ, allow the volume-
compressibility value to fall in the trend allowed by simple mixing behaviour. But when
Fe2þ substitutes Mg2þ, the gain in crystal-field stabilization energy of Fe2þ makes the
phase anomalously incompressible. Such stabilizations would be extremely sensitive to the
local (inhomogeneous) distortions of the coordination polyhedra, which could conceivably
influence the bulk pressure-dependent behaviour.

4.4.2. Volume compressibility and crystal-field splitting

The bulk modulus, K, of a mineral and the pressure dependence of its volume are
related as
K ¼ 2VðdP=dVÞ ¼ 1=b
where b is the volume compressibility.
At elevated pressures, the study of crystal-field spectra involves the inverse fifth-
power relationship which is expressed as:

DP =D0 ¼ ½R0 =RP 25

where D0 ; DP and R0 ; RP are the crystal-field splittings and the average cation – oxygen
distances for a transition-metal ion in the mineral structure at ambient and high pressures,
respectively.
Since R / 1=V 3 ; where V is volume and D / V 25=3 ; lnD ¼ 25=3lnV:
By differentiation,
dD=D ¼ ð25=3ÞdV=V ¼ ð25=3dÞr=r ¼ ð25=3ÞdP=K ð4-7Þ
where r is the density and K the bulk modulus or inverse of compressibility of the
356 Chapter 4

coordination polyhedron. K is related to the compressibility, b; of the coordination

polyhedron by:

K ¼ 1=b ¼ 2VðdP=VÞ or dD=dP ¼ ð25=3ÞD=K ð4-8Þ

Thus, the D values from high-pressure crystal-field measurement will help measure
the site compressibility or polyhedral bulk moduli of transition-metal-bearing minerals.
It should be remembered here that increased pressure causes an increase in D by a decrease
in M –O bond distance. As a result, energy separations between 3d orbitals increase with
rising pressure and the absorption bands are expected to move to shorter wavelengths
(higher energies), i.e., “blue shifted”.
The spectrally derived bulk modulus, K s ; is related to the crystal-field splitting, D;
as:

K s ¼ 5=3DðdP=dDÞ ð4-9Þ

From the above relations and the first-order EOS of Birch –Murnaghan the K0s can
be determined as:

K0s ¼ 2=3P½ðDp =D0 Þ7=5 2 ðDp =D0 Þ21 ð4-10Þ

The value of the polyhedral bulk modulus, K0s ; can be determined from the spectral
measurement of the crystal-field splitting parameters, D0 and Dp ; obtained at 1 bar and
elevated pressure, respectively.
A good agreement is observed between the polyhedral bulk moduli determined by
spectroscopy ðK0s Þ and by high-pressure XRD ðK0x Þ: The larger the coordination number of
a site, the greater its compressibility (Hazen and Finger, 1982). Again, sites having cations
of high CFSF values also show lower compressibility, e.g., [CrO6] octahedron.
When the transition-metal ions occupy regular or slightly distorted octahedral,
tetrahedral or cubic coordination polyhedra, the spectral method for determining the
polyhedral bulk moduli become best suited. For this reason, studies on the phases having
transition elements, such as garnet, spinel, perovskite and periclase, become fairly
accurate. Fortuitously, these are the major phases in the lower mantle. In contrast, the
values of the bulk moduli of polyhedra in olivines and pyroxenes are less accurate because
of their strong anisotropic character.

4.4.3. Vibrational spectroscopy

Spectroscopic techniques provide crucial information, often unique, on the

properties of materials at ultra-high pressures. Vibrational spectroscopy provides a
straightforward means for identifying phase transitions, as well as the mechanisms of such
transformations (e.g., soft modes and order parameters). Vibrational data can be used for
modelling the thermodynamic properties of such minerals at high P – T corresponding to
the deep Earth.
Principles of Techniques 357

Vibrational infrared spectra under high pressure reflect changes in the bonding
properties of crystals, glasses and melts. They provide information on crystal symmetry,
elucidate phase transitions, including the mechanisms involved, and directly determine the
lattice dynamical variables, which are important for calculating thermodynamic variables
(McMillan et al., 1989).
Infrared measurements may provide information on pressure-induced changes in
electronic excitations, including crystal-field, charge transfer and excitonic spectra of
insulators and semiconductors, inter- and intra-band transitions in metals and novel
transitions such as pressure-induced metallization (Hemley and Mao, 1997).
IR and Raman spectroscopies have been successfully employed for studying phase
transitions, molecular orientational ordering, strength of intermolecular interactions,
crystal structure and charge transfer in hydrogen. General reviews of the applications of
Raman and IR spectroscopy to Earth and planetary sciences are available (e.g., Sharma,
1990; McMillan et al., 1996). Excellent reviews on the theoretical techniques and
vibrational dynamics of mantle minerals are also available (e.g., Cohen, 1994;
Bukowinski, 1994).
It is possible to model the elastic properties of minerals as though they consisted of
strong springs between the atoms in the molecules and weak springs between different
molecules (see, e.g., Fig. 1.5b: Hemley & Mao, 1998). Raman and Brillouin are inelastic
light-scattering techniques that measure transitions to an excited state from the
ground state (Stokes scattering) and to the ground state from a thermally excited state
(anti-Stokes scattering). Raman spectroscopy probes the strong springs (optical modes)
and Brillouin spectroscopy the weak springs (acoustic modes). A stretching mode is one in
which the atoms vibrate in the direction of the line joining the atoms in the molecule and it
is thus a measure of the strong-spring frequency.
Raman experiments under pressure performed on all Group IV (elemental) and
III –V (compound) semiconductors showed a softening of shear waves (as in GaAs), which
is a precursor of the semiconductor to metal transition. With pressure, the charge transfer
from the cation to the anion also changes.
Vibrational spectroscopy not only provides means for identification and structural
characterization of minerals but also helps in our understanding of the thermodynamic
properties of minerals calculated via empirical or ab initio methods (which are used to
predict mineral stability and properties under various P – T conditions). Thus, one can infer
the thermodynamic properties including the EOS at high pressure and high temperature
from HP and HT infrared and Raman spectra.
The crystal (or molecule) exhibits several (say, N) modes of vibration, which can be
identified by a subscript i; which can take values from 1 to N:
p
n1 ¼ 1=2pc Ki =mi ð4-11Þ
where Ki and mi are the effective force constant and effective mass, respectively. The force
constant associated with a chemical bond correlates very closely with the length of the
bond. On first order-phase transition, the geometrics of the crystal and its vibrations
change. First-order phase transitions can be located by infrared or Raman spectroscopy
because (i) a discontinuous volume change should cause discontinuous frequency shifts
358 Chapter 4

and (ii) the accompanying symmetry change cause dramatic changes in band intensities
(even appear or disappear).
Vibrational frequencies typically shift at rates of about 0.5 cm21 (kbar)21
and, with care, band-centre frequencies can be located with ^ 0.2 cm21. Thus, a first-
order transition with DV=V as small as 0.03% (i.e., Da=a of about 0.01%, equivalent
to X-ray measurements accurate to 1 part in 104) should be clearly detectable in
this way.
Second- (or higher) order transitions do not produce such dramatic changes in
vibrational spectra. Discontinuous changes in V (first order) cause discontinuous
changes in n (and possibly intensity I) and, therefore, a discontinuous change in dV=dP
(second order) would be expected to produce a discontinuous change in dn=dP; and
possibly dI=dP:

4.4.3.1. Soft modes

Each vibrational mode of a crystal structure is associated with a specific
periodic distortion of the structure. When a high-temperature symmetric phase is
cooled, the frequency of the “soft mode” decreases. When it becomes zero, the
structure can hardly bear the distortion and transforms to a lower symmetry phase. The
word “soft” denotes crystal to yield to the displacements of atoms. Thus, in the high –
low transition of quartz, softening occurs in the normal lattice mode, occurring at
208 cm21 in the high-temperature form. This soft mode is Raman active. (Note: In a
zone-centre transition the lattice mode can be measured by Raman or IR spectroscopy,
although symmetry-dependent selection rules would define which modes are active or
inactive.)
The soft-mode frequency reaches zero at the Brillouin zone boundary, at points
with the appearance of new reflections in the diffraction pattern of low-temperature form.
Soft modes may be optic modes or acoustic modes.
Crystal-structure change through mode softening can be evaluated by the change in
some order parameter (e.g., Q) as a function of temperature. In quartz, for example, the
order parameter is related to the tilting angle h of SiO4 tetrahedra while, in the perovskite
structure, it is the angle of rotation w of the octahedra.
In the widely studied soft mode in displacive ferroelectric crystals, the frequency
tends to zero as the transition is approached. In passing through the transition, dn=dP
changes from negative to positive and the frequency rises again. By the nature of those
vibrations, the frequency is spectroscopically inactive in the second phase and is not,
therefore, open to study by these techniques beyond the transition. However, even when no
soft mode is accessible to spectroscopic study, there is still a wealth of information in the
vibrational spectra of the non-soft modes.
Some spectral features (in particular, the lattice side-band structure) approximate to
the “density-of-vibrational states”. These are used for deducing thermodynamic
parameters under high-pressure conditions. However, lattice side bands only approximate
to DOS function.
Principles of Techniques 359

Close to a phase transition, where an instability is imminent, there will be some

vibrational modes (usually shear-wave-type modes) with anomalously low frequencies at
or near some point in the zone boundary.

4.4.3.2. Pressure relationship

Pressure can be written as:
PðV; TÞ ¼ Pc ðVÞ þ Pvib ðV; TÞ þ Pel ðV; TÞ ð4-12Þ

where Pc ðVÞ is the 0 K pressure and Pvib and Pel are the vibrational and electron
contributions to the pressure, respectively. Neglecting the electron contributions, the above
can be modified as:
PðV; TÞ ¼ Pc ðVÞ þ Pth ðV; TÞ

where Pc is the reference curve which could be 0 K isotherm, representing rather the
determined room temperature isotherm, and Pth ðV; TÞ represents the difference in pressure
between the reference curve and the high P – T state.
For studies on displacive phase transitions, the pressure-induced ðdn=dP , 0Þ and
the temperature-induced soft modes ðdn=dT , 0Þ are studied, as exemplified by the types
of perovskite structure. Pressure-induced mode softening has been identified by Raman
study for the transition of stishovite ! CaCl2 structure at ,50 GPa (e.g., Kingma et al.,
1995).
Micro-Raman has high spatial resolution (, 1 mm or the diffraction limit of visible
light). It has proved useful in mineral identification in thin section.
Vibrational spectroscopy has also revealed the presence of Si – O –Si linkage
between tetrahedral SiO4 groups in b-Mg2SiO4 (McMillan and Akaogi, 1987) and ring
structures in coesite (Sharma et al., 1981). Vibrational data have proved useful in
calculating the specific heat and vibrational entropy of a phase (Lu et al., 1994). Perovskite
with the ideal cubic structure ðFm3mÞ would have three IR-active modes and no Raman
spectrum. From symmetry analysis:
Gvib ¼ 3F1u ðIRÞ þ F2u ðinactiveÞ ð4-13Þ

For cubic MgSiO3 perovskite, these modes are expected to occur in regions (e.g.,
Lu et al., 1994):
(i) 750 –1,000 cm21: asymmetric stretching of octahedral groups (SiO6),
(ii) 500 –700 cm21: deformation vibration of the octahedra, and
(iii) 250 –400 cm21: transitional motions of alkaline earth cation in the dodecahedral site.
With reduction of symmetry to orthorhombic Pbnm ðZ ¼ 4Þ; a splitting of IR-active
modes occurs with first-order Raman spectrum (e.g., Wolf and Bukowinski, 1987) (see also
Section 10.2.2.5):

Gvib ¼ 7Ag þ ðRÞ þ 7B1g ðRÞ þ 5B2g ðRÞ þ 5B3g ðRÞ þ 8Au ðinactiveÞ

þ 7B1u ðIRÞ þ 9B2u ðIRÞ þ 9B3u ðIRÞ ð4-14Þ

360 Chapter 4

Raman spectroscopy has helped identification of coesite (within pyrope grains) in

meta-sedimentary rocks from subducted continental crust and thus constrained the P – T
history of the assemblage. Micro-Raman has been used to identify coesite in eclogites from
central China, diamond inclusions in zircons and garnets from high grade (. 100 km)
metamorphic rocks and exsolution of garnet from clinopyroxene in high-pressure rocks.

4.4.3.3. Infrared
Because of the significant enhancement in ability to probe microscopic samples
provided by synchrotron sources (National Synchrotron Light Source (NSLS), for
example, at Brookhaven National Laboratory), studies under extreme pressures have
become possible. Synchrotron radiation sources have a very smooth and broad spectral
range extending from the hard X-ray region to very long wavelengths. The infrared region
can be measured with FTIR techniques. As discussed in an earlier section, when an FTIR
interferometer is coupled with special microscopes for high-pressure cells, a sensitivity up
to five orders of magnitude is gained relative to a grating system commonly used for HP IR
measurements. Again, in this system, the picosecond time structure provides important
advantages.
This facility helped in the discovery of materials under ultra-high-pressure, such as
the striking enhancement of the vibrational intensity in hydrogen at 150 GPa and the ionic
symmetric ice structure (with symmetric hydrogen bond) at 120 GPa. (Note: Above
100 GPa, however, diamond anvils usually break while releasing the pressure; special care
has to be taken to avoid it.)
IR spectroscopy helps study the fluid inclusions in diamonds and thus discover the
volatile content of the mantle. The OH sites in hydrous phases are characterized by Raman
and IR spectroscopy. In high-pressure hydrous phases, hydrogen bonding stabilizes the
structural framework (Williams, 1992).
Ultra-fast mid-infrared spectroscopy (at wavelengths from 2 to 10 mm) has gained
in popularity over the past 15 years because of its sensitivity to vibrational processes and
the orientation of molecules (see Seilmeir and Kaiser, 1993). A time-dependent infrared
spectrum of a sample provides detailed information on molecular vibrational energy flow,
conformational rearrangements, chemical reaction intermediates and transitory tertiary
structures of the system. IR pulses of picoseconds (1 ps ¼ 10212 s) to femtoseconds
(1 fs ¼ 10215 s) can be used to probe the molecular absorptions originating from transient
intermediates and stable products.

IR band-widths. With regard to band-widths the lattice and O – H stretch vibrations

display contrasting behaviour. The pressure-increased broadening of the FWHM of the A1g
peak is intrinsic and cannot be attributed to pressure gradients. The Eg lattice vibration and
the 360-cm21 mode display little change in FWHM with pressure. The FWHM of the O –H
stretch vibration broadens with pressure at a rate comparable with the lattice A1g mode.
However, this broadening is largely intrinsic and only a portion is due to non-
hydrostatic pressure distribution. The broadening of the band-width for the O –H stretch
vibration under non-hydrostatic conditions is similar to that observed for the infrared
Principles of Techniques 361

Figure 4.10. (a) Infrared spectra in the OH ¼ region for a natural single crystal of grossular garnet
(Ca3Al2Si3O12), and at various pressures. (b) the frequency change in relation to pressure derivative of mode
frequency (Hemley et al., 1998).

active vibration (, 3 cm21/GPa) (Kruger et al., 1989). The increased width is attributed to
increased anharmonic behaviour.
To illustrate infrared absorption in the OH region (i.e., around 3,600 cm21),
the spectra of grossular (Ca3Al2Si3O12) under different pressures are shown in Fig. 4.10a.
362 Chapter 4

The change in frequency with pressure ðdy =dPÞ shows a linear relationship
(Fig. 4.10b).
A recent study on magnon excitations in Sr2CuCl2O2 (Struzhkin et al., 1998), a
perovskite-type material, has provided unique information on magnetic excitations. The
pressure dependence of IR and Raman modes can be used to distinguish between magnon
and electronic excitations.

4.4.3.4. Raman spectroscopy

Unlike the fluorescence peaks, which appear at frequencies n ¼ DE indepen-
dent of the excitation laser frequency, Raman peaks appear at frequencies shifted
from laser frequency, nL ; by subtracting or adding DE; i.e., n ¼ nL 2 DE (Stokes),
or n ¼ nL þ DE (anti-Stokes). The symmetry of the system governs the selection rules,
ollowing which the Raman peaks appear. Change of bonding or electronic energy
evels affect the Raman peaks to shift.
For minerals, the most reliable detection of phase transitions is by IR and
Raman spectroscopy. Raman is more advantageous because of its higher spectral
resolution. Lattice properties of solids are best studied by Raman spectroscopy.
The spectra can be obtained on temperature-quenched samples while the sample is
under pressure.
Sharp Raman bands can be measured within seconds on micron-sized crystals.
From the spectral shifts of Raman bands with pressure, a number of important thermo-
dynamic properties, such as entropies, specific heats and Grüneisen parameters, and
even phase boundaries can be derived with high accuracy (see, for review, Chopelas,
1999).
The principles of Raman spectroscopy and related instruments are available
from McMillan and Hofmeister (1988). A schematic diagram of a confocal set-up
installed on a micro-Raman spectrometer (Dilor XY) was shown earlier in
Fig. 4.4d. Raman spectroscopy is a useful non-destructive tool for characterizing
mineral phases or fluid inclusions (McMillan, 1989), even on conventional
petrographic thin section (e.g., Gillet and Goffé, 1988). Raman spectroscopy is one
of the easiest techniques to set up for studying minerals, glasses and melts under
high pressure and temperature. McMillan and Wolf (1995) present an outline of
available Raman and IR data on silicate melts and glasses. The high-pressure
vibrational data of mineral phases relevant to Earth’s deep mantle are also available
(McMillan et al., 1996). Raman spectra of some mantle minerals are shown in
Fig. 4.11.
However, Raman data have so far been obtained up to 2,000 K and only a few
with simultaneous high temperature and pressure have been obtained by laser heating or
resistive heating in a DAC. Raman spectroscopy can provide information on important
types of transitions, such as crystal –crystal, crystal – melt, crystal – amorphous phase and
glass –liquid transitions.
Principles of Techniques 363

Some sources for Raman data of major mantle forming minerals or their analogues
as listed below:

Authors

a-Mg2SiO4 Gillet et al. (1991), Fiquest et al. (1992)

b-Mg2SiO4 Reynard et al. (1996)
MgSiO3 – ilmenite Reynard and Rubie (1996)
Garnets Gillet et al. (1992)
SiO2 polymorphs Gillet et al. (1990), Catex and Madon (1995)
Carbonates Gillet et al. (1993)

Characteristic Raman spectra of some minerals are presented in Fig. 4.11.

Principle. In its principle, an incident photon of energy Ep collides with a quantum of

lattice vibration of energy E1 and then enters the detector with energy Ed ; whence:
Ed ¼ E p ^ E 1
In the system, besides the energy conservation, the momentum is also conserved.
This is expressed in terms of phonon wave vector q ¼ 2=l; where l is the wavelength. q
plays the role of “quasi-momentum” and is conserved according to:
k2 ¼ k1 ^ q
where k is the wave-vector of light.
In the above equation, the ^ signs correspond to the absorption of a phonon and
creation of a phonon, respectively.
The lattice-mode velocities are given by the slopes of the curves and a typical
value is 103 m s21 for the acoustic modes. Since the speed of light is 3 £ 108 m s21,
the curve of EðkÞ for light is nearly a vertical line. Accordingly, the optical modes are
probed near k ¼ 0: The first-order interactions occur with multiple-phonon creation or
absorption, while the second-order corresponds to two-phonon interactions, etc. These
higher order effects allow other phonon states to be probed. Not all modes can be probed
because there are other restrictions on the interactions, which can be deduced from
symmetry considerations.
The incident beam is typically a laser and the Raman lines occur as satellite
peaks on both sides of the laser line. Since these peaks are relatively weak, and the
phonon energy is typically 0.1 eV, a double or triple spectrometer is used in an
arrangement such as that shown in Fig. 4.4d. A back-scattering geometry is most
frequently used.
Amongst the new generation Raman spectrometers are the micro-Raman
spectrometers. In these, the focusing of the excitation laser beam on the sample and the
collection of the Raman signal are achieved with the objective of an optical microscope
(Fig. 4.15; Gillet, 1996). Raman spectra can be obtained from a small sample size of a few
microns. Insertion of a confocal system in the spectrometer’s optics helps to discriminate
the Raman signal from thermal emissions or background fluorescence (from diamonds).
364 Chapter 4

Figure 4.11. Raman lines for common mantle minerals (from Bertka and Fei, 1997).

A charge-coupled device (CCD) can improve the signal/noise ratio. Small furnaces and
DAC can be adapted to micro-Raman spectrometers. With the latest generation of
furnaces, temperatures up to 2,700 K can be obtained routinely and the set-up can be
adapted on either synchrotron X-ray, Raman or Brillouin spectrometers (e.g., Richet et al.,
1993; McMillan and Wolf, 1995).
The observed intensities of high-temperature spectra are also corrected for the
temperature and frequency dependence of the first-order Raman scattering (Long, 1977) as
well as for second-order Raman scattering, which is intense at high temperature due to
increased multiphonon interactions (e.g., Gillet, 1996).
Principles of Techniques 365

Use of pulsed laser sources (e.g., YAG laser) coupled with chopped detectors
greatly improves the Raman signal against the noise of the thermal radiation from the
heated samples. However, some problems remain with the laser-heating technique using
either YAG or CO2 lasers. Accurate temperature measurements are far from achievable
and temperature only to the extent of , 10% error can be obtained (e.g., Boehler and
Chopelas, 1992).
If the thermal expansion during heating is total, then the thermal pressure is
maximum and is given by:

Pth ¼ f aKdT

where a is the coefficient of thermal expansion of the sample, K its thermal bulk modulus
and T the temperature.
The behaviour of Raman bands at pressure is expected to parallel that of the IR
bands since both sets are derived from the same atomic vibrations. Hofmeister et al. (1989)
presented the first far-IR (370 – 80 cm21) spectra of material above 60 kbar. The innova-
tions for the experiments were also detailed.

Anharmonic-mode parameters and intrinsic anharmonicity. The data obtained from IR

and Raman spectroscopy have been used to construct and constrain models for the heat
capacity and vibrational entropy of minerals, including high-pressure mantle phases
(e.g., Gillet et al., 1993a,b; Chopelas et al., 1994). These calculations are generally carried
out within the quasi-harmonic approximation.
The effects of intrinsic mode anharmonicity, i.e., the variation in mode frequencies
with temperature at constant volume, become important at high temperatures (e.g., Gillet
et al., 1991). Mode anharmonicity is evaluated by separately measuring vibrational-mode
frequencies at high temperature and at high pressure.
The values of the anharmonic parameters are mostly negative. For low-wave-
number modes (typically lattice modes), the anharmonic parameters, in general, have
higher absolute values than for high-wave-number modes (typically internal modes of ring
molecular groups such as SiOþ þ
4 or CO3 in silicates or carbonates; Gillet et al., 1993c). Soft
modes have anharmonic parameters (e.g., in a-quartz; Castex and Madon, 1995).
The change of vibrational frequencies with pressure is used to evaluate band
assignments based on 1 atm spectra. Mode Grüneisen parameters ðg ; dlnni =dlnVÞ and
their derivatives ðqi ; dlngi =dlnnÞ are computed. The initial values for the first
mode Grüneisen parameter range from 0.10 to 2.0, such that far-IR modes
(translations or rotations) generally have values larger than unity, whereas mid-IR
modes (internal vibrations) all have values less than unity.
From the observed frequency shifts with pressure and temperature, the isothermal
Gruneisen ðgi T Þ and isobaric anharmonic mode ðliP Þ parameters are calculated (Gillet
et al., 1989) as

gi T ¼ KT ðdlnni =dPÞT

giP ¼ 1=aðdlnni =dTÞP

366 Chapter 4

where ni is the vibrational frequency of the ith mode, KT the isothermal bulk modulus, and
a is the thermal expansivity.
The average Gruneisen parameters ðgÞ is given by:
X X
kgl ¼ giT Cvi = Cvi

The values of the isothermal and isobaric anharmonic parameters can be used to
obtain the intrinsic anharmonic mode parameter ai (Gillet et al., 1989)

ai ¼ ðdlnni =dTÞV ¼ aKT ðdlnni =dPÞT 2 ðdlnni =dTÞP ¼ aðgiT 2 giP Þ ð4-15Þ

In a quasi-harmonic approximation, pressure and temperature affect the vibration

frequencies only through the volume changes and in that case giT K ¼ giP : A given model
is anharmonic only if the parameter ai ¼ ðdlnni =dTÞV is different from zero.
Example: The mean intrinsic anharmonic parameter for majorite garnet is similar to
other garnets but is higher than that of an ilmenite-type structure. The effects of an intrinsic
anharmonic parameter on the position of the ilmenite þ majorite equilibrium can be
estimated using the thermodynamic database of Fei et al. (1990) and correcting the
enthalpies and entropies for the anharmonic contributions.
The occurrence of the first-order Raman spectrum results from the descent in
symmetry associated with the Brillouin zone folding arising from R- and M-point
instabilities within the cubic aristotype (e.g., Bukowinski et al., 1986). These modes are
also particularly sensitive to potential symmetry changes in the perovskite structure with P
and T (McMillan et al., 1996). The lowest lying modes in the Raman spectrum are
associated with coupled rotations of the linked SiO6 octahedra and these are expected to be
extremely sensitive to the degree of orthorhombic distortion of the perovskite structure; in
particular, they should soften if MgSiO3 perovskite undergoes a displacive phase transition
to tetragonal or cubic structures at high temperature (e.g., Hemley et al., 1989; McMillan
et al., 1996).

Si – O stretching mode. The 410 cm21 mode is the symmetric Si – O stretching mode,
ns (Si – O) (Sharma et al., 1981). On increasing pressure, Palmer et al. (1994) observed a
steep increase in ns(Si – O), which coincides with a large decrease in the Si – O – Si bond
angle (, 68 over 1 GPa interval). The closure of Si –O – Si angle can be correlated with
only slight increase in Si –O – Si bond distances caused by increased ionic repulsion as the
angle is forced to close.

4.4.3.5. Second-order Raman scattering: disordering

The basic signature in Raman spectrum of amorphization by loss of long-range
order is the vanishing of the external vibrational modes through the breakdown of the
q < 0 selection rule. This results in manifesting the DOS through the second-order Raman
scattering. This provides complementary information of the disorder in relatively more
rigid units in the structure. However, it should be noted that vanishing or broadening of the
modes may not necessarily correlate with the loss of long-range order.
Principles of Techniques 367

4.4.3.6. Non-linear optical methods

By employing non-linear optical techniques, one can perform three-photon
excitation of valence electrons to the conduction band or excitonic states in the vicinity
of the conduction band, which is above the one-photon threshold of the anvil.
In this class of experiments are the four-wave mixing experiments, such as coherent
anti-Stokes Raman scattering. These rely on the third-order susceptibility and are enhanced
when there are two-photon electronic resonances. Only a few HP studies have been
reported to date (e.g., Lipp and Daniels, 1991).

4.4.3.7. Brillouin scattering

Brillouin scattering is the interaction of light (photons) with thermally excited long-
wavelength lattice vibrations (phonons) in a crystal. Propagating acoustic waves produce
fluctuations in the refractive index from which light is scattered. Since the fluctuations are
moving at acoustic velocity, the scattered light is shifted in frequency by the Doppler
effect. Brillouin spectroscopy is similar to the more familiar Raman spectroscopy except
that the frequency shifts in Brillouin scattering is very small (, 1 –2 cm21) and the signal
intensity is weaker. The experimental challenge is to extract the weak Brillouin signal from
the intense, elastically scattered light.
Brillouin involves the measurement of acoustic phonons. In metals, the
measurements correspond to surface modes or plasmons. In transparent minerals, the
pressure dependence of the refractive index (dielectric function) can be investigated using
Brillouin scattering. Rayleigh scattering is a quasi-elastic scattering spectroscopy.
A Brillouin spectrum appears similar to a Raman one but the side peaks lie closer to
the laser line since the acoustic-mode energies are smaller (of the order of 10 meV). For
high resolution, multiple-pass interferometry is used. To make the Brillouin shifts
independent of refractive index in a DAC, the experiment is scattered in a forward-
scattering geometry, when the following relation holds:
p
Us ¼ Dnl0 = 2

where Us is the acoustic wave speed, Dn the change in the frequency of the light and l0 the
incident wavelength.
In the DAC, Brillouin scattering (Zha et al., 1993) and impulsive stimulated
scattering (Zang et al., 1993) have been used to obtain single-crystal elastic constants to
pressures as high as 24 GPa.
The Brillouin data can be analysed to give the bulk modulus approximately and can
be integrated to give the density as a function of pressure:
ð PB
rðPB Þ 2 rðPA Þ < ðCP =CV ÞU 21 dP
PA

This equation gives an approximation but Brillouin data probably give the most
reliable estimate of the volume.
368 Chapter 4

A coupling of the DAC with the Brillouin scattering technique allows measurement
of the entire set of elastic moduli for single crystals to a moderate pressure. Such
measurements on elastic moduli are important for the following reasons (Zha et al., 1998):
First, with the full complement of elastic moduli, one can evaluate the bulk and
shear moduli, EOS and other thermodynamic parameters.
Second, P-induced changes in elastic moduli can provide information on structure-
deformation mechanisms and also on how these changes correlate with the overall
mechanical properties of the materials.
Third, the pressure dependence of elastic moduli can reveal mechanisms of phase
transitions. As a precursor to phase transitions, softening of shear moduli provide evidence
for incipient lattice instabilities.
Fourth, elastic moduli measurements combined with XRD under high pressure can
be used for absolute pressure determination.
The extremely low background signal in light-scattering experiments using ultra-
pure synthetic single-crystal diamond anvils (Goncharov et al., 1998) is enabling the
extension of these light-scattering studies of electronic and magnetic excitations to a wider
class of materials, including opaque minerals and metals. Measurements of the two-
magnon excitations in Fe2O3 and NiO by high-pressure Raman scattering constrain the
magnetic properties of these materials at high density.

Brillouin spectroscopy: geometry. Using Brillouin scattering, acoustic velocities can be

determined from the frequency shift of light scattered from thermally generated acoustic
waves. Brillouin spectroscopy has been the only technique for elasticity-tensor
determination up to moderate pressures (e.g., , 25 GPa; Zha et al. 1993; Shimizu et al.,
1995). However, the technique has been successfully applied only to optically transparent
single crystals in hydrostatic media and not to metals and semiconductors at these
pressures, and its utility in more extreme conditions has not been demonstrated.
For Brillouin scattering, light from a frequency-stabilized laser is directed into the
sample through one of the diamond anvils. The scattered radiation passes through the
other anvil in a symmetric arrangement. A DAC and a large optical opening are used
in these experiments. The Brillouin scattered light is collected, spatially filtered and
passed through a tandem interferometer (e.g., Fabry – Perot), which eliminates the
overlap of successive interference orders. The frequency spectrum is recorded with a multi-
channel scaler. (Note: Further experimental details can be found in Zha et al. (1993,
1996).)
In the platelet-scattering geometry, light enters through one sample face and exits
through another parallel face. The acoustic-wave propagation direction is perpendicular to
the axis of the cell and co-planar with incident and scattered light. The equation relating the
acoustic velocity, V; to the measured frequency shift for this geometry is

V ¼ dnl0 ð2sinuÞ21

where l0 is the incident laser wavelength and u is the angle between incident scattered
light and the DAC axis at the outer diamond surface. The refractive index is not needed to
Principles of Techniques 369

determine the velocity in this case. By rotating the diamond cell around its axis, the
acoustic-velocity distribution within the sample plane can be completely characterized
(Zha et al., 1998).
The propagation of acoustic waves in anisotropic solids is governed by Christoffel’s
equation (Auld, 1973). Using Cardon’s solution of the cubic equations, this equation can
be written as follows (Rokhlin and Wang, 1992):
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2p c þ 2pj
rVj2 ¼ 2a=3 þ 2 cos j ¼ 0; 1; 2 ð4-16Þ
3 3

where r is the density, Vj are the acoustic velocities and the other parameters are

c ¼ arc cos{ 2 q=½2ðp=3Þ3=2 }

p ¼ a2 =3 2 b
q ¼ c 2 ab=3 þ 2ða=3Þ3
a ¼ 2Gij
b ¼ 2ðG212 þ G213 þ G223 2 G11 G22 2 G11 G33 2 G22 G33 Þ
c ¼ 2ðG11 G22 G33 þ 2G12 G13 G23 2 G11 G223 2 G22 G213 2 G33 G212 Þ
Gim ¼ Cijlm nj nl
Summation over repeated indices is implied; n ! [vectors n11 n21 n31 · · ·] specifies
the acoustic-wave propagation direction. Cijlm is the tensor of elastic constants which can
also be written in the contracted Voigt notations Cij ; where i; j ¼ 1; 2; …6 (Nye, 1985).
The established velocity averaging techniques (Voigt –Reuss –Hill) are used to
compute bounds on the elastic properties of a randomly oriented polycrystalline aggregate
from the single-crystal properties.
A pressure – volume EOS can be constructed from the Brillouin data by correcting
the bulk modulus and its pressure derivatives from adiabatic to isothermal conditions.

4.4.4. Ultrasound spectroscopy

In a crystal acoustic wave, velocities can be obtained by Brillouin spectroscopy, in

which an optical beam is scattered by an acoustic wave and the Doppler shift of the laser
frequency reveals the wave velocity. For Brillouin scattering, the crystal needs to be
colourless and unstrained. Recent breakthroughs have extended acoustic interferometry to
high P – I conditions in malti-anvil devices (e.g., Chen et al., 1998; Li et al., 1998;
Sinelnikov et al., 1998). Unlike Raman shifts, which are usually tens to thousands of wave
numbers (cm21) from nL, Brillouin shifts are usually less than 1 (cm21).
By measuring velocities of a primary (longitudinal) wave, Vp ; and two secondary
(shear) waves, Vs, as functions of orientation in a single crystal, the elasticity tensor
(compliance Sij or stiffness Cij ) can be determined. Acoustic-wave velotities and elasticity
are central to geophysical research.
370 Chapter 4

High-pressure dielectric properties can be determined from measurments of

refractive indices with optical interferometry (Hemley et al., 1991). Pressure-induced
changes in crystal symmetry are manifested by change in interfacial angles or
birefringence under crossed nicols.
Strongly coherent, collimated, monochromatic laser beams can be focused through
a DAC to provide a versatile microprobe with mm spatial resolution to multi-megabar
pressures (Hemley et al., 1987). Laser microprobes have offered the means for high-
pressure Raman, luminescence and Brillouin spectroscopies.
Because the nature of seismic waves is similar to elastic waves, it is crucial to know
precisely the longitudinal and transversal sound velocities and the associated elastic
properties such as the bulk modulus and shear modulus of various minerals at simultaneous
high-pressure and high-temperature conditions.
To derive seismic velocities at conditions of the deeper Earth, studies have been
performed in the past which normally determined the P or T dependence of velocity V: To
derive the second derivation, i.e., d2 V=dPdT; different data sets are required to be
combined and extrapolated, which results in relatively large uncertainties. To avoid such
extrapolation, ultrasonic-wave propagation at high pressure and temperature are
employed.
The sound velocities of a mineral are related to the slope of the dispersion curves
near the Brillouin zone centre: dv=dk where v is the mode frequency and k is the wave
vector (¼ 2p=a where a is the lattice constant). As expected from a continuous elastic
medium, the velocity must be proportional to the measured frequency:

V / va

The lattice constant versus pressure and the zero-pressure sound velocities have been
measured for a number of materials. The inelastic neutron-scattering data for MgO, Al2O3
and MgAl2O4 (Thompson and Grimes, 1978) show that the acoustic-mode dispersion
curves are nearly sinusoidal.
There have been advances, at ambient pressure, in resonant ultrasound spectroscopy
(Maynard, 1996) and in direct imaging of acoustic-velocity surfaces (Wolf and Hanser,
1995). There has been considerable progress in the study of sound velocity and elasticity at
high pressure. Ultrasonic sound-velocity measurements in the large volume press have been
pioneered recently using both single crystals to 6 GPa (Yoneda and Morioka, 1992) and
polycrystals to 12 GPa (Li et al., 1996). A recently developed gigahertz ultrasonic
interferometer has also been used with the large volume press (Chen et al., 1996) to obtain
selected single-crystal constants at simultaneous high pressure and temperature.
In ultrasonic interferometry, a burst of high-frequency signal is applied to the
piezoelectric transducer. The resultant elastic wave, propagating inside the sample,
produces a series of echoes. The travel time of an elastic wave through the sample is
measured by observing constructive and destructive interferences among overlapped
echoes as a function of career frequency. To determine the sound velocities, ultrasonic
interferometry measurements can be performed in which the signals from the buffer rod
and the sample interfaces are overlapped.
Principles of Techniques 371

4.4.4.1. Instrumentation
The method of ultrasonic experimentation can be detailed from that employed at
Bayreuth Geoinstitute (BG), which is as follows.
A HDAC is used equipped with heaters that are capable of achieving 10 GPa and
1,2008C. Typical sample dimensions in the DAC are 50 mm in height and 300 mm in
diameter. Pressure measurements are carried out using ruby fluorescence at room
temperature and Sm-doped strontium borate at high temperature. The frequencies that are
generally used by traditional ultrasonic measurements range from 10 to 100 MHz.
Only after the debut of the gigahertz ultrasonic interferometer has ultrasonic
interferometry on samples loaded in DAC become possible. The group at BG succeeded in
combining these two for the first time. A mechanical adapter is used to couple the gigahertz
ultrasonic interferometer to the DAC. The sound waves, which are generated by a piezo-
electrical transducer, are guided through the buffer rod into one diamond anvil and into the
sample in the DAC. The thickness of the sample and the travel time in the sample yield the
sound velocity. In order to determine the travel time accurately, the pulse-superposition
method is widely used. In this, the phase difference between the echo at the sample –
diamond interface and the echo at the free end of the sample is used to determine the travel
time. The precision of the travel time determined using this method can be less than 1024. In
addition, the wide band-width (from 400 MHz to 1.2 GHz) is required to ensure high
precision. Shen et al. (1995) at BG used DAC and a gigahertz ultrasonic interferometer to
measure the longitudinal and transversal sound velocities of minerals at HPT.
Bassett et al. (2000) have described instrumentation for measuring ultrasonic travel
time as well as the lattice parameter of a single crystal under high pressure (up to 6 GPa)
and temperature (2508C). The signal is carried from a thin transducer through a sapphire
buffer rod coupled to one of the two diamond anvils, designed for ultrasonic and XRD
measurements. The assembly also provides sufficient normal force to the buffer rod –
diamond interface to ensure that acoustic coupling is maintained without the use of a bond.
They chose to thin the transducer until the frequency corresponded to about
600 MHz. The signal traverses the diamond anvil and enters the single-crystal sample,
coupled to the anvil face cement, adhesion or a normal face. Interference of superimposed
waves reflected from the near and far faces of the single crystal is used to measure the
travel time of the sound waves in the sample. The sample size could be reduced through
reduced wavelength with GHz ultrasonic interferometry (Spetzler et al., 1993).
The two DACs make it possible to measure d-spacings at several different
orientations using a four-circle goniometer.
Measurement of travel times and lattice parameters at the same P – T conditions
allows the conversion of travel times to velocities and can provide simultaneous EOS.
This EOS can be used to determine the pressure vs. lattice parameter.

4.4.5. Fluorescence spectroscopy

In vibrational spectroscopy (e.g., infrared and Raman) the vibrational modes appear
near the Brillouin zone centre where the acoustic modes are zero. In the fluorescence
372 Chapter 4

method, the light interacts with the electronic transition and not with the vibration, thus
eliminating the Brillouin zone restriction. But this side-band spectrum should have
vibrational modes, having no inversion symmetry (i.e., with odd parity).
The acoustic modes found in the vibrational side bands of a fluorescing cation (such
as Mn4þ, Cr3þ or V2þ with 3d electrons) doped into the host lattice are monitored in
fluorescence spectroscopy. Trivalent chromium is commonly used. A sharp fluorescence is
observed in the case of d– d transition (2Eg ! 4A2g), which is spin, symmetry and orbitally
forbidden (e.g., Chopelas, 1996). This sharp electronic transition is vibrationally assisted
by odd parity vibrations of the host lattice because it relaxes the symmetry forbidden
selection rule.
The fluorescence of Cr3þ in octahedral coordination consists of a sharp central
“no-phonon” line onto which the vibrational modes are symmetrically superimposed. The
room-pressure fluorescence spectra of Cr3þ-doped five minerals are shown in Fig. 4.12
(read the caption). At a given frequency, the number of modes excited is determined by the
Boltzmann distribution function. Usually, there is more than one sharp peak near the
central peak. These may arise from the lifting of the degeneracy of the d-orbitals because
the 6-fold coordination site is not perfectly cubic (R2 line) or because the local lowering of
the site symmetry is caused by the substitution of an M3þ impurity for an M2þ host cation
(N-lines) and an attendant vacancy for charge compensation.
For Cr3þ cations, both T1u and T2u lattice phonons, irrespective of their location in
the Brillouin zone, appear in the spectrum. These lattice phonons, which include the
acoustic modes, appear as “side bands” on the low- and high-energy sides of the sharp
electronic transition at an energy removed from the pure electronic or no-phonon line
(corresponding to the energy of the phonon; Chopelas et al., 1996). In Cr3þ, the electronic
transitions of interest range from 685 to 700 nm.

4.4.5.1. Experiments
The experimental requirements for fluorescence spectrometry are very simple,
requiring only a laser, diamond cell and single monochromator with a photomultiplier
tube. The signal is strong and does not degrade even at high pressure and very small
samples can be used. This makes it possible to measure easily the velocities up to lower-
mantle pressures. The measured acoustic modes are spherically averaged.

4.4.5.2. Side-band fluorescence ultrasonic technique

Bulk and shear moduli can be calculated from sound velocities determined by the
side-band fluorescence method. These were found to be in excellent agreement with
those obtained by ultrasonic measurements to lower pressures (Chopelas et al., 1996)
(see Table 4.5). The side bands are representative of the phonons of unperturbed lattice.
The side-band method has several advantages over the more conventional methods
such as ultrasonic techniques, Brillouin scattering or even the newer impulse-stimulated
scattering technique. Interpretation of side-band method, however, suffers from ambiguity
in lower symmetry compounds, especially when iron is present even in small (# 1%)
amounts.
Principles of Techniques 373

Figure 4.12. Cr3þ-fluorescence lines in different mineral compounds, studied by Chopelas et al. (1996).
The R1 lines are set to zero to facilitate the comparison of the phonons of other materials. The central
peaks for these minerals as follows MgO–669.2 nm; MgAl2O4 –685.5 nm; Al2O3 –694.3 nm; YAG–688.8 nm;
pyrope – 698.9 nm; the R2 lines for YAG and pyrope are well resolved and indicated in the figures. The sharp
lines on the stokes sides of the central line in MgAl2O4 and Al2O3 spectra are N– lines and are indicated by
asterisks. The Al2O3 vibrational sidebands are significantly weaker relative to the R– lines at one atmosphere so
the spectrum is magnified by a factor of 10 to enhance the sidebands (q 1996 The Geochemical Society).
374 Chapter 4

TABLE 4.5
Comparison of pressure derivatives of the elastic moduli obtained by the sideband method to those obtained by
ultrasonic techniques (Chopelas et al., 1996)

Sideband Ultrasonics Reference

MgO Pmax (GPa) 36.5 3.2 Jackson and Niesler (1982)

K0 4.08 4.13 Yoneda (1990)
K 00 (GPa21) 20.036 20.022 Duffy and Ahrens (1995)
2.56 2.53
G00 (GPa21) 20.030 20.027
YAG Pmax (GPa) 57.5 0.3 Yogurtcu et al. (1980)
K0 5.16 4.44
K 00 (GPa21) 20.0036 2
G0 0.68 0.62
MgAl2O4 Pmax (GPa) 11.0a 1.0 Chang and Barsch (1973)
K0 5.05 5.15 Yoneda (1990)
K 00 (GPa21) 20.13 20.48
A) G0 0.072 0.51
Al2O3 Pmaxb (GPa) 53.0 1.0 Gieske and Barsch (1968)
G0 1.94 1.98
2.56 2.53
G00 (GPa21) 20.02 20.02

Pyrope Pmax (GPa) 28.0a 3.2 Webb (1989)

K0 4.5 4.7c
B) G0 1.73 1.81c

a
Phase change or change in compression mechanism at this pressure.
b
Only shear velocities were measured.
c
Pyrope with 36% iron at the Mg2þ site.

4.4.6. Photoluminescence spectroscopy

Photoluminescence spectroscopy has in principle a similarity to Raman

spectroscopy. The incoming photon of energy, Ep ; interacts with an electron in the
material, whereby energy of the electron is lifted from the valence band to the conduction
band. However, the electron experiences inelastic collisions with lattice vibrations and
impurities and the energy reduces to the minimum of the conduction band. This electron
recombines with a hole in the valence band, emitting a photon, which is detected. This
allows the band-gap energy ðEg Þ to be estimated.
In the conduction band, there are three principal minima, G; X and L: The G
minimum band corresponds very well with the valence-band maxima, as shown in the
middle and lower part of the figure. Under pressure, the G minimum moves upwards in
energy at a relatively fast rate of 100 – 150 meV/GPa. The energy of the L minimum
Principles of Techniques 375

increases to 1/3 of the G minimum. In contrast to these two, the X minimum decreases at
the rate of 25 to 215 meV/GPa.
For GaAs up to 5 GPa, the G level increases in energy to 0.6 eV, which is sufficient
to allow the X level to have the lowest energy level so that GaAs, which is normally a
direct-gap semiconductor, becomes an indirect-gap semiconductor. This electronic
transition (L – X crossover) is studied by transport measurements.
A shift in Raman lines with pressure is related to the stiffening of the lattice. With
the onset of phase transitions, the Raman lines “soften”. The changes in band gaps and
defect energies with pressure are measured by photoluminescence energy changes.
Because emission occurs from or to defect levels, the defects generated under pressure can
be studied by photoluminescence studies.
Pressure-induced defect luminescence can be found in metastably compressed
SiO2 and is related to non-black-body emission observed in shock-wave experiments.
The other example is intense visible luminescence with B1– B2 (NaCl – CsCl) transition
in CaO at 65 GPa (300 K) (Richet et al., 1989).

4.4.6.1. Photo-emission method

Probing of band structure is helped through interaction of electrons with the sample.
The other related methods are electron energy loss (EEL) and auger spectroscopy. The
photo-emission method includes fluorescence and phosphorescence, corresponding to
shorter- and longer-lived excited states (i.e., less than or greater than 1 ns, respectively).
The transitions are governed by Laporte Selection Rules (i.e., Dl ¼ ^1; where l is the
angular momentum quantum number).
The excitations may be associated with defects at near and across band gaps. The
excitation luminescence is used in pressure calibration such as in ruby (Al2O3:Cr3þ),
discussed in Section 4.4.5. By employing the tunable lasers (e.g., Ti-sapphire) and pulsed
excitation techniques, measurements by ruby luminescence have been extended to above
250 GPa (Goncharov et al., 1998). Also, samarium-doped yttrium aluminium garnet
(Sm-YAG) has been calibrated in this pressure range (Liu and Vohra, 1996).

4.4.7. X-ray diffraction

Diffraction techniques (X-ray and neutron) provide the most direct information
about the change in long-range order under pressure. These provide an averaging over
length scales of 1,000 Å, whereas the spectroscopic techniques (which provide vibrational
and electronic properties) reflect the information for the 10– 100 Å range. As pressure
increases, XRD planes move closer and the experiment measures the compressibility. At
still higher pressure, a phase transition to a new structure can be detected from the XRD
pattern. Under pressure, some X-ray amorphous materials may continue showing the
spectroscopic features of the crystalline state. An example is serpentine, for which the
Raman spectra will show sharp features in pressures when it is seen as X-ray amorphous
(Meade et al., 1992).
X-ray photons can be scattered elastically or inelastically. Elastically scattered hard
X-ray (, 1 Å) photons from different atoms can interfere constructively with the atomic
376 Chapter 4

planes in crystals, typically separated by , 0.5– 10 Å. The diffraction takes place
following the condition (Bragg’s law):

l ¼ hc=E ¼ 2dsinu

where l is the wavelength of X-ray, E its energy, d the interplaner spacing, 2u the
diffraction angle, c the velocity of light and h Planck’s constant. When u and E or l are
known, the values of d can be determined.
In DAC, the sample used to obtain a reasonable intensity of the diffraction peak
from a laboratory diffractometer set-up can be very small. To increase the intensity, a beam
from a synchrotron source is used. The energy of the diffraction peaks from the wide
spectrum of the synchrotron beam falling on the sample is measured. In this energy
dispersive diffraction (EDXD) study, the detector is set at a fixed energy (see Fig. 4.13).
Diffraction techniques provide the best measure for crystalline as well as
amorphous structures. An unambiguous description of atomic correlation between 0.5
and 10 Å, averaged over length scales of 100– 1,000 Å, can only be available by X-ray
studies. Spectroscopic techniques, however, probe only the vibrational and electronic
properties over a more limited range of distances and, to interprete the data, these generally
require a priori structural models. Again, in situ diffraction experiments allow direct
measurements on novel structures, which may be unstable on decompression.
The availability of synchrotron sources and energy-dispersive X-ray analysis have
reduced data collection times and increased the general capability of techniques by
permitting very fine collimation (, 4 mm) of the incoming beam (Ruoff et al., 1990).
XRD suffers limitation in a DAC experiment because the diamond-cell geometry
offers only a limited range of Q ð¼ 4psinu=lÞ values. For this reason, usually only the first
diffraction peak of the amorphous phase is recorded. However, the problem is generally
averted by use of an internal standard which remains crystalline under pressure, viz., gold,
which serves as a pressure marker.

Figure 4.13. Schematic presentation of diamond-cell sample configuration for EDXD with synchrotron radiation
(Mao and Hemley, 1996, q 1996 Overseas Publishers Association, Amsterdam).
Principles of Techniques 377

Advances in ultra-high pressure have evolved symbiotically with the development

of synchrotron techniques. For ultra-fine-level X-ray studies, high-brilliance and
low-emittance synchrotron X-radiation has become the primary tool. Experiments with
MA 8-type apparatus allow one to reach pressures up to 32 GPa and temperature up to
2,000 K (Funamori et al., 1996). However, it is quite difficult to take a high-quality XRD
profile using laser-heated DAC because of the limited volume of the heated sample and the
temperature gradient across the sample. The EDXD technique is used for probing
structural and strain variations in small samples under ultra-high P and T: The DAC
sample configuration for EDXD with synchrotron radiation is shown in Fig. 4.13.
In energy-dispersive XRD, Meade et al. (1992) worked with a higher Q range (up to
16 Å21) by changing the fixed Bragg angle and digitally subtracting the background
arising from Compton scattering from diamond anvils at higher Q values.
Non-hydrostatic stresses in a DAC, due to heterogeneous stress distribution, may
mimic the amorphization by broadening diffraction peaks. This is avoided by use of an
appropriate hydrostatic pressure transmitter such as a noble gas.
Notwithstanding its advantages, diffraction experiments are delimited in their scope
by the poor scattering power of amorphous materials and the small sample size
(, 10213 m3) in high-pressure experiments. Again, the broad Compton scattering from the
pressure-transmitting anvils (e.g., in DAC) overwhelms the diffuse scattering from small
samples. However, this Compton scattering can be damped by collimating the synchrotron
radiation (high brightness, low divergence) for dimensions of the order of 10 mm. By using
energy-dispersive diffraction and the high critical energy from a wiggler source, one can
probe a wide range of reciprocal space for weakly scattering amorphous materials
(Q ¼ 4p sinu=l; 1 , Q , 16:58 Å21; Meade and Hemley, 1991). The diffraction
technique has been used to study the medium-range order of silica glass and quantify its
variation with pressure.
Crystalline ! amorphous ðc ! aÞ transformation is best studied by X-ray
absorption spectroscopy (XAS). This is studied either in the mode of high-resolution
X-ray absorption edge structure (XANES) or extended X-ray absorption fine structure
(EXAFS). XANES is sensitive to local symmetry around the absorbing atom and EXAFS
is sensitive to the distances of the surrounding atoms.
Pressure can be monitored by the ruby-fluorescence method. Ruby-fluorescence
line-widths are used to detect non-hydrostatic conditions. Significant broadening does
not occur up to 15 GPa, indicating that quasi-hydrostatic behaviour exists up to this
pressure.
The pressure calibration in XRD (Liu and Vohra, 1996) — including the secondary
ruby scale, which is based on diffraction — are carried out in axial geometry. Above
12 GPa and room temperature, all pressure media solidify and therefore exert some degree
of non-hydrostatic stress on the sample.

4.4.7.1. Radial X-ray diffraction (RXD): deviatoric stress

Diffraction data obtained under non-hydrostatic stress conditions contain a wealth
of additional information on elasticity not contained in hydrostatic compression data.
Radial diffraction experiments help study the influence of non-hydrostatic stress on the
378 Chapter 4

lattice strains in the sample. Singh et al. (1998) showed that diffraction measurements on
polycrystalline samples under non-hydrostatic compression can be used to determine the
single-crystal elasticity tensor.
The hydrostatic pressure and uniaxial stress components are given by
sP ¼ ð2s1 þ s3 Þ=3
and
t ¼ s 3 2 s1
Generally, for all crystal systems, the spacing is a function of c; the angle between
the diffracting plane normal and the load direction (Singh, 1993). The geometry for radial
XRD of a uniaxially compressed sample in DAC is shown in Fig. 4.14 (Singh et al.,
1998a,b).
In this radial diffraction method, a micro-focus (4 –10-mm diameter) polychromatic
X-ray beam is passed through the sample under pressure (in DAC) in the radial direction. It
probes the lattice strain of the sample as a function of the angle ðcÞ between the diffraction
vector and the diamond-cell axis (load axis) (Singh et al., 1998). The energy-dispersive
XRD patterns (containing hkl notations) are collected at u0 steps (108, usually) of c from
0 to 908.
The difference in d-spacing obtained from C ¼ 08 and 908 patterns gives the
deviatoric strains:
Q ¼ ðd08 2 d908 Þ=3dP
where dP ¼ ðd08 þ 2d908 Þ=3 is the d-spacing under hydrostatic pressure. This quantity
is derived from the general expression dC ðhklÞ ¼ dP ðhklÞð1 þ ð1 2 3cos2 CÞQðhklÞÞ
(see Hemely et al., 1997). Thus, the d-spacings are seen to vary linearly with cos2 c;
the intercept dP ðhklÞ denotes the d-spacing under stress and the slope QðhklÞ is the lattice
strain under uniaxial stress condition (Singh and Balasingh, 1994). (Note: The pressure is

Figure 4.14. Geometry for radial X-ray diffraction of the uniaxially compressed sample in a diamond cell (Singh
et al., 1998a,b). s3 and s1 are axial and radial stresses. The diffraction from the sample is measured through the
gasket as a function of C by rotating the cell, 2u is the diffraction angle (Mao and Hemley, 1998, q 1998
Mineralogical Society of America).
Principles of Techniques 379

usually determined from the average dP using the EOS of the material under investigation;
Mao et al., 1990.)
The data can be used to determine the deviatoric stress component t (Meng and
Weidner, 1993):

t ¼ s1 2 s3 ¼ 6G , QðhklÞ . f ðxÞ

where f ðxÞ is a function of elastic anisotropy, x; G is the aggregate shear modulus and
kQðhklÞl is the average of Q over the measured diffraction peaks with indices hkl: Usually,
t increases with increase of P:
The maximum uniaxial stress, t; supported by a material is determined by its
strength: i.e., t # sn ¼ 2t; where sn and t are the yield and shear strengths of the material,
respectively. In high-pressure experiments, it is often assumed that t ¼ sn : However, in
general, t varies with the sample environment and the equality (von Mises condition) holds
only if the sample deforms plastically under pressure. The t versus pressure data for MgO
to 227 GPa (Duffy et al., 1995) was obtained by combining the measured d-values for the
reflections (200) and (220) in accordance with the equation of Singh (1993). Figure 12.20
shows that the t for MgO reaches as high as 20 GPa at 200 GPa.
As discussed in an earlier section, the stress in the sample at the origin of the
coordinates is given by a tensor (Eq. (5-27) in Section 5.3.1):

sij ¼ sp þ Dij

In terms of the measured d-shearing, the strain produced by Dij is given by

1D ðhklÞ ¼ ½dm ðhklÞ 2 dp ðhklÞ=dp ðhklÞ ð4-17Þ

where dp ðhklÞ is the d-spacing under equivalent hydrostatic pressure sp :

On equating the calculated strain to 1D ðhklÞ; one obtains:

dm ðhklÞ ¼ dp ðhklÞ½1 þ ð1 2 3cos2 CÞQðhklÞ

where C is the angle between the direction of the applied load and the diffraction-plane
normal and QðhklÞ is given by:

QðhklÞ ¼ ðt=3Þ½a{2GXR ðhklÞ}21 þ ð1 2 aÞð2GV Þ21 

GXR ðhklÞ is the shear modulus of the polycrystalline aggregate calculated under Reuss (iso-
stress) condition. The averaging is done only over the group of crystallites which
contributes to the intensity at the point of observation. GV is the aggregate shear modulus
calculated under Voigt (iso-strain) condition, a is the factor, between 0 and 1, which
decides the relative weights of the Reuss and Voigt conditions.
The expression for GXR ðhklÞ depends an the crystal system.
For the cubic system (Singh et al., 1998):

½2GXR ðhklÞ21 ¼ ½S11 2 S12 2 3ðS11 2 S12 2 S12 2 1=2S44 ÞGðhklÞ ð4-18aÞ
380 Chapter 4

where

GðhklÞ ¼ ðh2 k2 þ k2 l2 þ l2 h2 Þ=ðh2 þ k2 þ l2 Þ2

For the hexagonal system:

½2GXR ðhklÞ21 ¼ 1=2½ð2S11 2 S12 2 S13 Þ þ l32 ð25S11 þ S12 þ 5S13 2 S33 þ 3S44 Þ

þ l43 ð3S11 2 6S13 þ 3S33 2 3S44 Þ

where

l23 ¼ 3a2 l2 ðM 2 Þ; M 2 ¼ ½4c2 ðh2 þ hk þ l2 Þ þ 3a2 l2  ð4-18bÞ

and a and c are the lattice parameters of the hexagonal cell. The expressions are given
elsewhere for trigonal (Singh et al., 1996) and other systems (Singh et al., 1998b).
Single diffraction allows the separation of Bragg peaks in reciprocal space. This is
particularly important for single-crystal diffraction for the correct and unambiguous
identification of peak splittings in reciprocal space (angles between reflections) such as
those associated with the rutile – CaCl2 transition.

4.4.7.2. Density determination

Measurements of the density of minerals at high P can be obtained by a variety of
high-P XRD techniques. From the changes in density measurements, the EOS can be
determined. Single-crystal diffraction measurements still offer superior results compared
with those employing powder XRD. Therefore, single-crystal data have been
recommended by Angel and Ross (1997) as the basis for calculating the densities of
mineralogical models of the Earth’s mantle (supplemented only by data obtained by HP
powder XRD).

4.4.7.3. High-pressure XRD (powder) study: An example of MgSiO3 ilmenite

In their high-pressure powder XRD at RT on MgSiO3 ilmenite, Reynard et al.
(1996) used an energy-dispersive mode at the storage ring DCI (LURE, Orsay, France).
The sample was X-rayed in the DAC by a polychromatic X-ray beam, collimated by a
remote tungsten-carbide slit system to a 50-mm diameter spot centred on the pressure
chamber. The diffraction beam was analysed at 11.4398 2u between 5 and 50 keV by a
Canberra planar germanium detector (efficient area ¼ 50 mm2) with a resolution between
145 eV (at 5.9 keV) and 500 eV (at 122 keV). Silver foil placed between the diamonds was
used as a standard to calibrate the system in the operating geometry for the high-pressure
measurements on MgSiO3 ilmenite.

4.4.8. Mössbauer spectroscopy

Mössbauer spectroscopy has been advantageously employed in determining the

non-quenchable transformations in the neighbourhood of phase transition caused by P
and/or T changes. The basic principles of Mössbauer spectroscopy in relation to
Principles of Techniques 381

geochemistry and mineralogy have been discussed by many (see, for example, Mitra,
1992). The layout of the conventional DAC Mössbauer technique and the hyperfine
interactions involved in this technique are depicted Fig. 4.15.
Mössbauer study under pressures up to 10 GPa, employing a hydrostatic pressure
medium, causes no major problem but a temperature above 1,000 K normally lowers the
recoil-free fraction ð f Þ-value significantly and deteriorates the signal-to-noise ratio.

Figure 4.15. (a) Conventional diamond-cell Mössbauer technique (see King et al., 1978). (b) Hyperfine interactions
for the 57Fe nucleus, showing the energy levels of (1) bare nucleus, (2) electric monopole (isomer shift), (3) electric
quadrupole interaction (quadrupole splitting), (4) magnetic dipole interaction (hyperfine magnetic splitting). (from
McCammon, 1995a) (Source: Hemley et al., 1998, q 1998 Mineralogical Society of America).
382 Chapter 4

Therefore, the scope of high-pressure and high-temperature study by Mössbauer becomes

restricted.
Study of the short-lived intermediate structures or the mechanism of transformation
by Mössbauer spectroscopy has not yet been successful because its spectral counts need
more than half a day to accumulate to present a meaningful envelope. By keeping the
velocity of the Mössbauer drive constant, the absorption recorded in a single channel with
pressure and temperature can be noted, but this will have little applicability.
In gamma-ray resonance (Mössbauer) spectroscopy, the shift of the geometrical
centre of the 57Fe doublet may be written as:
dðT; PÞ ¼ d0 ðPÞ þ dSOD ðT; PÞ
where d0 is the chemical or isomer shift at 0 K and dSOD ðT; PÞ is the second-order Doppler
(SOD) shift. For Fe2þ
oct sites of chain silicates at ambient condition, the value of dSOD equals
0.23 mm/s.
The pressure derivative of d at constant temperature (›d=dPÞT equals ›d0 =dP þ
ð›dSOD =dPÞ: By adopting the Debye model, ð›dSOD =dPÞ can be estimated from Grüneisen
parameter g and the Debye temperature uD, taking into account that
ð›dSOD =›lnVÞ ¼ gð9kuD =16McÞFðT=uD Þ
In the above equation, V is the volume of the unit cell, k the Boltzmann constant, M the
mass of the resonant nucleus and c the velocity of light.
The Grüneisen parameter g ð¼ aV=bCV Þ may be obtained approximately by
calculating the volume compressibility at constant pressure (i.e., a) and the volume
compressibility at constant temperature ðbÞ: For example, a calculation for the value of g
can be made for hedenbergite using the high-temperature (Cameron et al., 1989) data of
hedenbergite and the heat capacity at constant volume CV (which was approximated by CP
calculated for diopside using the relationship of Krupka et al., 1985). The F-value in the
earlier equation can be formulated as
ð uD=T
FðT=uD Þ ¼ 1 þ 8=ðeuD=T 2 1Þ 2 24ðT=uD Þ4 x x3 dx=ðex 2 1Þ ð4-19Þ
0

In the above equation, x ¼ hn=kT where h is the Planck’s constant and n the
frequency of harmonic oscillators.

4.4.8.1. Pressure dependence of isomer shift (d0)

The value of isomer shift d (neglecting dSOD) is proportional to the ionic charge
density rð0Þ at the resonantly absorbing nucleus. For 57Fe, d decreases with increasing
rð0Þ: In general, rð0Þ may be considered as (Gütlich et al., 1978):
rð0Þ ¼ rc freeionð0Þ þ rov ð0Þ þ rval ð0Þ ð4-20Þ
In case of high-spin Fe2þ, rc free ion (0) is the 1s, 2s contribution of the free Fe2þ
core, rov(0) is the overlap contribution rov ð0Þ ¼ rc ð0Þ 2 rc free ion (0), where rc(0) is the
contribution of 1s, 2s, 3s orbitals of the effective Fe2þ core and rval(0) is the contribution
Principles of Techniques 383

from the 3d, 4s, 4p valence orbitals of Fe2þ and 2s, 2p orbitals of O22. The terms rov(0)
and rval(0) are expected to be dependent on compression. It should be noted that an effect
of pressure on d is possible only by a change in bond distances and bond angles around the
position of the resonantly absorbing nucleus.
Under pressure, any increase in rð0Þ values results mainly from the compression of
the valence orbitals due to bond shortening without changing rov(0) significantly.
The ›d ›P coefficient of a Fe (ground state 3s64s2) is 2 7.46 to 2 9.08 £
23
10 mm/s GPa.
In chain silicates, the electronic state of Fe2þ at M sites as well as the degree of
covalent Fe2þ ligand participation is comparable with those in FeF2 structure. The crystal
structure of FeF2 is of the rutile type, with Fe2þ being octahedrally coordinated. The
volume compression of the FeF6 octahedra in FeF2 is 4.1% between ambient pressure
and 4 GPa.

Isomer shift (d0) for sulfur bonding. The main factor determining the isomer shift for Fe
in octahedral coordination with S is apparently the spin state of Fe.
The calculated trends for the FeS102
6 cluster are for the density at the Fe nucleus to
increase as (i) the Fe – S distance is reduced, the charge density at the Fe nucleus,
r(0) ¼ 11,875.01e (a.u.)23 for R(Fe– S) ¼ 2.50 Å, reduces to quintet state, while the
charge density 11,875.22e (a.u.)23 for R(Fe – S) ¼ 2.26 Å, quintet state and/or (ii) the
high-spin quintet state is converted to low-spin singlet (r(0) ¼ 11,875.46 (a.u.)23 for
R(Fe –S) ¼ 2.26 Å, singlet).

Electron density changes. The electron density for the l ¼ 2 function within the Fe sphere
is larger in the singlet state than in the quintet (5.29 vs. 4.61 Fe 3d electrons) in the crystal-
field type orbitals. This change occurs because the 3eg " orbital, which is the crystal-field
orbital with the smallest amount of Fe 3d character (, 60%), is emptied in going from the
high- to the low-spin state and the 2t2g orbital (with , 80% Fe 3d character) becomes
populated (Tossell, 1977). However, along with this increase in total Fe 3d character, we
find an increased Fe 3d density at the Fe nucleus (140.96 vs. 140:51e (a.u.)23.) The
increased density in the Fe 3s orbitals is to be partially balanced by a reduction in density at
the Fe nucleus from the 2a1g bonding orbital (from 2.59 to 2:38e (a.u.)23). The change in
the 2a1g density occurs in spite of the very small (0.04 eV) change in 2a1g relative orbital
energy between the two spin states. The contributions from all other orbitals remain
constant (Tossell, 1977).

4.4.8.2. Quadrupole splitting (DEQ)

The DEQ for Fe2þ in octahedral coordination is expressed in terms of a reduction
function F as (Ingalls, 1964):

DEQ ¼ C a2 FðD1 ; D2 ; l0 ; a2 ; TÞ ð4-21Þ

where C is a constant ð¼ 2=7e2 Qð1 2 R0 Þðr 23 . 0Þ; D1 and D2 are the ground-state
splittings of the crystal-field levels, a2 is a covalency factor, which decreases with
384 Chapter 4

increasing covalency, l0 is the spin – orbit coupling constant for the free ion and T is the
temperature.
The reduction function, F; has two contributions, one from the valence electrons
(valence term) and the other from the influence of the crystal lattice (lattice term).
The values of D1 and D2 are related to the distortion of the coordination octahedron from
purely cubic symmetry. The valence term is highly T dependent and increases with
increasing distortion, eventually reaching a maximum value. The lattice term, which is
largely independent of temperature, also increases with increasing distortion but is of
opposite sign. The reduction function, F; therefore, increases initially with increasing
distortion to a maximum value but then decreases with further increase in distortion as the
contribution from the lattice term becomes greater.

Pressure dependence of quadrupole splitting. The 57Fe quadrupole splitting, DEQ ; is

determined by the maximum value Vzz of the second derivative Vii ði ¼ x; y; zÞ and the
asymmetry parameter h ¼ ððVxx 2 Vyy Þ=Vzz Þ: Here, V is the electrostatic potential at the
nuclear site. The contribution of h to DEQ is minor (maximum 15% for h ¼ 1).
In the case of high-spin Fe2þ in predominantly ionic compounds, Vzz can be
separated into three additive terms: (1) the valence contribution Vzzval ; which results from
the unbalanced population of 3d orbitals, (2) the lattice contribution Vzzlat ; which originates
from the non-cubic distribution of ligand charges and the most distant ions around the Fe2þ
site, and (3) Vzzov ; which accounts for the overlap between Fe2þ and ligand orbitals. It is
often difficult to arrive at conclusions from variation of DEQ alone about the individual
charges in these three components of Vzz : In principle, Vzzval provides the predominant
contribution to Vzz and determines the sign, Vzzlat is minor and opposite in sign. The sign of
Vzzov is also opposite to that of Vzzval :
At higher pressure, there is greater polyhedral distortion and enhanced participation
of covalency leading to higher values in ›d=›lnV and dDEQ =›lnV:
119
Sn Mössbauer measurements have been reported to about 29 GPa (Sterer et al.,
1990).

4.4.9. NMR spectroscopy

Information on structural parameters, such as inter-hydrogen distances, d(H – H),

and hydrogen bonding distances are provided by 29Si and 1H NMR spectroscopy, which is
element-specific and shows the relationship between intensity and concentration. Nuclear
magnetic resonance (NMR) (Lee et al., 1992) techniques have been used with DACs to
pressures 5.2 and 10 GPa, respectively.
NMR may also be an important tool for the study of hydrogen present at moderate
defect levels in silicates, both for determining absolute H levels in powders and for
investigating the distribution and structural chemistry of hydrogen-bearing defects.
29
Si has a natural abundance of 4.7%. Signals can be observed from the
sample using standard 29Si(1H) cross-polarization (CP) magic-angle spinning (MAS)
technique, wherein the detected 29Si magnetization is generated by polarization transfer
Principles of Techniques 385

from 1H nuclei. In addition, the CP-MAS signal intensity is sensitive to the 29Si – 1H
dipolar coupling, allowing the Si resonance to be distinguished by their proximity to
hydrogen.
29
Si{1H} CP-MAS and 1H NMR spectroscopy can provide structural and chemical
information on silicates synthesized at high pressure. With commercially available probes
fitted with small-diameter sample systems, useful signal levels can be obtained.
All the hydrous magnesium-silicates are readily detected by 29Si {1H} CP-MAS
and it is possible to obtain reasonable quantitative signal levels for several of the phases
(see, e.g., Phillips et al., 1997).
The signal intensity is highly dependent on the CP dynamics, which generally
varies between sites in the same phase, and the length of the polarization transfer period
(contact time, tc). The difference between two exponential functions that characterize the
relaxation of the 1H magnetization in the applied transverse field, T1P(H), and the cross-
relaxation between 1H and 29Si spin systems, TSiH, describes the variation of the signal
intensity with contact time as:
" ! #
1 tc tc
Iðtc Þ ¼ I0 exp 2 2 exp 2 ð4-22Þ
T T1PðHÞ TSiH
1 2 SiH
T1PðHÞ

Generally, TSiH varies among inequivalent Si, because it depends on the Si –H

heteronuclear dipole coupling and, therefore, the Si – H distances. As a result, the CP-MAS
intensity is not proportional to the number of corresponding Si nuclei unless the contact
time and relaxation constants satisfy the condition;
TSiH p tc p T1P ðHÞ
The development of MAS technique holds out the hope that NMR-based detection
of motional averaging may be capable of providing relaxation time data and thus,
indirectly, viscosity data at elevated temperatures for many silicate-melt compositions.
Efforts are currently being made to develop in situ high-T, high-P probes for use in NMR
spectroscopy for opening up the possibility of studies at elevated pressure in molten
silicate systems.
High-pressure NMR experiments on solids are carried out primarily for the
following two reasons:
(a) To separate the effects of density and temperature on various dynamic processes.
(b) To determine the P effects on chemical equilibria and reaction rates. This is discussed
below.
A combination of advanced 2D and 3D NMR techniques with high-pressure
capability represents a powerful new experimental tool enabling determination of the
protein folding.

4.4.9.1. Pressure effects on proton NMR spectra

Increasing pressure changes both the density and shear viscosity of liquids, which,
in turn, changes the relaxation times and chemical exchange.
386 Chapter 4

The relaxation time can be expressed in terms of the rotational correlation time, t;
with the following relation:

1 v0 t 4v 0 t
¼C þ ð4-23Þ
T1 1 þ v20 t2 1 þ 4v20 t2

where v0 is the Larmor frequency. The constant C is proportional to the proton second
moment expressed in angular frequency units.
At the phase-transition (say, from a to b) temperature T1 ; one can assume v0 t q 1
for the b-phase rotation. Then the expressions for the relaxation times are as follows:

ð1=T1 Þa ¼ 5v0 tC0 forv0 t p 1

ð1=T1 Þb ¼ 2Cb =v0 t forv0 t q 1

Assuming that rotational motion is a thermally activated process and that the
second moments are independent of temperature and pressure, one can obtain the
activation volume DV p ðTÞ at temperature from the pressure dependence of lnðT1 Þ:

T1 ðT; PÞ ¼ AðTÞexp{ ^ ½DV p ðTÞ=RTP} ð4-24aÞ

and the activation enthalpy, DH p ðPÞ; at constant pressure becomes

T1 ðT; PÞ ¼ BðPÞexp½^DH p ðPÞ=RT ð4-24bÞ

where the above equations have a plus sign when v0 t q 1 and a minus sign when
v0 t p 1:
A hysteresis may be observed in a – b pressure transition and b – a decompression
transition. For further insight into the scope of NMR studies under high pressure, the
readers are advised to consult Jonas (1982).

4.4.10. Thermoluminescence

The high-pressure thermoluminescent (TL) and photoluminescent (PL) techniques

have been described by Drickamer and his co-workers. With pressure, the trap depths were
seen to change. The glow curve peaks shift to higher temperature with increasing pressure.
From the glow curve data such as the glow curve maxima Tm and heating rate ðBÞ; the trap
depth ðEÞ can be calculated using the relationship of Hoogenstraaten (1958):
 
E Tm1 Tm2 B1 Tm2 2
¼ ln
K Tm1 2 Tm2 B2 Tm1

where Tm1 and Tm2 correspond to the glow curve maxima at heating rates B1 and B2
and k is Boltzmann’s constant. Strains in the lattice may cause trapping. The strain
may arise from the pressure of ions of different sizes and may extend over several
unit cells.
The effects of pressure on the TL are illustrated with some case studies.
Principles of Techniques 387

4.4.10.1. ZnS phosphor

ZnS phosphor contaning Mn2þ, Cuþ and Clþ has been investigated by Lang
et al. (1992). Glow peaks: The typical glow curves using the higher heating rate
(0.74 K s21) at pressures of 0, 1, 4 and 6.5 GPa are obtained. At ambient pressure,
two peaks at , 133 and , 236 K are seen. With increasing pressure as these peaks
increase in depth (grow in intensity,) new shallower peaks appear. The peaks also shift
to higher temperature with increasing pressure. With increasing pressure, this peak
grows in intensity while the others fade. At , 2.5 GPa, a fourth peak appears and
grows in intensity as it shifts to higher temperature with increasing pressure. The glow
peak associated with the trap initially at 0.49 eV is seen to decrease rapidly in
intensity and disappears by 1.5 Pa.
The glow peak for the trap initialy at , 0.24 eV showed an initial decrease in
intensity, then an increase by a factor of 2 –3 to , 2.5 GPa, after which it lost intensity
rapidly and disappeared by , 6.2 GPa. The glow peak for the trap that first appeared at
, 1 GPa increased in intensity up to , 4 –5 GPa and at 6.7 GPa it faded to nearly half the
intensity. The lowest energy trap showed its glow peak maximized at 5 GPa and decreased
by a factor of 1.5 at # 7.7 GPa.

4.4.10.2. Trap depths

The shifts of trap depths with pressure were observed. The lower temperature peak
gave a trap depth of 0.3 eV and the higher temperature peak a depth of 0.5 eV. At 1.5 GPa,
a very weak peak appeared with a trap depth of 0.15 eV and, above this pressure, both the
earlier peaks of (0.3 and 0.5 eV) disappear.
At ambient pressure, the excitation is from the top of the valence band to the
conduction band and the energy gap is 3.67 eV. The gap increases by 6.4 meV per GPa.
Therefore, the traps are apparently fixed in energy with respect to the top of the valence
band and the depth of the trap increases as the gap increases.
Above about 2.5– 3 GPa, excitation to the conduction band is no longer possible.
The deep level energies are often unattached to nearby band edges and do not follow them
as they move as a result of pressure (Hjalmarson et al., 1980). As the conduction band
increases in energy (Drickamer, 1990) vis-à-vis the valence band, deep levels originally at
conduction-band energies are exposed and excitation takes place to these levels. At higher
pressures, no significant increase in trap depth occurs due to a release from the traps to
these levels.

4.5. Computational simulations

4.5.1. Introduction

Nowadays, laboratory experiments are no longer indispensible because computer

simulations have become sufficiently precise in theoretical solid-state calculations. The
computational method even helps to develop important physical insights into what cannot
388 Chapter 4

be extrapolated from data or measurements. The two extremes of theoretical spectrum —

semi-empirical to first principles — reflect two views of the microscopic world.
In these, matter is perceived as composed of atoms and, to describe forces between
pairs or triples of atoms, a simple functional form is adopted. Parametrization is made such
that subsets of experimental data are reproduced with a fair degree of precision.
These methods have been successfully employed in ionic compounds and non-
transition metals. In the former, most of the energy is a sum of pair-wise (Coulombic)
interactions. These, however, fail significantly to explain the properties of iron at core
pressure, such as magnetism, and the elastic instability of bcc iron at high pressure.
Any first-principles theory would view solids as composed of nuclei and electrons.
In this construct, atoms and ions play no primary role. The simplest system of this can be
conceived to be a uniform electron gas with embedded nuclei. In this system, the total
energy consists of the kinetic energy of electrons and three distinct contributions to the
potential energy: (a) Coulomb interactions among nuclei and electrons, (b) electron
exchange and (c) electron correlation. The first involves sums over point charges and
integrals over the electronic charge density. Correlation accounts for the Coulomb
repulsion, whereas exchange embodies the Pauli exclusion principle. Exchange and
correlation reduce the total energy by reducing the Coulomb repulsion between electrons.
The pair potential is expressed as the sum of Coulomb energy.
Bukowinsky and Wolf (1986) have used Watson sphere-stabilized oxygen wave
functions in which the stabilizing sphere has a radius of 2.66 Bohr radii and a charge of
þ 1. For Si4þ, they have used Hartree – Fock wave functions. While computing the
dynamic properties, the vibrational-mode frequencies are obtained in the quasiharmonic,
adiabatic, rigid-ion approximation. The calculated elastic constants, ðCij Þ; correspond to
the long wavelength limit of the dynamical matrix (Maradudin et al., 1971)
Theoretical approaches can be categorized into three classes: (a) ab initio, (b) first-
principles approximations and (c) semi-empirical methods. The principles of these are
outlined below.

4.5.1.1. Ab initio methods

These methods are widely used in determining the interactions between orbitals and
charge density and also in exploring the properties of liquids, including transport. These
are more convenient to handle than the complex and elaborate first-principles calculations.
In one of their modifications, such as the tight-binding method, the charge density and
wavefunctions do not appear explicitly but the elements of the Hamiltonian and overlap
matrices are approximated by simple parametric functions that are so constructed that the
results of first-principles calculations are reproduced.

4.5.1.2. First-principles approximations

These methods are employed to solve the Schrödinger equation with a bare
minimum of approximations. In first-principles calculations, the system of interest is
represented as a collection of atomic nuclei and electrons and the total energy and the
forces on all the nuclei are obtained by solving Schrödinger’s equation determining the
electronic ground state.
Principles of Techniques 389

In calculations using first principles, one starts from the fundamental interactions
among electrons and nuclei, with little constraint of the experimental data. In the case of
ferroelectrics, the first-principles methods are based on density functional theory (DFT),
while some are based on Hartree –Fock theory. Kohn and Sham (1965) showed how to
compute the charge density and energy self-consistently, using an effective exchange-
correlation potential (VXC) that accounts for the quantum-mechanical interactions between
electrons.
The local density approximation (LDA) takes the exchange-correlation potential
from the uniform electron gas at the density for each point in the material.
The generalized-gradient approximation (GGA) includes the effects of local
gradients in density (Perdew et al., 1996). Given a form of exchange-correlation energy
EXC ðrÞ; one can find the self-consistent charge density and compute the energy for any
arrangement of nuclei (atoms). From the energies of a zero-temperature phase diagram,
phonon frequencies and elastic constants can be computed.
The calculations are based on DFT with the electronic exchange-correlation energy
treated using either the LDA or the more sophisticated GGA. First-principles calculations
on a large variety of materials, including transition metals, have been shown to give
accurate predictions of both static and dynamic quantities. Compared with LDA, more
successful results have been obtained by GGA in the studies of transition metals,
particularly iron.
Carr and Parrinello (1985) first showed how a liquid can be studied performing
first-principles molecular dynamics (MD). This was done by integrating Newton’s
equation of motion for the nuclei with the forces calculated from the first-principles ground
state. The first-principles MD technique that they pioneered uses the pseudopotential
approach, in which only the valence electrons are explicitly represented, with the orbitals
expanded on a plane-wave basis. Techniques of this kind have been widely used for
studying liquid metals and their accuracy in predicting both structural and dynamic
properties is well established (Holender et al., 1995).
In the approximation restricting only to the exchange-correlation potential, the
present state of the art employs what is known as the Linearized Augmented Plane Wave
(LAPW) method. First Principles Linearized Muffin Tin Orbitals (FP-LMTO) calculations
are also similarly precise.
The other approximation strategy is the pseudopotential method, in which some
physically motivated assumptions allowing rapid computation are made. However, all
these involve frozen-core approximation and pseudopotential approximation. In the
former, the structure and energies of the electronic core states are assumed to
be independent of crystal structure, pressure or other perturbations to the lattice. In the
latter, the potential due to the nucleus and core electrons is replaced by a
simpler, smoother object (i.e., the pseudopotential), which has the same scattering
properties.
First-principles theory is a complementary approach that has taken on increased
significance in recent years in exploration of the properties of Earth materials under
extreme conditions such as those of the lower mantle (23 –135 GPa). Elaborate electronic
calculations are much more reliable than traditional atomistic calculations, which are
390 Chapter 4

based on interatomic potentials (usually parametrized from existing experimental

information) (Karki and Stixrude, 1999).
One of the most accurate first-principles methods currently applicable to a wide
variety of materials in studying a wide range of properties is the plane-wave pseudo-
potential approach within the framework of DFT (Payne et al., 1992).
For computation, some have used the Cambridge Serial Total Energy Package
(CASTEP) and Cambridge Edinburgh Total Energy Package (CETEP) codes, based on the
DFT under two essential approximations (Payne et al., 1992). They first used LDA
followed by pseudopotential approximation. They also optimized the lattice constant and
internal structural parameters of a given structure at several pressures. From the stresses
generated by small deformations (strains) of the equilibrium, the elastic constants were
determined. The ionic positions were re-optimized in the strained lattice to account for any
coupling between strain and vibrational modes.

4.5.1.3. Density functional theory: Kohn –Sham equations

Traditionally, quantum chemistry is based on Hartree –Fock as a first approxi-
mation, followed by the inclusion of correlation effects. The correlated methods are
extremely computationally intensive and cannot be fully applied to crystals. This is
because an infinite number of many-body states can be formed even from a limited basis by
coupling different k-vectors.
DFT in the LDA or GGA gives accurate predictions for energetic properties of
closed-shell systems as well as ionic/covalent crystals and open-shelled transition metals
and transition-metals oxides.
In the frozen phonon method, one displaces atoms and computes the change in
energy as a function of displacement from which the potential surface for an instability or
phonon frequencies can be derived.
It should be noted that self-consistent methods are extremely computationally
demanding.
Many-body terms are important in understanding the relative stability of the
transition-metal structures. Exact ways of solving Schrödinger’s equation for solidus are
not available, hence DFT and LAPW may be used for solving the equations. The first-
principles DFT provides a powerful means for predicting the behaviour of solids
independent of experimental data. In DFT, the charge-density equation is solved self-
consistently along with the Schrödinger-type Kohn – Sham (1965) equations. The
equations are solved with the LAPW method. This is an all electron method — all core
and valence electrons are fully included but it makes no approximations to the shape of the
charge density or the potential.
In principle, the DFT is an exact theory of the ground state (Kohn and Sham, 1965).
It may involve some approximations such as LDA, which is used to the exchange
correlation energy functional and pseudopotential approximation. The latter recognizes
that the core electrons are essentially static (frozen) and do not vary their charge density
significantly under perturbations to the lattice. The pseudopotential mimics the potential
seen by the valence electrons because of the nucleus and frozen core electrons.
Principles of Techniques 391

Pseudopotential. Structural changes under pressure have been investigated theoretically

by using pseudopotentials (e.g., Chelikowsky et al., 1990a,b). The inter-atomic potentials
calculations have been used to explain the lattice dynamical and elastic instabilities of
AlPO4 (Keskar et al., 1994).

Example: AlPO4. Christie and Chelikowsky (1998) determined the structural properties
of a-berlinite, AlPO4, as a function of pressure via total energy using ab initio
pseudopotentials. The calculations were based on ab initio norm-conserving
pseudopotentials being rapidly convergent with a plane-wave basis (Troullier and Martins,
1990, 1991a,b).
The oxygen pseudopotentials were generated from the 2s22p4 non-spin-polarized
oxygen atomic valence ground-state configuration, considering a core radius cut-off of
1.79 a.u. for both the s and p pseudopotential components. The aluminium pseudopotential
was generated from the 3s23p13d0 aluminium atomic valence ground-state configuration,
using a core radius cut-off of 2.14 a.u. All the pseudopotentials were generated from
the 3s23p13d0 phosphorus atomic valence ground-state configuration, using a core radius
cut-off of 1.70 a.u. for all the pseudopotential components.
The pseudopotentials were transformed into the computationally efficient Klein-
man – Bylander separable form using the p-component as the local component for all the
elements. One electron Schrödinger equation was solved using an iterative diagonalization
technique (Troullier and Martins, 1991a,b). Plane waves up to an energy cut-off of 49 Ry
were included in the basis set. Two reduced k-points in the Brillouin zone were used to
construct the charge density for self-consistency in the potential for all calculations.

4.5.2. LCAO model

In the linear combination of atomic orbitals (LCAO) model, the valence band-
wave function is constructed from a suitable lattice sum of combinations of atomic
orbitals at each ionic site. When the atoms are far apart, the LCAO scheme gives an
accurate description of the energy spectrum: it is simply a superposition of the atomic
line spectra. This can be illustrated with the case of MgO, as discussed by Bukowinski
(1985).
Initially, the Mg 3s state is higher than the partially occupied O 2p state. As the
atoms are brought closer together, the potential barriers between them are lowered,
allowing the Mg 3s state to tunnel through to the oxygen atoms. This exchange of charge
will raise the O 2p energy while lowering the energy of the Mg 3s state.
In addition, a Coulomb interaction arises due to the unbalanced charges on the
atomic sites. An equilibrium state will be achieved when the Mg 3s and O 2p states have
the same energy, resulting in a totally degenerate valence state.
Meanwhile, the ionic charge increases with decreasing inter-atomic distance. When
the distance is sufficiently small to cause significant overlap among the valence orbitals on
neighbouring atoms, the degeneracy is lifted, the magnitude of the perturbation being
proportional to the crystal field and the overlap. The atomic orbitals hybridize and their
392 Chapter 4

energies are separated into bonding, non-bonding and anti-bonding levels. Since there are
only six valence electrons, the anti-bonding states remain unoccupied. They correspond to
the Mg 3s states. As the inter-atomic distance is further decreased, the hybridization
increases to the point where it is no longer possible to distinguish a simple superposition of
atomic orbitals. The valence charge density would now clearly exhibit the symmetry of the
crystal field. Whether a major part of the charge is localized in the manner of an ideal ionic
solid depends largely on the difference between the ionization energies of the original
neutral atoms.

4.5.2.1. Molecular dynamics simulation

MD simulations as a function of pressure and temperature have provided useful
microscopic insights in the study of the structure and dynamics, phase transitions and
thermodynamical properties. Any system for which the total energy can be calculated may
be simulated at finite temperatures and pressures. Matsui et al. (1994) have shown that, by
combining the MD method with quantum corrections to the MD values, it is possible to
simulate accurately the structural and thermodynamic properties of crystals over a wide
temperature and pressure range.
In the MD technique (e.g., Parrinello and Rahman, 1980), one starts with some
initial condition of the system of particles and then calculates its evolution as solids or as
liquids. The equations of motion are solved for the atomic positions and velocities in steps
of time of the order of a fs. Simulations are usually carried out on systems of a few tens of
atoms to even a million atoms with or without periodic boundary conditions. This is quite
adequate for a variety of phase-transition studies, including diffuse motion of atoms to
distances of the order of a nm.
Because of the huge computer time involved in the first-principles simulation,
which employs the Carr – Parrinello technique (Carr and Parinello, 1985), the
calculation is usually restricted to an order of 100 atoms. Semi-empirical potentials
are specifically useful for simulations of large, complex systems and a variety of
microscopic properties.
MD simulations on small systems of a few hundred atoms have also been extremely
valuable in providing a microscopic understanding of the structural and dynamical
phenomena, e.g., molecular rotation in solids, pressure-induced crystal ! amorphous
transition, prediction on new phases, diffusion and order – disorder transitions (Chaplot and
Sikka, 1993).
One can obtain fairly reliable semi-empirical inter-atomic potentials on the basis of
the experimental data on the structure and dynamics and use these potentials to calculate
the small differences in the free energies of the various structural phases with sufficient
accuracy to arrive at their phase diagrams with acceptable accuracy. The MD simulations
using the same potentials provide the atomic-scale mechanism of the phase changes as well
as the unknown structural details of the phases. Simulations can also help to investigate the
effects of Fe/Mg disorder and the point defects in the lattice.
Comparison of the results of simulation with those from the quasi-harmonic lattice
dynamical free-energy minimization is used to reveal the consequences of the
anharmonicity of lattice vibrations at high temperature. Through this simulation, it is
Principles of Techniques 393

possible to calculate the seismic velocities in the different phases at high temperature and
pressure conditions.
The MD simulations based on empirical potentials provide a powerful technique to
achieve a microscopic understanding of phase transitions and similar phenomena in
complex structures under extreme P – T environments. The results depend to a large extent
on the quality of the empirical potentials employed. It is here that the results from neutron
inelastic scattering and complementary data from diffraction and optical experiments
provide a basis for checking the potentials.
Matsui et al. (1994) have shown that, by combining the MD method with quantum
corrections to the MD values, it is possible to simulate accurately the structural and
thermodynamic properties of crystals over a wide temperature and pressure range.

4.5.2.2. Inter-atomic potential

In simulation studies, the empirical atom – atom pair potential, comprising Coulomb
and short-range interaction terms, can be used. The parameters are restricted by using
equal or nearly equal values for the same atom type in different crystals. The charges are
also expected to reflect the ionic state, although fractional charges less than their formal
values are required in the rigid-ion model. This effectively takes into account the
polarizability of atoms to reproduce the polar-optic-phonon frequencies as well as
the proper cohesive energies. The values of the potential parameters are obtained by the
requirement that the observed structure is in static and dynamic equilibrium. This requires
that each atom should be at a potential minimum, the internal stresses are zero and all
the eigen values of the dynamical matrix (squares of the normal mode frequencies) are
positive. Often the structure has to be relaxed slightly to ensure this requirement. The
relaxed structure and the known structure should be very close to each other. It may also be
required that the maximum of phonon frequencies is of the right order of magnitude or that
the elastic constants are reasonable or that the computed cohesive energy is of the right
magnitude. These latter conditions are not explicitly imposed but are used only as a
guideline. In principle, these conditions could be rigorously imposed but, for convenience,
parameters can be chosen that are nearly transferable between similar systems and have
some physical meaning reflecting the ionic state and radius, etc.

4.5.2.3. Tight-binding total-energy model

The Tight-binding total-energy (TBTE) is grounded in accurate first-principles
calculations. The parameters of the model have been fitted to LAPW band structures
and EOS of non-magnetic bcc, fcc and hcp phases of iron over a wide range of volumes
(40 – 90 bohr/atom). Cohen et al. (1994) showed that TBTE model precisely reproduces
LAPW results for Fe and accurately predicts EOS, phase stabilities, elastic constants and
phonon-dispersion curves of a number of transition metals. The TBTE model is non-
orthogonal and does not include pair potential or other structurally dependent non-band-
structure terms. The parametrization is not given in terms of coordination shells since these
are poorly defined in liquids and high-temperature solids. The interactions decay smoothly
with distance and vanish at 16.5 bohr. However, the TBTE model is observed to be
thousands of times more efficient computationally than LAPW calculations.
394 Chapter 4

The TBTE model used is based on a two-centre, non-orthogonal Slater –Koster

formulation. The hopping Hamiltonian and overlap tight-binding interaction parameters
between s, p and d orbitals (like sss, sps, etc.) are assumed to have the form:

Pi ¼ ðai þ bi Þexp½2c2i r f ðrÞ

where r is the inter-atomic distance and f ðrÞ ¼ {1 þ exp½ðr 2 r0 Þ=l21 is a cut-off function
with parameters r0 ¼ 16:5 and l ¼ 0:5 bohr. The Pi represents 20 parameters, 10 each for
the Hamiltonian and the overlap matrices. The interactions are relatively long range,
extending to more than three times the nearest-neighbour distance. However, many
tight-bonding models include the nearest interactions.

4.5.2.4. Potential-induced breathing model

Assuming that the electron and hole are localized, the total energy of the ground
state and the lowest excited state of the crystal is computed using the Potential-induced
breathing (PIB) model (Zhang et al., 1994). However, the PIB model can predict the
ground-state properties of the ionic solids very accurately with no adjustable parameters
(Isaak et al., 1990). In the PIB electron gas model, the crystal charge density is given
by overlapping charge densities of spherical ions which are stabilized with Watson
spheres. The total energy of the crystal is calculated with a LDA for the overlap kinetic
energy as well as the exchange and correlational functionals. It can be expressed as the
sum of three terms: the self-energy of each ion, the overlap energy and the Madelung
energy. The self-energies are the atomic (ionic) total energies. The overlap energy
includes the short-range electrostatic energy, the exchange-correlation energy and the
kinetic energy.

4.5.2.5. Variationally induced breathing model: MgO

The Variationally induced breathing (VIB) model is an ab initio Gordon-Kim
(1972) model based on DFT (Kohn and Sham, 1965). The principles of the simulation
study can be illustrated with the paradigm of lattice thermal conductivity of MgO
determined by using MD), a non-empirical ionic model (the VIB model) and Green –Kubo
theory (Cohen, 1998).
In this model, non-input from experiment is necessary. In MgO, the crystal charge
density is modelled as overlapping O22 and Mg2þ ions and the total energy is computed
using the LDA (Hedin and Lundqvist, 1971). Since O22 is not stable in the free state, the
ion is stabilized by a “Watson sphere”, a 2þ charged sphere included in the atomic
quantum computations (Mulhausen and Gordon, 1981). In the VIB model, the radii of the
Watson spheres is chosen by minimizing the total energy as a function of atomic positions.
Cohen (1998) and his associates found that this model does an excellent job for the thermal
EOS of MgO (Inbar and Cohen, 1995), gives melting behaviour consistent with other
potentials (Vocadlo and Price, 1996) and gives excellent results for vacancy formation and
diffusion in MgO. The methodology for arriving at the predicted thermal conductivity can
be further understood from the presentation of Cohen (1998).
Principles of Techniques 395

4.5.3. Electronic approximations: “Muffin tin”, KSS and Bloch’s theorem

The starting point for electronic approximation involves the one-electron

approximation, wherein the total electronic wave function is expressed in terms of
single-electron eigen states. In this approximation, each electron sees only an average
static field due to the charge distribution of other electrons and the total wave function is an
integration of single-electron eigen states.
Slater (1951) suggested that the exchange interaction could be averaged over the
electronic states of the system and that, for computing the exchange interaction,
the electron could be treated as a locally free electron with the actual local electron density.
Later, Kohn and Sham (1965) developed the Kohn –Sham – Slater (KSS) exchange
interaction:

Vex ¼ 23e2 a½3=8prðRÞ1=3

where rðRÞ is the electronic density and a an adjustable parameter with values smaller than
1.
For a single-electron state, U, one can obtain an equation with a simple local
potential VðRÞ as
½2h2 =2m7 2 þ VðRÞUðr; k; EÞ ¼ Euðr; k; EÞ ð4-25Þ
only when Vðr þ lÞ ¼ VðRÞ; and where l is a lattice vector and k is the wave vector of the
electron. Bloch’s theorem must also be abided by U as:
Uðr; k; EÞ ¼ fk ðr; EÞexpðik; rÞ ð4-26aÞ
where
fk ðr; EÞ ¼ fðr þ l; EÞ
In the muffin-tin potential approximation of the self-consistent iterations, the
atomic sphere radii are chosen such that the oxygen spheres are in contact with each other
and with the silicon spheres. This maximizes the fraction of the unit-cell volume that is
contained within the spheres and hence improves the convergence in terms of the number
of reciprocal lattice vectors that must be kept in the expansion of the Bloch states. This
choice of radii also turns out to be in fair correspondence with the ionic radii, as deduced
from the charge distribution in the crystal.
The solution Uðr; k; EÞ of equation (4-25) above is expanded in terms of basic
functions fðr; ki ; EÞ:
X
Uðr; k; EÞ ¼ Vðki Þfðr; k; EÞ ð4-26bÞ

where ki is the reciprocal lattice vector.

In the constant potential region, fðr; k; EÞ is represented by a single plane
wave:

fðr; k; EÞ ¼ expðik; rÞ ð4-26cÞ

396 Chapter 4

Inside the spheres, where the potential is spherically symmetric, fðr; k; EÞ is

expended in terms of atom-like function. Here, in the products of spherical harmonics and
solutions of the radial part of equation (4-25), the coefficients are so chosen that fðr; ki ; EÞ
is continuous across the sphere boundary.
An approximate solution of Schrodinger’s equation for a crystal is accomplished
with the help of the self-consistent symmetrized augmented plane wave (SAPW) method.
For a more detailed description of the method, see Bukowinski (1976).

4.5.3.1. Double exchange (DE) model

To explain the correlation between ferromagnetism and metallic electrical
conduction in Ca2þ-doped LaMnO3, Zener (1951) invoked a “double exchange”
mechanism. On the Mn ions surrounding the impurity Ca2þ ions, d-holes are produced.
These holes hop from one Mn ion to another Mn ion through a spin-independent transfer
mechanism.
If the ionic spins of two Mn ions are anti-parallel, this hopping process is
inhibited because a strong Hund coupling dominates for electrons on a given Mn ion.
Thus, hopping is dependent on the relative orientation of the ionic spins. The motion of
holes produces partial alignment of the ionic spins corresponding to lowering of kinetic
energy.
Now consider a lattice in which there are magnetic ions and mobile (itinerant)
electrons. The latter electrons may constitute a single band (ds band) and have strong
exchange interaction with localized spins. When the intra-atomic exchange integral ðJÞ is
far larger than the transfer integral, ðtÞ; i.e., J q ltl; a double exchange occurs.
A mobile electron hopping between two Mn ions has a transfer integral t and
interacts with spins of the Mn ions through the intra-atomic exchange integral J which may
be much greater than t: During hopping, the energy of the system changes by ^tcosðu=2Þ;
where u is the angle between the directions of the ionic spins in a semi-classical treatment.
In a more rigorous quantum-mechanical calculation, the u-spin relation is
1
S0 þ
cosu=2 ¼ 2 ð4-27Þ
2S þ 1
where S is the spin of a Mn ion and S0 is the total spin of the system. In a double exchange,
the interaction is proportional to cosðu=2Þ; not to cosu; as seen in the case of a Heisenberg
model.
The double-exchange phenomena are caused by the large Hund’s coupling JH : The
essence of double exchange is that, when an electron hops from site i to site j; its spin
parallel to Sic must change to become parallel to Scj : The hopping amplitude, tij ; thus
depends on the relative spin orientation (Ramirez et al., 1996). For two fixed sites i and j; it
is possible to choose phase factors so that tij ! ðtij =2Þ1 þ Sic : Scj =S2c ; tij cosðuij =2Þ:
The motion of mobile electrons can be described by the Hamiltonian:
X
H¼ tij cosðuij =2Þci cj
i;j
Principles of Techniques 397

where ci is the operator which creates (destroys) an electron with spin parallel to the
ionic spin at the ith site. Below a certain critical concentration of carriers, a system
(such as LaMnO3) has a ground state in which the spins are ferromagnetically ordered
but canted by some angle u: With temperature, the canting angle u would increase
or decrease to a transition temperature Ti when the system becomes a ferromagnet
or an anti-ferromagnet. Above the critical concentration, the system behaves as a
simple ferromagnet. The double-exchange physics implies that ferromagnetic order
increases the electron kinetic energy, thereby decreasing the effective coupling
strength.
The double-exchange phenomenon gives an obvious connection between electron
hopping and magnetic order –disorder in the spins. Disorder implies randomness in tij ;
which decreases below Tc or in a field. But this model suffers from the limitation that the
effects of bound states of holes around impurity ions are not considered.
An Mn3þ (3d4) ion has three electrons in the lower threefold degenerate d[ orbitals
and one excess electron in the upper twofold degenerate dz2 orbitals. The dz2 orbital on a
Mn4þ (3d3) ion is empty. By a strong Hund coupling, the d[ electron spins align to form a
localized spin with a magnitude equal to 3/2. The dz2 electrons, supplied by Mn3þ ions, are
itinerant in the ds band. They interact with the localized spins through the intra-atomic
Hund coupling but the ds electrons can hop from one ion to another by the transfer integral
of ds orbitals.
Now consider a lattice in which there are magnetic ions and mobile (itinerant)
electrons. The latter electrons may constitute a single band (ds band) and have strong
exchange interaction with localized spins. When the intra-atomic exchange integral is far
larger than the transfer integral, i.e., J q ltl; a double exchange occurs. In LaMnO3, the
electron – phonon coupling was proved to be strong (Millis, 1996). Lattice distortion splits
the on-site orbital degeneracy of the eg levels. This corresponds to the eg symmetry
distortions of the oxygen octahedra around a Mn site. A Mn site with no eg electron would
induce a breathing distortion of the surrounding oxygen ions and this breathing distortion
plays an important role in determining the compositional (i.e., x) dependence of the
structural-phase boundary.

DE-JT effects
Dynamic and cooperative. If there is a coupling between the electronic states and
vibrational modes, dynamic Jahn – Teller (JT) effect occurs. Through the cooperative JT
effect (CJTE), the phase transition is driven by interaction between the electronic states of
one of the ions in the crystals and the phonons. Measurements of heat capacities, elastic
constants and other properties offer clues for the CJTE phenomenon.
For a satisfactory explanation of various experimental data, the DE model is
combined with JT coupling, in which the electron – phonon ðe – pÞ interaction is
considered. It is observed that JT coupling reduces the Curie temperature ðTc Þ: This
combination model helps one understand the metal – insulator (MI) transition around Tc ;
namely, polarons localized by the combined effects of spin-polarons forming and the
dramatic increase of spin-disorder scattering above Tc :
398 Chapter 4

The combined effects of JT electron –lattice coupling and the double degeneracy
of eg orbitals in the CMR perovskites (e.g., manganites) in their metallic phase ðT , Tc Þ
were studied by Zhang et al. (1996). They found that, without the static JT distortion,
the carriers are electrons but when the static JT distortion is sufficiently large, the
carriers can be hole-like. It was also seen that JT distortion fluctuations contribute to
magnetoresistance at moderate to high temperatures. This is discussed further in Section
15.3.3.

4.5.4. LMTO method vs. APW and KKR

The linear Muffin tin orbital (LMTO) method is about two orders of magnitude
faster than the more rigorous (but only slightly more accurate) Korringa – Kohn – Rostoker
(KKR) and Augmented Plane Wave (APW) methods. Indeed, the LMTO method is the
linearized version of the KKR method. In it, the Dirac equation may be solved instead of
the Schrödinger equation. This yields an effective one-electron equation which is
essentially the Schrödinger equation with the inclusion of important mass velocity and
Darwin corrections; however, spin –orbit coupling is ignored (Skriver, 1984). Thus, this
semi-relativistic treatment allows the study of atoms with large mass. For sulphur and
oxygen (with low atomic mass), the relativistic treatment is unnecessary, whereas
relativistic corrections for iron are non-negligible but minor.
The LMTO method calculates pressures through application of the virial theorem
(Skriver, 1984).

4.5.4.1. Average pair correlation function for NN geometry: SiO2 glass

Average pair correlation functions at high pressure reveal significant changes in the
nearest-neighbour (NN) geometry in glass under compression.
An estimate of the real space correlation in the glass at high pressure may be made
from the Fourier sine transform of the structure factor.
ð
2
GðrÞ ¼ MðQÞQ½SðQÞ 2 1sinðQrÞdQ ð4-28Þ
p
where MðQÞ is a high-frequency filter that removes the finite truncation effects in the
transform. Because SiO2 glass contains more than one type of atoms, GðrÞ is a convolution
of all the Si – O, O –O and Si – Si correlations.

4.6. Shock pressure studies

Typical shock front velocities are of the order of 10 km s21 and the pressure ranges
from 10 to 500 GPa. Shock pressures can be very high, exceeding 600 GPa. The high
pressures in shocked materials are accompanied by very high temperatures, , 2,000 –
15,000 K. Shock waves are transient phenomena typically lasting , 1026 s. The upper
limits of dynamics pressure known today are: for light gas guns , 700 GPa and for
Principles of Techniques 399

laser-driven shock waves , 10 TPa. The pressure to break down nuclei into neutron stars
(i.e., , neutron binding energies nuclear volumes) could be , millions of electron
volts/(fermi)3 , 1022 GPa (Daniels, 1993).
The data on P – V – T EOS of shocked materials correspond to Rankine –Hugoniot
equations that follow from the conservation of momentum, mass and energy in the system
of the impact.
The other measurements that have been made under shock-compression
conditions include electrical resistance, laser XRD and spontaneous and coherent
Raman scattering. The material behind the shock front remains in a compressed and
heated state for several microseconds, which is long enough for millions of vibrations of
the atoms in the material. In this condition, the material may be regarded as a fluid
(see also Section 5.2.1). Hugoniot states are calculated by employing the impedance
match conditions to the measured initial density, impact velocity and shock-wave
velocity. A velocity-sensitive interferometer (VISAR (Velocity Interferometer System
for Any Reflector); Barker and Hollenbach, 1972) may be employed to investigate
compressive wave profiles.
Shock Hugoniot data on silicate rocks and minerals all demonstrate major shock-
induced phase transformations (e.g., Marsh, 1980). The Hugoniots of tectosilicates and
rocks composed chiefly of tectosilicates are all quite similar. For felspars, the measured
Hugoniots above 30 GPa can be interpreted in terms of hollandite structure (e.g., Sekine
and Ahrens, 1991). The Hugoniots for plagioclase and K-felspar have been recognized as
insensitive to composition. Shock-recovered samples of quartz and felspars from above
25 –30 GPa indicate transformation to diaplectic glasses, with densities and refractive
indices higher than fused glasses of the same composition (Velde et al., 1989). The density
increase may be due to the formation of diaplectic glass with 6-fold coordination of silicon
with oxygen. The condition for the formation of diaplectic glasses appears to be restricted
only to part of the high-pressure phase present in the Hugoniot state, which reverts to a
disordered diaplectic glass via solid – solid transition.
This page is intentionally left blank
 401

Chapter 5
(Crystalline) Materials Under High Pressure

5.1. Material properties

A material can be described using its macroscopic and microscopic properties. The
phenomena manifested by materials at pressures prevalent at the depths of the Earth are not
only first-rank problems of geosciences but stand at the forefront of modern condensed
matter physics (Hemley and Ashcroft, 1998). The core states in an atom remain sharp
delta-function-like states. These states are raised or lowered relative to their positions in
isolated atoms. The shifts are mainly due to the screened Coulomb potential from the rest
of the atoms in the crystal. The description of the former properties is obtained from its
thermodynamic behaviour. Through the elucidation of Boyles theory (1660), the pressure
ðPÞ; volume ðVÞ and temperature ðTÞ relationship was established.
For microscopic description, the basic Hamiltonian at the level of elementary
nuclear and nuclearcharge is expressed as:
^ ¼H
H ^ nn þ H^ en þ H^ ee ; ð5-1Þ

where H ^ ee are the kinetic energy of the nuclei and electrons, respectively, and the
^ nn and H
third term represents the Coulombic attractions. When the system is enclosed in a volume,
V; the stationary states of the fundamental Schrödinger equation can be represented by
^ CðVÞ ¼ EðVÞCðVÞ
H ð5-2Þ

The relation involves volume V; which is alterable by pressure. This quantum-mechanical

expression embodies a complex many-body problem. (Note: An atomic unit of pressure is
e2 =2a40 ; which corresponds to 14,720 GPa.)
The microscopic description of the properties of minerals, like other crystalline
solids, is seen to be governed by quantum mechanics, which governs the behaviour of
electrons and nuclei in solids. Again, the atoms in a mineral manifest space-group
symmetry. Hence, a knowledge of crystallography and solid-state physics becomes
important in understanding the mineral behaviour, especially with respect to the phase
transitions, equations of state (EOS), electrical and chemical transport, etc. However, in
minerals, it should be noted that interactions between electrons in atoms can be seen as
responsible for all such behaviour.
Conventional solid-state physics extends from electronic band theory, explaining
metals, insulators and semiconductors, to the theory of superconductivity and the quantum
402 Chapter 5

Hall effect. X-ray, neutron and light scattering have become powerful probes of structures
from microscopic to near-macroscopic length scales.
Quantum theory shows that the time-independent ground state of a system is given
by a complex anti-symmetric many-body wave function (whose square gives the
probability density of finding a particle in each point in space):

ð
rðrÞ ¼ dr2 dr3 dr4 · · ·Cðr; r2 ; r3 ; r4 ; r5 ; …ÞC p ðr; r2 ; r3 ; r4 ; r5 ; …Þ ð5-3Þ

In an atom or molecule, the eigenvalues are the well known energy levels of the electronic
system, which are modified by other atoms in the crystal.
The core states of atoms are sharp delta function-like states, which are
raised/lowered in energy with respect to their positions in isolated atoms. The valence
and conduction states broaden into bands. Under pressure, the bands broaden and become
different from those of atomic states (see Section 5.1.2).
Crystals must have a lower energy than the aggregate of the constituent atoms
separated from each other. The binding force between atoms is dominantly electrostatic
and the Pauli’s exclusion principle keeps the electrons apart. In ionic crystals, the binding
force is primarily due to the electrostatic attraction among ions (e.g., Naþ and Cl2) while
in covalent crystals the binding occurs through hybridization of valence electrons causing a
lowering of the energy of electrons (e.g., in diamond).
In metals, the binding force arises from embedding the atom cores in a sea of
itinerant electrons. In silicates, the bonds are nearly half ionic and half covalent. In van der
Waals bonding, the dipoles fluctuate on separated atoms or molecules and the forces occur
through local many-body exchanges and correlation interactions among electrons.
Excited state properties involve the energetic exciting of electrons out of their
ground-state configurations. These are depicted in optical spectra.
High pressure alters the nature of chemical bonds, electronic and crystal structures,
and thermal and mechanical properties of solids. Simple molecular solids may transform
into a polymeric phase before they become metals at high pressure (Maihiot et al., 1992).
Large molecules are stiffened by high pressure. That is how a droplet of pressure-frozen oil
(large molecule) becomes capable of denting a steel plate!!
Very soft materials manifest an increase in density by as much as 1,000% (i.e., 20
times) at pressures of several hundred GPa, while the incompressible ones may show up to
, 50% increase in density. With increasing density, electrons in all materials become
increasingly unstable and, above a critical density, the electrons delocalize into conduction
states, thus forming a metal. The metallization process occurs due to an electronic overlap
of valence and conduction bands.
Pressure may induce order, but it also can bring about disorder. Pressed beyond its
stability field, a crystalline matter may transform to an amorphous material. This
amorphization persists when the temperature is too low for recrystallization to the
equilibrium high-pressure crystalline phase. Such unusual metastable states manifest
varying degrees of disorder.
(Crystalline) Materials Under High Pressure 403

Much information on the bulk properties at high temperatures and pressures and on
single-crystal elasticity and strength anisotropy may be obtained by integrating the high-
pressure techniques, the scope of which are presented below (Table 5.1).
Pressure-induced phase changes or bonding changes follow several rules:
Rule 1
Open-structure collapse. Open structures stabilized by weak ionic or van der Waals forces
can easily collapse under pressure yielding to denser structures.

2:5 GPa
E:g:; KCl or NaCl ðBIÞ structure ! CsCl ðBÞ structure ðdenserÞ
P
ðdiscussed earlier in Section 3:1Þ

Rule 2
Valence number changes. A few valence bonds per atom are non-metallic, whereas those
with higher valence number are metallic. Under pressure, lighter elements behave like
heavier elements. Examples: Germanium (Ge) a semiconductor with open diamond
structure collapses to a white tin structure and becomes a metallic electrical conductor (like
aluminium). Silica (SiO2) (with SiO4 tetrahedra) at , 10 GPa and moderate T transforms to
stishovite (a TiO22-rutile structure) with hexavalent silicon.
In general, under pressure, an insulator transforms to a semiconductor or metallic state
(Mott transition). An exception to this rule is observed when high-valent cations transform
to lesser valency under pressure. Ferric iron is seen to reduce reversibly to ferrous iron at
,1 –2 GPa pressure (proved by 57Fe Mössbauer spectroscopy).
Rule 3
Effects on cooperative phenomena. An increase or decrease in magnetism and
superconductivity are seen under high pressure. For example, iron above 11 GPa (RT)
loses its ferromagnetic behaviour.
Rule 4
Reaction-path blocking: Certain reaction paths may be blocked. For example, poorly
crystalline graphite yields to ordinary cubic diamond but a hexagonal form of diamond

TABLE 5.1
The scope of different pressure techniques

Techniques Determination High-pressure range

XRD (hydrostatic) Lattice parameters Multi-megabar

Bulk moduli
RDX Shear modulus 200 GPa
Single-crystal elasticity tensor
Ultrasonic Velocities of VP and VS — their ,20 GPa
orientational dependence
Shock-wave Bulk elasticity HPT Hugoniot
Ab initio calculations Elasticity .300 GPa
404 Chapter 5

can only be prepared by subjecting highly crystalline graphite to a pressure of 13 GPa

and 1,5008C.
At high pressure, carbon takes the diamond structure (5 GPa), while silicon and
germanium take the white tin structure (10 GPa). White tin changes to a body-centred
tetragonal form with coordination number 8. Generally, the high-pressure forms of lighter
elements or compounds are suggested by low-pressure forms of chemically heavier
elements or compounds.

5.1.1. Thermodynamics, equilibrium and time interval

Thermodynamics provides a description of the equilibrium states of systems with

many degrees of freedom. It focuses on a small number of macroscopic degrees of
freedom, such as internal energy, temperature, density or magnetization, needed to
characterize a homogenous equilibrium state. In systems with a broken continuous
symmetry, thermodynamics can be extended to include slowly varying elastic degrees of
freedom and to provide descriptions of spatially non-uniform states produced by boundary
conditions or external fields. Since the wavelengths of elastic distortions are long
compared with any microscopic length, the departure from ideal homogeneous equilibrium
is small.
Thermodynamic equilibrium is produced and maintained by collisions between
particles or elementary excitations that occur at a characteristic time interval, t: In low-
temperature solids or in quantum liquids, t can be quite large, diverging as some inverse
power of the temperature T:
The mean distance l between collisions (mean free path) of particles or excitations
is characteristic velocity V times t: In solids, V is typically a sound velocity. Most
disturbances in many-body systems have characteristic frequencies that are of the
order t21 : If excited, they decay rapidly to equilibrium.

5.1.2. Many-body systems and broken symmetry

5.1.2.1. Crystalline symmetries: 5-fold symmetry, icosahedra and quasi-crystals

The system of crystals is classified based on 230 symmetries (see also Section
5.1.5). A periodic crystal is invariant with respect to a discrete set of translations only,
rather than to the continuum of translations that leave the high-temperature state
unchanged.
The concept of symmetry helps approach condensed matter phases, from high-
temperature fluids to low-temperature quantum crystals. A description of their symmetry
can be made in terms of order parameters, which can also be invoked to explain phase
transitions, elasticity, hydrodynamics and topological defect structures.
The concepts of broken symmetry and order parameters have emerged as unifying
theoretical concepts applicable not only to condensed matter physics but also to particle
physics and even to cosmology. Condensed matter physics has built on atomic and
(Crystalline) Materials Under High Pressure 405

molecular physics and also on classical and quantum mechanics. It also relies on statistical
mechanics and thermodynamics.
Along with these concepts, many-body theory can explain normal Fermi liquids,
electrons phonon-s, magnetism and superconductivity. Nature offers an unlimited variety
of many-body systems, from dilute gases to quantum solids to living cells and quark-gluon
plasma. These cover the subject which can be referred to as “hard” condensed matter
physics.
Mean field theories are set up. Mean field theory replaces the actual configurations
of the local variables (e.g., spins) by their average value and it neglects the effects of
fluctuations about the mean.
The densest packing of 12 spheres around a central sphere is icosahedral and this
rule may continue hierarchically. Since the early 1960s, many icosahedral clusters have
been found. The size of an icosahedron is limited by the increasing strain with increase in
size. At a certain size, a transition from icosahedron to cuboctahedron will probably occur.
The hierarchical packing may be obtained by arranging Penrose tilings in two or
three dimensions — a mathematical pattern that has the geometrical properties required of
a quasi-crystal. In 3D, Penrose tiles are obtuse and in acute rhombohedra form with angles
of 116.6 and 63.48 (see Mackay, 1998). However, hierarchy can also offer an alternative
to lattice repetition in providing an assembly of atoms with an infinite number of
almost identical or quasi-equivalent sites. Hierarchy has now appeared as a building
principle in a class of inorganic materials, the quasi-crystals. These are solids with 5-fold
symmetry as indicated by their diffraction patterns — a symmetry impossible for a
conventional pattern.
Quasi-crystals are a further step away from conventional crystals because they have
many centres of local icosahedral symmetry. We now may expect many more varied
structures beyond the austere domain of classical crystallography. Clusters of boron
suboxide, B6O, show icosahedral shapes, produced from hierarchial clusters with 5-fold
symmetry (usually forbidden to solid crystals). Without dislocated grain boundaries,
the glide planes in this boron suboxide are locked and so the particles are very
hard, suggesting promising technical applications (see Hubert et al., Nature, 391,
376 –378, 1998).

5.1.2.2. Broken symmetry

The macroscopic properties are governed by conservation laws and broken
symmetries. Associated with each broken symmetry are distortions, defects and dynamical
modes that provide paths to restore the symmetry of the original high-temperature state.
Magnetic systems have played a very important role in the development of our
understanding of broken symmetry.
These can reveal how the breakdown of symmetry occurs and results in generalized
elasticity. The phenomena we commonly observe involve an order of 1027 particles (e.g.,
as in a litre of water) and the motion of each of those particles can scarcely be observed.
However, we can observe microscopic variables, such as particle density, momentum
density or magnetization, and measure their fluctuations and response to external fields.
406 Chapter 5

It is these observables that characterize and distinguish the many different thermo-
dynamically stable phases of matter that account for why liquids flow, solids are rigid,
some are coloured, some are transparent, some are insulators and others are metals or
semiconductors, etc. Even the simplest atoms in aggregate can occur in different states. For
example, helium can be seen as gas, liquid or solid and also as a non-viscous superfluid at
low temperatures.

5.1.2.3. Electron excitations and band gaps

Presence of unpaired electrons builds up the magnetism. Hund’s rule maximizes the
net magnetic moment but decreases the total electrostatic energy. The formation of energy
bands (hybrid crystalline electronic states) may lead to intermediate- or low-spin magnetic
structures. In Fe3þ (3d5 : t2g3
and e2g electronic configuration) the net magnetic moment is
5mB (mB ¼ Bohr magneton) while that of Fe2þ ð3d6 : t2g 4
and e2g Þ is 4mB : The high-spin and
2þ 6
low-spin state of Fe ð3d Þ was shown in Fig. 4.5.
The origin of crystal-field splitting is not only due to the potential field of a point-
charge lattice but also due to bonding hybridization. Furthermore, d-states are not pure
atomic-like states but are dispersed across the Brillouin zone (energy varies with k). The
splitting is due to d –d interactions between next-nearest neighbours and to p –d and s– d
interactions between neighbouring (oxygen) ions. The d – d interactions lead to splitting
that varies as 1=r 5 : The splitting can also be effected by hybridization, which also mediates
the sign of the magnetic coupling J:
When the magnetic moments are unequally distributed over different sub-lattices,
ferrimagnetism results. From incomplete cancellation of aligned spins, a net spontaneous
magnetic moment arises. For ferromagnetic and anti-ferromagnetic phases, the magnetic
fields are aligned (below the Curie and Nèel temperatures), which gives rise to magnetic
splittings.
Excitations in a system can be induced by external (e.g., electronic or magnetic) or
internal (e.g., temperature) fields. In the presence of an intense external electric field, the
response can be a function of the magnitude of the field and the susceptibility is a function
of the field. This gives rise to a non-linear optical response (e.g., multi-photon excitations).
Electrons can be excited into extended or itinerant states (i.e., across the band gap) or the
excitations may be local (i.e., forming a localized electron –hole pair or exciton). The
electronic transitions involving the valence (bonding) band, the conduction (anti-bonding)
band and d-electron levels can be investigated by using optical spectra.
The study of the electronic structure of highly correlated transition-metal
compounds is important for a better understanding of a material and, in 1985, Zeanen,
Sawatzky and Allen (ZSA) proposed a theoretical phase diagram for it.
In addition to the on-site d – d Coulombic interaction ðUÞ employed in the Mott –
Hubbard theory, the ligand-valence band-width ðWÞ; the ligand-to-metal charge transfer
energy ðDÞ; and the ligand – metal hybridzation interaction are explicity included as
parameters in the model Hamiltonian. The high-energy-scale charge can be varied
moderately by external temperature and magnetic field but can be considerably affected
by pressure.
(Crystalline) Materials Under High Pressure 407

When an additional electron is added to a d-orbital, an energy (U; Hubbard)

increase occurs. For localization of the electron, the parameter that governs the tendency is
U=W; where W is the band-width. Mott or charge-transfer insulators (at low pressures)
become metallic under pressure because W increases with pressure and U decreases
because of increased screening. Thus, at high pressure, band theory is likely to be more
reliable. Again, Monte Carlo simulations (Gunnarson et al., 1996) predict metallic
behaviour when U=W , N; where N is the orbital degeneracy (5 in the d-orbital case).
This again shows that, at high pressure, band theory appears to be more appropriate.

5.1.2.4. Dielectric properties

The dielectric properties such as piezoelectricity, pyroelectricity and ferroelec-
tricity are structure dependent.
Piezoelectricity is manifested by electrical polarization caused by applied stress.
With the exception of point group 432, any crystal belonging to one of the 20 remaining
non-centrosymmetric point groups (that also contain a unique polar axis) is piezoelectric.
The motif common to many piezoelectric crystals is the tetrahedrally coordinated atom.
A common example of piezoelectric crystal is tourmaline, which is also
pyroelectric. In this, the silicon atoms in the tetrahedra are positively charged;
consequently, the extension along the polar axis would result in a negative polarity on
the (0001) face. The quartz form of SiO2 is a very important piezoelectric material used for
transducers and frequency-controlled devices.To cater for these applications, large quartz
crystals are grown by implanting seeds in large pressure vessels.
Pyroelectric materials are characterized by the presence of a spontaneous
polarization, Ps ; the magnitude of which is temperature dependent.
Ferroelectricity is manifested when the spontaneous polarization is capable of
reversal or re-orientation of its polar direction on application of an electric field.
Ferroelectric transition is either a displacive type or an order – disorder type. To understand
ferroelectric phase transitions, numerous experimental and theoretical studies are
undertaken. The origin of ferroelectric phase transitions in oxides is due to the anharmonic
potential surfaces caused by softening of the short-range repulsions by covalent
hybridization. These cause the atoms to move off-centre and towards each other.
Oxide ferroelectrics are studied for their (i) soft modes, using time-resolved
spectroscopy, (ii) atomic positions, by nuclear and X-ray studies, (iii) ground-state
potential surfaces, by electronic structure studies, (iv) electronic structure and
(v) macroscopic polarization.

5.1.2.5. Electronic and magnetic behaviour

The electronic and magnetic properties directly influence large-scale global
phenomena ranging from the initial differentiation of the planet, the formation and
transmission of the Earth’s magnetic field, the propagation of seismic waves and the
upwelling and downwelling of mass through the mantle (Hemley et al., 1998).
The pressure range within the Earth can compress the rock-forming silicates
and oxides by factors of 2– 3 and the molecular species and rare gases by well over an order
408 Chapter 5

of magnitude. Discrete magnetic and electronic transformations such as metallization and

magnetic collapse may also occur (e.g., Cohen et al., 1997).
Under pressure, volatiles can be bound as “valence lattices” in dense, high-pressure
phases, e.g., hydrogen in ice, mantle silicates and ferrous alloys. Pressure at depths may
dissociate Fe2þ –Mg2þ combination from oxide or silicate phases and incompatible
elements such as Fe and K may form alloys (Parker and Badding, 1996).
The valence and conduction states broaden into energy bands under higher
pressure. When an ion possesses a shell filled with electrons, it attains greater stability. The
interaction between ions through Madelung or strong electrostatic forces enhances the
crystal stability.

Band structures. In an insulator, the bands are filled and, due to the Pauli exclusion
principle, nothing can happen without exciting electrons to states above the gap. But
this excitation requires a large energy and, hence, in a small field no current can flow.
Metals have partially occupied states at the Fermi level and the current will flow.
In insulators, the highest occupied levels form the valence band, designated as Ev ;
and the empty energy levels form the bottom of the conduction band, Ec : The difference is
the band gap, Eg ¼ Ec 2 Ev : Crystals with band gaps between occupied and unoccupied
states should be insulators and those with partially filled bands should be metals.
From the intermediate states (formed by chemical doping), the electrons can be
excited into the conduction bands or holes in the valence bands and the crystal becomes a
semiconductor. In a non-magnetic system, each band holds two electrons. Thus, a crystal
with an odd number of electrons in the unit cell should be a metal since it will have at least
one partially filled band. Magnetic crystals that are insulators by virtue of local magnetic
moments are known as Mott insulators (Mott, 1990).
At energies intermediate between the valence and conduction bands, there can
occur localized states, which affect the optical and transport properties There is a possible
relationship between optical modes (high-frequency) and elastic properties (low-
frequency) acoustic modes.

5.1.2.6. Ionicity in bonding: Madelung forces

In crystals, atoms donate or accept electrons, resulting in ionicity. The ionicity is
driven by the increased stability attained through filling of the outer shell by electrons. For
example, oxygen having a nuclear charge (z) should have 1s2, 2s2 and 2p4 electrons but a
filled p-shell has six electrons. O22 ion is unstable in the free state but is stabilized by the
crystal field in an oxide. In an oxide or silicate crystal, the O grabs two more electrons to
form an O22 anion from another atom, usually a positively charged cation. This strong
electrostatic (called Medelung) interaction between the ions increases the crystal stability.
The ionic solids are dominated by electrostatic or Madelung forces between
charged ions. The alkali halides and alkaline earth oxides are proto-typical ionic solids.
Employing all modifications and the self-consistent methods, the EOS, elasticity, the
electronic and optical properties of these materials are obtained. In ionic crystals, the
Madelung energy is to be added to the overlap energies.
(Crystalline) Materials Under High Pressure 409

In atomic computation, most studies use a “Watson sphere” (Watson, 1958), which
is a charged sphere, usually of opposite charge to the ion.

5.1.3. Covalent bonding and hardness

Hard substances have a high number of strongly directed, covalent chemical bonds
per unit volume. Soft substances generally have fewer bonds per unit volume or bonds that
are weak or weakly directed, such as ionic or dipole attraction forces. Covalent (electron
pair) bond strengths vary between , 60– 90 kcal/mol for most elements present in hard
materials. The heavier elements generally offer more bonds per atom. A plot of hardness
measured by Knoop indenter vs. the bond energy per molar volume for various substances
is essentially linear.
The hardest materials are generally made of light elements, with diamond at the top.
Hard materials are brittle because the strongly directed bonds favour hardness but not
plasticity, which involves the inter-site motion of atoms. At high pressures, many brittle
materials become ductile.

5.1.3.1. Hardness and bulk moduli

Hardness ðHÞ of ionic and covalent materials increases with bulk modulus (Cohen,
1993). Diamond has the highest known bulk modulus, K ¼ 444 GPa, and it is also the
hardest material known, with its single-crystal H ¼ 90 GPa. It is followed by cubic boron
nitride (cBN) with corresponding values of K ¼ 369 GPa and single-crystal H ¼ 48 GPa
(Sung and Sung, 1996).
High-bulk moduli require high charges and small volumes; thus, tetravalent cation
dioxides could be hard. Although silicon is the smallest tetravalent cation, the common
forms of silicon dioxide are not hard because of their open structures, such as in quartz and
cristobalite phases ðNC ¼ 4Þ: But the denser phase stishovite has a much higher bulk
modulus, 298 GPa (Hemley et al., 1994). This value is much higher even than that of the
other common hard material, alumina, whose bulk modulus is 252 GPa. Indeed, among
polycrystalline materials, the hardness of stishovite (33 GPa) rivals those of the hardest
materials. The bulk moduli ðKÞ and the Knoop micro-hardness values of synthesized
compounds obtained by different workers were earlier presented in Table 3.5.
5.1.3.2. Phonon-s and band states
Most solids can be described by harmonic or by anharmonic approximations. In the
former, non-interacting phonon-s couple strongly and form a broad continuum band. In the
latter, strong interaction of phonon-s leads to the formation of states of bound quasi-
particles, coupled only weakly in the crystal. For example, in two-phonon- vibrational
spectroscopy, the quasi-particle is a biphonon- acting as a molecular oscillator. The
biphonon- forms a sharp peak separated from the origin of the broad two-phonon-
continuum band by the amount of its anharmonicity. The condition when the harmonic and
anharmonic cases overlap has been a problem of great interest in condensed systems. It is
postulated that if the relative magnitudes of the continuum bandwidth and the
anharmonicity could be varied so that the former would exceed the latter, the band
biphonon- would disappear into the continuum band.
410 Chapter 5

In a crystal, the band states are characterized by a continuous quantum number k; so

that the eigenvalues are 1ðkÞ and the eigenfunctions are represented by Bloch states
(see equations (4-27a) – (4-27c)):
fðk; rÞ ¼ UðrÞexp2ikr ð5-4Þ
where UðrÞ is a periodic function of position r: The eigenvalues (i.e., energy) as a function
of k are known as the band structure. By studying the band structure, densities of state
(DOS) and the charge densities, one can understand the nature of bonding and its changes
with chemistry, distortions and pressure.
Experimentally, bands can be studied by photoemission spectroscopy, which helps
determine the relative energies of emitted electrons as functions of input photon energy
and wavevector, or by employing X-ray spectroscopy.
In angle-resolved photoemission, the observed band structure is the energy
spectrum for removing electrons from the surface of the crystal, which is sometimes a
complex phenomenon — and may be broadened or shifted from the intrinsic energy levels
in the interior of the crystal.
In a non-magnetic system, each band holds two electrons and thus a crystal with an
odd number of electrons in the unit cell should be a metal because it will have at least one
partially filled band (except in Mott insulators). Magnetic crystals that are insulators by
virtue of local magnetic moments are known as Mott insulators (Mott, 1990). The real
energy states in a crystal are not single-particle eigenstates at each value of k; rather, there
is an energy spectrum which has more or less strong peaks at the quasi-particle energies.

5.1.4. Elasticity

According to Hooke’s law (Nye, 1985), the stress ðsÞ and strain ð1Þ for small
deformation in a crystal are linearly related by
sij ¼ Cijkl 1kl ; i; j; k; l ¼ 1; 2; 3
where the fourth rank tensor Cijkl is the elastic constant tensor.
Thus, the elastic constants can be determined directly from the computation of the
stress generated by small strains (Wentzcovitch et al., 1995). The cubic crystal has three
independent elastic constants, C11 ; C12 and C44 (in the Voigt notation).
The strained lattice (lattice vectors a0 ) used in determining the elastic constants is
related to the unstrained lattice ðaÞ by the relation a0 ¼ ðI þ 1Þa; where I is the identity
matrix. The strain tensor is
0 1
1 1=2 0
B C
1¼B @ 1=2 0 0C A ð5-5Þ
0 0 0
so that Hooke’s law gives
sxx ¼ C11 1; syy ¼ szz ¼ C12 1; syz ¼ C44 1; szx ¼ sxy ¼ 0:
(Crystalline) Materials Under High Pressure 411

For the lower symmetry of the strained lattice, the 4 £ 4 £ 4k-point mesh yields 20 special
k-points. The ion positions are still fixed by the symmetry so in the strained lattice only
electrons should be relaxed.
The elastic constants completely specify the elastic properties and acoustic
velocities of a single crystal. For the purpose of comparing with seismological data, it is
interesting to compute the elastic properties of an isotropic polycrystalline aggregate. The
bulk modulus of such an aggregate is well defined, whereas the shear modulus is inherently
uncertain, depending on the arrangement and shape of the constituent crystals (Watt et al.,
1976). The bulk modulus is related to the elastic constants by
K ¼ 1=3ðC11 þ 2C12 Þ ð5-6Þ
The isotropic shear modulus in the Hashin – Shtrikman averaging scheme (Hashim
and Shtrikman, 1962) is given by
G ¼ 1=2ðGU þ GL Þ ð5-7Þ
where the upper (U) and lower (L) bounds are, respectively,

5 18ðK þ 2C44 Þ
GU ¼ C44 þ 2 þ ð5-8Þ
Cs 2 C44 5C44 ð3K þ 4C44 Þ
and

5 12ðK þ 2Cs Þ
GL ¼ Cs þ 3 þ ð5-9Þ
C44 2 Cs 5Cs ð3K þ 4Cs Þ
where Cs ¼ ðC11 2 C12 Þ=2:

5.1.4.1. Elastic anisotropy

Anisotropy of crustal and mantle materials arises from the preferred
alignment (texturing) of the aggregate of intrinsically anisotropic minerals. For cubic
crystals, the elasic anisotropy is conveniently expressed in terms of Zener ratio ðAÞ; which
is the ratio of the shear moduli in the (100) and (110) planes in the [100] direction:
2C44 2ðS11 2 S12 Þ
A¼ ¼ ð5-10Þ
C11 2 C12 S44
For the elastically anisotropic case, A ¼ 1:
Under Reuss approximation, A can be directly measured from radial diffraction
experiments, without the use of any assumed bulk property (Singh et al., 1998; Merkel
et al., 2002) but, with increasing pressure, A value decreases (Karki et al., 1999).
For cubic materials, single-crystal elastic anisotropy and seismic anisotropy can be
related through the anistropy factor, A; defined as
2C44 2 C12
A¼ 2 1: ð5-11Þ
C11
For isotropic material, A ¼ 0:
412 Chapter 5

The anisotropy of a single crystal is determined by its elastic constant tensor.

For cubic phases, there are three independent elastic constants and, for hexagonal
phases, there are five. The elastic constants are related to the elastic wave (seismic)
velocities by the Cristoffel equation (Nye, 1985)
lcijkl nj nl 2 rV 2 dik l ¼ 0

where V is the velocity, r is the density, n is the propagation direction and Cijkl is the
elastic constant tensor and dik is the Kraenecker delta function. The elastic anisotropy
is determined by calculating the velocity of each of the elastic waves (one P and
two S) for all propagating directions. The data on elastic anisotropy have important
implications for the interpretation of seismological observations of the anisotropy in
terms of flow in the upper mantle (Tanimoto and Anderson, 1984). To determine the
elastic anisotropy of a polycrystalline aggregate, one must know the elastic constant
tensor of the individual crystals and the texture as specified by the orientational
distribution function.
The eigen values of the 3 £ 3 matrix yield the three unique elastic-wave velocities
for propagation direction n; whereas the eigenvectors yield the polarization directions
(Musgrave, 1970).

5.1.5. Elastic constants: crystal systems

The elastic constants are directly related to the inter-atomic potentials.

Consequently, the parameters of a potential are often determined from elastic constants.
In general, the number of independent elastic constants increases as the point-group
symmetry of the solid decreases: the most isotropic solids have the smallest number of
elastic constants. The highest symmetry that a three-dimensional (3D) crystalline solid has
is cubic symmetry and three independent elastic constants. The solid hexagonal symmetry
has only two independent elastic constants. In two- and 3D crystals of lower symmetry,
there are more elastic constants. For example, a 2D crystal with 4-fold symmetry (rather
than 6-fold symmetry, like a 3D crystal with cubic symmetry) has three elastic constants.
The numbers of independent elastic constants for 3D crystals are listed in Table 5.2.
For crystals of simple systems (e.g., cubic and hexagonal), the values for
room-temperature harmonic elastic constants and densities ðrÞ are shown in Table 5.3.

TABLE 5.2
Number of elastic constants for crystal systems

Crystal system No. of elastic constants

Triclinic 21
Monoclinic 13
Orthorhombic 9
Tetragonal 6 or 7
Rhombohedral 6 or 7
Hexagonal 5
Cubic 3
 (Crystalline) Materials Under High Pressure
TABLE 5.3
The elastic constants of some phases crystallizing in cubic and hexagonal systems

Name Ref (composition) r C11 C12 C44 Reference

Cubic crystals
Fluorite (CaF2) 3.180 1.64 0.53 0.337
Chromite (FeCr2O4) 4.450 3.225 1.437 1.167 Hearmon (1956)
Diamond (C) 3.511 9.320 4.112 4.167
Galena (PbS) 7.5640 1.020 0.380 0.250 Hearmon (1956)
7.5640 1.270 0.298 0.248
Gold (Au) 19.300 1.925 1.630 0.424 Neighbours et al. (1958)
Periclase (MgO) 3.583 2.963 0.951 1.559 Anderson et al. (1966)
Periclase (at 298 K) 5.390 2.230 1.200 0.790
Pyrite (FeS2) 5.016 3.818 0.310 1.094 Simmons et al. (1963)
Halite (NaCl) 2.162 0.487 0.131 0.127 Lewis et al. (1967)

Name (composition) r C11 C12 C13 C33 C55 Reference

Hexagonal crystals
Apatite 3.218 1.667 0.131 0.665 1.396 0.663 Hearmon (1956)
Beryl 2.68 2.800 0.990 0.670 2.480 0.658 Hearmon (1956)
Biotite 3.05 1.860 0.324 0.116 0.540 0.058 Alexandrov et al. (1961)
Cancrinite 2.460 0.520 0.086 0.124 0.826 0.238 Alexandrov et al. (1961)
Muscovite 2.790 1.780 0.424 0.145 0.549 0.122 Alexandrov et al. (1961)
Phlogopite 2.820 1.780 0.3020 0.152 0.510 0.065 Alexandrov et al. (1961)
b-quartz (873 K) 2.533 1.166 0.167 0.328 0.104 0.361 Hearmon (1956)
Sphalerite (ZnS) 4.089 1.312 0.663 0.509 1.408 0.286 Klerk (1967)

413
414 Chapter 5

The temperature derivatives of elastic constants are determined by using the value at room
temperature and the harmonic value of the elastic constants. Commonly, elastic constants
decrease with temperature but, for MgO, NaCl and other halides, C12 values are seen to
increase with temperature.
The experimental values of second-order elastic constants, G and K (in
1012 dyes/cm2), of different minerals (polycrystals) at room temperature (298 K) were
obtained by Cheng (1974).

Name composition G K
(a-Cr2O3) 1.298 2.321
Almandine 0.951 1.765
Almandine 0.943 1.770
Sp.Almandine 0.966 1.756
Sp.Almandine 0.961 1.777
Spess-alm-pyrope 0.049 1.750
Hematite (a-Fe2O3) 1.213 2.088
Fayalite (a-Fe2SiO4) 0.536 1.220
Fayalite (b-Fe2SiO4) 0.815 2.050
Spinel (MgAl2O4) 1.080 1.972
Fe–spinel (FeAl2O4) 0.853 2.103
Magnetite (Fe3O4) 0.773 1.769
Magnetite (Fe2TiO4) 0.263 1.210
Ferrosilite (FeSiO3) 0.608 1.018
Forsterite (Mg2SiO4) 0.797 1.281
Enstatite (MgSiO3) 0.788 1.066
Ortho-pyroxene 0.686 1.041
(Mg0.5Fe0.5)SiO3
(MnFe2O4) 0.692 1.851
(NiFe2O4) 0.713 1.823
a-quartz (SiO2) 0.447 0.378
SiO2 –rutile (SiO2) 1.696 2.93
Rutile (TiO2) 1.124 2.155

In a cubic crystal, there are only three unique elastic constants, C11 ; C12 and C44 :
The shear velocity in the 100 direction is calculated from ðC44 =rÞ1=2 and the compressional
velocity in the 100 direction from ðC11 =rÞ1=2 ; where r is the density. The shear velocity in
the 110 direction is calculated from {ðC11 2 C12 Þ=2r}1=2 and the compressional velocity
in the 110 direction from {ðC11 þ C12 þ 2C44 =2r}1=2 :
Values of A (equation (5-11)) greater than 1 signify that C44 is greater
than 1=2ðC11 – C12 Þ; whereas the opposite holds when A is less than 1. For gold,
the elastic anisotropy is large; A ¼ 2:9 at ambient pressure. An extrapolation
of ultrasonic data suggests that this should increase weakly with pressure (Duffy
et al., 1999).
At the transition, the C112 2 C12 instability gives rise to an anomalous
decrease in the shear-wave velocity, which provides a seismic signature that could be
diagnostic of the presence of a separate phase (e.g., free silica in the deep mantle
and D00 zone).
(Crystalline) Materials Under High Pressure 415

The estimates of elastic constants of minerals are generally taken from zero-
pressure experiments. However, the anisotropy can be strongly pressure dependent
(e.g., in MgO; Karki et al., 1997).

5.1.5.1. Cauchy relation and its violations

The Cauchy relation, defined as C12 2 C44 ¼ 2P; is valid when all inter-atomic
forces are central under static lattice conditions.
While studying MgO, Karki et al. (1997) observed that, as pressure increases, the
calculated value of C12 2 C44 2 2P decreases (Cauchy violation). The decrease
is relatively slow up to 100 GPa and then rapid between 100 and 150 GPa, as shown
in Fig. 5.1. The relatively faster decrease above 100 GPa is due to the slow increase
of C12 between 100 and 150 GPa. The initial pressure dependence of the
deviation from the Cauchy condition agrees fairly well with low-pressure ultrasonic
behaviour.
Since MgO remains a wide-gap insulator to pressures well beyond the deep-mantle
pressure (when covalent bonding or metallic bonding is less significant), the Cauchy
violations cannot be explained. The PIB model, with a zero-pressure value of 272 GPa for
C12 2 C44 2 2P; has also shown a very similar pressure dependence of the Cauchy
violation (Isaak et al., 1990).

Figure 5.1. Pressure variation of the Cauchy violation in MgO. The circles represent the calculated values. The
first-order extrapolation from ultrasonic data (Jackson and Niesler, 1982) is shown by the dashed line (Karki et al.,
1997; q 1997 Mineralogical Society of America).
416 Chapter 5

5.1.6. Born’s stability criteria: B1 ; B2 and B3

The mechanical stability of crystal lattices can be estimated from their elastic
moduli by the so-called Born’s stability condition based on the elastic lattice energy
(Binggelli et al., 1994). For the stability of a trigonal lattice, the following three conditions
need to be satisfied:

B1 ¼ C11 2 lC12 l . 0;

2
B2 ¼ ðC11 þ C12 ÞC33 2 2C13 .0

2
B3 ¼ ðC11 þ C12 ÞC44 2 C14 . 0: ð5-12Þ

For the elastic stability of a phase, the determinant as well as sequential principal
minor of the matrix of elastic stiffness coefficient ðcij Þ must be greater than zero
The first condition, B1 . 0; ensures stability with respect to the elastic waves in the
basal plane perpendicular to the c-axis; B2 . 0 implies positive compressibility and B3 .
0 is associated with shear acoustic modes in the y – z-plane (Terhune et al., 1985). When
one of these moduli vanishes, the initial crystalline structure becomes homogeneously
unstable against the corresponding fluctuations.
Under a given P; T condition, the lattice is elastically stable when all the three
conditions of elastic moduli are satisfied.
The values of the three parameters ðB1 ; B2 and B3 Þ can be estimated from the
calculated elastic moduli under compression.
For trigonal quartz-type phases at ambient pressure, six calculated values (in GPa)
obtained by different workers are

C11 C33 C44 C12 C13 C14 Reference

55.1 123.0 26.2 18.1 22.2 24.2 Tsuchiya et al. (2000)
64 118 37 22 32 2 Grimsditch et al. (1998)

Both B1 and B2 parameters increase regularly with pressure and are positive in all
pressure ranges up to the transition. However, B3 ; which indicates the shear stability of the
lattice, decreases and becomes zero at , 7 GPa as a result of decrease in C44 with pressure
(Fig. 5.2; Tsuchiya et al., 2000). The stability condition vanishes near the transition
pressure. This indicates that lattice instability from the shear softening of the quartz-type
lattice is induced by pressure (e.g., a-quartz study by Binggeli et al., 1994).
Tse and Klug (1991) reported a sudden decrease in the modulus B2 as a function of
time at the critical pressure. For a trigonal structure, B2 can be related to the stability of
volume compressibility which is given by

ðC11 þ C12 2 4C44 þ 2C33 Þ=B2 : ð5-13Þ

A discontinuous volume reduction is seen to occur at the first-order transformation

because the decrease in B2 to zero corresponds to the divergence in compressibility.
(Crystalline) Materials Under High Pressure 417

Figure 5.2. Calculated Born’s parameters of quartz-type lattice with pressure. The B1 ; B2 and B3 values are
shown in (a), (b), and (c) by solid circles, respectively. Earth value estimated from experimental
measurements is also plotted by an open circle. The lattice is elastically stable if these parameters are positive.
The B3 value decreases with pressure at 2 GPa and changes to negative near to amorphization pressure at
7 GPa. The result is related to the negative correlation of C44 with pressure (Tsuchiya et al., 2000, q
Springer-Verlag).

Hence, the violation of the condition B2 . 0 can be considered as caused by the

transformation.
For quartz, B2 softens at 22.3 GPa. For ice, B1 and B2 soften at the pressure of
, 0.9 GPa when becomes amorphous. This observation may suggest the question: Is an
amorphous phase a consequence of mechanical melting? MD calculation showed that, at
, 22.3 GPa, a-quartz would show an abrupt softening of B2 : This softening is due to
transition to a disordered phase. Employing first-principles pseudopotentials, it is seen that,
at , 30 GPa, B3 softens but B1 and B2 stiffen with pressure (Bingelli and
Chclikowsky, 1992).
418 Chapter 5

The result of Born’s stability criteria indicated that the volume collapse of the
quatz-type lattice with pressure-induced transformation originates from pressure-induced
shear elastic instability. The effect of the shear stress field decreases transition pressure and
enhances lattice instability as well as the uniaxial stress field. The quartz-type structure
transforms to the rutile-type structure on decompression when sxy is imposed (Tsuchiya
et al., 2000).

5.1.7. Thermoelasticity

From the experimental P – V – T data, the thermoelastic parameters can be derived

using a high-temperature Murnaghan EOS. Based on the thermodynamic identity

dP
KT ðP; TÞ ¼ 2V ð5-14Þ
dV T

and with KT ðP; TÞ expressed as (isothermal T ¼ 300 K):

dKT d2 K T
KT ðP; TÞ ¼ KT0 þ K 0T0 P þ ðT 2 300Þ þ PðT 2 300Þ ð5-15Þ
dT 0 dP dT

The P – V – T relation can also be defined by equation

21=b
b
VðP; TÞ ¼ VðO; TÞ 1 þ P ; ð5-16Þ
a

where

dKT
a ¼ K T0 þ ðT 2 300Þ
dT 0

and

d2 KT
b ¼ K 0T0 þ ðT 2 300Þ
dP dT 0

At high temperatures and pressures, a nearly uniform behaviour of a number of

thermoelastic parameters of oxides and silicates has been noted in experimental and
theoretical investigations. Above the Debye temperature, the product aKT has been
observed to be independent of temperature and pressure.
Important thermoelastic parameters such as the temperature derivative of bulk
modulus, dK=dT; the pressure derivative of thermal expansion, da=dP; and Anderson–
Grüneisen parameter, d T ; need to be derived experimentally at simultaneous high
pressures and temperatures. These thermoelastic parameters can be derived from
measurements of volume change of the crystallographic unit cell at simultaneous high
pressures and temperatures. Recent breakthroughs in high P – T single crystal X-ray
diffraction using a diamond-anvil cell (Zhao et al., 1995) provide most accurate
(Crystalline) Materials Under High Pressure 419

measurements of unit-cell dimensions at uniform temperature and sustained hydrostatic

pressure.

5.2. Atomic vibrations in crystals: phonon-s

In crystals, atomic vibrations propagate as weakly interacting waves with wave

vector k and frequencies vj ðkÞ: To each wave, one can assign an oscillator with frequency
vj ðkÞ:
Phonon-s. Each wave can behave as a particle (de Broglie) with energy equal to
Évj ðkÞ and momentum p equal to Ék: This quasi-particle, called a phonon-, is an
elementary component of sound energy with frequency v; just as a photon is related to
light (electromagnetic) energy. (Thus, wave ! quantum oscillator ! phonon-s.)
Phonon- gas. A solid may be conceived as a box containing phonon- gas and, like
molecules in ordinary gas, phonon-s collide with one another. The energy of phonon- gas is
the sum of the energies of individual phonon-s but the collision of phonon-s does not
conserve momentum. However, a crystal lattice does not always participate in the
collisions of phonon-s.
As the temperature falls, the number of high-momentum phonon-s becomes smaller
and smaller. At T p Q (Debye temperature), almost all phonon-s are close to the centre of
the first Brillouin zone.
By combining the dependence of heat capacity and mean free path on temperature,
one obtains

J , T 3 expðQ=TÞ:

Near absolute zero, the mean free path is enormously large. Even at extremely low
temperature, the number of phonon-s in a crystal is enormous, e.g., at ð1=10ÞQ (tenth
Debye) temperature. About 1020 phonon-s are present per cubic centimetre.
The phonon- gas is the main heat reservoir of a solid but the number of phonon-s in
a solid is not constant. High temperatures create more phonon-s and the number of phonon-
s is proportional to (the third power of) temperature. However, most of the phonon-s have
energies close to kB T:
There are acoustic and optical phonon-s. The velocity of phonon-s corresponds to
the speed of sound. A study of resonance absorption of light in a crystal would reveal the
properties of optical phonon-s.
Neutron scattering and phonon- study. Low-momentum phonon-s are ordinary
sound waves. Hence, by studying sound propagation in crystals, the properties of
individual phonon-s can be studied. The inelastic scattering of neutrons in crystals
provides very important information on phonon-s. In its motion through a crystal, a neutron
makes the atoms “swing” and creates sound waves and its own energy is lowered. The
change in the neutron’s energy is equal to the phonon-’s energy. This is what follows from
the dispersion law of phonon-s. Employing high-energy neutron sources, the obtained
phonon-’s energy spectra reveal the construction of solids.
420 Chapter 5

5.2.1. Elastic waves in crystals

The longitudinal ðVP Þ and shear ðVS Þ wave velocities of isotropic aggregates are
given by (see Section 2.6.2):
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffi
K þ 4=3G G
VP ¼ ; VS ¼ ð5-17Þ
r r

where r is the density and K and G are isotropic bulk and shear moduli, which are
determined from the single-crystal elastic constants ðCij Þ using the Hashin – Strikman
averaging scheme shown by equation (5-7). The bulk sound velocity is given by
(see Section 2.6.2)
sffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
K 4
VB ¼ ¼ VP2 2 VS2 ð5-18Þ
r 3

Temperature is expected to lower the density and seismic wave velocities, while addition
of Fe will increase the density and decrease the seismic-wave velocities.
The seismic-wave velocities of periclase in the lowermost part of the lower mantle
are lower than those of Mg-silicate perovskite by 3– 5%. Additional Fe has a strong effect.
Magnesio-wustite with XFe ¼ 0:20; a value appropriate for a pyrolite bulk composition,
has a shear velocity that is 15% lower than that of MgSiO3 perovskite.
It should be possible to detect a number of minor or secondary phases in the lower-
mantle magnesiowüstite, CaSiO3 perovskite and SiO2 silica — all have distinctive wave
velocities and should be seismically distinguishable.

5.2.1.1. Shock waves

Shock waves are intense compressional waves of short duration (typically
, 1026 s) created by the high velocity of projectiles or by intense light pulses. Shock
pressures can be high, exceeding 600 GPa. Shock pressure may accompany high
temperatures , 2,000– 15,000 K.
Shock-pressure measurements may involve P – V – T EOS. The data are based on
Rankie – Hugoniot equations that follow from conservation of momentum, mass and
energy in response to the impact. The observable quantities are US and UP ; VS is the shock-
propagation speed and VP the particle speed. The shock speed VS is measured by timing the
shock propagation over a known distance. The particle velocity VP is inferred from
collision properties of the impact or from the observation of free-surface velocity as the
shock wave emerges from the shocked material.
Shock measurements can be made for electrical resistance, laser X-ray diffraction
and spontaneous and coherent Raman scattering. Each shot in a shock experiment
generates one point in the EOS.

Shock waves in crystals. To study true shock waves at the atomistic level, non-
equilibrium molecular dynamics (NEMD) simulations with Newton’s equations of
(Crystalline) Materials Under High Pressure 421

motion are solved on the computer for thousands of strongly interacting atoms (Straub
et al., 1980). True shock waves exhibit steady profiles (density, velocity, stress and
energy) which accompany dissipative, irreversible flow of atoms in the direction
transverse to the planar-wave propagation. The NEMD profiles could also be
explained by a continuum constitutive model, which uses Navier – Stokes equations of
hydrodynamics.
Planar shock waves in a single crystal become steady waves through transverse
displacements of atoms, not by viscous flow as in fluid shock waves (Hoover, 1979) but
rather by plastic flow through the concerted slippage of atoms over each other.
In the Lennard –Jones (LJ) pair-potential solid, represented by fcc lattice, a
shock wave travelling in the (100) direction would result in slippage along one of the
four available [111] planes through emission at the shock front of a Shockely partial
dislocation. The wave leaves behind a stacking fault (the usual ABCABC… stacking
of a triangular-lattice, when the close-packed planes become ABABCA). For steady
planar shock waves, depending on their strength, the Hugoniot relation offers a general
statement of mass, momentum and energy conservation linking a given initial
equilibrium state.
To obtain the sound velocities from acoustic-mode formation, the following
relation can be used:

w a
V ¼ V0 ð5-19Þ
w0 a0

where the subscript zero represents values at one atmosphere.

Velocities vs. pressure can be used to calculate the pressure dependence of the
elastic moduli from the relation

KS ¼ rðVP2 2 4=3VS2 Þ; G ¼ rVS2 ð5-20Þ

where KS is the adiabatic bulk modulus, G the shear modulus and r the density.

5.2.1.2. Shock velocity and particle velocity

In shock-wave measurements, the paired variables are particle velocity ðUp Þ and
shock velocity ðUs Þ; from which, and with HugonistEOS, the pressure –density relations
are calculated.
A new pair of variables has been chosen to improve the accuracy of primary
calibration of anvil devices. They are the density ðrÞ measured with XRD and the acoustic
velocity ðVf Þ measured with the ultrasonic method or Brillouin scattering on the same
sample under the same compression (Mar and Hemley, 1998).
Pressure is derived from
ð
P ¼ Vf2 dr

The resultant P – r relation is a primary pressure standard.

422 Chapter 5

The metastable pressure – volume Hugoniot data may be converted to the

corresponding shock-velocity – particle-velocity ðVS – VP Þ data using the relation
(Liu, 1975):

VS ¼ V0 ½Pm =MðV0m 2 V m Þ1=2 ð5-21Þ

VP ¼ ½Pm ðV0m 2 V m Þ=M1=2 ð5-22Þ

where Pm and V m are the metastable Hugoniot pressure and volume, respectively, V0m the
initial volume of the high-pressure phase and M the molecular weight. It has been shown
by Ruoff (1967) that the data in the VS – VP plane can be represented by

VS ¼ C0 þ sVP þ s0 VP2 ð5-23Þ

and, if an EOS is assumed to have the following form:
0 00
K S ¼ K0S þ K0S P þ 1=2K0S P2 ð5-24Þ
then

K0S ¼ r0 C02
and
0
K0S ¼ 4s 2 1
0
where K0S is the isentropic bulk modulus at zero pressure, K0S is its first pressure derivative
evaluated at zero pressure and r0 is the initial density.
The method of converting the experimental Hugoniot to a new Hugoniot, centred at
a different initial condition, was demostrated by McQueen et al. (1963). In doing so, they
obtained

e g Vm g
P 12 e 21 2 1 ðE0e 2 E0m Þ
2 V V
Pm ¼ ð5-25Þ
g V0m
12 21
2 Ve
where P; V; E and g are the pressure, volume, specific internal energy and the Grüneisen
parameter, superscripts “e” and “m” denote the experimental and metastable Hugoniots
and subscript zero represents the initial condition. Some quantities have to be evaluated
before calculations from the foregoing equation can be performed. Firstly, determine the
change of the specific internal energy between the starting material and the high-pressure
phase at zero pressure:

DEtr ¼ E0e 2 E0m ¼ P½ðV m 2 V e Þ þ ðV0m 2 V0e Þ=2 ð5-26Þ

assuming that the temperature effect is negligible.
(Crystalline) Materials Under High Pressure 423

5.2.1.3. Shock-induced transitions

Shock-induced phase transitions are observed by noting: (i) the breaks or
discontinuities in the shock velocity ðVS Þ– particle velocity ðVP Þ curves, (ii) the abrupt
changes in sound velocity, (iii) the multi-wave structure in wave profiles of pressure or
particle velocity and by analyzing the shock-recovered samples.
The shock-treated sample data are compared with those obtained from in situ
experiments to know if the observed discontinuity in the Hugoniot P – V curve indicates
any new phase not observed by an in situ experiment. On the P – V – T surface, the shock
Hugoniot function chalks out a path different from an isotherm in static experiments.
Lattice disorder is observed in diverse minerals induced by stresses # 30 GPa (Horz and
Quaide, 1973). Pronounced lattice disorders for quartz and plagioclase felspar are observed
when shocked to , 22 GPa (Cygan et al., 1992). Conversion of graphite to diamond occurs
in rocks shocked to , 30 GPa, at a pressure distinctly short of melting (e.g., Koebertl et al.,
1995).
Most oxides and silicates (excepting MgO and Al2O3) undergo shock-induced
phase transitions in pressures up to 200 GPa (Ahrens, 1980).
Shock-induced phase transition can be of reconstructive or non-reconstructive type,
depending on whether the transition involves atomic diffusion or not. The latter type may
be completed within a micro-second time scale of a shock experiment. Martensite or
electronic transition belongs to this type. The martensitic nature of B1 – B2 transition in
NaCl shocked parallel to [111] occurs at pressure lower by 4 GPa than that shocked
parallel to [100] (Fritz et al., 1971).
In shock compression of Fe3O4 and a-Fe3O3, the transition is due to a change in
configuration from a high-spin state to a low-spin state. Mössbauer spectroscopic (and
XRD) study in DAC showed a transition to a non-magmetic state at , 50 GPa (Mao et al.,
1977). The volume change accompanying phase transition is predicted to be 13% based on
the density-cation radius systematics of the corundum structure. Both the crystal-field
splitting ðDÞ and the spin-pairing energy ðpÞ depend on the volume (Goto et al., 1982).
Under pressure, the following relations with the inter-atomic distance ðrÞ are assumed:

D / r 23:0
p / r1:3
Shock-compression experiments have been performed on a single crystal of
forsterite (Mg2SiO4) up to 170 GPa (by Syono et al., 1981). Extraordinary high values of
HEL (up to 12 GPa) were found to be direction dependent. The onset of phase transition
was at ,50 GPa, which is much higher than those known high-pressure transformations in
static experiments, such as , 20 GPa for olivine to modified spinel or to spinel transition
(Akimoto et al., 1976) and below 30 GPa for post-spinel dissociation to ilmenite or
perovskite type MgSiO3 plus MgO (Ito et al., 1982). The mixed phase region persists up to
,100 GPa, where the shocked state merged into the high-pressure regime (Syno, 1984).
Reconstructive-type phase transformations under shock compression preferentially
occur along the shear zone where temperature is high enough to promote the
transition, resulting in very heterogeneous textures in the shock-recovered materials.
424 Chapter 5

A heterogeneous yielding model has been developed for brittle, low thermal conductivity
materials, in which dynamic yielding occurs along the stress-concentrated sheared zone
(Syono, 1984).

5.3. Inelastic and non-hydrostatic states

From the 1920s, pioneers such as P.W. Bridgman, D. Griggs, H. Heard, M. Paterson,
W. Brace et al. worked on minerals and rocks under high-pressure –high-temperature
conditions. These involved inelastic processes such as brittle fracture, frictional sliding,
plastic deformation and dynamic metamorphism. The results have been applied to handle
diverse problems related to the rheology of Earth’s materials, such as the flexure of the
lithosphere, metamorphic textures, post-glacial rebound, deformation under partial
melting, forces driving plates, the physical nature of earthquake sources, seismic
anisotropy and so on. It is, however, now more clearly known that non-hydrostatic stresses
generated in high-pressure phase transitions are responsible for deep-focus earthquakes
(Kirby et al., 1991).

5.3.1. Stress states

A solid sample placed between a pair of flat faces of perfectly rigid anvils will
experience stress states which can be expressed along a set of three orthogonal axes ðx1 ; x2
and x3 Þ: The x1 and x2 axes are in a plane parallel to the anvil face, and x3 is along the load
axis. The centre of the anvil face is taken as the origin of the coordinates. For circular anvil
face, the stress state is cylindrically symmetric about the x3 axis. Thus, the stress component
s11 equals s22 : For a finite shear strength of the sample, s33 $ s11 : Since the anvil faces are
flat and parallel, the stress gradient in the direction of applied load (x3 axis) is absent.
The maximum shear stress for von Mises criterion of yielding leads to the relation

s33 2 s11 ¼ t ¼ 2ty ¼ sy

where t; ty and sy are, respectively, the uniaxial stress component, shear strength and yield
strength of the specimen. And the pressure sr ¼ ðs11 þ s22 þ s33 Þ=3 ¼ ðs11 þ t=3Þ
ð{ s11 ¼ s22 Þ: The value of t can be determined from the diffraction data.
The stress at the origin of the coordinates is given by a tensor:
0 1 0 1 0 1
s11 0 0 sr 0 0 2t=3 0 0
B C B C B C
sij ¼ B C B
@ 0 s11 0 A ¼ @ 0 sr 0 A þ @ 0
C B 2t=3 0 C A
0 0 s33 0 0 sr 0 0 2t=3

¼ sr þ Dij ð5-27Þ

The second term in equation (5-27) is the deviatoric stress component. The strain
produced by sij is a superposition of stress components. As the load on the anvil is
(Crystalline) Materials Under High Pressure 425

increased, sr increases and can reach a few hundred gigapascals. The magnitude of t is
normally a few % of sr : The strain produced by Dij is small and is adequately described
by linear elasticity theory. The adequacy of the linear elasticity theory is linked to
the precision of the lattice strain measurement, which is at best one part in 10,000
(Singh et al., 1998).

5.3.1.1. Non-hydrostatic stress

Under conditions of uniaxial non-hydrostatic stress, a phase transformation (or
distortion) could be observed to occur at a lower pressure than it would under hydrostatic
conditions. Experimentally, this has been shown by Zou et al. (1981).
The pressure-induced phase transition is reflected by changes in diffraction patterns.
It is possible to detect small distortions in the unit cell caused by the deviators from the
hydrostatic condition of the stress acting on the sample.
The measurement of unit-cell dimensions as a function of pressure gives the
pressure – volume relation (i.e., EOS) of a solid sample.
Ostapenko (1971) has shown that, for two polymorphs in hydrostatic equilibrium,
the denser modification will become more stable if a uniaxial stress of significant
magnitude is applied. A similar result has been shown to be true for second-order
transitions where the more rigid form (higher bulk and shear moduli) becomes the stable
phase if a sufficiently large uniaxial stress is applied (Ostapenko, 1973).
Under conditions of non-hydrostatic stress, the concept of thermodynamic
equilibrium is not valid. The local phase equilibria differ depending on the orientation
of the phase interfaces to the stress and, therefore, transformations occurring under
non-hydrostatic stress cannot strictly be located on an equilibrium-phase diagram.

Differential stress: examples. The rheological constraint at depth involves measurement

of small differential stresses at high P – T: Stress measurements at depth along the
San Andreas fault system suggest that plate-scale motion may result from shear stresses as
small as 10– 20 MPa (Hickman et al., 1988).
Microstructures of mantle xenoliths suggest that, with depth, as temperature and
mean stress increase, the differential stresses decrease from ,15 to ,5 MPa. (Note: At
70 km, mean stress is 2.2 GPa.) This is attained by large-scale isostatic adjustments
(Hager, 1990) and by the shape of the Earth’s geoid (Jeffreys, 1963). Moreover, the flow
strengths of crystalline solids weaken exponentially with temperature. (Note: At high
pressure, quartz may get weakened hydrolytically.)

Deviatoric stress. Deviatoric stress is the total stress with 1/3 the trace (sum of the
diagonal terms) subtracted from each of the diagonal terms. When the stress is deviatoric,
the sum of the principal stresses is zero. This reflects the shape-altering stresses and is the
driving force field for flow. A purely deviatoric stress can be represented in a coordinated
system where all the diagonal terms are zero and the off-diagonal terms are non-zero.
Deviatoric stress can be measured in multi-anvil apparatus from the broadening of
XRD lines. Uniaxial compression in DAC can ideally be used for quantitative study of
deviatoric stress at ultra-high pressures (Singh et al., 1998).
426 Chapter 5

The uniaxial component or deviatoric stress is a measure of t ¼ ðs1 2 s3 Þ: The

yield and shear strengths of a material, designated as sy and t; are related to the maximum
t; which the material can support as sy ¼ 2t ¼ t: This equility (Von Mises condition) is
satisfied only when the material is observed to deform under pressure. In high-pressure
experiments, the assumption commonly used is that t ¼ sy :
Shear stress ðtÞ is determined from the measurement of pressure gradients on the
sample (in a DAC) by

t ¼ ðh=2ÞdPðrÞ=dr

where h is the sample thickness. This analysis depends critically on the shape of the
diamond in the pressure cell and is only valid when the diamonds remain flat (no cupping)
and the sample continues to flow under loading (Spitzig and Leslie, 1971).
The stress difference between materials (assuming that the stress is equivalent to the
pressure) is obtained from the EOS of each material. This approach significantly
overestimates the deviatoric stress because it neglects the effect of the shear modulus on
the measured differential strains (Meng and Weidner, 1993).

Gold standard. The effect of deviatoric stress (i.e., departures from hydrostaticity) on the
unit-cell volume measurement and further on the volume-based pressure calibration is
directly proportional to K=G of the measured material (Meng et al., 1993). As gold has an
unusually high value of K=G (6.2) compared with most samples (1 –2), the measurement
of the unit-cell volume of gold is very sensitive to the existence of even a small amount of
deviatoric stress. At RT; the deviatoric stress in gold increases with increasing pressure
above 15 GPa. By 30 GPa, the calculated deviatoric stress is above 0.6 GPa. The deviatoric
stress decreases dramatically with increasing temperature and disappears by a temperature
of about 650 K. The pressure dependence of the uniaxial stress t for gold is obtained as

t ¼ 0:06 þ 0:015P;

where P is the pressure in GPa in (Duffy et al., 1999).

The method for determining the deviatoric stress in a diamond-anvil cell, the
deviatoric stress in gold as a function of pressure and temperature and also the deviatoric
stress effect on the volume-based pressure calibration have been discussed by Meng et al.
(1993). The elastic moduli of gold as a function of pressure is shown in Fig. 5.3b.

5.3.2. Crystallographic shear

Anion-deficient non-stoichiometry is accommodated in certain transition-metal

oxides by eliminating point defects by the so-called crystallographic shear, cs, which
provides a means of accommodating anion-deficient non-stoichiometry without introdu-
cing point defects and without change of cation coordination. Crystallographic shear
phases can be regarded as translation modulations of the parent structure, the translation
boundaries being cs planes.
(Crystalline) Materials Under High Pressure 427

Figure 5.3. The elastic moduli of gold as a function of pressure. (a) second order elastic moduli of gold as a
function of pressure. The doted lines show finite strain extrapolations of ultrasonic data (Hiki and Granato, 1966)
while the simple lines are obtained from study up to 37 GPa. The vertical axis stands for Cij (GPa). The symbols
show individual data points and solid lines are finite strain fits (Birch, 1978) to the present data combined with
ambient pressure data (Hiki and Granato,1966). Uncertainties are one standard deviation.

In the ordering of shear planes, elastic strain appears to play a crucial role.
Extensive relaxation around the shear planes seems essential for stabilizing some
structures. Isolated cs planes offer solid – solid interfaces.
5.3.2.1. Shear and deformational twinning

The extensional axis ð11 Þ is parallel to ½101 in the (101) plane of the tetragonal
phase of cristobalite. This plane consists of distorted six-membered tetrahedral rings, and,
in the high-temperature cubic phase, becomes a pseudo-close-packed (111) plane. This
direction is equivalent to ½1 12;
 the slip direction for the smallest possible “shear” of the
{111} layers. The cristobalite I ! cristobalite II phase transition can, therefore, be viewed
as a shear of the crystal structure. This involves the sliding of adjacent close-packed sheets
against each other, much like the martensitic-type transformation in close-packed metals.
Palmer and Downs (1991), in their single-crystal diffraction work on the high-
pressure phase of cristobalite, observed splits in single-crystal diffraction peaks (measured
in v and v=2u scans), indicative of deformation twinning. These deformation twins are
distinct from the tetragonal to monoclinic transformation twins. However, a co-existence
of both high- and low-pressure phases within the hysteresis region has been observed in
single-crystal X-ray experiments (Downs and Palmer, 1994) and in Raman spectroscopic
studies (Palmer et al., 1991).
5.3.3. Strain anisotropy in crystalline mass: e.g., hcp iron

Radial XRD (RXD) measurements (vide Section “Radial X-ray diffraction (RXD):
deviatoric stress” of Chapter 4) provide single-crystal elasticity information (Singh et al.,
1998a,b). In a laboratory experiment, the stress (non-hydrostatic) state of a sample is
determined by compressing the sample between the two anvils and by employing the RXD
technique.
428 Chapter 5

The results from RXD (zero-frequency) and multi-anvil (ultrasonic frequency)

measurements show good correspondence, bracketing a wide frequency range including
seismic waves. The K=G ratios obtained for hcp iron by the two methods under different
pressure experiments are (Mao et al., 1998):

Method Pressure (GPa) K=G

RXD 20–39 2.7 ^ 0.7
Ultrasonic 16.5 2.68 ^ 0.1

In the RXD measurements on hcp iron (Mao et al., 1998) the strong ðhklÞ
dependence of lattice strain reflects a strong elasticity anisotropy or an ðhklÞ dependence of
stress (Singh and Balasingh, 1994).
A strong hkl dependence of t; defined as t ¼ s3 2 s1 ¼ 1:5ðs3 2 sr Þ (s3 ¼ axial
stress; s1 ¼ radial stress) in the polycrystalline specimen may partially or fully account for
the observed lattice-strain anisotropy. Again, the development of the basal plane slip
texture common in hcp metals could conceivably lower the t of grains with their c-axis at
458 orientation.
The integrated study by Mao et al. (1998) reveals that the Earth’s inner core may
represent the low shear modulus of hcp iron close to melting or the existence of additional
components with low shear-wave velocities but with similar densities.
The seismic VS of the inner core is lower. If this softening of VS represents possible
Born –Durand pre-melting effects (Tallon, 1979) or partial melting, the observed shear
values would constrain the inner-core temperature to near-melting. The attenuation of
seismic waves may be affected by near-melting softening or by the presence of additional
low-velocity phases in the inner core.
Theories predict a small VP anisotropy (4% faster in the c than in the a direction) of
the inner core. This requires a perfect alignment of hcp iron crystals or the presence of a giant
single crystal (Stixrude and Cohen, 1995). A partial alignment of hcp iron crystals may be
sufficient for accounting the magnitude of the inner-core anisotropy (Mao et al., 1998).

5.4. Spontaneous strain

In phase transition, the spontaneous strain, 1; is defined in terms of the lattice

parameters. For example, for a cubic to tetragonal phase transformation, if the cubic phase
has lattice parameter a0 while the tetragonal phase has parameters a and c; the spontaneous
strain 1 is measured as
1 ¼ ða 2 a0 Þ=a0
which is proportional to Q2
If in the transition there is no change in the translational symmetry and the unit cell
merely changes its shape, no extra reflection would appear in the diffraction pattern. In
such a case, if no extra points appear in the reciprocal lattice, the transition does not
involve a change in the boundaries of the Brillouin zone. This type of transition is called a
zone-centre transition, where 1 / Q holds.
(Crystalline) Materials Under High Pressure 429

In cases where the translational symmetry changes with changes in the reciprocal
lattice and the boundaries of the Brillouin zone, they are called zone-boundary transitions.
Here, the relation 1 / Q2 commonly holds. When, in the transition (e.g., cubic )
tetragonal), a reduction in translational symmetry and the formation of a superlattice occur,
the relation 1 / Q2 holds.
The symmetry of the low-temperature form must be a subgroup of the
high-temperature symmetry. In other words, the low-temperature form is derived from
the high-temperature form by a loss of some symmetry elements, i.e., the structures are
topologically similar.
Excess free energy, expressed as G rather than DG; is defined as the excess over that
which the high form would have if the phase did not occur. Usually, the excess quantities
are stated as G; S; etc., rather than DG; DS; etc.

5.4.1. Spontaneous strain and order parameter

For a displacive phase transition involving a supergroup to subgroup relation, the

order parameter, Q; may couple to the lattice strain. The resulting excess lattice distortion
is characteristic of the phase transition and is usually expressed as the spontaneous strain, 1
(Carpenter, 1992).
The spontaneous strains for a tetragonal ! orthorhombic transition are given by
11 ¼ ða 2 ao Þ=ao
12 ¼ ðb 2 ao Þ=ao
and
13 ¼ ðc 2 co Þ=co
where a; b and c are lattice parameters of the orthorhombic phase, while ao and co are
linear parameters of the tetragonal phase extrapolated into the stability field of the
orthorhombic phase for P42 =mnm ! Pnnm (Carpenter et al., 2000) transition, which is
pseudo-ferroelastic in nature (Wadhawan, 1982).
The symmetry-breaking strain is
a2b
ð11 2 12 Þ ¼ ;
ao
which is proportional to the order parameter Q:
The volume strain is given by
Vs ¼ ðV 2 Vo Þ=Vo ;
where V is the unit-cell volume of the orthorhombic phase and Vo is the volume of
the tetragonal phase. For small strains, Vs ; 11 ; 12 and 13 are related approximately as
Vs ¼ 1 1 þ 1 2 þ 1 3 :
Of great current interest is the transformation in dense SiO2 (stishovite) from the
rutile to a CaCl2-type structure ðP42 =mnm ! PnnmÞ; which is thought to be a classic
displacive transition (Karki et al., 1997d). Single-crystal XRD measurements near
430 Chapter 5

50 GPa pressure in quasi-hydrostatic media have proven the P42 =mnm ! Pnnm
transition, which is largely displacive but exhibits hysteresis indicating weak first-
order character. A Landau model has been developed that quantitatively relates all of the
spectroscopic, elastic, structural and thermodynamic data for the Si – O system. The
elastic instability at the transition gives rise to anomalies in the Raman spectrum,
which are expected to be a general feature of such pressure-induced transition
(Hemley et al., 2000).
Since Q can never be measured directly, measurements of macroscopic properties,
such as the spontaneous strain, are essential for the thermodynamic description of the
phase transitions. This is illustrated below with a case of tetragonal to monoclinic
transition seen under pressure (Cristobalite I ! II; Palmer and Finger, 1994).
For the phase transition 422 ! 2, the active representation is of type E (Salje, 1990),
comprising an orthogonal pair of basis functions (see Table 5.4). The xzð13 Þ basis function
describes a shear parallel to a; such that the angle b departs from 908. The tetragonal
condition a ¼ b is relaxed in the monoclinic system and so an addition of improper strain
is allowed, involving extensions or contractions parallel to these two axes (i.e., non-zero 111
and 122 ). The strain is subject to the constraint that there shall be no net volume change and,
therefore, the condition to be satisfied is 111 ¼ 2122 (Palmer and Finger, 1994).
The alternative basis function yz ¼ 2xzð15 ¼ 214 Þ is a shear parallel to [110], i.e.,

the ð110Þ plane of the tetragonal cell, leaving a unique (diad) axis perpendicular to the
 shear plane. There is no longer any constraint for g ¼ 908 and so an additional
ð110Þ
improper strain, 112 ; is allowed. The resulting unit cell has a ¼ b ¼ g – 908 (because of
123 and 113 ). This is better described by a conventional monoclinic unit cell, with b
perpendicular to the ð110Þ shear plane of the tetragonal phase and with a and c lying within
this plane. It can be clearly seen that the deformation can be described by the xz-basis
function but not by the yz – xz basis function.
Palmer and Finger (1994) described the spontaneous strain that violates the
symmetry of the super group 422. In addition to this symmetry-breaking strain ð1sb Þ; extra
and non-symmetry-breaking strains ð1nsb Þ are allowed, which couple to the A
representation of 422. These strains do not affect the off-diagonal (shear) terms of the

TABLE 5.4
Spontaneous strain relations for the subgroup 422 (after Salje, 1990)

Active Repn. Sub-group Spontaneous strain No. of elastic domains Basis functions
Proper Improper

A2 4 1 Z
B1 222 11 ¼ 212 2 ðx2 2 y2 Þ
B2 222 16 2 xy
E 2 15 11 ¼ 212 4
E 2 15 ¼ 214 16 4 x, y, xz, yz
E 1 14 ; 15 16 ; 11 ¼ 212 8

Note: The monoclinic subgroups correspond to an orientation with the unique axis parallel to [010].
(Crystalline) Materials Under High Pressure 431

strain tensor but may modify the on-diagonal terms and, hence, the unit-cell volume.
One can expect symmetry-breaking strains parallel to a and b; but not c: Therefore, the
axial strains may be expressed as

111 ¼ 1nsb sb
11 þ 111

122 ¼ 1nsb sb
22 þ 122 ð5-28Þ

133 ¼ 1nsb
33

The non-symmetry-breaking strains must conform to the high-symmetry point

group 422 and, therefore, the condition becomes

1nsb nsb nsb

11 ¼ 122 – 133 :

The relation between two symmetry-breaking strain components (i.e., for no volume
change) is

1sb sb
11 ¼ 2111 :

From these relations one arrives at the following:

1
1nsb
11 ¼ ð1 þ 122 Þ ð5-29Þ
2 11

1
1sb
11 ¼ ð1 2 122 Þ:
2 11
The phase relation involves a shear plus a volume change. The measured strain
tensor, S; can, therefore, be separated into a symmetry-breaking tensor:
0 1 0 1 0 1
111 0 113 1nsb
11 0 0 1sb
11 0 113
B C B nsb C B sb C
S¼B
@ 111 122 0 C B
A ¼ @ 111 1nsb
11 0 C B
A þ @ 111 21sb
11 0 C
A ð5-30Þ
111 122 133 1nsb
11 1nsb
11 133 1sb
11 21sb
11 0

The symmetry-breaking strain tensor may be diagonalized to reveal the orientation

(i.e., the shear plane) and magnitude of the corresponding strain ellipsoid.

5.5. Strain tensor

Under a hydrostatic pressure, the change in the unit-cell parameters makes up the
strain and this phenomenon is represented by a second-rank tensor (Nye, 1957). From each
increment of pressure, the strain can be calculated (with certain uncertainties).
The three principal coefficients of the strain, 11 ; 12 and 13 (conventionally,
l11 l . l12 l . l13 l) obtained by diagonalization of the strain tensor represent the fractional
change in length along the principal axes of the strain tensor. The isothermal linear
432 Chapter 5

coefficient of volume compressibility, b; is given by

b ¼ Vo 2 Vp =Vo ¼ 2ð11 þ 12 þ 13 Þ ð5-31Þ
since for compression the principal strains are all negative.
For second-rank tensors, the visualized strain surface is an ellipsoid with principal
axes of length ð1 þ 11 Þ; ð1 þ 12 Þ and ð1 þ 13 Þ:
In monoclinic felspars, the three principal strains are unequal and one of the
principal axes is along the diad b-axes. Hence, the other two principal axes must lie in the
(010) plane. It is seen that in alkali felspars about 60– 70% of the volume compression is
brought about by a linear compression along the (100) plane normal. However, the same
degree of anisotropy is observed in plagioclases.

5.6. Bulk modulus of ionic compounds

Under ambient conditions, the bulk modulus of ionic compounds is given by a

general relationship:
K / Za Zc =V
where Za and Zc ; are the formal anion and cation charges, respectively, and V is the specific
volume per ion pair (Anderson and Nafe, 1965).
Under pressure, packing efficiency increases at phase transition, when the cation
coordination number ðNc Þ increases. As the structure becomes more compact, compression
becomes increasingly difficult and the bulk modulus rises. High-pressure phases may thus
be hard materials. The archetypal example is the transformation of graphite to diamond or
of quartz to stishovite.

5.6.1. Molar volume

The inverse relationship between bulk modulus and molar volume V was first
formulated by Bridgman (1923). This relationship was used for predicting bulk moduli of
mantle minerals where the data were not available (e.g., Hazen and Finger, 1979). The
basic expression is KV ¼ constant. However, anomalous compression behaviours known
through recent studies in (Mg,Fe)2SiO4 spinels and Ca(Mg,Fe)Si2O6 clinopyroxenes do
not conform to the above relationship (Hazen 1993; Zhang et al., 1997). They found that
Mg-rich members with smaller molar volume were more compressible at high pressures
than their Fe-rich counterparts of larger molar volume.
To explain this discrepancy, a comprehensive study comprising the determination
of the crystal structure type, valence state of cations, electronic configuration, polyhedral
coordination number as well as the cation substitution becomes a necessity.

5.6.2. Shear modulus: mantle perovskite

Compared with the determination of bulk modulus and the EOS at high pressure,
the shear modulus, G; is more difficult to determine. In geophysics, shear modulus is
(Crystalline) Materials Under High Pressure 433

an important quantity as it is related to the longitudinal ðPÞ and shear ðSÞ velocities as
(see also equation (5-20)):

Vp ¼ ½ðKs þ 4=3GÞ=r1=2 ; Vs ¼ ðG=rÞ1=2 ð5-32Þ

Using a PIB model (Cohen, 1987), a value of 190 GPa for the G of MgSiO3
perovskite was determined under ambient conditions. Brillouin scattering measurements
(Yeganeh-Haeri et al., 1989) showed this value as 184.2(^ 4.0) GPa (both Voigt – Reuss –
Hill bounds).
The single-crystal elastic moduli (in GPa) for MgSiO3 perovskite at zero pressure
using the theoretical PIB model and the experimental Brillouin scattering spectroscopy
indicate that the value of G under room T; P equals that of the lower mantle at 1,000-km
depth (Dziewonski and Anderson, 1981). If the lower mantle is perovskite-rich, the
pressure and temperature effects offset each other at that depth (40 GPa and 1,900 –
2,300 K). At 1,071-km depth, pressure G equals 255 GPa (at 300 K). These indicate that
dG=dT < 0:04 GPa/K is consistent with geophysical data when a perovskite-rich lower
mantle is assumed. The measured values of dG=dT for 15 silicates fall in the range of
2 0.08 to 2 0.014 GPa/K (Sumino and Anderson, 1989).

5.7. Magnetic features

5.7.1. Ferromagnetism

In ferromagnets, interactions between the dipoles favour a state in which the dipoles
align along the same direction and long-range magnetic ordering occurs at Tc : The
spontaneous magnetization is zero above Tc ; where thermal motion overwhelms any
magnetic alignment.
In other words, ferromagnetism arises from a spontaneous alignment of the atomic
magnetic dipoles (spins), driven by magnetic interactions between the atoms. But magnets
of certain crystal structure show an inability to attain magnetic order, solely by virtue of the
geometrical arrangement of the magnetic atoms. These are geometrically frustrated. The
high-temperature supercounductivity is believed by some to be due to a kind of frustrated
magnetic state (Anderson, 1986).
Ferromagnetism is generally considered to be geometrically unfrustrated. For
discussion on ferromagnetism phase transition, see Section “Ferromagnetic phase
transition” of this chapter.

5.7.1.1. Curie temperature

The pressure dependence of Tc is a strong function of the doping level and is
reminescent of the effects observed in high-temperature superconductors (Neumeier and
Zimmermann, 1994), where the total charge-carrier concentration n increases with
pressure. Under pressure, the value of nðPÞ could be non-zero and enhance the
ferromagnetic double-exchange coupling.
434 Chapter 5

TABLE 5.5
dTC =dP and dTN =dP some ferromagnetic materials

Compound dTC =dP (K/GPa) Source

Sr Ru O3 perovskite < 25.7 Neumeier et al. (1994)

Ni þ 32 White and Geballe (1979)
Fe ¼0 White and Geballe (1979)
Co ¼0 White and Geballe (1979)
SC3ln þ 1.7 Grew et al. (1989)
YFe5O12 þ 12.5 Kafalas et al. (1971)
dTN =dP (K/GPa)
Ca12x SrxMnO3 ,þ 4 Neumeier et al. (1994)

The dTc =dP values of some ferromagnetic materials are shown in Table 5.5.
Large values of dTc =dP suggest that strong magneto-volume effects will be observed in
these materials. The largest dTN =dP and dTc =dP values occur for the highest electrical
resistivity.
The ferromagnetic double-exchange interaction is presumably more strongly
pressure dependent than direct or super-exchange (assumed to be responsible for the
anti-ferromagnetism) in similar compounds.

5.7.2. Ferrimagnetism

In a ferrimagnet (e.g., magnetite), there are two non-compensating sub-lattices of

local dipoles called MA and MB ; with MA – MB : The interactions between MA and MB
favour their anti-parallel orientation. Again, a long-range ordering occurs at Tc ; with the
dipoles MA aligned in one direction and the dipoles MB oriented in the opposite direction.
At any temperature below Tc ; the spontaneous magnetization is the algebraic sum
of the contributions arising from each of the two sub-lattices. In most cases, at any
particular temperature, the contribution of one of the sub-lattices is larger in value so that
the resulting spontaneous magnetization increases steadily as T is lowered.
In a few cases, the two contributions cancel out at a temperature called the
compensation temperature, Tcomp : Between Tc and Tcomp ; the contribution of one of the
sub-lattices dominates. Below Tcomp ; the contribution of the other sub-lattices dominates.
The spontaneous magnetization is first positive, then zero and finally negative as the
sample cools down. This behaviour, predicted by Louis Neel in 1948, is observed in metal
oxides and also in some molecule-based magnets containing organic molecules
(Mathonière et al., 1996).

5.7.3. Spin states of iron

In accordance with Hund’s rule, the ground-state electronic configuration is the one
with maximum spin multiplicity, i.e., S ¼ 2: The state with maximum S is stabilized
through exchange energy. This state with maximum spin will have the most parallel
electrons and, hence, the greatest exchange stabilization energy.
(Crystalline) Materials Under High Pressure 435

In the case of Fe2þ-bearing compounds, such as FeO, FeS2 and FeS, the electronic
configuration is [Ar] 3d6 and the levels consist of two sets of orbitals t2g and eg ; which are
split according to the octahedral ligand field 10 Dq (or D0 ).
In the HS state, the electrons occupy the orbitals according to Hund’s rule and
electrons in 3d6 are spread over as t2g"3
e2g" t2g"
1
: In the LS state, the electronic configuration
becomes t2g" t2g# and the total 3d magnetic moment is zero (in XES, the Kb0 satellite
3 3

emission line vanishes).

In the case of an Fe3þ-bearing compound, such as Fe2O3, the electronic
configuration is [Ar] 3d5 and the configuration of the 3d orbitals is t2g" 3
e2g" in the HS and
3 2
t2g" t2g# in the LS state to a first approximation.
The exact electronic configuration and magnitude of the magnetic moment would
depend on the nature of the ligand field, crystal structure and 3d band structure. However, a
finite moment is expected in the case of lower-spin Fe2O3 (in XES emission spectrum,
a weak Kb 0 satellite shows up).
Because the high-spin phase should be the high-entropy phase, the increasing
temperature should promote the high-spin magnetic phase rather than the low-spin non-
magnetic phase. Increasing temperature leads to disordering of the local moments.

5.7.3.1. Electronic/magnetic ordering: examples

Pressure-induced electronic and magnetic order – disorder effects are known in
condensed matter research (Kelso and Banerjee, 1995). To site a few examples: Verwey
transition in Fe3O4, spinels (Kaakol et al., 1992), electron delocalization (insulator-to-
metal transition in VO2; Pintchovski et al., 1978) and spin unpairing as in Co3O4 and
CaFeO3 (Mocala et al., 1992). At high pressure when DV becomes positive, metallization
and spin pairing may occur.

5.7.3.2. Magnetic collapse: oxides and perovskites

As pressure and temperature are varied, a number of transitions are expected. These
may occur together in several groups, such as Mott transition, transitions between band
states and localized states, high-spin –low-spin transitions, structural-phase transitions and
magnetic-ordering transitions.
Magnetic collapse can occur if the crystal-field splitting becomes larger than the
exchange splitting. In such a case, the occupancy of state changes and the magnetic
moments collapse. The magnetic collapse occurs by band broadening with pressure, with
very little change in the crystal field. Indeed, the crystal-field t2g – eg splitting is seen to
remain the same at all pressures. A uniform decrease in effective DOS with pressure rather
than a change in CF splitting (which is low) gives rise to magnetic collapse.
The magnetic properties at very high pressure of FeO, MnO, CoO, NiO and
FeSiO(pvs)
3 have been investigated by Cohen et al. (1997) using the first-principles linear
muffin-tin orbital (LMTO) and linearized augmented plane wave (LAPW) electronic
structure methods within the GGA. They computed the anti-ferromagnetic moments
for cubic rock salt (B1)-structured FeO, MnO, CoO and NiO and found that all
four compounds exhibit magnetic collapse. For FeO, magnetic collapse occurs at relatively
436 Chapter 5

high pressures, whereas for CaO the transition pressure is relatively low (see also Section
12.2.1).
(Mg,Fe)SiO3 perovskite containing up to 15% Fe is considered to be the most
common mineral in the Earth. In this phase, Fe is in a high-spin state in the A-site and a
low-spin state in the B-site at geophysically relevant pressures (135 GPa). In the A-site,
magnetic collapse does not occur until 1 TPa whereas, in the B-site, the magnetic
moment decreases smoothly from high spin at low pressures to low spin at high
pressures. Again, the B-site is much smaller than the A-site. The B-site band-width
becomes much larger when Fe is in the B-site; this leads to different magnetic behaviour.
When the magnetic collapse occurs, it is governed by the band-width rather than by the
crystal field.
Transition-metal sulphides, having a greater overlap of larger sulphur anions, will
show larger band-widths and, consequently, lower pressure transitions. For example, in the
FeS2 –MnS2 system, Fe2þ is in a low-spin state, whereas dilute Fe2þ in MnS2 iron is in a
high-spin state because of the larger molar volume of MnS2 (compared with FeS2). Under
moderate compression (12 GPa), the Fe2þ ions undergo magnetic collapse to the low-spin
state. Substitution in a smaller site will also cause a compression effect on Fe2þ to induce
magnetic collapse at lower pressures. This occurs in Mg(Fe)O, magnesiowüstite, in
which iron undergoes magnetic collapse at lower pressures than for FeO. This is
because Mg2þ ion is smaller in size than Fe2þ ion. Also, Fe2þ ion in Mg-wüstite will
collapse at lower pressures than Fe2þ ion dissolved in Fe liquid. The magnetic collapse of
Fe2þ iron in Mg-wüstite may be responsible for the anomalies observed in the D00 zone
above the core.
Magnetic collapse in Co2þ would make it more siderophile. This explains the
relative depletions of Ni and Co in the silicate mantle. Thus, ultra-high-pressure
experiments help one to a better understanding of the partitioning of elements in the lower
mantle.

5.7.3.3. Magnetism in phase stability

Magnetism can affect phase stability, lattice distortions, elasticity, EOS, and
vibration frequencies.
Magnetism arises from electron spin, which is a vector quantity, behaving like
quantum angular momentum. The orbital angular momentum of electron leads to orbital
magnetic moments. The orbital moments become important in f-orbital electrons, such as
present in rare Earths and lanthanides. When the localized f-electrons interact with
delocalized band-like states interesting phase transitions occur with pressure, such as with
Ce. d-State electrons get more delocalized than f-state electrons, which are more localized.
But localization is also important.
The antisymmetric property of the wave function leads to the Pauli exclusion
principle, without which the electrons would fall into the nucleus and no atom will exist!
The opposite spin property allows a maximum of two electrons to exist in an orbital. Since
electron is indistinguishable one can consider the quantum states to be paired rather than
the electrons themselves. In order to lower the total potential energy the electrons are
(Crystalline) Materials Under High Pressure 437

required to have the same spin. The lowering of potential energy by lining up the spins is to
raise the system’s kinetic energy. The electronic potential energy favours magnetic while
the electronic kinetic energy favours non-magnetic electronic structure.
Under pressure electrons are pushed close together, the bands become wider and the
material becomes non-magnetic. The total energy change between magnetic and non-
magnetic system is known as exchange energy.
From low temperature as the temperature is raised the magnetic moment directions
on each atom will fluctuate and at critical temperature, Tc, in ferromagnetic the moments
will disorder.
In antiferromagnets the magnetic moments lie in opposite directions in alternate
sites. Under pressure the kinetic energy is lowered. Some ferromagnets under pressure
become antiferromagnet, e.g., fcc iron.

5.7.3.4. Magnetic frustration

At the lowest free energy state some have non-collinear spins, and it becomes
impossible to transform lattice with perfect antiferromagnetic pattern, (i.e., neighbours
with opposite pointing spin), e.g., fcc and hcp structures. Such properties are responsible
for anti-Invar effect (high thermal expansivity) as in fcc iron.

5.8. Polyhedral changes

It is known that a closed 3D figure made of rigid triangles can be squeezed or

stretched into a new shape without distorting the faces. The concept of flexible polyhedra
retaining their volume constant has destroyed the belief that a given set of edge lengths
should yield a finite number of shapes.
The formula of Heron, the Greek mathematician (of Alexandria) says that area, x; of
a triangle with side lengths ða; b; cÞ must solve the polynomial:

16x2 þ a4 þ b4 þ c4 2 2a2 b2 2 2a2 c2 2 2b2 c2 ¼ 0:

The volume of a tetrahedron has to a satisfy similar but more complicated

polynomial. For an octahedron, the polynomial involves 16th powers of the volume.
The volume of any polyhedron might also solve some version of Heron’s polynomial.
The volume of a polyhedron with fixed side lengths could only change by
jumping from one solution of the polynomial to another but if the polyhedron
change is gradual, the volume cannot change suddenly; it has to remain constant. A
flexible four-sided figure, for example, can change its area without changing the side
lengths.
The compression of the individual octahedral configuration is strongly dependent
upon both the geometry of the polyhedron itself and the types of, and the connectivity to, the
neighbouring polyhedra. The dominant mechanism of compression is usually one
of polyhedral rotation and the change in interpolyhedra bond angles. Conversion of
corner-shared MO6 octahedra to various structural patterns is shown in Fig. 5.4.
438 Chapter 5

Figure 5.4. Transformation of corner-shared (MO6) octahedra (a) to edge-shared octahedra (b) to corner and
edge-shared octahedra (c) to face-shared octahedra (d). The structure consists of an infinite array of corner-shared
octahedra (a), and infinite strings of edge-shared octahedra (c).

The bonding geometry controls the pressure-induced changes of the material’s

physical characters. Generally, with pressure, the bond-length distortion decreases as the
anions tend to approximate close packing.
The degree of polyhedral distortion at high pressure can be characterized with
distortion parameters defined (Renner and Lemann, 1986) as
Bond-length distortions (BLD):
100 X
n 
ldi 2 dl
BLDð%Þ ¼ ð5-33Þ
n i¼1 d

Edge length distortion (ELD):

100 X
n
ðxi 2 x Þ
ELDð%Þ ¼ ð5-34Þ
n i¼1 x

Angular distortions (AD):

100 X
N
ADð%Þ ¼ laj l ð5-35Þ
N j

Xm
ai 2 aideal
aj ¼ ð5-36Þ
i¼1
aideal

where di and xi are the bond length and edge length, respectively, d and x are the average
bond length and edge length in a polyhedron, a is the polyhedral angle and aideal is the
polyhedral angle of a regular polyhedron.
The bulk modulus of a mineral, K; represents the pressure dependence of its molar
volume, V; as
K ¼ 2VðdP=dVÞ ¼ 1=b;
(Crystalline) Materials Under High Pressure 439

where b is the volume compressibility of the crystal. At elevated pressure, K generally

increases.
Usually, under increasing pressure, SiO4 tetrahedra become more regular, as do the
MgO8 polyhedra whereas AlO6 octahedra become more distorted. MO6 and SiO4 polyhedra
dictate the compressibility of the structure hosting these. A high-valence state of cations in
octahedra would lead to a more incompressible structure. The compressibility of Si –O
bonding is of great interest because of the transition between 4- and 6-fold coordinated
silicon in the Earth’s transition zone and lower mantle. The constituent polyhedra of mineral
phases in the upper and lower mantle are shown in Fig. 5.5 (see caption). The high
incompressibility of Si –O bonding makes it difficult to determine the bulk modulus of SiO4
tetrahedra occurring in the mantle minerals. However, experiments at extended pressures by
different workers have demonstrated the Si – O bonding to show a significant compressibility
(although over a broad range) (Kudoh and Takeuchi, 1985, Hazen and Finger, 1989; Hugh-
Jones and Angel, 1994; Zhang et al., 1997, Zhang et al., 1998).
The crystal volume for an experiment above 20 GPa is more than five times smaller
than for an experiment at pressures lower than 10 GPa.
The macroscopic bulk modulus is due to polyhedral compression and tilting of
polyhedra about shared-corner linkages. The bulk modulus is related to pressures as
Kp kdl3 =S2 zc za :
where Kp is the bulk modulus of the polyhedron, kdl is the mean distance between the
cation and anion, zc and za are the charges on the cation and anion, respectively, and S is a
scaling factor equal to 0.50 for oxides and silicates.

Figure 5.5. Major constituent polyhedra of the mantle mineral phases. In the upper mantle, silicon is coordinated
to four oxygens and metal (M) atoms (Mg, Fe, Ca,…) to six oxygens. In the lower mantle, the oxygen coordination
of silicon ranges from 4 to 6, and that of metal atoms from 6 to 12 (after Madon, 1992).
440 Chapter 5

Change in the volume of coordination polyhedra is related to the bond parameters,

such as the M– O distance ðdÞ; coordination number ðnÞ; cation radius ðrÞ; cation and anion
charges ðZc and Za Þ and ionization ðS2 Þ: The volume changes in the direction of a; b and g
in anistropic crystals can be represented as

dV n
av ¼ 1=v < 120 £ 1026 8C21 ð5-37Þ
dT s2 zc zz

dV d3
bv ¼ 21=v < 0:133 £ 1026 bar21 ð5-38Þ
dP s2 zc za

dV 3ðr2 2 r1 Þ
gv ¼ 1=v < ð5-39Þ
dX d

where X ¼ atomic fraction of large cation.

The compressibility in isothermal conditions is defined as
1 dV
bT ¼ 2 :
V dP T

An increase in the coordination number (by transition, etc.) will increase the
polyhedral compressibility. Compression of cation polyhedra in oxides is a fundamental
property of the type of polyhedron, essentially independent of the structure type.
The oxides and silicate minerals suffer structural changes. The changes are
controlled by the following the observed rules:
(A) With pressure, the lattice volume change is negative because crystal compresses
mostly along the axes.
(B) The metal – oxygen bond distance, d; shows an average compressibility, b: In the case
of a crystalline phase:
b ¼ ð21=dÞdd=dP < 0:044ðd3 =S2 Zc Za Þ £ 106 bar21 ðsee Fig: 5:6ðaÞÞ:

(C) The polyhedral compressibilities (and change in O – cation – O angles)

correspond to the total volume compression. The relationship between the bulk
modulus and volume for a variety of marerials is shown in Fig. 5.6(a) (Hazen and
Finger, 1979).

5.8.1. Elasticity of MgO6 and SiO6 octahedra: MgSiO3 ilmenite

The calculated Si – O, Mg – O and O – O bond lengths and the polyhedral volume

decrease continuously with increasing pressure.
The polyhedral bulk moduli obtained from fits to third-order Birch – Murnaghan
EOS are
(Crystalline) Materials Under High Pressure 441

Figure 5.6(a). The bulk modulus-volume relationship for polyhedra a variety of materials (Hazen and Finger,
1979, q 1998 Mineralogical Society of America).

for SiO6 octahedron:

K ¼ 338 GPA ðwhen K 0T0 ¼ 5:05Þ
and, for MgO6 octahedron:
K ¼ 172 GPA ðwhen K 0T0 ¼ 3:96Þ:
Thus, MgO6 octahedra are , 2 times more compressible than SiO6 octahedra.
The degree of distortion (QE and s 2 ) in both MgO6 and SiO6 octahedra decreases
under compression, but the distortion of the former decreases more rapidly with pressure
(Fig. 5.6(b)).
The ilmenite structure consists of alternating MgO6 and SiO6 octahedral layers
normal to the c-axis. The high elastic anisotropy of ilmenite is connected to the large
contrast in the compressibility of the two types of octahedra, namely MgO6 and SiO6.
Compression along the c-axis is jointly determined by the more compressible MgO6 and
less compressible SiO6 octahedra, whereas compression along the a-axis is only controlled
by the less compressible SiO6 octahedra (Karki et al., 2000).

5.8.2. Anisotropic deformation: decompression

Cubic crystals deform isotropically while the deformation of anisotropic crystal is

determined by the anisotropy of bond strength and shearing. Negative linear thermal
expansion is common in monoclinic and triclinic crystals.
Highly ansiotropic deformation in oblique-angle crystals, such as felspars,
amphiboles, pyroxenes and micas, cause decompression of rocks, increase in fluid
permeability and ore deposition through metamorphism. The anisotropy of felspars and
442 Chapter 5

Figure 5.6(b). Pressure variation of distortion parameters (a) quadratic elongation (QE) and angle variance (s2)
for MgO6 and SiO6 octahedra.

quartz ða ! bÞ transitions causes volume change by 0.6% (at 5738C), and causes
decompression of host rocks such as granite, etc. Similarly, highly anisotropic thermal
expansion of calcite is responsible for the thermal decompression of marbles.

5.8.3. Radius ratio and coordination changes

The ratio of the ionic radii of cation ðRc Þ and anion ðRa Þ determines the
coordination number of the cation. When this ratio ðRc =Ra Þ ranges between 0.414 and
0.732, octahedral coordination occurs. In the case of tetrahedral (4-fold) coordination, this
ratio is less than 0.414 while, for cubic (8-fold) and dodecahedral (12-fold) coordination,
this ratio falls below 0.732.
In the high-pressure (denser) polymorphs, the coordination number of Si and Mg
increases (especially in perovskite structure). This is in consequence of the increase in
radius ratio for most cations, which show less compressibility compared with the highly
polarizable oxygen anions. Divalent cations (except low-spin ones) are more susceptible to
a pressure-induced increase in coordination number than are the trivalent cations. Thus, at
high pressure, divalent high-spin cations, e.g., Fe2þ, Mn2þ, Co2þ and Cr2þ, acquire a
coordination number greater than six. These cations occur in the A-sites of perovskites in
the lower mantle; whereas higher-valent states, such as Ti4þ, Ti3þ, V3þ, Cr3þ and Fe3þ,
remain in octahedral sites throughout the mantle and stabilize the B-sites of perovskite and
also the structure of magnesiowüstite.
In the Earth’s deep interior as pressure increases the coordination number, the
cation – oxygen distance increases. Thus, high-pressure phases show greater compressi-
bility than their low-pressure polymorphs. The volume reduction with pressure causing
discontinuous structural changes is countered by the increase in cation coordination.
Transformations such as those of MgSiO3 from pyroxene to garnet to perovskite result in
large density increases (from increased atom-packing efficiency).
(Crystalline) Materials Under High Pressure 443

5.8.4. Five-fold coordination: silicon and titanium

Five-fold coordination of silicon by oxygen (vSi) has been predicted by theoretical

studies of the effects P and T on silicate melts and glasses (Jin et al., 1994; Badro et al.,
1996) and is inferred experimentally by 29Si NMR and vibrational spectroscopy in alkali
silicate liquid and glasses (e.g., McMillan et al. 1994; Farber et al., 1996). vSi has been
proposed as playing a role in the dissolution of silicates (e.g., Xiao and Lasaga, 1996).
Although vSi is suspected in high-pressure silicates, no crystalline phase of silica
containing vSi have yet been observed. It may form part of the network of high-pressure
amorphous phases and be hypothetized in high-temperature SiO2-rich melts or glasses.
The first-principles electronic structure method and inter-atomic potential studies
show that vSi species are gradually replaced by ivSi and viSi with increasing pressure (e.g.,
Kubicki et al., 1993). Recent molecular-dynamic simulations on a-quartz under non-
hydrostatic stress found evidence for a crystalline phase of SiO2 composed entirely of SiO5
groups (Badro et al., 1996). The detail of the SiO5 polyhedron is shown in Fig. 5.7. Badro
et al. (1997) performed theoretical calculations in silica as a function of non-hydrostatic
stress. Molecular-dynamics calculations reveal a crystalline-to-crystalline transiton from
a-quartz to a vSi- bearing phase at high P in the presence of deviatoric stress. The structure
of the phase possesses P32 21 space-group symmetry. First-principles calculations within
the local-density approximation, as well as molecular dynamics and energy minimization
with inter-atomic potentials, show this phase to be mechanically and energetically stable
with respect to quartz at high pressure.
Upon decompression, the vSi phase reverts to a-quartz through an intermediate
four-coordinated phase and an unusual isosymmetrical phase transformation. The change
in Si – O bond length and Si – O – Si angle with molar volume on decompression of the
penta phase is shown in Fig. 5.8. The calculated powder X-ray diffraction pattern of SiO2
penta for a relaxed structure (P ¼ 16 GPa) is shown in Fig. 5.9. The results suggest the

Figure 5.7. Detail of the SiO5 polyhedron showing the position of the atoms and bond lengths calculated by LDA
(P ¼ 16 GPa) (from Badro et al., 1997, q 1997 American Physical Society).
444 Chapter 5

Figure 5.8. Structural variations in the penta phase as a function of volume during decompression Si–O–Si bond
angles (top) and bond distances (bottom) (Badro et al., 1997, q 1997 American Physical Society).

importance of the application of non-hydrostatic stress conditions in the design and

synthesis of novel materials (Badro et al., 1997).
Five-coordinated silicon is presumed to be present in an intermediate complex
during dissociation and polymerization processes of silicates (e.g., Lasaga and
Gibbs, 1990) and also during viscous flow and diffusion processes in silicate melts
(Zue et al., 1991).
(Crystalline) Materials Under High Pressure 445

Figure 5.9. Calculated powder X-ray diffraction patter of the SiO2 penta for the relaxed structure (P ¼ 16 GPa)
(from Badro et al., 1997, q 1997 American Physical Society).

Kanazaki et al. (1991) and Kudho and Kanazaki. (1998) synthesized a-CaSi2O5
phase at high P; T (12 GPa/1,5008C) and found the existence of a square-pyramid 5-fold
coordination of silicon in the structure.(see Figs. 5.10 and 5.11).
Assuming 1.4 Å as the ionic radius of oxygen, the ionic radius of [5]Si is
obtained as 0.33, which is intermediate between 0.26 Å for [4]Si and 0.40 Å for [6]Si
(Shannon- and Prewitt, 1969). Penta-coordinated silicon has been reported in organic
compounds. In all these structures, the [SiA5] groups are more or less distorted trigonal
bipyramids.

Figure 5.10. Square-pyramid coordination environment about Si3. Numbers indicate bond valences. The sixth
Si–O distance is indicated by broken line (Kudoh and Kanzaki, 1998, q 1998 Springer-Verlag).
446 Chapter 5

Figure 5.11. (a) [5]Si coordinated by five oxygens atoms with Si3 –O6 being vertical. (b) As above with Si3 –O6
being horizontal (Kudoh and Kanzaki, 1998, q 1998 Springer-Verlag).

The square pyramid 5-fold coordination of Ti by O has been reported for

Ba2TiSi2O8 (Moore and Louisnathan, 1967) and for Na2TiSiO5 (Nyman and O’Keeffe,
1978). In the case of the TiO5 square pyramid, the four longer Ti –O bonds are tetragonally
placed and the fifth oxygen atom makes a shorter Ti –O bond, forming a highly compressed
square pyramid.

5.8.5. Thermal expansivity and deformation equivalence (a=b)

The mean linear coefficient of thermal expansion, a ða ¼ av =3Þ; where av is the
coefficient of volumetric expansion of a coordination polyhedron formed by essentially
ionic bonds, has simple reversible relations to Pauling bond strength (cf. equation (5-37)):

1 dd n
a1000 ¼ < 4:0ð4Þ 2 £ 1026 8C21 ð5-40Þ
a dT S Zc Za

where d is the cation – anion separation, n is the coordination number of the cation, Zc
and Za are cation and anion charges, respectively, and S2 is the ionic-bond coefficient,
which is equal to , 0.5 for oxygen compounds, as stated earlier. Thus, all types of
polyhedra with a definite cation and anion have the same parameter a; which can be
used for predicting the effect of T on the polyhedra and thus does not depend on
structural bonds in the polyhedra. Furthermore, all coordination polyhedra with similar
values of Pauling bond strength have a similar coefficient a (see Fig. 5.12).
For example, in octrahedra with divalent Mg, Co, Fe, Cd, Mn, Ca, Ba and Sr, the
thermal expansion ða1;000 Þ equals 14 ^ 1 £ 1026 8C21.
Molar volume at 1 bar is expressed as function of T:
ðT
0
Vð1; TÞ ¼ V1;Tr exp aðTÞdT ð5-41Þ
Tr
(Crystalline) Materials Under High Pressure 447

Figure 5.12. The relationship between the mean linear polyhedral thermal expansion vs the pauling bond strength.

0
where V1;Tr is molar volume at 1 bar, Tr is the transition temperature and aðTÞ is the
thermal expansion, depending on temperature as

aðTÞ ¼ a0 þ a1 T þ a2 T 2 þ a3 T 3 : ð5-42Þ

The thermal expansivity at ambient pressure, sT0 ; can be assumed as

aT0 ¼ a þ bT:

All crystalline substances can be represented in P – T – X space by surfaces of a

constant molar volume (isochroic surfaces). For many substances, the isochroic surfaces
coincide with isostructural surfaces.
The deformation equivalence is measured as the ratio of thermal expansivity, a;
with compressibility b: The minerals in a single zone have a similar deformation
equivalence, a=b: This a=b value increases with pressure in depth. As a consequence of a
decrease in the temperature gradient, the expansion slows down and the proportion
of compression increases. The a=b relation obtained for a polyhedron (Hazen and
Finger, 1982) is

a=b < 90 n=d3 bar=8C:

448 Chapter 5

Evidently, the higher coordination numbers ðnÞ of the atoms in deep-seated phases
determine their greater density and deformation equivalence, a=b: All Mg, Fe2þ, Al and
Fe3þ octahedra have a=b < 65 bar/8C. Many silicate minerals reveal a – b inverse
relationships. Even rutile, having a more covalent bonding, demonstrates an inverse
relationship.
In subducting slabs, the rocks in depth zones under higher temperature and pressure
expand and compress simultaneously. The average a=b increases with depth of the zone
ðhÞ; rock density ðrÞ; P-wave velocities within the zone ðVP Þ and the mean symmetry of the
minerals in the zone.
Molar volume as a function of pressure and temperature can be calculated using the
Murnaghan equation:
21=K 0P
K0 P
VðP; TÞ ¼ Vð1; TÞ 1 þ P ð5-43Þ
KT

where KT is isothermal bulk modulus expressed as

KT ¼ 1=ðb0 þ b1 T þ b2 T 2 þ b3 T 3 Þ

and K 0P is the pressure derivative of the bulk modulus, which, in some cases, has a
temperature dependence

K 0P ¼ K 0P Tr þ K 0PT ðT 2 TrÞ lnðT=TrÞ

K 0P Tr is the pressure derivative of the bulk modulus at transition temperature, ðTrÞ and K 0PT
is its temperature derivative.
The specific temperature dependence of the bulk modulus of oxides and silicates is
available from Saxena et al. (1993). It is advisable to check that employment of the
Murnaghan equation and the Birch– Murnaghan equation gives close results in the studied
P; T range.

5.8.6. Volume compressibility: negative

Under hydrostatic pressure ðPÞ; most materials contract in all directions. Therefore,
the volume compressibility ð2dV=V dPÞ; area compressibility ð2dA=A dPÞ and linear
compressibilities ð2dL=L dPÞ are all positive. Materials are thermodynamically forbidden
to have negative volume compressibilities but some crystals having negative linear
compressibilities have been known; e.g., lanthanum niobate, cesium dihydrogen phosphate
(Prawer et al., 1985) and some others.
Materials having negative area compressibilities might be used as electrode
clamps to provide a sensitivity increase by an order of magnitude for ferroelectric
pressure sensors. However, materials with negative compressibilities are rare.
The pressure-induced expansion of the negative-area-compressibility material forces a
ferroelectric sheet to increase in area when pressure is applied. (Note: negative
compressibility may be relevant in muscular hydrostats, such as those found in worms.)
(Crystalline) Materials Under High Pressure 449

5.8.6.1. Relative compressibilities

Relative compressibility can be measured quite precisely from relative volumes of
crystals at several pressures:

b1 =b2 < ½ðVP =V0 Þ1 2 1=½ðVP =V0 Þ 2 1; ð5-44Þ

where b1 and b2 are compressibilities of two different crystals,1 and 2, and V0 and VP are
their unit-cell volumes measured at ambient pressure and high pressure.
Small compressibility differences in compositionally similar sites of wüstites
(Hazen, 1981), felspars (Angel et al., 1988), pyroxenes (McCormick et al., 1989),
wadsleyites (Hazen et al., 1990) and silicate spinels (Hazen, 1993) have been
documented.

5.8.7. Thermodynamic parameters and EOS

One of the most important thermodynamic parameters for high P – T calculations is

the unit-cell volume of a mineral. Its variation with P and T is described by its EOS with
three components: (1) compressibility at ambient temperature, (2) thermal expansivity at
ambient pressure and (3) cross-derivative terms describing the effect of pressure on
thermal expansivity, equivalent to the effect of temperature on compressibility. For
accurate thermodynamic calculations, each component should be measured at relevant
P – T conditions. EOS measurements are useful for constraining lower P – T metamorphic
reactions.
The EOS presents the thermodynamic relations between V; T; P and other external
fields. The EOS (when b is constant) can be written as a Taylor series or polynomial
expansion of P in terms of V: Birch– Murnaghan EOS is capable of treating compression
of up to factors of , 2 (i.e., V=V0 ¼ 0:5). For hydrogen and helium V=V0 equals 0.15.
EOS data play a central role in efforts to describe the mineralogical structure and
convective dynamics of the Earth’s deep interior. Experiments defining the molar volume
as a function of pressure clarify the nature and character of phase transitions that may occur
in the crust and mantle and, furthermore, provide valuable constraints on models of
bonding in minerals.

5.8.7.1. P – V – T data and EOS

An EOS reflects the underlying interaction potential among the ions and electrons
that make up a crystal. The P – V – T EOS plays a central role in study of the Earth’s deep
interior. Using the EOS, one can calculate the densities and bulk-wave velocities of
candidate materials at lower-mantle P; T conditions. By fitting the densities and wave
velocities to seismic observations, one can test models for the bulk chemistry of the lower
mantle (e.g., Bina, 1995; Jackson, 1998). In addition, the EOS enables us to determine the
depth dependence of important thermoelastic parameters, such as the thermal expansivity
and temperature sensitivity of the bulk modulus. The latter is necessary for comparing
seismic-bulk velocity profiles with laboratory data.
450 Chapter 5

Recently, new methods using a diamond cell combined with laser or resistance
heating have been used to obtain P – V – T EOS data directly at lower-mantle P – T
conditions (e.g., Fiquet et al., 1998; Shim et al., 1998; Saxena et al., 1999).
Under pressure, the changes in volume V is expressed in terms of the bulk
modulus, K:
Vðdp =dv ÞT ¼ 2KT :

An expansion of the bulk modulus in terms of pressure is

KðPÞ ¼ K0 þ K 00 P þ 1
2 K 00 P2 þ · · ·
where K0 is the value for P ¼ 0; K 00 is the first-pressure derivative at P ¼ 0 and so on.
Using only the first two terms and integrating the Murnaghan equation (for a
crystal) empirically becomes
0 0
P ¼ K0 =K 00 ½ðV0 =VÞK 0 2 1 ¼ K0 =K 00 ½ðd0 =dÞ3K 0 2 1 ð5-45Þ
where d is the lattice parameter.
The values of K0 and K 00 are obtained from precise measurements of elastic
constants; and X-ray data for the dðPÞ fit of this equation (within the empirical error for
P , 0:2K0 ). A typical value of K is 75 GPa. To describe dðPÞ at higher pressure, higher-
order equations can be used but experimental data of K at high pressures are often not
readily available.
5.8.7.2. Birch EOS
The Birch – Murnaghan EOS (Birch, 1947) has often been used to fit isothermal
compressional data. When the cell volume data do not follow any isothermal experimental
path, the Birch– Murnaghan EOS for diverse temperatures can be fitted to the P – V – T
data. It is important to indicate that any VðP; TÞ data point should be reached by taking a
realistic thermodynamic path. A standard way to fulfil the thermodynamic necessity is to
heat the ambient volume V0 to a temperature T and then compress the expanded volume
Vð0; TÞ along the isothermal to reach the VðP; TÞ: The modified Birch –Murnaghan
equation is thus written as
P ¼ 3KT f ð1 þ 2f Þ5=2 ½1 2 3=2ð4 2 K 0 Þf þ · · · ð5-46Þ
where
 2 300Þ;
KT ¼ KT0 þ KðT K 0 ¼ dK=dP; K ¼ dK=dT
ð
f ¼ 1=2½ðVT =VPT Þ2=3 2 1; VT ¼ V0 exp av ð0; TÞdT

where V0 is the cell volume at ambient conditions, VT at high T and VPT at high P – T
conditions and thermal expansion at zero pressure is av0 ¼ av ð0; TÞ ¼ a þ bT 2 c=T 2
(T in kelvin) and f is called the Eulerian strain (see equation (2-9)).
In thermal expansion, the C=T 2 term and the high-order derivatives of bulk
modulus K 00 ; K and d2 K=dPdT are generally ignored. The equation is a modified
(Crystalline) Materials Under High Pressure 451

isothermal Birch – Murnaghan EOS by replacement of K0 with KT and substituting V0 =VP

with VT =VPT so that temperature effects are incorporated.
Generally, four iterative steps, where each step has different constraints on starting
parameters, can be taken to achieve the final fitting results:
Step 1: fix av0 and fit KT0 ; K 0 ; K
Step 2: fix KT0 ; K 0 and fit av0 ; K
Step 3: fit av0 ; KT0 ; K 0 ; K simultaneously
Step 4: Combine all data and fit av0 ; KT0 ; K 0 ; K simultaneously.
The Birch equation assumes that the underlying potential can be represented as a
series in ð1=r 2n Þ: The commonly used third-order Birch equation (with parameters V0 ; KT0
0
and KT0 ) includes n ¼ 1; 2 and 3.
However, there is no fundamental reason to expect ð1=r 2n Þ to represent inter-atomic
interactions well, so it is not surprising that the Birch equation is not perfect.
The Vinet and Holzapfel equations converge smoothly to a constant at large
volumes, being consistent with physically based potentials.

5.8.7.3. Equations of state: density ratio

The Birch– Murnaghan EOS is widely used amongst geoscientists. The two
important pressure – density relationships used are as follows.
The first-order Birch– Murnaghan EOS for many solids is approximately

P ¼ 3=2K0 ½ðV0 =Vp Þ7=3 2 ðV0 =Vp Þ5=3  ð5-47Þ

where K0 is the bulk modulus at zero pressure (i.e., at 1 bar, ø 1024 GPa) and V0 and Vp
are volumes at room and elevated pressure, respectively. Considering ðV0 =Vp Þ equals
ðr=r0 Þ; the following relations hold:
Birch – Murnaghan (third order):

P ¼ ð3=2ÞK0 ½ðr=r0 Þ7=3 2 ðr=r0 Þ5=3 {1 þ 3=4ðK 00 2 4Þ½ðr=r0 Þ2=3 2 1} ð5-48Þ

Birch – Murnaghan ( fourth order):

P ¼ 3=2K0 ½ðr=r0 Þ7=3 2 ðr=r0 Þ5=3 {1 þ 3=4ðK 00 2 4Þ½ðr=r0 Þ2=3 2 1

þ 3=4½K0 K 000 þ ðK 00 2 4ÞðK0 2 3Þ þ 35=9½ðr=r0 Þ2=3 2 12 }

The EOS can be expressed as a function of density via C ¼ r=r0 :

The second-order Birch –Murnaghan EOS appears as

3KT0 7=3
P¼ ðC 2 C 5=3 Þ ð5-49Þ
2
452 Chapter 5

Likewise, the third-order Birch– Murnaghan EOS becomes


3KT0 7=3 3
P¼ ðC 2 C 5=3 Þ 1 2 ðK 0T0 2 4ÞðC2=3 2 1Þ ð5-50:1Þ
2 4
The third-order Birch – Murnaghan equation has been widely used to decribe the
isothermal compression of mantle minerals
" #( " #)
3 V0 7=3 V0 5=3 3 V 2=3
Pst ¼ KT0 2 1 2 ð4 2 K 0T0 Þ 0
21 ð5-50:2Þ
2 V V 4 V
where KT0 is the isothermal bulk modulus, K 0T0 is the pressure derivative of the bulk modulus
and V0 is the volume. The subscript 0 refers to ambient conditions (1 bar and 300 K).
The Vinet EOS comes as
3 0
P ¼ 3KT0 C 2=3 ð1 2 C 21=3 Þ exp ðK 2 1Þð1 2 C 21=3 Þ ð5-51Þ
2 T0
These equations accurately describe the compressional behaviour of real materials.
However, better precision is attained by ab initio calculations, which provide not
only a test data set but also enable us to calculate the internal energy of a system. These,
therefore, provide independent means of calculating g through the Mie –Grüneisen relation.
The volume at simultaneous pressure and temperature can be inferred from the
Mie –Grüneisen EOS:
P300 K ¼ ðPth 2 Pth 300 K Þ
where P300 K is the static pressure at 300 K, obtained by using available experimental room
temperature compression data to a third-order Birch –Murhaghan EOS:
" #" !#
3 V0 7=3 V0 5=3 3 0 V0 2=3
P300 K ¼ K0 2 1 þ ðK 0 2 4Þ 21 ð5-52Þ
2 V V 4 V

where Pth and Pth 300 K represent the thermal pressure at a given pressure and temperature
and the thermal pressure at 300 K, respectively. Pth is related to the vibrational Helmholz
free energy Fvib : as

dFvib
Pth ¼ 2
dv T

The usual experimental technique employs multi-anvil apparatus and synchrotron

radiation (e.g., Martinez et al., 1996). In this set-up, P and T can be controlled and
measured, large sample volumes can be employed for high-quality data and nearly
hydrostatic pressure distribution is achieved. The synchrotron radiation can be relatively
easily handled.
The isothermal compression data can also be analyzed using the Murnaghan EOS:
V K 0T0 1
¼ 1þ P 2 0 ð5-53Þ
V0 KT0 K T0
(Crystalline) Materials Under High Pressure 453

The two EOS have different assumptions concerning the second-order pressure derivative
of the bulk modulus K 00T0 : K 00T0 is zero in the Murnaghan EOS, while it has a non-zero
negative value in the Birch – Murnaghan EOS. It is often assumed that K 00T0 has more effect
on the refinement of K 0T0 :

5.8.8. Bulk moduli: isothermal and isentropic

Silicates may be modelled as ionic compounds with bond strength determined to a

first approximation by Coulombic forces. Bridgeman (1923), for example, demonstrated
empirical inverse correlations between bulk modulus and molar volume and described his
results in terms of an electrostatic model of inter-atomic forces. Bulk modulus – volume
relationships have since become useful for predicting the behaviour of oxides, halides,
silicates and many in the isomorphous and isoelectronic series (Anderson and
Anderson, 1970).
The isothermal and isentropic bulk moduli are distinguished by employing the
relation

dV
KS ¼ 2V ð5-54Þ
dP S

where Ks is the isentropic bulk modulus at reference P (usually 1 bar), V is molar volume
and ðdV=dPÞS is the change in molar volume with pressure at constant entropy.
Conversion of KS to the isothermal bulk modulus ðKT Þ can be calculated from the
relationship:

KS
KT ¼ ð5-55Þ
ð1 þ agTÞ

where a is the coefficient of thermal expansion and is defined as

1 dV
a¼ ð5-56Þ
v dT P

with V the molar volume, T the temperature at constant pressure and g is the thermal
Grüneisen parameter, which is defined as

d ln u aKs
g¼ ¼ ð5-57Þ
d ln V rCP

with u the Debye temperature and CP the specific heat at constant pressure.
As a function of pressure, the P – V EOS provides a determination of density r and
bulk modulus KS (or KT ), or bulk-sound velocity VB ð¼ KS =rÞ:
The stability and thermodynamic properties can be derived from the quasi-
harmonic lattice dynamics employing parameter-free pair potentials derived from the
MEG formulation.
454 Chapter 5

By neglecting second- and third-order derivatives, the bulk modulus at room

temperature can be expressed as
KT ¼ KT0 þ ðdKT =dTÞP ðT 2 300 KÞ
whereas the unit-cell volume is given by
ð
VT0 ¼ V0 exp aT0 dT :

Compare this with the molar volume shown by equation (5-43).

5.8.8.1. K of mineral mixture: Reuss bound and Voigt bound

Watt et al. (1976) have shown that, for crystalline materials, the bulk modulus ðK p Þ
of a mixture is bounded by the sum of the bulk moduli ðKi Þ as a function of volume fraction
ðvi Þ and the sum of compressibilities ðKi21 Þ as a function of volume fraction:
!
X n X n
p
KR ¼ vi =Ki # K p # vi Ki ¼ Kvp
i¼1 i¼1

The Reuss bound ðKRp Þis calculated for the case of equal stress throughout an isotropic
mixture, whereas the Voigt bound ðKvp Þ is calculated for uniform strain in an isotropic
aggregate. (Note: In the case of a melt, the real condition lies somewhere between the ideal
conditions of uniform stress and uniform strain in the structure.)

5.8.8.2. Crystal-field spectra

The velocities of seismic waves are influenced by the bulk modulus and rigidity
parameters. From the measurements of pressure-induced variation in CF spectra of
transition-metal-bearing minerals, the polyhedral bulk moduli of silicate and oxide
minerals can be determined.
From the spectrally determined bulk modulus at room pressure, K0s ; one obtains the
first-order Birch – Murnaghan EOS as

K0s ¼ 2=3P½ðDP =D0 Þ7=5 2 ðDP =D0 Þ21

where D0 and DP are the crystal-field splitting parameters at room and elevated pressure,
respectively. The polyhedral moduli determined from high-pressure X-ray data conform
closely to those obtained from crystal-field spectra at high pressures. The higher
coordination sites in a crystal system in general show higher compressibilities (Hazen and
Finger, 1982) whereas a large CFSE of a cation at a site reduces the compressibility. For
example, the [CrO6] octahedron shows poor compressibility because of very large CFSE of
Cr3þ at octahedral sites.
In the upper mantle, olivine and pyroxenes are anisometric and the value of
polyhedral bulk moduli shows polarization dependence. The cations in these show a
tendency to order in very distorted coordination polyhedra leading to pleochroic spectra,
which makes the pressure-dependence spectral evaluation difficult. In the lower
mantle, fortuitously, the major phases are cubic or their polyhedral sites are also
(Crystalline) Materials Under High Pressure 455

cubic, e.g., spinel, garnet, perovskite and periclase. In the determination of the bulk
moduli, the pressure dependence of the spectral shift, therefore, becomes more reliable
(see Section 4.4.2).

5.8.9. Velocity –volume relationship

To interprete the lateral velocity anomalies in terms of lateral temperature

anomalies, the following relationship in logarithmic form is useful:
ðd ln r=d ln VÞ
dT ¼ xdV ð5-58Þ
a
where dV is the velocity anomaly, a the thermal expansion coefficient and r the density,
while the velocity variation is due to a temperature variation. The isobaric value of ðd ln 
r=d ln VÞP yields the isothermal value. (Note: lnðr=r0 Þ ¼ 2lnðV=V0 Þ:)
The linear relationship between velocity and volume requires that the value
of ðd ln r=ln VÞT increases with pressure. The linearity with temperature requires that
ðd ln r=d ln VÞP decreases with increasing temperature.
The velocity – volume functions of alkali halides NaCl and KCl are nearly the same
at room pressure and temperature (near or above the Debye temperatures of these halides).
Hence, the isobaric and isothermal derivatives of ðd ln r=d ln VÞP are also nearly the same.
The value of the elastic (or anharmonic) contribution to ðd ln r=d ln VÞP appears to
increase from the 1-atm values as pressure increases.
The relationship between the isothermal and isobaric values of ðd ln r=d ln VÞ can
be obtained from

d ln v d ln v d ln v
¼a 2 ð5-58:1Þ
dT V d ln r T d ln r P
From the above equation, one can derive the hypothesis that ðd ln r=d ln VÞP
increases with depth (Yuen et al., 1993). The linear velocity –volume relationship
requires that ðd ln r=d ln VÞT increases with pressure or decreasing volume. The isobaric
ðd ln r=d ln VÞP should increase with depth as intrinsic anharmonicity decreases with
depth.
Chopelas et al. (1996) observed that the velocities of sound in minerals (pyrope,
YAG, Al2O3, MgAl2O4 and MgO) are linear with volume as long as no phase change or
change in compressional mechanism occurs.

5.8.10. Velocity –density relationship: rules

Usually, four velocity – density systematics are used in the geophysical interpret-
ation of seismic velocities in the Earth’s interior: (a) the VP2 – r relationship (Birch, 1961),
 relationship (Anderson, 1967), (c) the VB2 – r relationship (Wang, 1968)
(b) the f – ðr=MÞ
and (d) the bulk modulus – volume relationship (Anderson and Anderson, 1970). These are
discussed in the following.
456 Chapter 5

(a) Birch Law (1961). The relationship between VP and r for various silicates and oxides
takes the form of a linear equation VP ¼ aðMÞ þ br where aðMÞ and b are material
parameters. Similar relationships between Vs and r are also fairly established
(Liebermann, 1970).
 relation. Anderson’s (1967) seismic EOS, based on data for a set of
(b) The f – ðr=MÞ
rocks and minerals with M  values in the range 18– 34, is

 ¼ 0:048f0:323^0:12 ðr0 is density at zero pressureÞ:

r0 =M

(c) VB2 – r relation. A linear relationship between bulk velocity VB ð¼ ðfÞ1=2 Þ and r is
established (Wang, 1968). VB is nearly proportional to VP and should depend on M  as
well as on r; just as VP does.
(d) Bulk modulus – volume relationship. The scaling relationship between bulk modulus
K0 and specific molar volume V0 at ambient conditions is in the form of
K0 V0x ¼ constant. For constant crystal structure, x < 1; when M  is constant, the
value of x is taken as < 4 (Shankland and Chung, 1974).
Anderson and Nafe (1965) presented an analysis of K0 V0x relationships, in which the
data of quartz, forsterite, basalt, etc. were included. They also derived the relation:

V0i < rðx21Þ=2

where V0i is the ambient velocity for mode i ði ¼ 1; 2; 3Þ: The velocity is seen as a nearly
linear function of density.

5.8.11. Stretch densification

A solid increases in density when stretched along an axis of negative linear

compressibility. Materials showing negative compressibilities (when hydrostatically
compressed) in one or more dimensions have been discussed by many authors (Baughman
et al., 1998). These are shown to have negative Poisson’s ratios. (Note: Poisson ratio is the
ratio of a lateral contraction to a longitudinal elongation produced by a tensile stress.)
However, a few crystals with a negative Poisson’s ratio show a negative linear
compressibility. Plastically deformed foams and honeycombs are known to provide either
negative Poisson’s ratios or Poisson’s ratios whose sums exceed unity about a stretch
direction. Some of these crystals decrease in volume and expand in two dimensions when
stretched in a particular direction and increase in surface area when hydrostatically
compressed. These show the property “stretch-densified”. The known stretch-densified
phases, such as cesium dihydrogen phosphate and lanthanum niobate, are monoclinic
phases but no stretch-densified triclinic phases have been identified.
Stretch densification may be modelled from a wine-rack-like deformation mode
(Baughman and Galvao, 1993). Molecular mechanics calculations suggest that
ferroelasticity (and associated shape-memory behaviour) should occur in combination
with negative linear compressibilities. Negative linear compressibilities may result from
various structures comprising helical chains: (i) single helices, (ii) oppositely wound
(Crystalline) Materials Under High Pressure 457

helices (like the finger cuff) and (iii) the network of left- and right-handed helices. A tensile
stress in the helical axis direction decreases the volume (of the finger cuff) and the
cylinder-direction angle between the helices becomes , 109.468.

5.8.12. Compressibility and Si – O –Si bending

Shallow Earth materials mainly compress by bond angle bending whereas highly
symmetric closest-packed deep Earth materials compress mainly by bond shortening. For
example, Si –O –Si angle bending correlates with the compressibility of the SiO2
polymorphs (Hemley et al., 1994), which all seem to lie on the same DV – DðSi – O – SiÞ
trend (Downs and Palmer, 1994).
The compressibility of an individual silica structure is simply related to the
displacements of the kSi –O –Sil0 angles from their global equilibrium value (,1448); the
further from the equilibrium angle, the stiffer the structure.
In most framework silicates, the principal compression mechanism is T – O – T bond
bending (i.e., framework distortion) coupled with the compression of alkali – oxygen and
alkaline earth – oxygen bonds. Cation – oxygen-bond compression was shown to be
strongly influenced by Coulombic effects, whereby the bond compression becomes
inversely proportional to the cation charge. Thus, large monovalent and divalent cation
sites, typical of felspars, felspathoids, zeolites and other framework aluminosilicates,
display significant compression (Hazen and Finger, 1989). The M-cation affects the
bending energy of the T– O – T angle through the charge of the M-cation and the M –O
bond length. If the bridging oxygen is bonded to three or more atoms, the energy required
for bending the T– O – T angle is significantly increased (Geisinger et al., 1985).
The compression mechanism of the alkali felspars is dominated by the compression
of the alkali-containing channels. The compression pathway may result from the T –O – T
angle-bending energies that are linked to the alkali cation bonding. The most compressible
direction is the one which narrows the channels containing the alkali cations.
A large family of dense structures can be constructed starting with a close-packed
(or nearly so) array of oxygen atoms. Distinct structures are obtained depending on the
ordering of the Si atoms in the octahedral sties. This produces chains of SiO6 octahedra
with different degrees of kinking. The structure with no kinks is that of stishovite, which
has the CaCl2-type structure.

5.8.12.1. Ionic compressibilities

From the ionic radii in a crystal structure, ionic radius vs. bond strength systematics
have been developed. If the bond strength is defined as Z=N; where Z is the valence of the
cation and N is the coordination number, the ionic radii for Si, Al, Mg, Cr, Fe and Ca given
by Shannon and Prewitt (1969) can be shown to be a linear function of the log of the bond
strength, thereby satisfying equations of the form:
R ¼ ab logðZ=NÞ;
a and b are the fitted constants and R is the Shannon and Prewitt radius. Knowledge of a
few ionic-radius values for a particular atom in a polyhedra allows one to predict
458 Chapter 5

accurately the ionic radius for any other combination of valence and coordination number,
provided all the ions have the same spin state.
Empirical relations between the volumes of coordination polyhedra and the ionic
compressibilities have been proposed by many workers and have been used with
considerable success in predicting crystal structures at high pressures. However, Hazen
and Finger (1979) found that the relations are less dependent on structure types. Moreover,
the observed bond compressibilities, b; can be modelled by an equation of the form:
b ¼ 0:217 b2 ;
where the constant b is derived from linear regression and is the slope of the ionic radius –
log bond strength curve for that type of ion. Since this equation is independent of the
structure type and includes information about the coordination number, it provides a
general means for estimating the ionic compressibility of various coordination polyhedra.
The values are calculated from the relationship of Hazen and Prewitt (1977) and the
observed values. The agreement obtained is remarkably good (Kudoh et al., 1992).

5.9. Free and thermal energies: phase boundaries

Thermodynamic parameters are used to verify the consistency of phase-equilibrium

data and also to extrapolate the phase boundaries beyond the limited P – T space of the
experiments (e.g., Chopelas et al., 1994a). Estimates of the phase boundaries using the
thermodynamic parameters complement phase-boundary measurements at high P and T:
Delineation of phase boundaries helps to model the mantle compositions and temperatures
at the seismic discontinuities.
The thermodynamic parameters which are mostly employed for estimating phase
boundaries are the following:
DH : change in enthalpy
DS : change in entropy
CP : heat capacity
DV : change in volume across the phase transition.
Enthalpies are measured by calorimetry (e.g., Akaogi and Ito, 1993). Entropies and
heat capacity are derived from spectroscopic measurements using statistical thermodyn-
amics. Volumes are derived from compression measurements at room temperature and
thermal-expansivity systematics (e.g., Chopelas and Boehler, 1992b), i.e., from elastic
constants, EOS and thermal expansivity. Changes in DS for transitions by P and T changes
can be evaluated by using vibrational models. This has been done for several materials
(e.g., Chopelas et al., 1994; Chopelas, 1996). An interdependence between entropy and
volume in a given phase at different P – T conditions can be seen by considering the
equation for the entropy of a gas:
V2
DS ¼ nR ln
V1
where n is the number of moles, R is the universal gas constant and 1, 2 represent the two
states.
(Crystalline) Materials Under High Pressure 459

This simple relationship depicts that entropy and volume are not independent
variables and this relation remains valid even in the case of a solid (mineral). The changes
in DS are quite small (of the order of a few %) but are large enough to change the
topography of a phase-diagram calculation over long extrapolations.
A phase diagram is the map of the stability domains of the co-existing phase in the
P – T plane and the slope of the co-existing boundary of phases is represented by the
Clausius –Clapeyron equation:
DP DS
¼
dT DV
where DV and DS are the changes in volume and entropy at the transition.
Close examination of the relationship between S and V for the forsterite to b-phase
transition (Chopelas, 1991) showed that DS=DV changed by an amount much smaller than
the experimental error over a large temperature and pressure range.
At the point of a phase change in bulk matter, the internal energy change DU from
solid to liquid must exactly balance the contribution of entropy to the total energy change,
TDS: Only at P and T where these two are equal can solid and liquid co-exist in
equilibrium. This is the relation that yields the co-existence curve of the solid – liquid phase
diagram. The equality of DU and TDS ensures that the solid and liquid forms are equally
likely to be found.

5.9.1. Free energy

At high pressures, the atoms organize themselves more efficiently, tending to pack
into a higher coordination number. Stable (thermodynamically) phases are those, which
manifest minimum energy. The internal energy as a function of crystal volume is related to
the free energy and pressure as
P ¼ 2ðdA=dVÞT ¼ 2dE=dT; at T ¼ 08 K;
where A is the Helmholtz free energy ðA ¼ E 2 TSÞ; E the internal energy and S the
entropy.
Gibbs free energy relates the extrinsic variable ðP and TÞ to intrinsic properties
ðS and VÞ of a material body. The Gibbs free energy of a pure phase and end members of
solid solution at P – T conditions equal
ðT ðT ðp
0 0 CP
GðP; TÞ ¼ H298 þ CP dT 2 T S298 þ dT þ V dP ð5-58:2Þ
298 298 T 1

where CP is the heat capacity expressed as

CP ¼ a þ bT þ cT 2 þ dT 2 þ eT 3 þ f T 20:5 þ gT 21
The Gibbs free energy for solid solution GSS can be expressed as

GSS ¼ x1 G1 þ x2 G2 þ RTðx1 ln X1 þ x2 ln x2 Þ þ Gex

460 Chapter 5

where G1;2 is the Gibbs free energy of solid-solution end members 1 and 2. x1 is a molar
ratio in a phase, x2 ¼ 1 2 x1 and Gex is excess Gibbs free energy of the solid solution
described by polynomial (Redlich – Kister model):
Gex ¼ x1 x2 ½A0 þ A1 ðx1 2 x2 Þ
where A0 and A1 are the pressure- and temperature-dependent parameters.

5.9.1.1. Free energy change and phase boundary

At the phase boundary, there is no change in the free energy ðDG ¼ 0Þ and the locus
of the phase boundary can be written as
DS DH
P¼ T2 : ð5-59Þ
DV DV
Setting the Gibbs free energy change, DG; to zero, the following equation is employed for
the determination of phase boundaries:
ðP
0 0 0
DGP;T ¼ 0 ¼ DHP;T 2 TDSP;T þ DV dP ð5-60Þ
0

where DH is the enthalpy of transition, T is temperature, DV is the volume change across

the phase boundary and P is the pressure. The other symbols carry their usual meaning.
Further, if one assumes that the molar-volume change ðDVÞ is not a strong function
of pressure, then equation can be further simplified to
DG ¼ DH 2 TDS þ PDV:
The entropy DS in phase-boundary calculations can be estimated by selecting one point
along a measured phase boundary, solving for entropy using the equation and then
calculating the remaining phase boundary using this value. Absurd values are obtained if
the chosen point does not fall on the phase boundary. However, vibrational spectroscopic
data and statistical thermodynamics allow entropy to be determined independently.
The general condition for equilibrium is expressed as
RT ln K þ DGT;P ¼ 0
where K is a ratio of products of the activities of products and reactants and the Gibbs free
energy change, DGT;P ; is approximated by the following expression (e.g., Gasparik, 1994):

DGT;P ¼ DHT0 2 TDS0T 2 CT 1=2 þ ðDVT0 2 bPÞP ð5-61Þ

In the above expression, C is a parameter capable of expressing the heat-capacity
differences, particularly those arising from disorder, and b can express the difference in
compressibilities. The parameters can be used only in those cases where the differences in
the heat capacities or compressibilities significantly affect the phase relations and could not
be omitted. The effects on the phase relations arising from differences in thermal
expansions are found to be less significant and thus not necessary to include.
(Crystalline) Materials Under High Pressure 461

The free-energy change DG determines the chemical equilibria (relating reactants

and products):

DG ¼ RT lnðam an · · ·=ax ay · · ·Þ:

For most mineral reactions, DG ranges , 50 kcal (200 kJ) each side of zero.
Under pressure, a change in volume of 1 cm3 corresponds to a change in energy of
10 kcal (40 kJ). This small volume change can bring about a large shift in chemical
equilibria. Phase equilibria are shifted by pressure when inter-atomic bonds are broken and
reformed.

5.9.1.2. Volume change and DH

The thermodynamic relationship that exists between pressure ðPÞ; temperature ðTÞ;
volume change ðDVÞ and heat absorbed in melting ðDHÞ is

dT=dP ¼ TDV=DH:

High-density forms are, in general, favoured at high pressure. Since DH can be

positive or negative, the effects of pressure and temperature may reinforce or oppose
stabilization of the high-pressure phase. It is seen that the melting points of iron increase by
about 1008C (2128F) under pressure of 5– 10 GPa. At 1 bar (105 Pa), NaCl melts in an iron
crucible but, at 10 GPa pressure, iron can be melted in an NaCl crucible. This is because
the DH for the melting of iron is very small compared with the DH for melting of NaCl.

5.9.1.3. Activation volume and activation enthalpy

When the activation volume ðV p Þ is considered equal to the formation volume of
a vacancy, and the vacancy is considered as a cavity in a solid under pressure, the
semi-empirical model of O’Connell (1977) can be used, wherein
0
V p ¼ V0p ð1 þ PK 00 =K0 Þ21=K 0

where V p is the activation volume at pressure P; V0p is the activation volume at zero
pressure and K 00 is the pressure derivative of K0 : K 00 is defined as 4/9 of the bulk modulus
ðKÞ of the matrix.
At low pressure (e.g., 1 GPa), the normal stress at the inter-phase boundary will be
low and could even be tensile owing to the volume change of transformation. At higher
pressure (e.g., 15 GPa), there will be a high compressive stress normal to the boundary.
Changes in the structure of the inter-phase boundary arising from differential stress
can result in changes in the activation volume. Because there are no estimates of V p
(or its pressure dependence) for diffusion across inter-phase boundaries in minerals, Rubie
and Ross (1994) used an empirical model (O’Connell, 1977) with V p decreasing from
12 cm3 mol21 at 1 bar to about 4 cm3 mol21 at 15 GPa (Kirby et al., 1996).
The relationship between the Gibbs free energies for the perovskite and enstatite
phase (Mg, Fe)SiO3 at low and high pressure are schematically shown in Fig. 5.13
as a function of the configurational coordinate describing the phase transformation.
462 Chapter 5

Figure 5.13. The relationship between the Gibbs free energies for the perovskite and enstatite phase of
(Mg, Fe)SiO3 at low and high pressure are schematically illustrated as a function of the configurational coordinate
describing the phase transformation. It is assumed the value of Ep ; the activation energy, is equivalent at low and
high pressure. The energy of transformation at zero pressure, Etr is shown for comparison (Knittle and Jeanloz,
1987a, q 1887 AAAS).

It is assumed that the value of Ep ; the activation energy, is equivalent at low and high
pressure. The energy of transformation at zero pressure, Etr ; is shown for comparison.
For silicate perovskite, the activation enthalpy ðH p ¼ Ep þ PV p Þ is plotted in
Fig. 5.14 as a function of pressure (assuming that the activation volume varies between 1
and 30 cm3 mol21). The top scale indicates the depth of the Earth that corresponds to the
pressure given on the bottom scale. The extrapolated values of H p for perovskite are
compared with a range of values found for olivine. At 670 km depth, if V p for perovskite is
small (1 – 10 cm3 mol), then H p is still small (# 300 kJ/mol) at comparable pressures.
Fig. 5.14 illustrates the importance of determining V p in order to determine H p for minerals
deep in the Earth (Knittle and Jeanloz, 1989)

5.9.1.4. Communal entropy: fluid

Kirkwood (1950) introduced the concept of communal entropy ðDcom STV Þ; which
represents the additional entropy a model fluid acquires when its molecules are free to
move around the whole available volume, compared with a crystal or glassy state in which
(Crystalline) Materials Under High Pressure 463

Figure 5.14. The activation enthalpy ðH p ¼ Ep þ PV p Þ as a function of pressure for silicate perovskite.
The activation volume is assumed to vary between 1 and 30 cm3 mol21. The top scale indicates the depth in the
Earth corresponding to the pressure shown on the bottom scale. The extrapolated values of H p for perovskite are
compared with a range of values found for olivine. If V p for perovskite is small (1–10 cm3 mole), then at 670 km
depth, H p is still small (#300 kJ/mol) in comparison with a representative silicate such as olivine at comparable
pressures (Knittle and Jeanloz, 1987, q 1987 American Geophysical Union).

the molecules are naturally constrained. An alternative communal entropy can similarly be
defined as the entropy difference between the single-occupancy (SO)-cell model and the
free fluid at the same temperature and pressure ðDcom STP Þ: These two communal entropies
are related as

Dcom STV ¼ Dcom STP þ R ln PV=RT ð5-62Þ

At constant T and vanishing P; the Gibbs free-energy difference between the SO-cell
model and the free fluid is exactly RT: The SO-cell model undergoes phase transition at
pressure Pp ; up to which the EOS is given by
pffiffiffi pffiffiffi
PpSO ¼ ½1 þ p 2ðV0 =VÞ4=3 þ ðp 2Þ2 ððV0 =VÞ3 þ ðV0 =VÞ6 Þ=V p ð5-63Þ

where PpSO is a measure of the reduced pressure in hard-sphere units of kT=s 3 (where k is
Boltzmann’s constant), V p is the reduced volume in units of N s 3 and V0 is the minimum
volume at close packing.
464 Chapter 5

For a thermodynamic hard sphere (fluid), the pressure, at a fixed temperature and
volume, is directly related to the collision frequency as

pffiffiffiffi
Pp ¼ ð1 þ nC ðzÞ p=3tÞ=V p ;

where nC is the number of collisions in time, t: Only first neighbours collide in the stable
crystal region.

5.9.1.5. Heat capacity, entropy and phase boundaries

To explain the T-dependence of the heat capacity of solids, Einstein (1907)
proposed the quantization of vibrational energy (as a single characteristic frequency). He
unified the nascent theory of radiation quanta (of Planck) with the thermodynamics of
solids in his groundbreaking paper “Planck’s theory of radiation and the theory of
specific heat” (Einstein, 1907). Here, he demonstrated that if the atomic vibrations are
quantized (in accordance with Planck’s nascent theory) then the heat capacity of a solid
will be temperature dependent rather than be a constant, as given by the Dulong –Petit
law of classical thermodynamics. Taking clues from heat capacities, Einstein plotted the
heat capacity of diamond as a function of temperature to show that atomic vibrations in
solids are quantized. In Einstein’s picture, the heat capacity increases monotonically
from zero with increasing temperature (Fig. 5.15). The temperature (x-axis) is scaled to
the Einstein temperature qE ¼ 1,320 K. The heat capacity (y-axis) is given in
cal mol21 K21). Debye generalized the quantization rules to include all lattice vibrations
(like standing waves). In the modified form, the heat capacity in simple systems
increases as a power of the temperature. The powers correspond to the dimensions
involved, viz. for 3D systems the power is 3 and for 2D and 1D systems the powers are
2 and 1, respectively.
The heat-capacity measurement provides information on the quantized nature of
vibrational structure. The constant volume heat capacity, CV ; can be estimated from the

Figure 5.15. Heat capacity of diamond vs. Einstein temperature (Einstein, 1907).
(Crystalline) Materials Under High Pressure 465

relations
ða
ex
CV ¼ 3Nk x2 gðnÞdn ð5-64Þ
0 ðex 2 1Þ2
where N is the number of atoms in the unit cell, k is the Boltzmann constant, n is the
frequency of vibration, x is hn=kT; h is Planck’s constant and gðnÞ is the DOS (vibrational
model).
The thermodynamic properties and EOS parameters for phases stable at P – T
conditions of the mantle are presented by Saxena and Shen (1992), and Saxena et al.
(1993). For these, the phase-equilibria data, calorimetric measurements and relationship
between CP ; CV ; thermal expansion a; and compressibility b are taken into account.
(Note: CP ¼ CV þ a2 VT=b:)
The constant volume heat capacity, CV ; can be converted to the constant pressure
heat capacity, CP ; using

CP ¼ CV þ TV a2 KT ð5-65Þ

where a is the thermal expansivity and KT is the isothermal bulk modulus. To obtain
entropy, the following is integrated over temperature
ð T1
CP
dS ¼ dT ð5-66Þ
0 T

where T is the temperature under study.

Variations of DH with temperature can be estimated using the heat capacity
obtained from equation (5-65) in
ðT
0 0
DH ðTÞ ¼ DH ðT0 Þ þ DCP dT ð5-67Þ
T0

Changes in volumes can be estimated by first using the bulk modulus and its pressure
dependence in the third-order EOS (Birch, 1978). The volumes are then corrected for
temperature with 1-atm thermal expansivity systematics (e.g., Chopelas and Boehler,
1992). This method circumvents the need to compress the materials at high temperatures
where the bulk moduli are poorly known.
At the point of a phase change in bulk matter, the internal energy change, DU; from
solid to liquid must exactly balance the contribution of entropy to the total energy change,
TDS: Only at P and T where these two are equal can solid and liquid co-exist in
equilibrium. This is the relation that yields the co-existence curve of the solid – liquid phase
diagram. The equality of DU and TDS ensures that the solid and liquid forms are equally
likely to be found.
Heat capacity cannot be measured by performing calorimetry at pressure. This
problem can be circumvented by measuring the vibrational spectra of minerals at mantle
pressures and by utilizing these data to calculate heat capacity and entropy as a function of
both temperature and pressure (e.g., Kieffler, 1980, 1982). The mode-Grüneisen
466 Chapter 5

parameters ðgi Þ obtained from the same experiments permit direct calculation of gth as a
function of pressure.
Kieffer’s (1979) lattice dynamics model accurately reproduces heat capacity CV
and entropy S at 1 atm for forsterite (Akaogi et al., 1984) and fayalite (Hofmeister, 1987).
The pressure dependence of the other properties is computed when CV ðPÞ are known or
can be calculated.
Employing the heat capacity ðCP Þ; thermal expansion ðaÞ and density of solids and
melts ðrÞ; an adiabatic P – T trajectory can be calculated from the relation:

aT
ðdT=dPÞS ¼ ð5-68Þ
rCP

The geotherms corresponding to MORB adiabats have been determined as 1,3008C

(Nisbet et al., 1993). This adiabat intersects the olivine– b-phase transition at 1,3958C
and 13.7 GPa (410 km). In this transition, the reaction is exothermic, while the g-spinel
to perovskite þ Mg-wüstite phase transition is an endothermic reaction and a
temperature decrease should occur across 660 km. However, the phase boundaries
representing equilibrium reactions should run in either direction (up for low pressure or
down for high pressure) with the perturbing P and T; and the sign (positive or negative)
of the changes in enthalpy and entropy should change accordingly. This also signifies
that the olivine to b-phase transition is exothermic and the transformation of g-spinel to
perovskite þ Mg-wüstite is endothermic.

5.9.2. Thermal-expansion coefficient

The effect of pressure on the thermal-expansion coefficient is determined (Birch,

1952) by the relation:
dT
V
a ¼ a0
V0

where dT ð; qÞ is the second Grüneisen parameter, which is assumed to be independent of

temperature and pressure above Debye temperature (e.g., Anderson et al., 1991). a0 and V0
are the thermal-expansion coefficient and molar volume at some reference state. The
thermal-expansion coefficient at higher temperatures and pressures is calculated using the
above equation along an isothermal. The volume dependence of thermal expansion in the
case of alkali halides has been expressed as (Yagi, 1978):

a=a0 ¼ ðV=V0 Þd0

where d0 ¼ 2ð1=aKÞðdK=dTÞP : From static P – V – T measurement, d0 is directly

measured and a systematic difference of d0 is observed between alkali halides with
NaCl structure ðd0 ¼ 2 – 3Þ and those with CsCl structure ðd0 ¼ 6 to ,7Þ: The volume
(Crystalline) Materials Under High Pressure 467

relation is expressed as
ðT
V ¼ V0 exp aðTÞdT ð5-69Þ
T0

This relation is corroborated by the experiments conducted by Fei et al. (1992) on

(Mg0.6Fe0.4)O magnesiowüstite in the range of T ¼ 1,100 –2,000 K and P reaching up to
30 GPa.
Temperature increases the volume while pressure decreases it. The relations are
defined by the thermal expansion coefficient, a; and the bulk modulus:
1 dV 1 1 dV
a¼ : or ¼ : : ð5-70Þ
V dT K V dP
The variation of frequency with pressure provides a basis for estimating entropy vs.
pressure, which directly yields thermal expansivity through the Maxwell equation:
dV dS
¼2 ð5-71Þ
T P dP T

and the thermal expansivity is then

1 dV
a¼
V dT P

To calculate a from the variation of entropy vs. pressure, the molar volume of the phase
must be known at the pressure and temperature of interest. At higher pressures, the bulk
moduli ðK0 Þ and their pressure derivative ðK 00 Þ for each of the phases are used to calculate
the volumes at various pressures. Any uncertainty in volume determination does not
significantly contribute to uncertainty of expansivity. An accurate computation of the
entropy is more important than the volume.
The thermal expansivity is generally inversely proportional to the bulk modulus
ðK0 Þ: This is because gth and V=CV vary little among the minerals. It is reasonable that a
very incompressible mineral such as stishovite will have the lowest a-values whereas the
more compressible ones (e.g., orthoenstatite) will have one of the highest a-values at RT
(see Table 5.5).

5.9.2.1. a values: spectroscopic vs. volumetric

Since the vibrational mode frequencies of crystals depend solely on the variation
of volume, the volume thermal expansivity ðaÞ; isothermal bulk modulus ðKT Þ and
the pressure and temperature dependences of frequency ni of a vibrational mode are
related by
aKT ¼ 2ðdvi =dTÞP =ðdvi =dPÞT ð5-72Þ
The value of ðdvi =dTÞP can be reliably measured through experiments, followed by KT ;
ðdvi =dPÞP and a; using the above relationship if the values of the other three quantities are
available.
468 Chapter 5

At RT the normal modes or vibrations of a crystal lattice at or below , 300 cm21

are predominantly active in the crystal. This can be seen in the anti-Stokes spectrum where
the modes above 300 cm21 have almost no intensity. As temperature is increased, the
intensity of the higher energy modes increase in the anti-Stokes spectrum since more bonds
(modes) become active in the expansion.
At high temperatures, the exact value of a is underestimated. When the higher
energy vibrations become active, the pressure shifts of these modes will increase.
However, the high-energy modes of the high-pressure polymorphs are inter-coupled
and thus are not separable into independent contributions from various polyhedral units
(e.g., MgO6 octahedra and SiO4 tetrahedra).
Using the thermodynamic Maxwell relation (equation (5-71))

ðdS=dPÞT ¼ 2ðdV=dTÞP

the values of thermal expansities ðaÞ can be determined. In this case, the entropies at high
pressures are derived using a statistical method and spectroscopic data.
The spectroscopically determined thermal expansivities of minerals correspond
very well with those derived from volumetric data. The results of some minerals are
tabulated in Table 5.6 (Chopelas, 2000).

TABLE 5.6
Comparison of spectroscopically determined thermal expansivity a with those derived from volume
measurements (Source: Chopelas, 2000)

Mineral Spectroscopic (1025 K21) Volumetric (1025 K21)

Forsterite 2.40a 2.72b

b-Mg2SiO4 1.89a 2.01c
g-Mg2SiO4 1.84a 1.68d
MgO 2.79a 3.11e
Stishovite 1.33(8)0 14.2 g
MgSiO3, Orthoen. 3.25(10)0 2.2 –4.77 h
MgSiO3, High Clen. 2.59(10)0 Not measured
MgSiO3, Majorite 2.24(9)0 2.36
MgSiO3, Ilmenite 1.7(1)0 –
MgSiO3, Perovskite 1.8(1)a 1.72i
a
Chopelas (1996).
b
Kajioshi (1986).
c
Suzuki et al. (1980).
d
Suzuki et al. (1979).
e
Isaak et al. (1989).
f
Chopelas (2000).
g
Fei et al. (1990), Ito et al. (1974a) and Ito et al. (1974b).
h
See Chopelas (2000).
i
Funamori et al. (1993), Utsumi et al. (1995), Wang et al. (1994).
(Crystalline) Materials Under High Pressure 469

5.9.3. Grüneisen parameter (g)

The Grüneisen parameter ðgÞ is used for assigning constraints on geophysically

important parameters such as the P; T dependence of the thermal properties of the mantle
and core, the adiabatic temperature gradient and the geophysical interpretation of
Hugoniot data.

5.9.3.1. Mode Grüneisen (M-G) parameter

The first mode Grüneisen parameter is calculated as
d ln ni
gi ¼ 2 ð5-73Þ
d ln V
where ni is the frequency of the ith vibrational mode and V is the corresponding unit-
cell volume. This is equivalent either to gi ðVÞ ¼ 2ðV=ni Þðdni =dVÞ or to gi ðPÞ ¼
ðKT =ni Þðdni =dPÞ:
Calculation of the first- and second-mode Grüneisen parameters, gi and qi ;
from frequency and pressure determinations is analogous to deriving bulk modulus KT
solely from volume and pressure measurements.
The second Grüneisen parameter is calculated from
qi ¼ d ln ni =d ln V
Equivalently,

qi ¼ ðV=gi Þðdgi =dVÞ ¼ 1 þ gi 2 ðV 2 =gi ni Þðd2 ni =dV 2 Þ

qi ¼ ðKT =gi Þðdgi =dPÞ ¼ g i 2 K 0 ðKT2 =gi ni Þðd2 ni =dP2 Þ:

5.9.3.2. Thermal Grüneisen parameter (gth )

The thermal Grüneisen parameter ðgth Þ has been discussed in the following in
relation to the spectroscopic M-G parameter. If gi for the mode represents all the vibrations
of the crystal (Gillet et al., 1998), then one can write
gi , gth ¼ g0 ðV=V0 Þq
where g0 represents the extrapolated value of gth at zero pressure with the volume
dependence of gth explicitly given by parameter q:
The Grüneisen thermodynamic parameter is calculated as
aK0T V0
gth ¼ ð5-74Þ
CV
where a is the 1-atm thermal expansion coefficient, K0T is the isothermal bulk modulus at
1 atm, V0 is the molar volume at 1 atm and CV is the molar-heat capacity at constant
volume ðCV ¼ CP 2 a2 K0T V0T Þ: The specific heat at constant pressure ðCP Þ is related to
the thermal-expansion coefficient ðaÞ and bulk modulus as CP ¼ Cv þ a2 K0T :
470 Chapter 5

This parameter is of significant interest to geoscientists because it determines the

limitations on the thermoelastic properties of the lower mantle and core. It is directly
related to the EOS.
P The spectroscopically derived weighted average Grüneisen parameter is kgl ¼
Ci gi =Ci ; where Ci is the Einstein heat capacity of mode i and gi is the Grüneisen
parameter of mode i: A comparison P of the weighted average of the spectro-
scopic Grüneisen parameters kgl ¼ Ci gi =Ci Þ with the thermal Grüneisen parameter
gth ¼ ðaKT V=CV Þ is presented in Table 5.7 (Chopelas, 2000). Discrepancies between
kgl and gth have been observed in all minerals studied and kgl is always 10 – 15% lower
than gth ; suggesting that many materials cannot be well described as Debye solids.
The dependence of gth on Fe content in olivine series is discussed in Section 6.5.2
(see Fig. 6.30).
In the vibrational spectrum of silicate minerals, all the high-energy modes are
associated with the very incompressible SiO4 tetrahedra and have very low mode gth values.

5.9.3.3. Density and Grüneisen parameter

The density ðrÞ and the Grüneissen parameter ðgÞ are related as

ðg=g0 Þ ¼ ðr=r0 Þ2q

where q is an arbitrary constant (Anderson, 1968). The EOS along the solidus is obtained
when r is determined from P: To obtain rðPÞ; the bulk modulus – pressure curve KðPÞ is
used. An empirical KðPÞ is obtained from the shock-wave Hugoniot.

TABLE 5.7
P
Comparison of the weighted average of the spectroscopic Grüneisen parameters kgl ð¼ Ci gi =Ci Þ with the
thermal Grüneisen parameter gth ð¼ aKT V=CV Þ (Source: Chopelas, 2000)

Mineral kgl gth

a
Forsterite 1.19 1.29
b-Mg2SiO4 1.29b 1.39
g-Mg2SiO4 1.10c 1.25
MgO 1.47d 1.52
Stishovite 1.40e 1.34
MgSiO3 Orthoenstatite 1.20e 1.28
MgSiO3 High Clinoen 1.09e 1.22f
MgSiO3 Majorite 1.32e 1.28f
MgSiO3 Ilmenite 1.24e 1.22f
MgSiO3 Perovskite 1.43d 1.42

Parameters for all minerals for calculation of gth are available in Chopelas (2000).
a
Chopelas (1990).
b
Chopelas (1991).
c
Chopelas et al. (1994).
d
Chopelas (1996).
e
Chopelas (2000).
f
Using a at room T determined by Chopelas (2000).
(Crystalline) Materials Under High Pressure 471

5.9.3.4. Debye model

Using Einstein and Debye approximations, the Grüneisen parameter in the Debye
model, gD ; bears the relation:
gD ¼ 2d ln QD =d ln V;
where QD is an effective Debye temperature (equivalent to the Einstein temperature, QE ;
by the relation QD ¼ 5=3QE for T . QD ).
Spectroscopic data and theoretical predictions on silicate perovskite indicate that
gth of Mg-silicate perovskite is 1.7– 2.0 (Hemley, 1991), while QD values range from 725
to 1,200 K (Stixrude and Bukowinski, 1990). However, the vibrational DOS is not well
represented by a Debye model.

5.9.3.5. Anderson – Grüneisen parameter

The variation of a with pressure is characterized by the Anderson – Grüneisen
parameter gT;S ; which is defined as
1 dKT
gTS ¼ 2 ð5-75Þ
aK T dT P

which is nearly constant at Earth’s internal temperature conditions. When gT;S is

independent of P and T; it equals ðd ln a=d ln VÞP :
Intregation of first and second Grüneisen parameters with respect to temperature at
constant volume would lead to the Mie – Grüneisen expression for g:
Pth V
g¼ ð5-76Þ
Eth
(Pth ¼ thermal pressure and Eth ¼ thermal energy).
In principle, Pth and Eth can be calculated from ab initio free-energy calculations
(e.g., Vocaldo et al., 1999, 2000).
In the Mie –Grüneisen approach, the total pressure, Ptot ðV; TÞ; can be expressed as a
sum of the static pressure, Pst ; i.e., isothermal compression at 300 K and the thermal
pressure increases along an isochore, DPth :
Ptot ðV; TÞ ¼ Pst ðVÞ þ DPth ðV; TÞ:
Empirical EOS for both (Mg,Fe)SiO3 and (Mg,Fe)O, based on an anharmonic
Einstein model, were developed by Jeanloz and Knittle (1989). These enable one to predict
the densities of both the phases under lower-mantle pressures and temperatures. Above the
Debye temperature, a decrease in gT is expected (Anderson et al., 1990) and hence the
measurements need to be extended to higher temperatures (e.g., . 900 K).

5.9.3.6. Vinet equation

The Vinet equation (Vinet et al., 1987) is derived from a scaled approximate form
for the energy:
EðrÞ ¼ 2DEð1 þ ap Þ exp½2ap 
472 Chapter 5

where
r 2 r0
ap ¼ ;
l
and DE is the binding energy, and r is the length per electron. This gives

PðxÞ ¼ 3KT0 ð1 2 xÞx22 exp½3=2ðK 0T0 2 1Þð1 2 xÞ

where x ¼ ðV=V0 Þ1=3 : The energy can be expressed as
4KT0 V0
E ¼ E0 þ 2 2V0 KTT0 ðK 0T0 2 1Þ22
ðK 0T0 2 1Þ2

½5 þ 3K 0T0 ðx 2 1Þ 2 3x exp½23=2ðK 0T0 2 1Þðx 2 1Þ

Vinet EOS works surprisingly well for a wide range of types of materials and for
compressions of up to h ¼ 0:1:

5.9.3.7. Holzapfel equation

The Holzapfel EOS (Holzapfel, 1996) is similarly given by

PðxÞ ¼ 3K 0T0 x25 ð1 2 xÞ exp½ðcx þ c0 Þð1 2 xÞ

where c0 and c are chosen to give K 0 and the limiting Fermi gas behaviour as x ! 0:
If c ¼ 0; one gets a three-parameter ðV0 ; KT0 ; and K 00 Þ EOS that behaves better at
extreme compression (Hama and Suito, 1996):

25 5 3
PðxÞ ¼ 3KT0 x ð1 2 xÞ exp½c0 ð1 2 xÞ ¼ 3KT0 x ð1 2 xÞ exp ðKT0 2 3Þð1 2 xÞ
2

5.9.3.8. Logarithmic equation

The logarithmic equation EOS (Poirer and Tarantola, 1998) give at third order:

V0 K 0T0 2 2 V0 2
P ¼ KT0 ln þ ln
V 2 V

For extremely compressible matter, e.g., hydrogen, the Vinet EOS is more accurate
than the Birch equation. For most of the fits, the V0 is fixed at a known value of
23.0 cm3 mol21 (Silvera, 1980). The best fit given by the Vinet EOS corresponds
well with the fits using Birch or Holzapfel but the logarithmic EOS fails completely
(Cohen et al., 2000).
5.9.3.9. Microscopic and macroscopic
For assessing the microscopic origin of thermodynamic properties such as thermal
expansivity and entropy, the M-G parameter gi (obtained from least square fits of the high-
pressure mid-infrared and Raman data; Chopelas and Boehler, 1992) is important. The
(Crystalline) Materials Under High Pressure 473

value of g0i (value at zero pressure) is comparatively large, i.e., it has an anharmonicity at
low pressure.
The M-G parameter relates the pressure and volume derivatives of the frequency of
interest. In most cases, the frequency of the far-IR bands depends linearly on volume and
the frequency of mid-IR bands can be related to either volume or pressure.
The g varies as a function of pressure and volume (e.g., Poirier, 1991) as

0 1
d2 ðPV 2x=3 Þ
VB dV 2 C ðx 2 2Þ
g¼2 B@
Cþ ð5-77Þ
2 dðPV 2x=3
Þ A 3
dV

Thus, g behaves differently with different values of parameter x and the Grüneisen
parameter is a direct function of the chosen EOS, which defines P as a function of V:
In the absence of reliable high-P; high-T experimental data, thermoelastic
parameters, such as the Grüneisen parameter, cannot be reliably obtained from
approximate descriptions. Rather, they can be derived from rigorous, highly accurate,
quantum-mechanical free-energy calculations.

5.9.4. Thermal expansion and crystal-field changes

Temperature has a 2-fold influence on a crystal structure. Increased thermal

motions cause increased amplitude of atoms vibrating about their crystallographic
positions and thermal increase in inter-atomic distances causes a decrease in the value of
crystal-field splitting, D; at elevated temperature. From these considerations, the
relationship between the crystal-field shifting and the thermal expansivity is known to be

5=3
DT D0
¼ ¼ ½1 þ aðT 2 T0 Þ25=3 ð5-78Þ
D0 DT

Because of thermal vibration, the absorption bands broaden and, due to the
expansion of the M –O distance, the band centres move to slightly longer wavelengths, i.e.,
are “red-shifted”. Thus, high pressures and elevated temperatures show compensatory
effects on band maxima of absorption bands in crystal-field spectra. But the cumulative
effect of both these causes an intensification of the absorption bands through the effects of
increased covalency and increased vibronic coupling.

5.9.5. Radiative-heat transport

Minerals absorbing radiation in the near-IR and visible regions control the
radiative-heat transport mechanism in the mantle (Clark, 1957).
474 Chapter 5

The energy transfer of photons through a grey body (i.e., one in which absorption by
photons is finite, non-zero and independent of wavelength) is given by (Stacky, 1969)
16n2 ST 3
Kr ¼ ð5-79Þ
3a
where Kr is the effective radiative conductivity, n is the mean refractive index, T is
temperature, a is the mean absorption coefficient and S is the Stefan’s constant. S relates
the power E (i.e. the rate of energy emission per unit area) of an ideal black body to
absolute
dE
¼ ST 4 ð5-80Þ
dT
temperature, T; as follows.
In octahedrally coordinated Fe2þ, the 2t2g ! 4eg ligand-field transition occurs at
near-IR. Absorption of black-body radiation by this transition is believed to retard
significantly the radiative heat flow in the upper mantle (Shankland et al., 1974). However,
the large increase in the 2t2g ! 4eg band energy due to an HS ! LS transition of Fe2þ in
the lower mantle may make the lower-mantle iron (II) mineralogy much more transparent
to black-body radiation and, hence, an increase in the thermal conductivity of the lower
mantle.

5.9.6. Thermal pressure: Eularian strain

Thermal pressure can be described by using the Debye model (Jackson and Rigden,
1996), and by employing the following relationship:
gðVÞ
DPth ¼ ½Eth ðV; TÞ 2 Eth ðV; T0 Þ ð5-81Þ
V
ð u=T
9nRT j3 dj
Eth ¼ ð5-82Þ
ðu=TÞ3 0 ej 2 1

h
V
g ¼ g0 ð5-83Þ
V0

g0 2 gðVÞ
u ¼ u0 exp ð5-84Þ
q
In the above relations, Eth is the vibrational energy for a given volume and temperature, R is
the gas constant, g is the Grüneisen parameter ðq is the volume dependence of the
Grüneisen parameter ðq ¼ d ln g=d ln VÞÞ; which is assumed to be constant, n is the
number of atoms per formula unit and u is the Debye temperature.
The Debye approach (equation (5-84)) provides a description of thermal pressure
without the truncation problem that can arise when one uses a polynomial expansion
(Crystalline) Materials Under High Pressure 475

(Jackson and Rigden, 1996). This enables determination of thermoelastic parameters and
their pressure and/or temperature dependence in an internally consistent fashion (Shim and
Duffy, 2000).
The pressures and their uncertainties are derived from room-temperature – volume
measurements across the sample. Elastic models of samples in the laser-heated diamond
cell indicate that the thermal pressure for perovskite at temperature above 1,300 K may be
3 –5 GPa.
In materials at high temperatures and low strain, there is a large volume dependence
of the thermal pressure (Wolf and Jeanloz, 1985). Birch’s (1978) normalized pressure, F; is
given as (see equation (5-52)):

F ¼ P½3f ð1 þ 2f Þ5=2 21 ; ð5-85Þ

2=3
where f is the Eularian strain measured as f ¼ 1=2½ðV0 =VÞ 2 1:

5.9.6.1. Thermal pressure as a function of volume

The temperature derivative of the isothermal bulk modulus at constant volume is
obtained from the thermodynamic identity:
dKT dKT dKT
¼ þ aKT ð5-86Þ
dT V dT P dP T

A non-zero value of ðdKT =dTÞV would suggest a volume dependence on thermal pressure.
The thermal pressure as a function of volume at a constant temperature is expressed by
ðT ¼ 300 KÞ:
V0
Pth ¼ a þ b ln ; ð5-87Þ
V
where
ðT
a¼ aKT dT
300

and
dKT
b¼ ðT 2 300Þ
dT
For an earlier discussion involving EOS, see the second half of Section 5.8.7.3.

5.10. Phase transitions

The kinetics of phase transitions may elucidate the dynamics of the process in the
Earth. Phase transformation in minerals defines the fields of their stability as a function of
P; T and their intensive variables. At any P; T; the stable phase is one in which the free
energy is the lowest. However, transitions do not always occur at thermodynamic
476 Chapter 5

equilibrium pressures since kinetic factors are also involved. For this reason, it is usual to
observe that, in an experiment of increasing pressure, a higher pressure than the true
equilibrium pressure, Peq ; is needed to induce transition. Again, when pressure is
decreased, the reverse transition does not occur at Peq ; but at a pressure below the
equilibrium pressure.
To study electronic and phase transitions, several techniques, e.g., Raman and
Brillouin scattering, absorption and luminescence measurements, energy dispersive X-ray
diffraction, and conductivity measurements, are employed. The phase transitions observed
in minerals include displacive phase transitions, cation-ordering transitions (e.g., Al –Si
and Na – K) and orientational order –disorder phase transitions. The different types of
phase transitions observed in minerals are given in Table 5.8 (Dove, 1997), which offers
tabulated examples of phase transitions in minerals that occur with a change in either
temperature or pressure. This was extracted from an electronic search of the Science
Citation Index for the years 1981 – 1996. In some cases, details such as symmetry change
may still be unknown or uncertain (marked with a “?”). The transformational behaviour of
some materials may be affected by the presence of sample impurities, sample treatment or
kinetic factors, which can account for some observed uncertainties (source: Dove, 1997).
Reconstructive phase transitions have been observed in complex silicates such as
olivine and pyroxene. The phase-transition behaviour of framework silicate minerals is of
considerable interest. The concomitant development of mean-field theories of phase-
transition behaviour, microscopic computer modelling and new experimental techniques
probing a range of length scales and dynamical phenomena have facilitated a profound
re-evaluation of the structural behaviour of complex silicates.
Many common rock-forming aluminosilicates exhibit diverse structural behaviour
with changing temperature and pressure such as cation order – disorder processes (e.g., Fe–
Mg, Al –Si) and elastic instabilities that lead to displacive phase transitions. The coupling
between individual order parameters, often by means of a common lattice strain, can make
the overall behaviour very complex. Many transitions involve distortions of the alumino-
silicate framework. In the context of phase transition, it should, however, be noted that Al
and Ca enter into solution in major ferromagnesian phases and modify the phase
relationship. But this role could not yet be described in detail from experimental studies. In
the case of non-reconstructive-type phase transition, the results of dynamic and static
experiments agree satisfactorily.

5.10.1. Mixed and quasi-stable phases

However, in reconstructive phase transition involving time sufficient enough for

atomic diffusion, the scenario becomes different as, for the onset of phase transition,
an overriding pressure is required and a wide mixed phase region is observed before the
high-pressure phase region. This may be illustrated with quartz.
Phase transition in quartz starts around 13 GPa, which is higher than the
equilibrium transition pressure to stishovite phase at , 9 GPa, although with no indication
of the appearance of the coesite phase. A wide mixed-phase region is observed to persist up

TABLE 5.8
Examples of phase transitions (after Dove, 1997)
Material
Quartz, SiO2
Cristobalite, SiO2
Tridymite, SiO2
Leucite, KAlSi2O6
Albite, NaAlSi3O6
Anorthite, CaAl2Si6O16
Kalsilite, KAlSiO4
Kaliophilite, KAlSiO4
(Crystalline) Materials Under High Pressure
TC or PC Change Comments Reference
848 K P62 22 ! P31 21 Two-stage displacive phase transition involving an Castex and Madon
intermediate incommensurate phase (1995); Dolino and
Vallade (1994)
530 K, 1.2 GPa Fd3m ! P41 21 2; First-order displacive phase transitions involving Dove et al. (1995);
P41 21 2 ! P21 zone-boundary instabilities Palmer et al. (1994)
748 K, 623 K P63 =mmc ! P63 22; A number of displacive phase transitions occur on Cellai et al. (1995)
P63 22 ! C2221 cooling. The two given here involve zone-centre
instabilities
960 K Ia3d ! I41 =acd; Two-stage displacive phase transition, the first being Dove et al. (1995)
I41 =aCd ! I41=a a ferroelastic phase transition. Other materials with
the leucite structure but different chemical compo-
sition can undergo other displacive and order–
disorder phase transition
1,250 K C2=m ! C 1 Ferroelastic phase transition. An Al –Si ordering Xiao et al. (1995b)
transition follows at lower temperatures but, because
this does not involve a further symmetry change, it
does not lead to a distinct transition temperature.
Substitution of Kþ for Naþ suppresses the
ferroelastic phase transition and is only weakly
dependent on temperature
560 K I 1 ! P1 Displacive phase transition involving a zone- Daniel et al. (1995);
boundary instability. Substitution of Sr2þ for Ca2þ Phillips and Kirkpa-
allows a ferroelastic phase transition although the trick (1995)
ordering transition to an I2=m phase
A preliminary study indicates the presence of one or Capobianco and
more phase transitions, but details remain sketchy Carpenter (1989)
1,000 K P63 22 ! P63 ð?Þ Apparently a zone-centre transition, but details are Cellai et al. (1992)
sketchy
477

TABLE 5.8 (continued)
Material
Calcite, CaCO3
Soda niter, NaNO3
Akermanite, Ca2 MgSi2 O7 ,
and related melilites
Cordierite, Mg2Al4Si5O18
Perovskite
Titanate, CaTiSiO5
Staurolite
Colemanite,
CaB3O4(OH)3.H2O
Chlorapatite, Ca5(PO4)3Cl
Cryolite, Na3AlF6
Langbeinite, K2Cd2 (SO4)3
Natrite, (Gregoryite), Na2CO3
478
TC or PC Change Comments Reference
1,260 K, 1.5 GPa  ! R3m;
R3C  Orientational order– disorder phase transition invol- Dove et al. (1997);
 ! P21 =c ving the carbonate molecular ions. The ordering Fiquet et al. (1994)
R3c
involves doubling of the size of the unit cell. An
additional phase transition occurs at 2.2 GPa
560 K  ! R3m
R3c  Orientational order– disorder phase transition invol- Harris et al. (1990)
ving the nitrate molecular ions. The ordering
involves doubling of the size of the unit cell
343 K  1 m ! Inc
P42 Incommensurate displacive transition. A possible Brown et al. (1994);
phase transition to another commensurate phase at Webb et al. (1992)
low temperature has not yet been identified
P6=mcc ! cccm Al–Si ordering transition Redfern et al. (1989a);
Thayaparam et al.
(1996)
1,384 K, 1,520 K Cmcm ! Pbnm; Displacive phase transition involving tilt of TiO6 Guyot et al. (1993);
Pm3mð?Þ ! CmCm octahedra, with evidence of phase transition to Redfern (1996)
tetragonal and cubic phases at higher temperatures
497 K C2=c ! P21 =a Zone-boundary displacive phase transition Bismayer et al.
(1992);
Zhang et al. (1995)
Cmm ! C2=m Al-vacancy ordering transition Hawthorne et al.
(1993)
270 K P21 =a ! P21 Ferroelectric phase transition Gallup and Coleman
(1990)
620 K P63 =m ! P21 =a Ferroelastic phase transition Bauer and Klee (1993)
820 K Immm ! P21 =n Spearing et al. (1994);
Yang et al. (1993)
P21 3 ! P21 21 21 Transition temperature depends on composition, Boeriogoates et al.
with several possible substitutions of the Cd2þ cation (1990);
Percival (1990)
760 K P63 =mmc ! C2=m Ferroelastic phase transition involving the softening Harris et al. (1993,
Chapter 5
of the C44 elastic constant 1995, 1996);
Swainson et al. (1995)
 (Crystalline) Materials Under High Pressure
Ilvaite, CaFe2Si2O2(OH) 346 K Pnam ! P21 =a Phase transition driven by ordering of electrons on Ghazibayat et al.
the Fe sites (1992);
Ghose et al. (1989)
Sodalites Sodalites of different composition can undergo Depmeier (1988,
displacive phase transitions, Al– Si ordering phase 1992)
transitions, and phase transitions involving orienta-
tional ordering of molecular ions in the large cavity
coupled to displacive distortions of the sodalite
framework
Brucite, Mg(OH)2 6–7 GPa Possible phase transition involving ordering of Catti et al. (1995);
the H atom Duffy et al. (1989a)
Garnets Several postulated cation-ordering phase transitions Hatch and Griffen
based on different observed ordered structures (1989)
Gillespite, BaFeSi4O10 1.8 GPa P4=ncc ! P21 21 2 First-order phase transition, mostly displacive in Redfern et al. (1993,
character but also involving some changes in 1997)
coordination
Ferrosilite, FeSO4 1.4–1.8 GPa C2=c ! P21 =c Displacive phase transition. Similar transitions are Hugh-Jones et al.
found in other pyroxenes (1994); Shimobayashi
and Kitamura (1991)
Arcanite, K2SO4 860 K P63 =mmc ! Pmcn Orientational ordering of SO22
4 anions Miyake et al. (1981)
Sanmartinite, ZnWO4 – P2=c ! P1 Jahn–Teller phase transition as a function of of Redfern et al., (1995);
scheelite, CuWO4 composition Schofield et al., (1994)
Chiolite, Na5Al3F10 150 K P4=mnc ! P21 =n Displacive phase transition Spearing et al., (1994)
Schultenite, PbHAsO4 313 K P2=c ! Pc Ordering of the hydrogen bond Wilson (1994)

479
480 Chapter 5

to , 50 GPa. The observed Hugoniot of the high-pressure phase is explained as due to

stishovite (see Graham, 1973).
Again, a quasi-stable phase with a higher free energy may survive because of
kinetic factors. The best-known example for such a case is offered by carbon, for which the
stable phase with minimum free energy is graphite and the metastable phase is diamond.
Therefore, “diamonds are not for ever”, but its change is so slow that it can hardly be
perceptible even over a million years!
Similarly, Si, on release of pressure from the metallic state, remains in a metastable
state, which has a volume intermediate between that of the normal semiconducting silicon
and the high-pressure metallic phase. The energy bands in silicon show the empty metallic
conduction band and the filled valence band, formed by hybridization of 3s and 3p
electrons.

5.10.2. Lattice disorder

Temperature may affect the transition by introducing disorder into the lattice, which
introduces a strong scattering of the conduction electron by the local lattice defects. Upon
melting when the disordering is large, the bands become so diffuse as to make the
transition from one phase to another continuous rather than discontinuous.
Lattice disorder introduces strong scattering of the conduction electron by local
lattice defects. For a given crystal structure and chemical composition, the relative
positions of the various electronic bands are determined mainly by the inter-atomic
distances. The temperature effect operates only via lattice disorder and excites photons
(which are indirect ones), causing blurring of the bands. This blurring is sufficient to
destroy the narrow gap existing between the valence and the conduction bands in
semiconductors.

5.10.3. Silicon: b-tin ! hcp

Computer modelling showed that the simple hexagonal phase of Si would be a

high-T superconductor, and this has later been confirmed by experiment (Erskine et al.,
1987). Silicon under pressure at 12 GPa shows a transition to b-tin structure, followed by a
simple hexagonal phase at 14 GPa. At still higher pressures, transitions to the hexagonal
close packed (hcp) and face centred cubic (fcc) take place.

5.10.4. Cation distribution and order – disorder

The transitions are structural phase transitions or isostructural transitions where the
valence state of the atom or the nature of the spin density wave state changes with pressure.
A crossing of crystal-field levels with pressure may even occur that can be presented using
an angular overlap model.
The minor cation substitutions that control the order – disorder variants in mantle
minerals are presented in Table 5.9.
(Crystalline) Materials Under High Pressure 481

TABLE 5.9
Minor cation substitutions and order–disorder variants that might affect EOS of mantle minerals (compiled by
Hazen and Yang, 1999)

Mineral and composition Cation substitutions Order–disorder

3þ
Olivine (Mg,Fe)2SiO4 Ca, Fe Mg–Fe
Wadsleyite (Mg,Fe)2SiO4 Fe3þ, H Mg–Fe
Spinel (Mg,Fe)2SiO4 Fe3þ, Ti Mg–Si
Pyrope (Mg3Al2Si3O12) Ca, Fe, MgSi–2Al Mg–Si; Ca–Mg
Majorite (Mg,Fe)SiO3 Ca, Al, Fe3þ Mg–Al– Fe3þ –Si; Ca –Mg
Clinopyroxene (Mg,Fe,Ca)SiO3 Na, K, Al, Fe3þ, Ti Mg–Al– Fe3þ –Ti– Si; Ca–K–Na
Orthopyroxene (Mg,Fe)SiO3 Ca, Al, Fe3þ Mg–Al– Fe3þ –Ca
Perovskite (Mg,Fe)SiO3 Ca, Al, Fe3þ Mg–Al– Fe3þ –Si
Anhydrous B (Mg,Fe)14Si5O24 Al, Fe3þ Mg–Fe2þ –Al –Fe3þ –Si
Magnesiowustite (Mg,Fe)O Fe3þ, vacancies Fe2þ –Fe3þ vacancies

5.10.5. Incommensurate phases

Incommensurate phases occur in systems when competing periodicities of the two

lattices, such as basic and atomic, show misfits when packed together.
An incommensurate quantity is a periodic distortion of an otherwise regular lattice
(displacive incommensurability).
The incommensurate wave is made up of an amplitude wave (amplitudon) and a
phase wave ( phason). The spatially modulated electron density forms the charge-density
waves (CDW). The periodic distortion accompanying the CDW (due to interaction
between the conduction electron and the lattice) is responsible for the incommensurate
phase.
Incommensurate phases are commonly encountered in metal oxides, sulfides and
other materials where point defects (vacancies) order themselves, giving rise to
superstructures. Many insulating solids exhibit incommensurate phases. The phase
transitions generally occur in the order: normal ! incommensurate ! commensurate as
the temperature is lowered.
A modulated structure is described as a periodic or partly periodic perturbation of a
crystal structure with a repetition distance appreciably greater than the basic unit-cell
dimensions. For the incommensurate phase, a unit cell cannot be defined. No unit cell can
contain an exact period of both the wave and the underlying crystal structure.
Accompanying the phase transitions, anomalies in electrical resistivities of the
chalcogenides are observed.

5.10.6. Order of transition: first order and second order

First-order transition is associated with significant hysteresis. (Note: Hysteresis

effects are used to characterize the “order” of a transition.)
482 Chapter 5

The Clausius – Clapeyron equation describes the thermodynamics at a first-order

transition:

dP DS DH
dT DV TDV

Second-order transition is associated with some disordering process. The ordering

parameter becomes unity for a perfect order while at perfect disorder it is zero. In second
order transition, DV and DS have zero values.
Landau’s theory provides the basis for second-order transitions. In second-order or
structural transitions, the symmetry of the crystal changes discontinuously. Second- and
higher-order transitions are often referred to as continuous transitions. In a second-order
transition, the soft-mode frequency is zero at Tc while, in a first-order transition, the change
of phase occurs before the mode frequency reaches zero.

5.10.7. Order parameters

Examples: For ferromagnetic to paramagnetic transition the order parameter is

magnetization.
For ferroelectrics such as BaTiO3, the order parameter is polarization.
For phase transition of SrTiO3, the order parameter is the angle of rotation of the
oxygen octahedra. In this transition, one of the optic modes of SrTiO3 exhibits softening
behaviour (i.e., soft modes). A soft-mode behaviour under pressure has been examined
by Samara (1984). It is known that not every phase transition is associated with a soft
mode. Phase transitions in some ferroelectrics may result from lattice dynamical
instability.

5.10.8. Superlattice ordering

Superlattice ordering of point defects has been found in metal halides, oxides,
sulfides and other systems.

5.10.9. Structural changes

On heating, CsCl will transform to NaCl structure, while a distorted perovskite

would transform to a cubic form. For the former, reconstructive transition (Buerger, 1951)
is invoked.
In displacive transition, only small changes in the arrangement of coordination
polyhedra occur.
Structural transitions can be ferrodistortive — with no change in the number of
formula units in the unit cell (e.g., ferroelectric materials) and anti-ferrodistortive — with
changes in the number of formula units in the unit cell (e.g., both ferroelectric and
anti-ferroelectric materials).
(Crystalline) Materials Under High Pressure 483

5.10.10. Phase changes: principles and types

In high-pressure phase-transition studies, the structure of the heavier element

compounds at ambient conditions usually correspond very well to the high-pressure
polymorph of the lighter element compounds in the same group of the Periodic Table.
Isostructural compounds are different only in the types of cations. It is possible that
compounds having smaller cations require higher pressure to undertake the same type of
phase transition at the same temperature. In such cases, repulsion in a polyhedron
dominates over the transition.
A phase transition is called first order when the transition occurs with a
discontinuous change in the structure and entropy, and, hence a latent heat, at the transition
temperature. It is called second order where the structure of the low-temperature phase
merges continuously with that of the higher temperature phase at the phase transition, with
no discontinuous change in the entropy and hence no latent heat.
Displacive-phase transitions involve small motions of atoms to change the
symmetry of the crystal structure (see Fig. 5.16). For example, in the case of quartz,
cristobalite and leucite, the phase transitions involve small translations and rotations of the

Figure 5.16. Displacive phase transitions in minerals. In each case a is the low-temperature phase and b is the
high-temperature phase. In these the displacive phase transition occur as a result of rotations and translations of
the nearly rigid tetrahedra (Dove, 1997, q 1997 Mineralogical Society of America).
484 Chapter 5

(Si,Al)O4 tetrahedra (Fig. 5.16; Dove, 1997). In the case of displacive transitions, an
octahedra may rotate about the [001] axis (e.g., TiO6 in SrTiO3) while, in some, the
octahedra tilt by different amounts about all three axes (e.g., TiO6 and SiO6 octahedra in
CaTiO3 and MgSiO3, respectively).

5.10.10.1. Thermal transformations

The thermal transformations affecting a crystal are guided by the following
three rules:
Rule 1
The effect of temperature is scalar and the deformation of a crystal can be described by a
second-rank tensor.
Rule 2
Anharmonicity of thermal oscillations of atoms cause an increase in interatomic distance
affecting the thermal expansion. More anharmonic oscillations with higher amplitudes
(i.e., weaker bonds) manifest greater thermal expansion.
Rule 3
An increase in the thermal motion (vibrations, rotations and jumps) of atoms, etc., allows
an increase in the crystal symmetry through raising the vibrational symmetry of the atoms,
ultimately leading to more symmetrical high-temperature modification.

5.10.10.2. Soft modes

Each vibrational mode of a crystal structure is associated with a specific periodic
distortion of the structure. When a high-temperature symmetric phase is cooled, the
frequency of the “soft” mode decreases. When it becomes zero, the structure can hardly
continue with the distortion and transforms to a lower symmetry phase. The word “soft”
denotes crystal that yields to the displacements of atoms. Thus, in the high – low transition
of quartz, softening takes place in the normal lattice mode, occurring at 208 cm21 in the
high-temperature form. This soft mode is Raman active. (Note: In a zone-centre transition,
the lattice mode can be measured by Raman or IR spectroscopy, although symmetry-
dependent selection rules would define which modes are active or inactive.)
The soft-mode frequency reaches zero at the Brillouin zone boundary, at points
with the appearance of new reflections in the diffraction pattern of low-temperature form.
Soft modes may be optic modes or acoustic modes.
Crystal structure changes through mode softening can be evaluated by the change in
some order parameter (e.g., Q) as a function of temperature. In quartz, for example, the
order parameter is related to the tilting angle, h; of SiO4 tetrahedra while, in the perovskite
structure, it is the angle of rotation, w; of the octahedra.
5.10.10.3. Order parameter (h). Free energy and transition temperature
The free energy of the low-temperature phase can be written as a power series in the
order parameter h as (Dove, 1997):

GðhÞ ¼ G0 þ 1=2 Ah2 þ 1=4 Bh4 þ · · · ð5-88Þ

(Crystalline) Materials Under High Pressure 485

where the parameters A and B are constants and G0 is the free energy of the system for
h ¼ 0 (Dove, 1993). Usually, GðhÞ is independent of the sign of h and, therefore, only
contains terms with even powers of h:
Equation (5-88) represents an expansion of the free energy about a maximum value
in the low-temperature phase and is, therefore, expected to be valid only for small values of
h; i.e., only close to the phase transition.
For the free energy of equation (5-88) to represent a phase transition, it is necessary
that the value of A changes sign at the transition temperature so that it is positive for
temperatures above the transition temperature Tc and negative for those below. The
simplest implementation of this condition is to assume that A ¼ aðT 2 Tc Þ: It is also
assumed that we only need to consider the smallest number of terms in the expansion so
that we can rewrite equation (5-88) as

GðhÞ ¼ G0 þ 1=2 aðT 2 Tc Þh2 þ 1=4bh4 ð5-88:1Þ

where a and bð¼ BÞ are positive constants. The equilibrium condition dG=dh ¼ 0 applied
to equation (5-88) leads to the predictions that h ¼ 0 for T . Tc ; that there is a continuous
(second-order) phase transition at T ¼ Tc and that at, lower temperatures, h is non-zero
and has the temperature dependence:

aðTc 2 TÞ 1=2
h¼ ð5-89Þ
b
When the constant b is negative, the form of the free energy gives a discontinuous (first-
order) phase transition and the expansion of the free energy must be taken to a higher order
(see Dove, 1993, Appendix D). The order parameter can be described by the above
equation for temperatures down to 100 –200 K, when the effects of the Third Law of
Thermodynamics becomes important. This relation holds good at temperatures below Tc
but, at very close to Tc ; it takes the general form

h ¼ aðTc 2 TÞb ð5-90Þ

A term of the form PV could be added to the free energy, noting that the change in
volume at a phase transition usually scales as DV / 21=h2 : At increasing pressure, the
value of h2 at 0 K is likely to increase.
Since the electrons responsible for the JT effect are on the outside of the ions, they
interact strongly with the lattice, giving rise to structural-phase transitions (at high
temperature). The major products of electronic-phase transitions may be classed under:
(a) Ferromagnets. In these, long-range magnetic order below the Curie temperature
induces a spontaneous magnetization, Ms ; resulting in many magnetic domains, each
with a Ms vector oriented in a direction different from that in adjacent domains.
(b) Ferroelectrics. These are characterized by different orientational states below Tc but
the spontaneous polarization is induced by cooperative crystallographic distortion.
In these, the domain boundaries can be controlled by an external field.
(c) Ferroelastics. In these, a cooperative crystal distortion induces a spontaneous strain
below Tc : The domain boundaries of a ferroelastic can be controlled by applied stress.
486 Chapter 5

5.10.10.4. Landau theory

Landau theory is effectively employed in describing the thermodynamics of phase
transitions (e.g., Salje, 1992). Odd terms in the Landau free energy force the phase
transition to be first order.

Ferromagnetic phase transition. Landau theory works well over a wide range of
temperatures for displacive-phase transitions such as ferromagnetic-phase transitions.
The mean-field theory of ferromagnetism predicts that the magnetization will vary as
lTc 2 Tl1=2 and the magnetic susceptibility will vary as lTc 2 Tl21 ; regardless of the
specific details of the magnetic ordering. Landau theory predicts identical behaviour.
For magnetic-phase transition at temperatures close to transition temperature, the
magnetization is found to vary as lTc 2 Tlb with b < 0:38; and the susceptibility as
lTc 2 Tlg with g < 1:3:
Landau theory gives a good approximation to the free energy. The framework of
Landau theory can allow the relationship between different phase transitions in the same
material to be understood, such as when there is an Al –Si ordering-phase transition and a
displacive-phase transition. By employing Landau theory, many questions may be
answered by symmetry arguments. In Al – Si ordering-phase transition, the energy required
to form Al – O –Al linkages is quite large (, 40 kJ mol21 or more; Dove et al., 1996) and
the cations only disorder in equilibrium at temperatures well above the melting points.
Strangely, there exists a wide range of ordering temperatures, even to low temperatures,
and there is also some mechanism that allows disordering for a phase transition to occur.

Ferroelastic transition. For ferroelastic phase transitions, one may consider only the
strain-order parameter coupling. This leads to re-normalized elastic constants, which
reflect the inverse susceptibility. In the pressure range where the orientation strain
ellipsoids remain invariant, the strain-order parameter coupling also remains invariant.
Such high-pressure phase transition can, therefore, be treated in the same Landau manner
as many high-temperature structural-phase transitions.

Ferroelectric-phase transition. In PbTiO3 perovskite, Pb2þ and Ti4þcations are off-

centre along [001] to generate a ferroelectric-phase transition. (Note: In a ferroelectric-
phase transition, small changes in the atomic positions generate a macroscopic dielectric
polarization.) BaTiO3 is well known for this ferroelectric-phase transition. In the high-
temperature cubic phase, the Ti4þ atoms are potential-energy maxima.
The potential-energy minima for the Ti4þ cations are located away from the central
site along the eight k111l directions. In the high-temperature phase, the Ti4þ cations hop
among the eight different sites. When the Ti4þ cations lie preferentially in the sites in the
positive c-direction, the ferroelectric-phase transition occurs. Still, four of these remain.
Therefore, on further cooling, there are subsequent phase transitions leading to all Ti4þ
cations occupying the same single site in the unit cell. Ferroelectric-phase transitions that
involve the ordering of a proton between the two sites on a double-well hydrogen bond,
such as in KH2PO4, are a further extreme example of this (e.g., Lines and Glass, 1977).
(Crystalline) Materials Under High Pressure 487

Such transitions may be considered as displacive, with small atomic displacements,

or else as an order – disorder transformation. But in the displacive case, very little of the
entropy is configurational, whereas in the order –disorder case the entropy is mostly
configurational.
Displacive-phase transitions can be understood in terms of soft-mode theory, which
developed from a better understanding of lattice dynamics using inelastic neutron-
scattering techniques (Ghose, 1985, 1988).
The theory stands on the observation that, on cooling toward the transition
temperature, the frequency of the lattice vibration falls to zero. A vanishing frequency
implies a vanishing restoring force against the corresponding deformation. For this reason,
it is called a soft mode. The atomic displacements associated with the soft mode are the
same as the deformation of the structure in the low temperature.

Displacive transition: polymorphism. In displacive transition, the primary bonds in the

structure are distorted. Symmetry change occurs, usually between high- and low-
pressure/temperature forms (e.g., between high and low quartz at 5738C). In this transition,
as the temperature decreases, the high-symmetry structure becomes unstable relative to
some specific distortion. A displacive can be continuously monitored in terms of
temperature (or pressure) and bond angle. This transition is usually fast and involves only
small changes in energy. No change in translational symmetry occurs, i.e., the unit cell is
essentially the same for both the polymorphs. This is known as zone-centre transition.

Reconstructive transition
Olivine(a) ! spinel(g). In olivine structure, oxygen atoms constitute a hexagonal close-
packed array with Mg, Fe occupying half of the octahedral sites and Si occupying one-
eighth of the tetrahedral sites. The close-packed oxygen layers are parallel to (100). In
spinel structure, the oxygens are nearly close-packed cubic. Transition from olivine to
spinel does not call for a coordination change of cations but the linkages between the cation
polyhedra become more compact, which may account for the density increase by , 8%.
The volume reduction implies a reduction in the effective radius of oxygen.
The mechanism for olivine– spinel transformation may be martensitic in nature and
hence diffusionless. This involves the passage of partial dislocations on alternate close-
packed oxygen layers to convert the hexagonal close-packing of olivine to cubic close
packing in spinel. For this transformation, cation displacements (i.e., synchroshear) are
needed.
In the martensitic mechanism, close-packed layers in olivine and spinel are parallel,
i.e., the topotactic relation of (100)ol parallel to (111)sp holds.

b-Olivine ! spinel(g). The structural relation between b-(wadsleyite) phase and

g-(spinel) phase is such that one can be transformed to the other by the passage of
partial dislocations with Burgers vector equal to this glide operation. A glide-plane
displacement converts the existing glide to a mirror and vice versa. The stacking fault,
which thus results, shows a local structure of the other phase. Therefore, a stacking fault in
488 Chapter 5

a b-phase has the spinel structure. This is how stress-generated dislocations provide
nucleation sites for phase transformations to proceed.
b – g transition depth in a subducting slab: Plunging velocity. The nucleation and
growth mechanism in a subducting slab can be activated beyond the cut-off temperature of
, 7008C. When the velocity of the downgoing slab is high, the overpressure will be high
and olivine becomes metastable. When the temperature goes beyond the cut-off
temperature, the rate of transition to spinel will be rapid enough to increase enormously
the change in free energy (i.e., DG), causing implosive transition. This energy release as
seismic waves causes the deep-focus earthquakes in downgoing slabs. The depth for
transition to occur decreases with decrease in plunging velocity. A greater subduction rate
means a greater depth at which the b –g transition will take place.

Driving forces for phase transition. There are two mechanisms for driving the phase
transition to occur. First, a force should exist to distort the structure locally and, secondly,
some interaction should exist to give a coupling between local ordering processes. The
operation of the forces can be illustrated with the cases of BaTiO3 and aluminosilicates.
In BaTiO3, the Ti4þ cation hops between sites of local potential energy minima that
give rise to the local distortions of the structure at low temperatures. The Ti4þ cations in
neighbouring unit cells interact and force each other to order in the same way.
Similarly, in many aluminosilicates, the large cavities formed by the framework of
linked SiO4 and AlO4 tetrahedra are occupied by cations such as Kþ and Ca2þ. In these
phase transitions, both displacement of the cations from the centres of the cavities and
collapse of the framework are involved. It is possible that the cations rattling around in
their cavities will like to order and also that some mechanism should develop for ordering
the cations in the neighbourhood as well. Coupling of these would bring forth the
transformation of the structure.
For displacive-phase transitions in silicates, there are two aspects for the driving
forces.
First, there is coupling between local ordering or deformation and the neighbouring
atoms that allows long-range ordering, described by parameter J in the standard paradigm.
In silicates, it arises from the stiffness of the tetrahedra that leads to a local deformation
propagating over large distances.
Second, there is a longer range force that drives the actual deformation, which is
described by the doubled-well potential VðhÞ; discussed in Section 5.10.10.6. below.

5.10.10.5. Landau order parameter

For a structural distortion, change in free energy is associated with the changes in
enthalpy and entropy. The latter two can be measured calorimetrically.
The variation in free energy can be related to the interaction energies between the
atoms. Again, the macroscopic properties such as strain, optical birefringence and site
occupancy change the thermodynamic properties.
The Landau order parameter, Q; is related to the change in some macroscopic
property through the phase transition. Change of order parameter with temperature
describes the thermodynamic process of the ensuing phase transition. For example, in the
(Crystalline) Materials Under High Pressure 489

transition from tetragonal to orthorhombic structure, the optical indicatrix changes from
uniaxial to biaxial, while the change in birefringence (which is equivalent to the degree of
transformation) is directly proportional to Q; the order parameter. A discontinuity in the
change in Q is marked by the critical temperature, Tc ; of the phase transition.
Strict symmetry rules define the form of order parameter in relation to the change in
symmetry. The correct form of the order parameter and its relationship to certain physical
properties for a given change in symmetry is available in standard tables. Usually, the
measured properties scale as Q or Q2 :

5.10.10.6. Origin of doubled-well potential, V(h)

There are three contributors to the doubled-well potential, VðhÞ; in silicate phases
(Dove et al., 1995).
First, the long-range interactions, mostly arising from interactions between the
highly polarizable O atoms. These are attractive and tend to contract the structure to the
densest state possible, but the collapse is thwarted by short-range pepulsive interactions.
Second, the short-range interaction between a cation (e.g., Kþ or Ca2þ) occupying
a large cavity site and neighbouring O anions. However, a collapse of the cavity about the
cation may take place and the effect may propagate over large distance.
Third, the energy with the Si –O –Si (or Al – O – Al) bond angle, which is ideally
, 458. Bond angles differing from this value will have a higher energy. For high-
temperature cristobalite, this angle enlarges to 1808 (Schmahl et al., 1992) and this is
associated with a disorder with neighbouring tetrahedra, which tend to rotate to reduce this
angle (Swainson and Dove, 1995). To have as many bonds as possible to lead to an ideal
bond angle, the structure should undergo displacive-phase transition. When the Si – O –Si
bond angle is near the ideal value (, 458), the energy associated with this bond will oppose
phase transition or else phase transition will involve the rigid-unit mode (RUM) distortion
(of SiO4 tetrahedra) with the smallest distortion of the Si –O – Si bond angle (Dove et al.,
1995). RUM is discussed below.

5.10.10.7. Rigid-unit mode: “split atoms” and energy spectra

The existence of RUM in framework silicates is not trivial. Each tetrahedron has six
degrees of freedom ðFÞ: Each corner has three constraint equations that link it to the corner
of the connected tetrahedron, so the number of constraints per tetrahedron is also six. Thus,
the connected tetrahedra in a framework silicate, constrained with F ¼ C (total number of
constraints), should have no modes of deformation. Some computational methods have
been developed to determine all RUM for a given framework structure, taking account all
possible wave vectors (Hammonds et al., 1994).
Each tetrahedron is taken as a rigid body and atoms shared by two tetrahedra are
counted as two separate atoms, called “split atoms”. Any mode of deformation rotating or
translating the rigid tetrahedron may cause the split atoms to separate. This concept can be
incorporated into the formalism of molecular lattice dynamics (Dove, 1993, Chapter 6).
The RUM are then the vibrational modes calculated to have zero frequency. The number of
RUM in any structure is usually small (but greater than zero) compared with the total
number of wave vectors (Hammonds et al., 1996). On constraining tetrahedra as perfectly
490 Chapter 5

rigid holding no intertetrahedral forces, the RUM are the modes with zero frequency but, in
reality, the inter-tetrahedral forces will be non-zero and these will lead to an energy
spectrum for the RUM (e.g., in cristobalite, the RUM energies are of the order 0– 1 THz;
Dove et al., 1995).). In the high-T phase of SiO2, the Si –O – Si bond angle is near to
the ideal value and a few rigid-unit mode (RUM) distortions do not involve change in
this angle.

5.10.11. Pressure-induced order – disorder

Effects of pressure on atomic order – disorder in crystalline phases have received

much attention because of their importance in geophysics, solid state physics and material
science (see Hazen and Navrotsky, 1996). Pressure-induced order – disorder phenomena
play a key role in the energetics, crystal chemistry and physical properties of solids such as
minerals, ferroelectic materials, alloys, fullerenes and high-temperature superconductors.
Order – disorder transition may alter crystal symmetry, causing changes in electrical and
thermal conductivity, vibrational spectra and elastic moduli. Systematic studies on atomic
order –disorder, compressibility and crystal chemistry of mineral phases are essential to
obtain more realistic EOS for mantle phases that exhibit ordering-dependent properties.
Cation order– disorder can significantly modify high-pressure behaviour. In
addition to structural and compositional factors, cation order –disorder also plays an
important role in determining the elastic properties of crystalline phases. Even the
properties of ferroelectrics and cuprate superconductors may be tuned by pressure. Cation
disorder can increase compressibility (Hazen and Yang, 1997).
Pressure-induced ordering is an unexpected phenomenon. Garnet group minerals,
which are cubic in structure, occur above 400 km. At higher pressures, they become
tetragonal through an ordering of the contents of the octahedral and dodecahedral sites
(Angel et al., 1989). Similarly, a large degree of ordering is seen in wadsleyite, b-(Mg,Fe)2
SiO4 (Finger et al., 1993). High-pressure cation ordering has been observed in olivine
(Aikawa et al., 1985), wadsleyite (Finger et al., 1993), garnets (Hazen et al., 1994) and in
anhydrous phase-B (Hazen et al., 1992). This aspect has been discussed in sections relating
to some of these minerals.
Hazen and Navrotsky (1996) reviewed the effects of pressure on order –disorder
reactions and demonstrated that many phases display a significant volume disordering:

DVdis ¼ Vdisordered 2 Vordered:

Silicates with Mg – Fe ordering commonly have DVdis up to 0.5%, while values

exceeding 2% obtain for some mixed-valence oxides and sulfides. Some (e.g., Liebermann
et al., 1977) used velocity – density systematics to infer that cation ordering affects spinel
elasticity. Cation ordering can also influence the EOS; e.g., the bulk modulus of
stoichiometric pseudo-brookite-type MgTi2O5 is seen to vary by 6%, depending on the
ordered state of Mg and Ti in two different octrahedral sites.
AB2 O 4 spinels ðFd3mÞ have two octahedrally coordinated cations for
each tetrahedrally coordinated cation. “Normal” spinels are fully ordered ([4]A[6]B2O4).
(Crystalline) Materials Under High Pressure 491

For the intermediate form, ([4]A0.33B0.67) [6]

(A0.67B1.33)O, maximum disorder occurs on
both tetrahedral and octrahedral sites.

5.10.11.1. Fe –Mg ordering in silicates

Recent high-pressure studies of the phases appearing in the system MgO –FeO –
SiO2 including wadsleyite (Finger et al., 1993), anhydrous B (Hazen et al., 1992) and
olivine (Aikawa et al., 1985) have indicated that pressure may induce significant Mg – Fe
ordering. Pressure-induced ordering may play a significant role in cation distributions,
phase equilibria and element fractionations in the mantle. In orthopyroxene, ordering
may affect its elasticity (Bass and Weidner, 1984) and thermochemical properties
(Chatillon-Colinet et al., 1983).
The Fe– Mg order – disorder equilibria in pyroxenes and amphiboles can be used as
a powerful means to determine temperature – time paths of metamorphic and igneous rocks
on the Earth and the Moon. This principle can help the use of thermodynamic calibrations
of heterogeneous phase equilibria between co-existing minerals in xenoliths entrained
in basaltic and kimberlitic magmas and help to clarify the oxidation state of the Earth’s
upper mantle.
An extension of such a study can include the crystallographic controls of Fe3þ and
2þ
Fe in lower-mantle phases and thus can allow modelling of the oxidation state of the
lower mantle. This would help to answer questions pertaining to the Earth’s early
evolution.
The degree of ordering of Mg and Fe between two octahedral sites, M1 and M2
(in olivine/orthopyroxene), is expressed by the distribution coefficient:

KD ¼ ðFeM1 =MgM1 Þ=ðFeM2 =MgM2 Þ:

Orthopyroxenes, having two very different octahedral sites, when allowed to

equilibrate at low temperature, order strongly. Fe orders into the more distorted M2-site,
while Mg prefers the smaller M1-site. KD values as large as 0.50 have been reported
for samples annealed at temperatures below 5008C, whereas samples heated to 1,0008C
and rapidly quenched typically have KD values between 0.2 and 0.3. Above 1,0008C,
most Mg-silicates manifest nearly complete disorder (Virgo and Hafner, 1969) (see
Section “Intra-crystalline Mg –Fe ordering” of Chapter 6).
A completely ordered mineral has a KD ¼ 1; whereas the completely disordered
one has KD ¼ 0: The KD values of some minerals are cited below as examples:

Minerals KD Reference
Olivine 1.8 Finger and Virgo, 1971
Wadsleyite 2.7 Finger et al., 1990
Grünerite 2.3 Finger, 1969

Fe –Mg distribution: three-component lower mantle. In the lower mantle, the atomic
proportions of Fe and Mg in the co-existing phases of perovskite and magnesiowüstite may
be discussed with reference to a simplified formula relationship as described below.
492 Chapter 5

Writing f ¼ Fe/(Fe þ Mg) in atomic ratio, with subscript 1 for perovskite and 2 for
magnesiowütite, the relations are found to be (Stacey and Isaak, 2000):

r1 ¼ 4104 ð1 þ 0:272f1 2 0:012f12 Þ kg m23

r2 ¼ 3209 ð1 þ 0:701f2 2 0:061f22 Þ kg m23

and

r ¼ ð1 2 xÞr1 þ xr2
for a volume fraction x of magnesiowütite. The two molecular weights are m1 ¼
100:389 þ 31:542f1 and m2 ¼ 40:304 þ 31:542f2 ; so that

½ð1 2 xÞr1 f1 =m1 þ xr1 f2 =m2 

f ¼ ð5-91Þ
½ð1 2 xÞr1 =m1 þ xr2 =m2 

If the lower mantle is composed only of (Fe,Mg)SiO3 perovskite and (Fe,Mg)O

magnesiowütite, then the overall ratio is tightly constrained to 0.220 ^ 0.005
(independently of other assumptions). Arguments based on cosmic abundance and the
composition of meteorites, xenoliths and peridotites strongly suggest that f is # 0.11 for the
whole mantle (O’Neill and Palme, 1998). The apparent high value of f ð¼ 0:22Þ for lower-
mantle mineralogy probably suggests the signature of CaSiO3 perovskite in the lower
mantle. Therefore, one must contemplate a three-component model for a more definitive
explanation of the results for the lower mantle.

5.10.11.2. Structural disordering and twinning

In crystals, four types of structural disorders involving atoms are encountered, as
discussed below.
(1) Substitutional disorder. This is the most common type of disorder seen in the mineral
kingdom, covering felspars, ferromagnesian silicates, spinels, carbonates, etc. This
disorder is also seen in non-stoichiometric crystals in which defects occur as missing
atoms (vacancies), e.g., wüstite Fe12xO and oxide superconductor YBa2Cu3O72x
(Hazen, 1990) or as interstitial excess O, as in La2NaO4þx (e.g., Chaillout et al.,
1989).
(2) Positional disorder. Static positional disorder, for example in albite (NaAlSi3O8), Na
atoms, occupy four distinct mean positions in different unit cells, depending on the
local arrangement of Al and Si (Winter et al., 1977). This contributes to the thermal
vibrational disorder.
(3) Rotational disorder. Rotation of CO3 groups along an axis is seen in rhombohedral
carbonates (Ferrario et al., 1994); molecular crystals such as H2 (Mao and Hemley
1994) and C60 (Fischer and Heiney, 1993) show rotational disorder under pressure.
(4) Distortional disorder. Quartz (a-SiO2) can distort from a high-symmetry form in
more than one equivalent way (Kihara, 1990); the perovskite-type structure also show
distortional disorder.
(Crystalline) Materials Under High Pressure 493

Beyond a critical orientational disorder, a system evolves to an amorphous state.

This is consistent with the critical disorder model of amorphization proposed in the context
of ion implantation-induced amorphization (Riviere, 1977).

5.10.11.3. Free energy and order parameter (Q)

The free energy ðGÞ of a phase with a possible iso-symmetric transition has been
expressed by Christy (1994) in terms of an order parameter, Q; representing some
structural distortion as

G ¼ aQ þ bQ2 þ cQ3 þ dQ4 þ · · · ð5-92:1Þ

with an extensive variable such as temperature and or pressure. When some of the
coefficients in the above relation are changed, a phase transition could be introduced
between structures of the same symmetry but with distinct stable values of Q: Christy also
showed that such iso-symmetric transitions are necessarily first order and that when the
free energy varies with P and T; the transition line may terminate at a critical point in P – T
space. Around the extrapolation of the critical point in P – T space, a crossover region
should occur of very rapid change in order parameter. The crossover line is defined as the
locus of a minimum in d2 G=dQ2 (where G ¼ free energy and Q ¼ order parameter).
However, when there are no thermodynamic discontinuities, there should be no
phase transition. However, in experiments, distinguishing a true crossover from a weak
first-order iso-symmetric phase transition often becomes difficult.
In the equation (5-92.1), Q is assumed to remain homogeneous (i.e., the structural
distortion remains constant in space) but when there is a mixing of two structural states
(in unit-cell scale), the free energy is lowered and the order parameter Q becomes
inhomogeneous. In two sublattices, Q may be denoted as Q1 and Q2 ; which, in a simple
expression for G; would be related as

G ¼ ð1=2Þ{aðQ1 þ Q2 Þ þ bðQ21 þ Q22 Þ þ cðQ31 þ Q32 Þ þ dðQ41 þ Q42 Þ}

þ ðl=2ÞðQ1 2 Q2 Þ2 ð5-92:2Þ
The last term involves coupling between the two sublattices Q1 and Q2 ; higher order
coupling terms may be omitted. In a homogeneous case, Q1 ¼ Q2 and equation (5-92.1)
reduces to equation (5-92.2).
Equation (5-92.1) can be recast in terms of two variables, S and D; ½S ¼
ðQ1 þ Q2 Þ=2 and D ¼ ðQ1 2 Q2 Þ=2 as follows:
 
G ¼ aS þ bS2 þ cS3 þ dS4 þ {ðb þ 2lÞD2 þ dD4 } þ 3cD2 S þ 6dD2 S2 ð5-93Þ

Evidently, the first term is equivalent to equation (5-92.2), representing the free energy of
the system if Q is homogeneous with the average value S: The second term is equivalent to
the Landau expansion for the free energy of a zone-boundary transition with order
parameter D; for which the symmetry rules allow only even powers in the order parameter.
The last terms are the symmetry-allowed linear-quadratic and bi-quadratic coupling terms
between the order parameters of the iso-symmetric and zone-boundary transitions.
494 Chapter 5

By equation (5-93), the stable structure predicted is governed by the sign of the
coefficient of D2 : Again, if ðb þ 2lÞ , 0; then D ¼ 0 becomes a local maximum and the
minima are symmetrically placed on either side of D ¼ 0: Therefore, if b becomes more
negative at reduced temperature or pressure, the high-symmetry phase will undergo a
second-order transition to the cell-doubled structure. Depending on l; this equilibrium line
may intersect either the crossover in the high-symmetry phase or a segment of a first-order
iso-symmetric transition line.

Tricritical/first-order transition. Higher order terms in equation (5-93) may be involved

when one of the transitions from the low-symmetry to the high-symmetry phase are either
first-order or tricritical in the Landau sense. For example, in anorthite, the transition is
tricritical at room pressure (Redfern and Salje, 1987) whereas the transition at high
pressure is first order in character (Hackwell and Angel, 1993, 1995). The phase-diagram
topology may be characterized for the zone-boundary transition by an equilibrium line to
be near-isobaric at high pressure and near-isothermal at high temperature.

5.10.11.4. Order parameter (Q) and strain (1) in phase transition

When one tries to employ an extension of the Landau presentation of phase
transitions to conditions of high pressure, the problem arising relates to the excess volume.
When volume change is relatively small, a high-temperature phase transition normally
accompanies a volume strain (Carpenter, 1992) in proportion to the square of the order
parameter Q (i.e., V1 / Q2 ).
In a phase transition such as cristobalite I ! II, both the symmetry breaking
strain, 1sb (E representation, equation (5-30)) and the non-symmetry-breaking strain
(A1 representation, volume) are proportional to Q2 in the lowest order. This relation reflects
the dependence on the square of the macroscopic order parameter Q for the tetragonal to
monoclinic phase transition. Because the strain components change in proportion to each
other (with P and T), the overall spontaneous strain ellipsoid changes in size, although not
in orientation.
From different strain components, a total scalar spontaneous strain can be validly
calculated (because of the constant 1 – Q coupling) as (Salje, 1990):
vffiffiffiffiffiffiffiffiffiffiffiffiffi
u 6
uX
1tot ¼ t 12i ð5-94Þ
i¼1

where 1i are the components of the spontaneous strain tensor derived (see Section 5.4).

5.10.12. Isosymmetric transitions

In recent years, high-pressure, single-crystal diffraction experiments have attained

high precision, which has enabled the detection of a number of apparent phase transitions
(cross/across) at high pressures manifesting no detectable symmetry change. In this cross
through the phase transition, the space-group symmetry remains unchanged since the
atoms within the unit cell occupy the same Wyckoff sites before and after the transition.
(Crystalline) Materials Under High Pressure 495

Such transitions have been termed “iso-symmetric” (e.g., Christy, 1995). This
phenomenon is obviously seen in amorphous systems and gas –liquid transitions. Such
iso-symmetric transitions have also been detected in some complex framework structures
(e.g., orthopyroxene, clinopyroxene, anorthite, etc.) showing large degrees of internal
structural freedom (Angel, 1996).
It is known that the free-energy changes associated with simple compression of a
single phase are far greater than those associated with cooling to low temperature.
Therefore, modest pressures can result in greater modifications to crystal structures and
properties than do temperatures and new transitions may occur (Angel, 1996).
Nevertheless, the displacive-phase transition characteristics in minerals at high
pressures often correspond fairly to high-temperature transitions. For example, the ferroic
transition at high temperature in ilvaite, CaFe3O8(OH), shows an order-parameter
behaviour which is similar to that obtained at high pressure (Finger and Hazen, 1987). In
co-elastic crystals (Salje, 1990), high pressures lead to an increase in elastic stiffness tensor
and a coupling between strain and the order parameter of the transition.
Potasium titanyl phosphate (KTP) undergoes strongly first-order (3% volume
change) iso-symmetric phase transition at , 5.7 GPa, as does sodium-doped KTP (Allen
and Nelmes, 1996). Such transitions are driven by significant changes in electronic
structure, as exemplified by the orthorhombic ! orthorhombic transition in (La,Ba)CuO4
(Paul et al., 1987) and in complex framework structures. A large number of internal
degrees of structural freedom are suppressed under pressure and some dynamic motion of
the larger cations leads to iso-symmetric transitions (e.g., in anorthite).
5.10.12.1. Energetics of iso-symmetric transition
The energetics of phase transition between the high-symmetry structures in terms of
an order-parameter approach show that the stability of the cell-doubled low symmetry of a
phase arises simply from the development of inhomogeneity in the order parameter.
Although the structural differences between the high-symmetry and low-symmetry phases
and between the two high-symmetry structures are quite minor, the thermodynamic
consequences of these changes can be quite significant. This is most clearly seen in the
intersection of the crossover line with the low-to-high-symmetry transition boundary in
anorthite (Hackwell and Angel, 1995). The high-pressure P1 , I 1 transition (in anorthite)
is marked by the disappearance of the superlattice reflections (first-order transition).
On the phase diagram, the phase boundary is almost isobaric.

5.10.13. Growth rates

The growth rate during an interface-controlled reconstructive polymorphic phase

transition, involving diffusion across the inter-phase boundary, can be described by
(Carbon and Rosenfeld, 1981)

x0 ¼ K0 T exp½2ðH p þ PV p Þ=RT½1 2 expðDGr =RTÞ ð5-95Þ

where K0 is a constant, T is absolute temperature, H p is the activation enthalpy, V p is

the activation volume for growth, DGr is the Gibb’s free energy change for reaction
496 Chapter 5

(must be negative) and R is the gas constant (Christian, 1975). The first exponential is a
kinetic factor describing the thermally activated diffusion of atoms across the inter-phase
boundary. The rate of this process increases rapidly with temperature but decreases with
increasing pressure, assuming that DV is positive. The second factor in brackets depends
on the thermodynamic driving potential in the system.
DGr can be approximated wherein DP is the overstep of pressure beyond
equilibrium and DV is the transformation-volume change at the conditions of reaction.
Hence, this factor is zero at equilibrium (where DGr ¼ 0) and, therefore, the growth rate is
zero. This factor approaches unity as DP increases.
For example, with increasing pressure at constant temperature, the growth rate of
spinel during transformation from olivine first increases (due to thermodynamic factors)
and then decreases (due to kinetic factors).

5.11. Charge distribution in ionic solids: valence and core states

In an ideal ionic solid, the valence charge is completely localized around an anion.
Deviations from complete localization, manifested by reduced ionicities or by definite
valence band-widths, indicate a measure of covalency in the bands. The distribution of the
theoretical valence charge is expected to provide a more direct qualitative picture of the
valency deduced from the band structure.
This can be illustrated with the example of MgO, as discussed by Bukowinski
(1980).
The valence and core charge densities within the Mg and O spheres, computed at
V0 =V ¼ 1:0 (Figs. 5.17 and 5.18) show no more than 5% of the valence charge is located in
the Mg sphere; its shape is suggestive of overlap tails from the oxygen ions. An analysis
into spherical harmonics shows that the valence charge is primarily of p-like character on

Figure 5.17. Core and valence charge density of MgO in the Mg sphere at V0 =V ¼ 1:0; a0 is the Bohr radius
(Bukowinski, 1980, q 1980 American Geophysical Union).
(Crystalline) Materials Under High Pressure 497

Figure 5.18. Core and valence charge density of MgO in the O sphere at V0 =V ¼ 1:0 (Bukowinski, 1980,
q 1980 American Geophysical Union).

both ions, with small amounts of other angular momenta induced by the crystal field. The
small amount of valence charge in the Mg sphere may be understood as a consequence of
the effective repulsion that arises from the orthogonalization of the valence states to the Mg
2p core states. Since the lowest empty states are at the bottom of the conduction band, the
exchange repulsion of the valence charge is very efficient. Further evidence of this is found
in the small effect that compression has on the amount of valence-charge overlap with the
Mg2þ ion.
The core states of Mg and O have small but finite amplitudes at the sphere radii,
indicating a certain amount of core – core overlap (Figs. 5.17 and 5.18). This is, of course,
one of the sources of the repulsive potential between the Mg and O ions and is the reason
why the O 2s and Mg 2p orbitals had to be treated as band states. In spite of this, the core
density in the Mg sphere is practically indistinguishable from that of the free Mg2þ ion.
Thus, in a first approximation, MgO could have valence electrons with the Mg2þ ions. The
true picture is somewhat more complicated because of the presence of angular momentum
components other than l ¼ 1 and because more than one electronic charge is distributed
outside the spheres. Thus, although the charge on the Mg ion is close to the nominal þ 2,
the remaining charge is not entirely contained within the O sphere. Changing the radii has
some effect on the charge distribution but this cannot completely eliminate the charge
outside the spheres. With the model of equal touching spheres, the O and Mg core charge
densities are approximately equal at the point of contact. Changing the sphere radii would
transfer an unreasonable amount of core charge into the constant potential region. A model
which best describes the calculated charge distribution would consist of Mg2þ ions that
overlap with O cores that are of similar spatial extent. In addition, six valence electrons are
distributed throughout the unit cell in such a way that only about 80% of the valence charge
may be identified with the O site. This interpretation is in essential agreement with the
experimental charge density (Adams, 1978).
However, a more accurate treatment of many-body effects is not likely to change
the qualitative model presented above. Good agreement of the result of Bukowinski (1980)
498 Chapter 5

with the charge density obtained with the empirical pseudopotential method and the
measured charge density further supports this conclusion.

5.11.1. Ionic solid under compression: MgO

Compression should increase the valence charge density within the oxygen sphere,
while the total charge within the sphere would decrease. However, the apparent
incompressibility of the valence charge around the oxygen core may be a direct
consequence of the high-potential energy that results from accumulating so much excess
charge.
In the case of MgO, a high potential in the neighbourhood of the O core and the
strong repulsion of the Mg ion (due to the large band gap) explains the low polarizability of
MgO. The valence charge has virtually “no place to go”. Added to this is the fact that the
cores of O overlap with the Mg ion. These facts together account well for the large bulk
modulus observed in MgO. In MgO, there is a small concentration of valence charge
around r ¼ 0:5 Bohr radii from the Mg site (Bukownsiki, 1980). The valence charge is also
distributed throughout the whole unit cell and, in the neighbourhood of lattice sites, it
resembles the corresponding atomic-valence states. In MO language, the valence states
may be said to be composed of Mg 3s states and O 2p states (see Section 5.11).
Indeed, the computed electronic band structure and charge density coupled with
experimental data suggest that the MgO valence electrons are distributed throughout the
unit cell. This charge is possibly localized mostly with O ion, while an electronic charge is
distributed between the two atomic spheres of Mg and O. However, a small part of this
charge is localized near to the Mg site, where it mimics the Mg 3s state.

5.11.1.1. Band-gap change: implication in lower mantle

The band gap increases with pressure in the high-pressure regime of the Earth’s
mantle. Conduction band gaps play important roles in the thermal and electrical
conductivity of the mantle. As compression increases, MgO attains a higher transparency
to thermal radiation. At a temperature of a few thousand degrees, the thermal radiation
shows the peak energy to be an order of magnitude lower than the MgO band gap. Thus, at
the lower mantle, MgO should be a good conductor of radiative heat and a strong insulator
to electricity.
Such a change in properties with band-gap change is of tremendous significance
since the lower mantle is presumed to host a mixture of simple oxides, which would behave
the way of MgO. However, in many such oxides, the systematics and models developed
from zero-pressure data are likely to become unreliable when extrapolated to high
pressure. In particular, cation substitution and compression can result in a substantial
change in electron distribution around the neighbouring anion core. For example, the
presence of Fe2þ cation impurity will form intrinsic d-levels and compression to lower-
mantle densities will increase the overlap of the Fe2þ d-elections. Thus, as pressure
increases, the Mg-silicates tend to become increasingly transparent to thermal radiation.
(Crystalline) Materials Under High Pressure 499

5.11.2. High-spin – low-spin transition

For transition-metal ions of d4 to d7 systems, high-spin and low-spin states are

possible. The high-spin state is usually stable in oxides and silicates at normal pressures,
except for the Co3þ ion. Since the ionic radius of the low-spin state is smaller than that of
the high-spin state, an increased pressure will enhance the low-spin state by spin-pairing.
The condition for HS ! LS transition under pressure is that DGðPÞ ¼ 0: Usually, a
contraction of the metal –ligand distance produces a large increase in the crystal-field
splitting.
In a crystalline field, the degeneracy of Fe2þ 3d orbitals is lost and in an octahedral
field they split into two sets of orbitals called t2g and eg : The difference between these two
is the crystal-field splitting (10 Dq, commonly designated as D), which increases with
decreasing Fe –O bond length, R: The ionic bonding model predicts the relation:
10 Dq / 1/R 5.
The six d electrons (each with spin quantum number S ¼ 1=2) of the Fe2þ cation
can couple to give states with spin S ¼ 2; 1 and 0. Because of exchange energy, both the t2g
and eg orbitals are split into spin-up ( " ) and spin-down ( # ) sub-orbitals. The difference
between these two sub-orbitals is expressed as Uex : Usually, Uex (for the t2g orbitals) ¼ Uex
(for the eg orbitals) but this is not always so. As can be seen in Fig. 5.19, the high-spin state
ðS ¼ 2Þ will be the most stable as long as Uex . 10 Dq. A decrease in Fe– O bond length
will increase 10 Dq and eventually Uex will be smaller than 10 Dq. Thus, the low-spin
ðS ¼ 0Þ state will be most stable. From a different point of view, as the Fe– O bond length
gets smaller, Uex will decrease because of greater Fe– O covalency.

5.11.2.1. Energy change in spin transition

The difference of the total energy in the crystal between the high- and low-spin
states is given by the change of the crystal-field stabilization and the spin-pairing energy
in the transition-metal ion (see Fig. 5.19). The total energy ðWÞ of the low-spin state is
given by
2WLS ðVÞ ¼ WHS ðVÞ þ SN½PðVÞ 2 DðVÞ
where S is the number of spin pairings (in octahedral coordination for d4 and d7 systems, it
is 1 and for d5 and d6 systems, it is 2), N is the number of transition-metal ions in a crystal
and P is the spin-pairing energy.
The high-spin state ð6 A1g Þ at normal pressure has the crystal-field splitting, D
smaller than the spin-pairing energy, P: When D exceeds P at high pressures, the low-spin
state ð2 T2g Þ becomes more stable than that of high-spin Fe3þ (r ¼ 0:645 Å). The volume
change associated with the spin transition can be evaluated as 13% from the plot of the
cell volume against the ionic radii systematics for the corundum structure sesquioxide (Fig.
3.10). This observed relationship is generally valid where both D and P depend on the
volume of the crystal.
In accordance with Griffith (1956), P can be written with Racah parameters for d4,
d , d and d7 systems as 6B þ 5C; ð15=2ÞB þ 5C; ð5=2ÞB þ 4C; respectively. The spin-
5 6

pairing energy is proportional to the Racah parameter B; since the ratio B=C is almost
500 Chapter 5

Figure 5.19. Schematic of occupancies of states (a) high-spin and (b) low-spin ferrous iron (Courtesy:
R.E. Cohen).

constant (Tanabe and Sugano, 1954). The Racah parameter B describes the effects of
repulsion between electrons of a given ion. An increase in the degree of covalency between
a metal ion and its ligands by increased pressure would be accompanied by the spreading
out of the electron charge cloud and, therefore, reduction of the repulsion and the value of
B: For the volume dependence of D and P; the simple power-law formulae hold as
D ¼ D0 ðr=r0 Þ2m
P ¼ P0 ðr=r0 Þn ;
where m and n characterize the volume dependence of the crystal-field splitting and spin-
pairing energies, respectively. In the case of Cr3þ in Al2O3, the value of n is estimated from
the pressure dependence of B as about 1.0 (Goto et al., 1979). For Fe3þ in Fe2O3, the values
of D0 and P0 are obtained as 21.86 and 4.07 £ 10212 erg, respectively (Lehmann, 1970).
In the case of Fe2þ in MgO, the value of m is estimated to be 3.0 from optical data at high
pressure (Sankland, 1968).
The EOS of low-spin form is derived by using the derivation of 2WLS ðVÞ with
respect to V as

2dWLS ðvÞ dDv dPðvÞ
PLS ðvÞ ¼ þ SN 2 ð5-96Þ
dV dV dV
According to Ohnishi’s (1978) theory, the Birch –Murnaghan EOS adopted for high-spin
state is
PHS ðVÞ ¼ ð3=2Þ½ðV=V0 Þ27=3 2 ðV=V0 Þ25=3 K0 x{1 þ ð3=4ÞðK 00 2 4Þ½ðV=V0 Þ22=3 2 1}
Internal energy change. The free energy of HS ! LS transition at constant T and P is
DG ¼ DU 2 TDS þ PDV
DU: The change in internal energy DU for HS ! LS transition is simply the energy
difference between 5 T2g and 1 A1g states of Fe2þ cations (5 T2g is the spectroscopic state
(Crystalline) Materials Under High Pressure 501

resulting from the high-spin ðt2g Þ4 ðeg Þ2 configuration and 1 A1g is the spectroscopic state
arising from the ðt1g Þ6 configuration). From spectroscopic measurements or electronic
structure calculations, the energy difference can be estimated.
The 5 T2g !1 A1g electronic transition is spin-forbidden so no absorption band is
expected in Fe(II) oxides and silicates. But the energy difference ðDEÞ between the 5 T2g
and 1 A1g states of Fe2þ, as calculated from ligand field theory, is

DE : ð5 T2g 21A1g Þ ¼ 5B þ 8C 2 20 Dq:

where B and C are the Racah parameters describing the interelectronic exchange and
repulsion energy. However, the B and C Racah parameters for Fe2þ in oxides and silicates
are not well known. The free ion value of B is 1,058 cm21 (Lever, 1968) but for solids it
should be less by a factor b; the nephelauxetic ratio. That is

BðsolidÞ ¼ bB ðfree ionÞ:

For divalent transition-metal ions, b for oxides and silicates is about 0.9. In a free
ion C ¼ 3:7B and B ¼ 1,058 cm21. Taking all values together, 5B þ 8C becomes
32,950 cm21.
The optical absorption spectral band of Fe2þ in regular octahedral coordination
in oxides and silicates gives the value of 10 Dq. However, due to the dynamic JT effect,
the line may split. For example, in (Mg, Fe)O, the 5 T2g !5 Eg transition is split into bands
at 11,600 and 10,000 cm21. Therefore, 10 Dq is either 11,600 or 10,000 cm21. When the
dynamic JT effect is small, a single band occurs at 11,000 cm21(Mao and Bell, 1972).
For Fe2SiO4 spinel, 10 Dq equals 11,000 cm21 at R(Fe –O) ¼ 2.16 Å. Therefore,
the HS –LS transition energy, DU; is calculated as 10,950 cm21 at R (Fe – O) ¼ 2.16 Å.
How the DU energy will change with R (Fe – O) is estimated by assuming 10Dq0 /10Dq ¼
(R/R0 )5. The resulting values for DU as a function of R (Fe –O) are given in Table 5.10
below.

TABLE 5.10
Internal energy estimated from HS ! LS transition in octahedral Fe2þ

R(Fe–O) (Å)
2.16 2.05 1.95 1.85
21 a
Estimates from ligand-field theory (cm )
10 Dq 11,000 14,300 18,300 23,900
DU 10,950 4,346 23,654 214,850
Estimates from SCF–Xa –SW MO calculation (cm21)a
10 Dq 11,130 14,826 18,040 23,520
DU 10,760 3,243 25,241 216,210

Note: 10,000 cm21 ¼ 119.6 kJ/mol, and 1 eV ¼ 8,066 cm21. DU can also be estimated from first-principles
electronic structure calculations.
a
Calculated using b ¼ 0:9; B ¼ 1,058 cm21, C ¼ 3:7B:
502 Chapter 5

The electronic structure of an (FeO6)102 cluster as a function of Fe – O bond length

can be calculated using the self-consistent field Xa-scattered wave (SCF-Xa-SW) method.
The theory behind the SCF-Xa-SW method is given by Johnson (1973) and Slater (1974).
The molecular orbital diagram for an (FeO6)102 cluster with an Fe –O bond length
of 2.16 Å is shown in (Fig. 5.20). The calculations were carried out using a spin-
unrestricted formalism that takes into account the different exchange potentials for spin-up
ðaÞ and spin-down ðbÞ electrons. Using the “transition-state” formalism (Slater, 1974), the
energy of the HS ! LS transition was calculated for (FeO6)102 clusters with decreasing
b b
Fe –O bond lengths. The 5 T2g !1A1g energy is given by 2½1ðt2g Þ 2 1ðeag Þ; where 1ðt2g Þ and
b a 3 b
1ðeg Þ are the t2g and eg one-electron orbital energies in the configuration ðt2g Þ ðt2g Þðeag Þ1 :
a a

The results are in good agreement with the rough estimates obtained by using
ligand-field theory. Fitting the SCF – Xa –SW calculated energies to the ligand-field

Figure 5.20. Self-consistent field X 1 – SW molecular orbital diagram for an (FeO6 )102 cluster at
R(Fe–O) ¼ 2.16 Å. Orbitals indicated by dashed lines are unoccupied.
(Crystalline) Materials Under High Pressure 503

relation gives

DU ¼ 5B þ 8C 2 20 Dq0 ðR0 =RÞ5

given 5B þ 8C ¼ 33,370 cm21 and 10 Dq0 ¼ 11,480 cm21 with R0 ¼ 2:16 Å. The SCF –
Xa – SW calculated value for 5B þ 8C decreases by about 10% when the Fe– O bond
length is decreased from 2.16 to 1.85 Å. Also, the value for 10 Dq closely follows the
10
Dq0 =10 Dq ¼ ðR=R0 Þ5 dependence.

5.11.2.2. Spin-pairing in the lower mantle

In the lower mantle, low-spin Fe2þ ions are likely to exist, as has been
experimentally determined on a host of oxide phases by pressures overlapping those
present in the lower mantle. Molecular orbital calculations (Tossel, 1976) also indicated
that high-spin to low-spin transition could take place in Fe2þ in FeO in this mantle region
(Sherman, 1988). MO calculations also showed that, at depths greater than 1,700 km, Fe2þ
in magnesiowüstite would largely exist in low-spin state (Sherman, 1991). Low-spin Fe2þ
ions may also exist in the perovskite structures in the lower mantle (Williams et al., 1989).
However, above the spin-pairing transition point, low-spin Fe2þ may have a smaller
ionic radius than Mg2þ; this possibly would lead to a reversal of melting-point
relationships. This change to low-spin configuration of Fe2þ would affect the magnetic
properties of the lower mantle.
In the lower mantle, the transition of (Mg, Fe)O from B1(NaCl) to B2(CsCl) takes
place along with the change in electronic structure of Fe2þ in silicate perovskite. In both
the phases, Fe2þ occupy eight-coordinated sites. The change in internal energy, DU; for
the HS ! LS transition of 8-fold coordinated Fe2þ(the 5 Eg !3 T1g ), is estimated from
ligand-field theory as

D ¼ 210 Dq þ 6B þ 5

calculating the difference between 5 Eg !3 T1g states. The calculated HS ! LS transition

energy for 8-fold coordinated Fe2þ as a function of the Fe – O bond length is given in
Table 5.11.
The table indicates that any iron that is partitioned into the silicate perovskite
phase will always be in the high-spin state, assuming that Fe2þ cations occupy only the 8 to
12-fold coordination sites. The calculations presented earlier suggest that the majority of
Fe2þ cations in (Fe,Mg)O will not be in the low-spin state until a depth greater than
1,700 km is reached (see Fig. 5.21).
Magnesiowüstite (Mg, Fe)O and silicate perovskite (Mg, Fe)SiO3 phases dominate
in the lower mantle, where the ambient pressure is sufficient to effect a high-spin ðS ¼ 2Þ to
low-spin ðS ¼ 0Þ transition.

Band broadening. Under pressure, the collapse of the high-spin magnetic state is caused
by the band broadening due to shorter nearest-neighbour distances, not by an increase in
the crystal-field splitting. The change in bond character from ionic to metallic would affect
the mineral stability. As the charge moves out of the bond direction, the shapes of the
504 Chapter 5

TABLE 5.11
HS ! LS energy (in cm21) for 8-fold coordinated Fe2þ

R(Fe–O) (Å)
2.30 2.16 1.95 1.85
21
10 Dq cm 7.143 10.476 11.528 14.731
DU (cm21) 16,186 12,853 11,801 8,598

All calculated by using B ¼ 1,058 cm21, C ¼ 3:7B and b ¼ 0:9:

transition-metal ions and oxygen ions would change. This will affect the phase diagrams
and elasticity.
Under high pressure transition-metal elements lose their properties and behave as
different elements and the chemical behaviour drastically changes when the valence and
other bonding electronic behaviour lose their significance in bonding.

5.11.3. Pressure dissolution and substitution

Dissolution of pyroxene in garnet has very little effect on thermal expansion but
substitution of iron and magnesium in the pyrope – almandine join seems to have a large
effect.
When d0 is calculated, using the ðdK=dTÞP values as reported by Soga (1967), its
values become 5.3 for pyrope and 6.4 for almandine. For the pyroxene — garnet solid

Figure 5.21. Calculated PT-curve for the high-spin to low-spin transition of FeO. Also shown is an approx.
mantle geotherm (assumed adiabat). The HS ! LS PT-curve crosses the mantle geotherm near 1700 km depths.
(Crystalline) Materials Under High Pressure 505

solution — it is observed (Yagi et al., 1987) that, with increasing pyroxene component, the
bulk modulus seems to decrease and also that the thermal expansion of majorite is
expected to be similar to that of garnet with the same Fe/Mg ratio.

5.12. Amorphization

Materials lacking translational or orientational long-range order are called

amorphous phases. Although configurational disorder leading to excess entropy should
render amorphous phases greater stability than ordered ones, this is not observed to be
so. This is because, at temperature much below the melting temperature, the increase
in entropy causes lowering of Gibbs free energy. This energy is, however, lower than
the increase in the internal energy necessary for destroying long-range order by
distortion.
Thus, at high temperature, the entropy term predominates up to the melting
temperature and internal energy becomes less significant in determining the structural state
of the material. Crystalline material with artificially created defects would transform to
amorphous phase, thereby lowering its free energy. A common example is the
amorphization or glass formation by radiation. In such a situation, the melting curve
shows a downward trend with pressure. This pressure-induced amorphization (PIA) is
termed “pseudo-melting”.
Meteorite craters offer impact melted rock glass, called “diaplectic glasses”. These
contain dense pressure phases as stishovite or hollandite, formed at the high-pressure
regimes in the Hugoniot curve.
High-pressure studies are critical for identifying new equilibria and metastable
states that can be accessed as amorphous materials and compressed to smaller volumes.
Recent studies of inorganic liquids and glasses under compression documented changes in
nearest-neighbour geometry and vibrational spectra (Hemley et al., 1986; Williams and
Jeanloz, 1988; Durben and Wolf, 1990). Just as crystalline compounds are modified by the
high-pressure phase transitions, so the amorphous materials undergo structural transitions
under pressure. In recent years, the intermediate-range order observed in amorphous
materials has drawn considerable attention (Gaskell et al., 1991).
Studies on PIA have demonstrated that, at higher densities, potential energy dictates
structures, not the entropy. At higher pressure, internal energy dominates, not entropy, and
entropy does not favour the state. Unlike amorphization, through thermal quenching the
kinetic processes are inhibited in a PIA.

5.12.1. Pressure-induced amorphization

A large number of materials which exhibit amorphization when subjected to

static or dynamic high pressures have been investigated. The pressure-amorphized
materials are now termed as “glass without fusion”. In shock compression, the fusion
and glassy structure may be a consequence of the high temperature and strain rates
associated with it. However, an examination of the short- and medium-range order in
506 Chapter 5

some of the pressure-amorphized materials suggests that these could be structurally

different from the glasses obtained from quenching the melts. In this context, Raman
spectroscopy has been extremely useful in providing insight into the evolution of
disorders in the preceding crystalline phase. The residual short-range order in the
amorphous state has also been deeply probed. It is found that the distribution of bond
lengths and bond angles in the pressure-amorphized-state is narrower than those known
in melt/quenched glasses.
The crystalline to amorphous transformation (c ! a) opens up the question of
clarifying (a) the relationship between these pressure-induced amorphous phases and
the conventional glasses quenched from high-temperature melts and (b) the unusual
mechanical processes that have been documented during metastable transitions in hydrous
silicates. The latter may play an important role in generating deep-focus earthquakes in
subduction zones (Meade and Jeanloz, 1991).
The pressure-amorphized state is metastable and is believed to be resulting from the
kinetic hindrance of equilibrium-phase transitions. Slow kinetics associated with the
molecular reorientation and translation across a phase transition leaves the system trapped
in the metastable amorphous state. Directional bonds and non-hydrostaticity are some of
the important factors that determine the amorphization pressure. The residual order in the
pressure-amorphized materials is quantitatively different from quenched glasses. The
existence of inhomogenous disorder in potash alum supports the applicability of a critical
disorder model of amorphization. PIA was first observed in ice (Mishima et al., 1984).
Sharma and Sikka (1996), however, suggest that this phenomenon may have been known
earlier. Conversely, amorphization of crystalline phases stable at high pressure may occur
on decompression. Liu and Ringwood (1975) showed that the cubic CaSiO3 perovskite,
which is stable at high pressure (,16 GPa), transforms to an amorphous phase during
decompression.
Since the first report on ice (Mishima et al., 1984), a number of compounds have
been found to exhibit amorphization at high pressure with widely different bonding
natures, such as covalent SiO2 (Hemley et al., 1988) and AIPO4 (Kruger and Jeanloz;
1990), ionic LiKSO4 (Arora and Sakuntala, 1992) and Ca(OH)2 (Kruger et al., 1989),
van der Waals SnI4 (Sugai; 1985). PIA features have been observed in quartz-type forms of
GeO2 and AlPO4, and also in framework silicates, pyroxenes, olivines and hydrous
silicates.
Amorphization is brought about by different mechanism in different materials. For
example, in quartz (SiO2), the breaking of Si – O bonds due to bending of the Si –O –Si
angles beyond their energetic limit has been proposed as the cause for PIA (Hazen et al.,
1989). Dimerization of the tetrahedral molecules is believed to be the cause of PIA in SnI4,
whereas orientational disorder of sulphate ions is identified as reponsible for
amorphization in a number of binary sulphates.
PIA has been reviewed by Richet and Gillet (1997) with reference to differential
stress, crystalline transformations, compression mechanisms and shearing processes. They
also discussed the thermodynamics of amorphization and mechanistic interpretations, with
special reference to elastic and dynamic instabilities and shearing processes.
(Crystalline) Materials Under High Pressure 507

5.12.1.1. Metastability and reversible amorphization

A phase, while transforming from one crystalline state (c) to another crystalline
state (c0 ) may get trapped in an intermediate metastable amorphous state (a) due to slow
kinetics of c– c0 transition. Further pressurization may accelerate the a –c0 transition (see
Fig. 5.22). As an example, high-density amorphous ice is seen to transform to crystalline
ice VII when further pressurized to 4 GPa (Hemley et al., 1989).
Amorphization may be reversible or irreversible. Reversible transition with
significant hysteresis is seen in cases of compounds having a dissimilar type of
bonding among different groups of atoms such as FeSiO4, AlPO4, Ca(OH)2, LiKSO4,
SnI4, etc. Irreversible amorphization is seen in the cases of ice, quartz and
Ca2Al2Si2O8 (Williams and Jeanloz, 1989). Examples of monatomic phases with
directional bonds showing PIA are graphite (Goncharov et al., 1992) and sulphur (Luo
and Rouff, 1993).

5.12.1.2. Non-hydrostatic pressure and amorphization

The presence of non-hydrostatic pressure also appears to have a role in driving the
PIA. Freezing of the most commonly employed pressure-transmitting media such as
methanol : ethanol (1 : 4) mixture at 11 GPa in a gasketted diamond-anvil cell leads to the
development of a non-hydrostatic component of pressure in the cell. Hence, in some
systems exhibiting PIA above 11 GPa, a non-hydrostatic pressure may be responsible.
Under non-hydrostatic conditions, the Raman lines broaden much more rapidly and the
amorphization pressure is lowered. The non-hydrostatic condition leads to shear stress and
asymmetrical distortion of the molecular units. These accelerate the growth of disorder and
thus effectively lower the pressure of amorphization. Lowering of amorphization pressure
from 23 to 14 GPa is reported in graphite when no pressure-transmitting medium is used
(Goncharov, 1992). A similar observation has been made with sulphur (Luo and Ruoff,
1993).

Figure 5.22. A schematic three level diagram for c ! a, and a ! c0 transformations.

508 Chapter 5

5.12.2. Disordering and amorphization: Raman scattering

In the process of amorphization, the bond length and bond angle show a distribution
other than being unique. As the amorphous phase lacks periodicity, the disappearance of
lattice or the external vibrational modes in the Raman spectra is used to identify the PIA
transition (Deb et al., 1993). The basic signature of amorphization by loss of long-range
order in Raman spectrum lies in the vanishing of the external vibrational modes through
the breakdown of the q < 0 selection rule. This results in manifesting the DOS through the
second-order Raman scattering. This provides complementary information on the disorder
in relatively more rigid units in the structure. However, it should be noted that vanishing or
broadening of these modes may not necessarily correlate with the loss of long-range order.
The disorder is manifested in the broadening of the internal modes of the strongly
bound polyatomic groups or molecular ions. An increase by a factor of , 5 in the width of
Raman lines (Klug et al., 1986) or of infrared absorption (Kruger et al., 1989), associated
with the internal modes, is observed across PIA.
Investigation of a number of binary sulphates shows that the high-pressure
crystalline phases have disorder which is in the form of distinct orientations of sulphate
ions. This results in the splitting of the non-degenerate symmetric stretching mode of the
sulphate ions as seen in the Raman spectra (Arora and Sakuntala, 1992). Growth of these
disorders eventually leads to amorphization. Raman spectra arising from the polyatomic
units such as sulphate ions correspond to that of the bond lengths and bond angles.
It should, in principle, be possible to obtain information about such short-range order
from the analysis of the spectra.

5.12.2.1. Non-bonded atoms and steric hindrances

PIA can also be brought about by the structural frustration caused by kinetic
impedance and steric hindrances (Sharma and Sikka, 1996). The steric constraints arise
due to the reduction of non-bonded inter-atomic distances under pressure, when a
significant modification of the molecular shapes is kinetically inaccessible. A general
correlation is noted between the pressures of phase transformations and the limiting
distances of non-bonded atoms. This happens when the repulsive energy cost for further
squeezing those non-bonded atoms far exceeds the energy cost of the distortion
of polyhedra. This leads to a phase transition to relieve the steric strain in the structure
(Sikka et al., 1994).
For AlPO4, the relevant non-bonded distances are for the non-bonded O· · ·O atoms.
For these, the largest steric limit is Pauling’s van der Waals separation of 2.8 Å, and the
smallest extreme limiting value is 2.6 Å. But a phase transition is not generally initiated at
the van der Waals limiting distance. The O· · ·O distances decrease with pressure and,
between , 10 and 15 GPa, these reach a plateau value of , 3 Å (corresponding to the range
where c/a ratio shows a plateau). At 29 GPa, there occurs a few O· · ·O contacts which are
, 2.58 Å, i.e., approaching the limit of 2.6 Å, at which distance the nucleation of the
disordered phase occurs.
(Crystalline) Materials Under High Pressure 509

5.12.2.2. Memory glass: AlPO4

The most common mineral known to manifest PIA is quartz, which, when subjected
to a pressure of ,15 –30 GPa at ambient temperature, undergoes slow amorphization.
Kingma et al. (1993), however, have observed a new phase preceeding amorphization. An
isostructural mineral, berlinite (AlPO4), similarly shows PIA at , 20 GPa. This mineral,
however, behaves as a memory glass’ (e.g., Kruger and Jeanloz, 1990; Chaplot and Sikka,
1993; Polian et al., 1993).
A memory glass would recrystallize to the original crystallographic state upon
release of pressure (i.e., quenching). Since the pioneering work of Kruger and Jeanloz
(1990), this property has been noted in several materials iso-structural with quartz.
To investigate the mechanism of memory effect in berlinite, Tse and Klug (1992)
and Chaplot and Sikka (1993) used inter-atomic potentials to perform molecular dynamics.
They showed that the O – P– O bond-angle distribution remains close to the original in
quartz structure and the PO4 tetrahedra remain four-coordinated even when severely
distorted. All these perform the observed memory behaviour. Some, however, noted it as
polymorphic crystal – crystal phase transition (Gillet et al., 1995).
The crystallographic characters and changes under pressure of berlinite were
earlier discussed under Section “Side-band fluorescence ultrasonic technique” of
Chapter 4.

5.12.3. Solid – liquid (melt) stability boundary

The thermodynamic solid – liquid (melt) stability boundary is determined by the

equality of free energies of the two phases. The mechanical stability of a crystal can be
deciphered from the calculation of elastic constants at several pressures and temperatures.
At a given temperature, the maximum pressure of stability is determined via the Born
stability criteria. Mechanical instability occurs when a combination of elastic constants
violates one of the Born stability conditions (see Born and Huang, 1956).
In a semi-empirical approach, the quasi-harmonic lattice dynamics may be
combined with the Lindemann criterion for melting to compute a thermodynamic melting
line. The mechanical instability line due to the violation of the Born stability condition
C11 2 lC12 l . 0 (see Born and Huang, 1956) and the theoretical thermodynamic curves
are compared with experiment. The temperature where mechanical instability occurs is
mostly higher than the thermodynamic melting point.
In a molecular dynamics study a mechanical instability due to the softening of the
elastic modulus C66 ð¼ C11 2 C12 Þ of ice structure under high pressure has been proposed
by Tse (1992).

5.12.3.1. Law of melting: Lindemann

The law of melting in its differential form is

d lnTm =d lnr ¼ 2ðg 2 1=3Þ ð5-97aÞ

where Tm is the melting temperature, r is the density and g is the Grüneissen parameter.
510 Chapter 5

The relation relies on the postulation that, on melting, the long-range order of the solid
state breaks down.
Lindemann theory offers the slope of the temperature of melting ðTm Þ with pressure
as

dTm 2Tm 1
¼ g2 ð5-97bÞ
dP KT 3

where g is the Grüneisen parameter at a particular V and KT is the isothermal bulk modulus
at the T and V of the triple point. Again, g ¼ aC KT V=CV ; where aV is the volume
coefficient of thermal expansion and CV is the specifics heat at constant volume.
Equation (5-97a), known as the Lindemann theory of melting, can be related to g as

d ln Tm
g¼ ð5-98Þ
ln r

For melting relationships, equation (5-97a) or (5-98) seems to hold good in many cases.
Tm ðPÞ can be determined by first calculating PðVÞ and Tm ðVÞ: The thermal EOS
appropriate to temperatures above Tm0 (where Tm0 is the temperature of melting at ambient
pressure) is

PðV; TÞ ¼ PðV; Tm0 Þ þ Pth ð5-99Þ

where the thermal pressure Pth is evaluated for T . Tm0 along the melting curve.
The variation of g with volume along the liquids is assumed to be
q
gðrÞ V
¼ ð5-100Þ
g ð r0 Þ V0

The values of g0 (at P ¼ 0) of iron phases are presented by Anderson and Isaak (2000).
The melting temperature equation is obtained by substituting equation (5-98) in
equation (5-97b) and integrating (Anderson, 1995, p. 286) so that
2=3   q 
Tm V 2g0 V
¼ exp 12 0 ð5-101Þ
Tm0 V0 q V

where Tm0 is the melting temperature at the beginning volume, V0 at P ¼ 0:

In some of the melting experiments, particularly those involving shock loading, the
sample may not remain in the initial phase or in the single phase before melting. Often,
decomposition or transition to another phase may precede the melting depending on the
experimental conditions. This would also influence the observed melting temperature.
Lower melting temperatures observed in an experiment may be largely accounted for by
the level of defects. The true level of defects in a real sample in experiments could be much
higher, depending on the sample history.
Mechanical melting is a consequence of the softening of elastic moduli.
(Crystalline) Materials Under High Pressure 511

Activation volume and melting temperature. The activation enthalpy H p as a function of

the melting temperature Tm (at 1 bar) can be expressed as
H p ¼ aT m
where a is a constant. This relationship seems approximately valid for thermally activated
processes such as diffusion for similar structure but different composition (Frost and
Ashby, 1982). Therefore, by incorporating the melting temperature Tm ; the relation
becomes
x0 ¼ K0 T exp½2ðaTm þ PV p Þ=RT £ ½1 2 expðDGr =RTÞ ð5-102Þ
p
The activation volume V can also be related empirically to Tm (Poirier, 1985) for
processes such as lattice diffusion and creep. However, the activation volume for a grain-
boundary diffusion process (for example, the growth of spinel in olivine) will differ from
the activation volume for lattice diffusion. Therefore, its relationship to Tm is not clear.
There are at present no estimates for the activation volume of diffusion across inter-phase
boundaries in minerals and there are no reliable models for the pressure dependence of this
process. However, the activation volume V p has been noted to decrease significantly with
pressure increase (Kirby et al., 1996).
This page is intentionally left blank
 Section D

Mineral Systems
This page is intentionally left blank
 515

Chapter 6
MgO – FeO– SiO2 (MFS) System: Olivines and Pyroxenes

6.1. Introduction

Iron and magnesium are the major metal elements in the mantle composition. Mg2þ
is a simple closed-shell ion and it forms simple ionic bonds in oxides and silicates, whereas
Fe2þ is an open-shell, magnetic and very non-spherical ion that forms complex ionic –
covalent bonds.
The divergent bonding properties of Fe– O and Mg – O in different phases, such
as olivine, clinopyroxene and spinel, are manifested in anomalous compression
behaviour at high pressures. This has strong implications for the properties of the
Earth’s deep interior. An accurate determination of the Mg – Fe elasticity relationship
in the mantle minerals such as olivine, clinopyroxene and spinel would provide
explanations for the properties of the 670-km (24-GPa) seismic discontinuity. The
schematic phase diagram of the MgO – SiO2 system under lower transition-zone
pressure (24 GPa) is shown in Fig. 6.1.
The phase equilibrium in the MgO – FeO – SiO2 system at pressures and
temperatures of the mantle transition zone have been studied for many years and several
phases are found to be stable at these P – T conditions such as (Mg, Fe)2SiO4 (olivine,
b-spinel, g-spinel (Mg, Fe)SiO3 pyroxene, ilmenite, perovskite, majorite, (Mg, Fe)O
(magnesio-wüstite) and SiO2 (stishovite, St). The phase diagrams for the system MgO –
FeO – SiO2 at 1,1008C at various pressures are shown in Fig. 6.2a – f. With a delimited
composition range, (Mg, Fe)2SiO4 spinel dissociates into perovskite and magnesio-wüstite
(Mg, Fe)O, extending over the compositional range (Fe/(Mg, Fe) ¼ 0 – 0.22, which covers
the likely mantle compositions (Anderson, 1989).
The thermodynamic data set for MgO, FeO and SiO2 have been obtained using all
available phase-equilibrium data, thermochemical data (heat capacity, enthalpy of
formation and transition) and thermophysical data (thermal expansion). The thermodyn-
amic properties of the phases in the system MgO – (Al2O3)– SiO2, determined by several
workers are presented in Table 6.1 (Ito, 1984).
The P –T phase diagram for the Mg2SiO4 polymorphs (a-, b- and g-phases) is
shown in Fig. 6.3a. Using these data, the phase relations at Mg2SiO4 composition are
constructed as shown in Fig. 6.3b. Mg2SiO4, an olivine under pressure (at fixed
temperature), transforms first to spinelloid (b) and then to spinel (g), which in turn
decomposes to a mixture of MgSiO3 perovskite and MgO.
516 Chapter 6

Figure 6.1. Schematic phase diagram of the MgO–SiO2 system based on multianvil work (Ito and Takahashi,
1987). At about 24 GPa (670 km depth) the composition and melting temperature of MgSiO3-perovskite are near
the eutectic (Boehler, 2000, q 2000 American Geophysical Union).

Mg2SiO4 – Fe2SiO4 system. This system was investigated early on by Ringwood and
Major (1970) and Akimoto (1972) at temperatures of 1,273 and 1,473 K and pressures up
to 18 GPa. Yagi et al. (1979) studied the system under 15 –70 GPa and 1,0008C. Later
workers (Katsura and Ito, 1989; Fei et al., 1991) determined the isothermal phase relations
which, in the neighbourhood of 1,6008C, appear as in Fig. 6.3b. For the Mg2SiO4 – Fe2SiO4
system, the isothermal (1,6008C) phase diagram for the post-spinel transformation is
shown in Fig. 6.3c.
At MgSiO3 composition, the pyroxene (a chain silicate) first decomposes to a mixture
of b-phase plus stishovite (the rutile polymorph of SiO2 with Si in 6-fold coordination),
which then recombines first to form MgSiO3 ilmenite and finally MgSiO3 perovskite (Fig.
6.4 after Anderson, 1989). At high temperature and intermediate pressure, a field of garnet
appears. This is a tetragonal phase with a largely ordered distribution of Mg and Si on
octahedral-sites, analogous to YIG and YAG and familiar to material scientists.
A striking feature of the phase diagram is the negative P – T slope of the perovskite-
forming reactions. Since the perovskite-bearing assemblage is denser, it must have higher
entropy. This is almost certainly related to the simultaneous presence of silicon in
octahedral coordination (with longer, weaker Si – O bonds than in an SiO4 tetrahedron) and
magnesium in a roughly eight-coordinated central-site in the orthorhombic (distorted)
perovskite. The high entropy of perovskite has been confirmed by lattice vibrational
modelling.
In the phase diagram of MgSiO3 (Fig. 6.4), the negative P – T (Clapeyron) slope of
perovskite-forming transitions has important geophysical implications: when dP/dT
is sufficiently negative, mixing between the upper and lower mantle is inhibited.
Ito et al. (1990) determined the value of dP/dT as , 20.004 GPa/K, which is near to what
is required to prevent two-layer convection. If dP/dT has an intermediate value of
2 0.002 GPa/K, an intermittent mixing of the upper and lower mantle is possible
(Machetel and Weber, 1991).
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 517

Figure 6.2. (a –f). Phase diagrams for the system MgO–FeO–SiO2 at 1,1008C and various pressures from 19
to 26 GPa. Sp ¼ (Mg12xFex)2SiO4; Mw ¼ (Mg12xFex)O magnesio-wüstite; I ¼ (Mg12xFex)SiO3 ilmenite;
P ¼ (Mg12xFex)SiO3 perovskite (see Ito, 1984, 1989; q 1984 Terra Scientific).
518 Chapter 6

TABLE 6.1
Thermodynamics properties of phases in the system MgO–(Al2O3) –SiO2

Phase Q
DHf,970 (kJ/mol) 0
S970 (J/mol/K) 0
V298 (J/bar) a £ 105 KT (kbar) K0

Beta phase (Mg2SiO4) 232.0 268.3 4.052a 3.2b 1,720c 5

Corundum 0 176.8 2.558b 2.6d 2,500e 4
Enstatite (Opx) 269.5 382.5 6.264f 3.2d 1,070i 4
Enstatite (Cpx) 261.2 376.3 6.034 3.2 1,280 4
Fosterite 262.0 275.0 4.360g 3.5g 1,280e 5
Ilmenite (MgSiO3) 24.1 186.2 2.635a 2.5h 2,100i 4
Ilmenite (Al2O3) 95.8 165.7 2.635 2.5 2,100 4
Majorite 52.3 756.6 11.431j 1.4 1,180 4
Mg-Tschermak Pix 27.0 380.5 5.888k 3.2 1,070 4
Periclase 0 79.0 1.124l 4.2l 1,610a 4
Perovskite (MgSiO3) 65.0 189.7 2.446a 4.4 2,440m 4
Perovskite (Al2O3) 139.7 169.2 2.446 4.4 2,440 4
Pyrope 281.9 764.3 11.320n 2.7o 1,740e 4
Spinel (MgAl2O4) 222.5 270.7 3.971p 2.7d 1,950e 5
Spinel (Mg2SiO4) 221.3 260.7 3.965a 2.6b 1,820c 5
Stishovite 45.0 103.0 1.401a 2.3q 3,060r 5
a
Jeanloz and Thompson (1983); b Akaogi et al. (1984); c Weidner et al. (1984); d Skinner (1966); e Sumino and
Anderson (1984); f Chatterjee and Schreyer (1972); g Hazen (1976a); h Ashida et al. (1988); i Weidner and Ito
(1985); j Angel et al. (1989); k Gasparik and Newton (1984); l Hazen (1976b); m Yeganeh-Haeri et al. (1989);
n
Newton et al. (1977); o Hazen and Finger (1978); p Robie et al. (1978); q Ito et al. (1974); r Weidner et al. (1982).

The activation energies of the MgO(NiO) –SiO2 system obtained by different

workers are given in Table 6.2.

6.1.1. Stability of binary oxides and ternary phases

The ternary phases as seen in the binary oxide systems, MgO – SiO2 (olivine,
modified spinel, spinel, pyroxene, ilmenite and perovskite structure) and MgO – GeO2
(olivine, spinel, pyroxene, ilmenite and corundum structure), are in contrast to the small
number of ternary phases encountered in the nickel-bearing silicate and germanate
systems: NiO – SiO2 (olivine, spinel) and NiO –GeO2 (spinels only).
The MgO – SiO2 system under pressure greater than 20 GPa (in DAC) manifests
two transformations.
MgSiO3 (ilmenite) ! MgSiO3 (perovskite)
Mg2SiO4 (spinel) ! MgSiO3 (perovskite) þ MgO (periclase).
With pressure, the ternary phases will become unstable with respect to the more
dense binary oxides. This happens at a pressure when the standard free energy of formation
of the ternary phase from oxides is small. This free energy may result from an oxide-ion
transfer reaction (acid –base reaction). The reaction becomes more exothermic with
decreasing ionic potential (charge/radius) of the divalent ion and less exothermic with
increasing covalency of the divalent metal –oxygen bond. Unless ternary phases have
considerable stability with respect to binary oxides at atmospheric pressure, one sees few
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 519

Figure 6.3. (a) P, T phase diagram for the Mg2SiO4 polymorphs (a: olivine, b: modified spinel phase, g: spinel).
Solid lines from Akaogi et al. (1984), dashed lines from Suito (1977). (after Akaogi et al., 1984, q 1984
Mineralogical Society of America.) (b) P – T diagram of end member Mg2SiO4 (from Chopelas et al., 1994),
q 1994 Springer-Verlag. (c) Isothermal ðT ¼ 1; 6008CÞ calculated phase diagram for the Mg2SiO4 –Fe2SiO4
system. The crystalline structures are a (olivine), b (b-phase), g (g-spinel). The experimentally determined
boundaries (Katsura and Ito, 1989) are the dashed curves (after Akaogi et al., 1989, q 1989 American
Geophysical Union).

new high-P polymorphs. Evidently, this generalization is true as long as the binary oxides
are the denser assemblages.
For (particularly high-pressure) polymorphs in the MgO – SiO2 system, it has
been found that changes in pressure dependencies of the Raman modes indicate reversible
discontinuities in the compressional behaviour. This complicates extrapolation of
the physical properties of these materials measured at low pressure to mantle conditions
520 Chapter 6

Figure 6.4. P – T phase diagram for MgSiO3 composition. The approximate velocity of P waves (km/s) is
indicated below the names of the mineral phases. The arrows show the direction in which the phase boundaries are
expected to move when Al2O3 is added (after Anderson, 1989, q 1989 Blackwell Scientific Ltd).

(e.g., Chopelas and Boehler, 1992). But in a compressibility study up to 50 GPa, it has been
seen that the bulk modulus and its pressure derivative are in agreement with the elastic
properties measured at near-ambient conditions by Brillouin spectroscopy and ultrasonics,
i.e., K0 ¼ 18:3GPa and K 00 ¼ 5:4 (Weidner et al., 1984).

6.1.2. MgO – FeO –SiO2: thermodynamic data and phase equilibria in the mantle

The MgO – FeO – SiO2 (MFS) system is the most investigated system in planetary
sciences. The thermodynamic data on activation energies, enthalpies, entropies, heat
capacities, thermal expansion and conpressibilities of end-members of the solid solutions
are presented in Tables 6.3 and 6.4 (Fabrichnaya, 1995).
The thermodynamic properties of end-members have a greater influence on the
calculated phase diagram than the mixing parameters of solid solutions. The difference

TABLE 6.2
Silicate activation energies

Activation energy (kJ/mol) Reference

Mg2SiO4 ol ! sp 325 (^17) Sung and Burns (1976)

Ni2SiO4 ol ! sp 113 (^20) Hamaya and Akimoto (1982)
SiO2 qtz ! coes ,150 Naka et al. (1976)
SiO2 amorphous ! coes ,200 Naka et al. (1976)
Olivine (creep) 520 (^40) Kirby (1983)
Forsterite (creep) 669 (^29) Jaoul et al. (1981a)
Olivine (oxygen self-diffusion) 372 (^13) Reddy et al. (1980)
320 (^40) Jaoul et al. (1981b)
Olivine (silicon self-diffusion) 376 (^42) Jaoul et al. (1981a)
Olivine (static annealing) 323 (^18) Kohlstedt et al. (1980) and Karato (1981)
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 521

between the mixing energies is more important than their absolute values. One can obtain
mixing parameters from these experimental relationships (for details, see Fabrichnaya,
1995). The data on varying phase equilibria determined by different authors are cited below.
Nature of phase equilibria Authors
Divalent phase equilibria
ol þ b, b þ g Katsura and Ito (1989)
ol þ g Nishizawa and Akimoto (1973)
Trivariant exchange equilibria data
Px þ g Nishizawa and Akimoto (1973)
ol þ Mw, b þ Mw, g þ Mw Fei et al. (1991)
Divalent equilibria
Pv þ Mw þ St Ito et al. (1984)
Ito and Takahashi (1989)
g þ Mw þ St, Pv þ Mw þ g Ito and Takahashi (1989)
g þ St þ Mj, Px þ Mj Ohtani et al. (1991)
g þ St þ Ilm Ito and Takahashi (1989)
Exchange equilibrium
Pv þ Mw Fei et al. (1991)

TABLE 6.3
Molar volume V (cm3 mol) at 1 bar, T ¼ 298.15 K and thermal expansion (1/K) a ¼ a0 þ a1 T þ a2 T 21 þ
a3 T 22 for minerals in the MgO–FeO–SiO2 system (Fabrichnaya, 1995)

Mineral V a0 £ 105 a1 £ 108 a2 £ 103 a3 £ 101

Mg2SiO4
Forsterite 43.67 2.01 1.39 1.627 2 3.38
b-Forsterite 40.54 2.319 0.904 2 3.966 7.496
g-Forsterite 39.65 1.225 1.104 2.496 2 5.11
Fe2SiO4
Fayalite 46.28 5.67285 0.163147 2 2.5186 2 16.1331
b-Fayalite 43.14 8.774 2 0.2609 2 33.98 44.35
g-Fayalite 42.02 8.897 0.2803 2 34.26 38.13
MgSiO3
Orthoenstatite 31.276 3.871 0.44633 0.34352 2 17.2784
Ilmenite 26.35 2.2704 0.682087 2 1.8087 3.851
Perovskite 24.447 4.802 0.45677 2 8.5134 1.1908
Majorite 28.5 1.1019 0.75075 6.0403 2 10.267
FeSiO3
Orthoferrosilite 32.95 4.95565 0.87617 2 11.856 15.6019
Ilmenite 27.60 2.2704 0.682087 2 1.8087 3.851
Perovskite 25.59 2.627 1.5198 0.0 2 0.429
Majorite 29.425 3.0836 0.6659 2 6.106 6.453
MgO
Periclase 11.25 3.64 0.835 0.85 2 9.5
FeO
Wustite 12.25 2.60186 1.46789 2.78637 2 4.27635
SiO2
Stishovite 14.01 0.23 1.2 6.2 2 11.3
522 Chapter 6

TABLE 6.4
Compressibility (1/bar) b ¼ 1=KT ¼ b0 þ b1 T þ b2 þ T 2 þ b3 T 3 ; K 0 ¼ ðdK=dPÞT and K 00 ¼ ðdK 0 =dTÞP for
minerals in the MgO–FeO–SiO2 system (Fabrichnaya, 1995)

Mineral b0 £ 107 b1 £ 1010 b2 £ 1014 b3 £ 1017 K0 K00 £ 104

Mg2SiO4
Forsterite 7.427 1.24 0.69 1.702 5.2 2.0
b-Forsterite 5.51282 0.92017 0.8849 1.1529 4.3 3.0
g-Frosterite 5.07778 1.3371 22.9854 2.7822 4.3 6.0
Fe2SiO4
Fayalite 6.63158 2.0846 21.0135 1.16683 4.0 2.2
b-Fayalite 5.54 0.7947 7.069 20.4957 4.0 5.0
g-Fayalite 3.86 3.978 26.304 0.8275 4.0 8.0
MgSiO3
Orthoenstatite 8.892 1.35844 3.1613 1.14126 4.2 1.5
Ilmenite 4.504 0.82987 20.38777 1.2095 4.0 3.0
Perovskite 3.7184 0.23496 2.2508 0.0 4.1 0.0
Majorite 5.8168 2.521 211.0 5.58636 4.0 10.0
FeSiO3
Orthoferrosilite 9.27714 3.70379 210.1 2.9097 4.2 0.0
Ilmenite 4.75282 0.82987 20.38777 1.2095 4.0 3.0
Perovskite 3.367 1.01 0.0 0.0 4.0 0.0
Majorite 5.4858 0.806 20.46 1.567 5.6 20.0
MgO
Periclase 5.875 1.101 1.37 0.48 4.17 0.4
FeO
Wüstite 5.3304 0.932 0 0 4.0 0
SiO2
Stishovite 2.954 0.8961 23.29 2.331 6.0 10.0

The phase diagram of the MgO –FeO – SiO2 system for the composition close to
pyrolite (XSiO2 ¼ 0.4 mol%, Fe/(Fe þ Mg) ¼ 0.12) is presented in Fig. 6.5.
The sharp phase transition of spinel to perovskite þ Mg-wüstite occurs at a bulk-
olivine composition with XFe ¼ 0.1 and at P ¼ 23.5 GPa, which corresponds to the sharp
discontinuity at ,670 km. This bulk composition will yield an assemblage of perovskite
of XFe¼0.02 and Mg-wüstite ¼ 0.2. The mineralogical constitution of the peridotite lower
mantle is inferred (volumetrically) to be 70% perovskite, 18% Mg-wüstite, 8% CaO-rich
phase with a small amount of stishovite and Al2O3.
At 670-km discontinuity, the dissociation of spinel is completed within a very small
depth interval. This would make the discontinuity very sharp, especially in the upper
portion. The dissociation of majorite, on the other hand, proceeds over a fairly large depth
range, ,80 km. All these features are in agreement with seismic models, in which the 670-
km discontinuity is seen as large and sharp, but there is still a gradient of velocities to the
depth of 750 or 800 km (Walk, 1984).
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 523

Figure 6.5. Phase diagram of the FeO–MgO–SiO2 system for the composition close to pyrolite XSiO2 ¼ 0.4
mol%, Fe/(Fe þ Mg) ¼ 0.12. The next phase assemblages are indicated by numbers: 1: ol þ g þ Px, 2:
o1 þ b þ Px, 3: b þ Px þ St, 4: b þ Px þ Mj, 5: b þ St þ Mj, 6: b þ g þ Mg, 7: g þ Mj, 8: g þSt þ Mj,
9: g þ Pv þ Mj, 10: g þ Px þ St, 11: b þ g þ Pv, 12: b þPv, 13: b þ Pv þ Mj, 14: b þ Pv þ Mw, 15:
g þ Pv þ Mw, 16: g þ St þ Ilm. Dashed lines restrict the range of possible geotherms (adapted from
Fabrichnaya, 1995, q 1995 Springer-Verlag).

6.2. MFS system in Mars

The Martian bulk (MB) composition, projected into the MFS system at pressures
of 11, 13.5 and 15 GPa and derived from experiments by Bertka and Fei (1996),
is presented in Fig. 2.27a. At 11 GPa, a tie-line connecting co-existing olivine
and clinopyroxene passes through the bulk composition. By 13.5 GPa, both olivine
and clinopyroxene shift towards higher Mg numbers (see Section 6.3.3), stabilizing Mg-
wüstite with Mg #30. The bulk-MB composition falls within the triangle defined by the
co-existing olivine, clinopyroxene and Mg-wüstite. At 15 GPa, this situation remains
valid with Mg-Wu shifting to a higher mg number, co-existing with b-phase and
clinopyroxene.
However, in the Mg2SiO4 – Fe2SiO4 binary system, an increase in pressure causes
a shift towards high Mg-content in olivine, b-phase and g-spinel (Fei et al., 1991). This
explains why magnesio-wüstite is absent in the Earth’s transition zone. But if the mantle is
iron-rich, as Mars is, magnesio-wüstite may co-exist with b-phase in the transition zone.
Using an oxide mix, Bertka and Fei (1996) observed assemblages containing
majorite, g-spinel, magnesio-wüstite and stishovite at 17 GPa (1,2008C) and 19 GPa (1,200
and 1,7508C). At this pressure range, pyroxene breaks down. A Mg4Si4O12 pyroxene
is transformed directly to majorite or, at lower temperatures, to b-phase þ stishovite or
524 Chapter 6

g-spinel alone. Addition of Fe lowers the transformation pressures and increases the stability
field of majorite co-existing with g-spinel and stishovite.
At high temperatures, pyroxene may transform directly to majorite (Akaogi et al.,
1987). At 24 GPa (8008C), MB composition would show both magnesio-wüstite and
stishovite to be stable along with Mg –Fe silicate perovskite, CaSiO3 perovskite and
majorite. The model of the seismic discontinuities in the Martian transition zone may well
be affected by the presence of magnesio-wüstite plus stishovite. In Fig. 2.27, in the MgO –
FeO – SiO2 system at 24 GPa pressure, the MB composition is seen to fall within the
triangle defined by the co-existing Mg – Fe silicate perovskite, stishovite and magnesio-
wüstite. The terrestrial spinel lherzolite (KLB composition), representing Earth’s upper-
mantle composition, falls within the triangle defined by the co-existing Mg –Fe silicate
perovskite, stishovite and magnesio-wüstite (see also Fig. 2.28).
The Martian lower mantle should manifest a transition of spinel to perovskite plus
magnesio-wüstite and stishovite. This transition reaction has a negative slope in P – T
space and is very susceptible to variations in the marstherm. Thus, stishovite may be
stable in the Martian lower mantle, although it is possibly absent in the Earth’s
lower mantle.

6.3. Mg-olivines

Magnesium-rich olivines ((Mg12xFex)2SiO4) are the major minerals of the Earth’s

upper mantle down to the transition zone (e.g., Ringwood, 1991). With increasing
pressure, they transform into the denser b-phase (modified spinel) and then g-phase
(spinel). The diffraction pattern of Mg2SiO4 at 69 GPa shows the continued presence of
b-spinel along with a-olivine (Fig. 2.27).
Mg-olivines are components of the forsterite (Mg2SiO4)– fayalite (Fe2SiO4) solid-
solution series. The structure of olivine as viewed in the (100) plane is shown in Fig. 6.6.
In this single chain Mg – Fe silicate series — the M– O distance (d ) at octahedral M1
(d ¼ 2.095 Å)-site is shorter than at octahedral M2 (dM2 ¼ 2.131 Å)-site. In olivine
structure, there is scope for the entry of trivalent and monovalent cations through coupled
substitution. Such possibilities are depicted in Fig. 6.7. The M1-site, having lesser volume,
has greater force constants, i.e., KðM1Þ . KðM2Þ (Stanek et al., 1986). The relative
volume change as related to force constant is

 
dVðM2Þ . dVðM1Þ KðM1Þ 3
¼ ø 1:4
dP dP KðM2Þ

However, Mössbauer spectroscopic studies up to 3.0 GPa pressure did not show any
significant change in the hyperfine parameters.
Lack of cation ordering in olivine is explained by CFT. The d splittings of Fe2þ at
both M1- and M2-sites are seen as equal but a weak-site preference for Fe2þ for M2 is
explained by dynamical Jahn –Teller effect, by which the ground state is considered as
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 525

Figure 6.6. Olivine structure as viewed in the (100) plane (from Andrault et al., 1995, q 1995 Springer-Verlag).

the doubly degenerate 3dxz and 3dyz level. (In contrast, the positive sign of Vzz determined
by Stanek et al. (1986) suggests that the ground state must be 3dxy-singlets for both sites.)

6.3.1. Olivines in the mantle: pyrolite model

The 410-km discontinuity at the top of the transition zone is attributed to the
transformation of olivine (the most abundant mineral of the upper mantle) to wadsleyite
(b-(Mg, Fe)2SiO4). Seismological observations indicate that the discontinuity is narrow,
frequently less than 10 km wide and locally as narrow as 4 km.
Seismological studies of the 410-km discontinuity reveal that the increases in wave-
speed are smaller and more abrupt than expected from an extrapolation of the laboratory
behaviour of these minerals to mantle conditions. Irifune and Issuhiki (1998) asserted that
exchanges of Mg and Fe between olivine and other mantle minerals work as an important
factor in resolving this apparent discrepancy. They found that the iron content in olivine
526 Chapter 6

Figure 6.7. Possible substitution types for tri- and monovalent cations in olivine structure, (a) (Si4þ,
Mg2þ) $ (Al3þ, Cr3þ) type substitution, (b) (Si4þ, Mg2þ) $ (Al3þ, Al3þ) and 2(Mg2þ, Mg2þ) $ 2(Naþ,
Al3þ) type substitution (from Taura et al., 1998).

changes significantly with increasing pressure. This is in consequence of the formation of a

relatively iron-rich majorite phase at these pressures.
According to the pyrolite model, olivine accounts for only 60% of the mantle by
volume. Olivine from the shallow mantle has been reported to have an approximate
composition of Fo89(Mg0.89 Fe0.11)SiO4, which transforms from a-phase to wadsleyite (b)
to ringwoodite (g) phases under progressive pressure. Ultimately, it disproportionates to
perovskite and magnesio-wüstite. The seismic discontinuity at 670 km is due to the
transformation of (Mg, Fe)2SiO4 to a mixture of perovskite and magnesio-wüstite:

ðMg; FeÞ2 SiO4 ! ðMg; FeÞSiO3 þ ðMg; FeÞO

olivine perovskite magnesio-w€ustite

For calculating the density and seismic-velocity profile, a fixed composition for
mantle olivine is generally presumed. Pyrolite or peridotite composition gives the
density and velocity profiles which generally agree with the seismologically derived
profiles of the upper mantle and the transition zone. To account for the observed
velocity jump at 410-km depth, the olivine content is assumed to range between ,30
and ,60 vol%.
In pyrolite, all phases grow increasingly Mg-rich with depth. Olivine above the
a –b transition becomes enriched in Mg relative to Fo89. In consequence to the Mg
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 527

enrichment of a-phase, the a –b transition is shifted to higher pressures. Thus, the onset of
the transition in pyrolite is postponed to greater depths relative to Fo89.
The next abundant minerals in pyrolite are orthopyroxene, clinopyroxene and
garnet, which successively dissolve into the following one with increasing depth.
Unfortunately, in most models of the mantle, olivine polymorphs are studied in isolation
from pyroxenes and garnets.

6.3.1.1. Mg2SiO4 –Fe2SiO4 system: binary loop in the mantle

The phase diagrams for the Mg2SiO4 – Fe2SiO4 system tend to suggest that olivine
of mantle composition with Mg/(Mg þ Fe) ¼ 0.9 would transform to wadsleyite through a
binary loop of 10– 20-km width (Katsura and Ito, 1989; Akaogi et al., 1989).
The presence of garnet and pyroxene tend to narrow the olivine – wadsleyite loop by
buffering the Mg/Fe ratios of the co-existing (Mg, Fe)2SiO4 minerals (Stixrude, 1997). The
Mg2SiO4 –Fe2SiO4 loop could also be as narrow as 4 km for mantle olivine at a
temperature ,1,773 K (Fig. 6.8; Goodfinnsson and Wood, 1998) at 410-km discontinuity
at the top of the transition zone (Helffrich and Wood, 1996).
The mineralogy and chemistry of the transition zone, in particular, govern the
dynamics of connective flow of the heat and matter in the interior. This flow, in turn,
drives the tectonic evolution of the surface, including the occurrence of volcanism and
seismicity.

a – b – g transformation. a-olivine ((Mg, Fe)2SiO4), the most abundant mineral in the

upper mantle, transforms to modified spinel (b-phase) and spinel (g-phase) structure at

Figure 6.8. Pressure-composition diagram for the system Mg2SiO4 –Fe2SiO4 from Katsura and Ito (1989)
showing the olivine-wadsleyite loop at 1,773 K. The symbols indicate compositions of coexisting olivine and
wadsleyite in the experiments. The pressures of the experiments have been adjusted so that the points fall close to
the two-phase loop. The cross denotes approximate propagated standard errors in the compositions and 5%
uncertainty in pressure. All pressure adjustments are within this uncertainty (Gudfinnsson and Wood, 1998,
q 1998 Mineralogical Society of America) (see also Figure 6.10).
528 Chapter 6

high pressure (,13.5 and ,18 GPa at 1,400 and 1,5008C, respectively) and then
decomposes to an assemblage of MgSiO3-rich perovskite and (Mg, Fe)O ferropericlase
(at ,23 GPa and 1,6008C).
In this transformation, the oxygen sublattices transform from hexagonal close-
packed in olivine to face-centred cubic in spinel. This transformation is believed to
occur by shear restacking of the oxygen ions (Fig. 6.9) effected by large shear stresses
and/or with a large driving force (far from Clapeyron) (see Section 6.5.1). The P – T
pseudobinary diagram of Mg2SiO4 – Fe2SiO4 (at 1,6008C) is shown in Fig. 6.10a. The
high-pressure spinel phases, wadsleyite and ringwoodite, are seen in shocked
chondritic meteorites.
Because Si is four-coordinated and Mg and Fe are six-coordinated in both olivine
and spinel structures, denser phases are achieved through changes in the linkages of
coordination polyhedra. The volume changes of the a –b and b –g transitions are small at 8
and 2.5%, respectively. In fact, the olivine and spinel structures are related closely enough
that transformations could take place through intermediate mixed phases (Madon and
Poirier, 1983) (see Fig. 6.10b).
Polymorphic transformation under near-equilibrium conditions in the a þ g or
a þ b stability fields by diffusion-controlled growth is most likely to occur at high
temperature. It is experimentally seen that, at 1,2008C, the a-olivine to b-phase and
b-phase to g-spinel transitions take place at 14 and 18 GPa, respectively (Katsura and Ito,
1989), i.e., under conditions corresponding approximately to those of the 400- and 520-km
seismic discontinuities. However, the olivine to modified-spinel and the post-spinel
transformations are believed to underlie the two major seismic discontinuities at about
400- and 670-km depths in the mantle, respectively.

Figure 6.9. The (100) plane of olivine showing the shear mechanism for the transformation to spinel.
The positions of some of the oxygen ions in the close-packed plane of the hcp sublattice are shown by circles. The
oxygen centered at A is in the plane above the other five oxygens and the shear 1/12[013] along AB moves
this oxygen to form (dashed circle) a spinel (see Poirier 1981, 1991, q 1991, Cambridge University Press).
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 529

Figure 6.10. (a) Mg-silicate phase transformations at 1,6008C at varying depths, shown with respect to
Mg/(Mg þ Fe) ratio. (b) The same with changing temperature and with reference to the geotherm (after Madon,
1992, q 1992 Academic Press).
530 Chapter 6

a – b transition: thermodynamics. For any occurrence of the co-existing olivine and

wadsleyite, the following thermodynamic relations may be considered:
Mg2 SiO4 ¼ b-Mg2 SiO4 ð6-1Þ
ðolivineÞ ðwadsleyiteÞ

Fe2 SiO4 ¼ b-Fe2 SiO4 ð6-2Þ

ðolivineÞ ðwadsleyiteÞ

For the Mg2SiO4 component, the chemical potential shows the following relationships:

mwadsl ol
Mg2 SiO4 ¼ mMg2 SiO4 ð6-3Þ

and

mol o ol
Mg2 SiO4 ¼ mMg2 SiO4 þ RT ln aMgaSiO4 ð6-4Þ

mwad o wad
MgSiO4 ¼ mMg2 SiO4 þ RT ln aMgaSiO4 ð6-5Þ

Assuming the standard state chemical potential, mO Mg2 SiO4 ; to be equal to the molar free
energy of the respective pure phase at the P, T of consideration, and then using the relations
(6-4) and (6-5) in relation (6-3), one can obtain the relationship between equilibrium
pressure, P, for co-existing olivine– wadsleyite solid solution and the activity of the
Mg2SiO4 component in the two phases as follows (Wood, 1990):
ðP
DV 0 dP ¼ 2RT ln awad ol
Mg2 SiO4 þ RT ln aMg2 SiO4 ð6-6Þ
P0

When P 0 is the pressure of equilibrium of pure Mg2SiO4 wadsleyite and olivine at

temperature T, the relation becomes:

DV 0 ðP 2 P0 Þ ø 2RT ln awad ol
Mg2 SiO4 þ RT ln aMg2 SiO4 ð6-7Þ

Since P and P 0 are very close, they make DV 0 be treated as a constant. Wood et al. (1996)
calculated the DV 0 to be 2 2.24 at 14.5 GPa and 1,773 K, which corresponds to the 410-km
discontinuity.

6.3.1.2. Fe –Mg in a – b phases: ordering

Sawamoto and Horiuchi (1990) synthesized a single crystal of wadsleyite of
composition (Mg0.9Fe0.1)2SiO4 at 2,0008C and 18 – 20 GPa and found ordering of Mg and
Fe on the octahedral-sites.
The apparent partition coefficient of Fe and Mg between the two phases (a and b)
is determined as K 0ða=bÞ ¼ {Xa =ð1 2 Xa Þ}={Xb =ð1 2 Xb Þ}; where Xa and Xb denote
mole fractions of iron in phases a and b, respectively. The values of Fe – Mg
partition coefficients between co-existing phases have been determined by several workers
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 531

TABLE 6.5
The enthalpy and entropy in Mg-silicate phases and MgO in solid–solid phase transitions (Chopelas et al., 1994)
0
Reaction DH975 (J/mol) DS975 (J/mol) DS180
1273

b!g 9,080a 27.2 25.5

g ! pv þ MgO 99,000b þ9.5 þ9.5
b ! pv þ MgO 108,080c þ2.3 þ4.0
a
Akaogi et al. (1984); b Akaogi et al. (1984); c Akaogi et al. (1989).

0 0 0 0 0
(see e.g., Takahashi and Ito, 1987) such as K (a(gt) , K (a(cpx) , K (a(opx) , K (b(gt) , K (b(cpx) and
0
K (a(b), where gt, cpx, opx stand for garnet, clinopyroxene and orthopyroxene, respectively.
The stability field of b-phase expands into the a- and g-fields when H2O and Fe
are present (i.e., San Carlos olivine containing 11% Fe) (Kagi et al., 1997). The b- and
g-phases show different IR bands because of their difference in symmetries, i.e.,
orthorhombic vs. cubic.

6.3.1.3. Thermal properties of MgO –SiO2 system

The thermal properties (enthalpy and entropy) of b-Mg2SiO4, g-Mg2SiO4, MgO
and MgSiO3 perovskite in the MgO – SiO2 system with reference to the corresponding
reactions are presented in Table 6.5.
The P – modulus diagram of end-member Mg2SiO4 is shown in Fig. 6.11 (Chopelas
et al., 1994). In the diagram, the phase boundary for the iron-free system was calculated
from thermodynamic data and from the equations as presented in Table 6.5. This resulted
in a triple point at ,23 GPa (230 kbar) and 2,260 K.

Figure 6.11. Bulk (Ks) and shear (G) moduli of a-Mg2SiO4 (open symbols) and b-Mg2SiO4 (filled symbols).
Lines are third order finite strain fits to the experimental data. Ambient pressure data of Isaak et al. (1989) were
used in fitting the a-Mg2SiO4 data (Zha et al., 1998a).
532 Chapter 6

The calculated Clapeyron slopes for the transition are (Chopelas et al., 1994):
b!g 50 ^ 4 bar=K
g ! pv þ MgO 225 ^ 4 bar=K
b ! pv þ MgO 27 ^ 3 bar=K
An addition of iron to Mg2SiO4 significantly changes the topography of the phase
diagram, mainly by lowering the pressure of b- to g-Mg2SiO4 transition.

6.3.2. Nucleation rates

For predicting the grain-boundary nucleation rates of b- and g-(Mg, Fe)2SiO4 (in
the subduction zone), Rubie and Ross (1994) used activation enthalpy for growth as a
function of the melting temperature at 1 bar for the relation:
N ¼ K0 T expð2fDGphom =kTÞ exp½2ðaTm þ PV p Þ=RT
where N is the rate of nucleation on grain boundaries, K0 is a constant, f is a shape factor,
DGphom is the activation energy for homogeneous nucleation and k is the Boltzmann
constant. When the strain energy is zero ð1 ¼ 0Þ; it corresponds to the case when the
volume change of transformation is completely accommodated by ductile flow rather than
by elastic strain.
For solid-state first-order transformations, volume change accompanying trans-
formation is often large, resulting in the development of localized stress (Rube and
Thompson, 1985) and the growth is controlled by slip deformation of the outer rim.
The activation energy for homogeneous nucleation is given by
DGphom ¼ 16pg3 =3ðDGv þ 1Þ2 ;
where g is the interfacial free energy, DGv is the free energy change of reaction per unit
volume and 1 is the strain energy.
For the olivine– spinel transformation, the only available estimates of nucleation
rates are for the composition Ni2Si2O4 (Rubie et al., 1990). The rates of grain-
boundary nucleation of g-Ni2Si2O4 are estimated at 3.6 –3.7 GPa and 825 –9808C.
Because of the large uncertainties involved in extrapolating the limited
experimental nucleation-rate data to subduction-zone conditions, Rubie and Ross
(1994) calculated transformation kinetics using different nucleation-rate models:

Model 1: The strain energy (1) is taken as zero.

Model 2: The strain energy (1) is maximized. This occurs when olivine deforms elasti-
cally in response to the volume change accompanying nucleation of b or a.

The elastic strain energy is approximated using the relation:

1 ¼ 2Ka fðV2 2 Va Þ2 =3V2 with f ¼ 3G2 =ð3G2 þ 4Ka Þ
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 533

where Ka and Va are the bulk modulus and molar volume of (a) olivine and G2 and V2 are
the shear modulus and molar volume of the product phase (b or a). The molar volumes,
shear and bulk moduli are calculated from the data of Akaogi et al. (1989) and Anderson
(1989, p. 105).

6.3.3. Elasticity

Olivine becomes Mg-rich with increasing pressure. At 13.5 GPa, the Mg numbers,
i.e.,

 
Mg
100
Fe þ Mg

in olivine should be in the 92 – 93 range, equivalent to 410-km depth. At this depth, the
modified spinel has a composition of Mg number ¼ 88 –89 only.
The olivine content in the upper mantle has been the subject of debate because of
the earlier lack of data on the elastic properties of olivine at transition temperatures and
pressures. Seismic velocity profiles alone provide the direct constraints on the deep mantle.
Seismic data and laboratory measurements of acoustic-wave velocities in (Mg, Fe)2SiO4
are considered to simulate the 410-km discontinuities. Using Brillouin scattering, acoustic
velocities can be determined from the frequency shift of light scattered from thermally
generated acoustic waves. Zha et al. (1995) recorded Brillouin spectra using an Arþ laser
and a tandem Fabry – Perot interferometer. The pressure derivative of the bulk modulus
determined from the Brillouin data is consistent with that determined solely from X-ray
diffraction (Downs et al., 1996). Comparison of laboratory measurements of the acoustic-
velocity contrast in the a – b system with the magnitude of the seismically observed
discontinuity at 410 km provides a way to constrain the olivine content of the mantle at this
depth. The elastic properties of (Mg0.9, Fe0.1)2SiO4 obtained by several workers are given
in Table 6.6.

TABLE 6.6
Elastic properties of (Mg0.9, Fe0.1)2SiO4

Property a-phase b-phase

a
Ks (GPa) 129(1) 174(1)b
G (GPa) 79(1)a 110(1)b
(›Ks/›P)T 4.2(2) 4.8(2)c
(›G/›P)T 1.4(1)e 1.7(1)c
(›Ks/›T)P (GPa/K) 20.017(1)a 20.018(3)d
(›G/›T)P (GPa/K) 20.014(1)a 20.014 to 20.024e
a b c d
Isaak et al. (1989, 1992); Sawamoto et al. (1984); Gwanmesia et al. (1990); Fei et al. (1992) and
e
Meng et al. (1993).
534 Chapter 6

The aggregate bulk modulus (Ks) and shear modulus (G) are calculated at each
pressure from the average of the Voigt and Reuss bounds, which are the limiting values of
the moduli of a random, polycrystalline aggregate obtainable from single-crystal elasticity
data (Watt et al., 1976). The variation of bulk and shear moduli of a- and b-Mg2SiO4 under
pressure is presented in Fig. 6.11.
The aggregate shear moduli of a- and b-polymorphs of Mg2SiO4 are shown in
Table 6.7. Duffy et al. (1995) employed Brillouin scattering in a DAC up to 16 GPa on a
pure forsterite crystal and found dK ¼ dP ¼ 4:2 ^ 0:2: The pressure derivative of shear
modulus (dG/dP) gives a values of 1.71 (Zhang et al., 1993), using G ¼ 77.6 GPa. Duffy
et al. (1995), however, report the values as G ¼ 81.6 and dG/dP ¼ 1.4.
The temperature effects on the values of Ks and G (i.e., dKs/dT and dG/dT) of
forsterite are obtained as 0.016 and 2 0.014 GPa/K, respectively, by Isaak et al. (1989) in
the temperature range 300– 1,700 K.
The temperature dependence of the shear modulus, dG/dT, for 15 silicates
determined by Sumino and Anderson (1989) fall in the range of 2 0.08 to 2 0.014 GPa/K.
The a- and b-phases of (Mg0.9Fe0.1)2SiO4 are assumed to have the same value of dG/dT
around 2 0.014 GPa/K while, for silicates, oxides and about 80 halides, these values are
mostly less than 2 0.02 GPa/K. Of these, MgO shows the largest dG/dT value:
2 0.024 GPa/K while a value of 2 0.014 GPa/K has been used for b-Mg2SiO4 by Duffy
et al. (1995) in their calculation.
A summary of the bulk moduli and Pmax for olivine determined by several workers
using different methods are presented in Table 6.8.
The axial linear compressibilities of (Mg12x, Fex)2SiO4 olivines determined at
room pressure from compression measurements on varying values of x are presented in
Table 6.9.
Elastic moduli of single-crystal forsterite (a-Mg2SiO4) at 3 –16 GPa and 295 K,
using Brillouin scattering in a DAC were reported by Duffy et al. (1995).

6.3.3.1. San Carlos olivine

Zha et al. (1998) performed both Brillouin scattering and crystal XRD
measurements on a single crystal of San Carlos olivine at each pressure up to 32 GPa to
obtain the EOS and elasticity independently. The absolute pressure could be calculated

TABLE 6.7
Aggregate elastic properties of Mg2SiO4 polymorphs

Property a-Mg2SiO4 b-Mg2SiO4

Kos (GPa) 129a 170 (2)

G0 (GPa) 79a 115 (2)
K0os 4.2 (2) 4.3 (2)
G00 1.4 (1) 1.4 (2)

a
Isaak et al. (1989) [source: Duffy et al., 1995].
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 535

TABLE 6.8
Summary of bulk-moduli data for olivines
0
Reference K0 (GPa) K0 Pmax (GPa) Method

Forsterite
Kodoh (1985) 122.6 4.3 15 Compression
Syono and Goto (1982) 134 3.5 93 Shock wave
Will et al. (1986) 135.7 3.98 30 Compression
Andrault et al. (1995) 128 ^ 8 4.0 42 Compression
Fa17 Andrault et al. (1995) 134 ^ 10 4.0 34 Compression
Fa66 Andrault et al. (1995) 136 ^ 10 4.0 20 Compression
Fayalite
Hofmeister 126.8 5.0 42.5 IR spectroscopy
Kudoh 130.5 4.0 14 Compression
Williams et al. (1990) 133.0 4.0 42 Compression
Andrault et al. (1995) 125 ^ 6 4.0 10 Compression

from thermodynamic relations, which compared well with the ruby fluorescence pressure
scale.
The pressure dependence of elastic moduli is shown in Fig. 6.12. The shear moduli
(c44, c55 and c66) exhibit stronger non-linear pressure dependences than longitudinal
moduli (c11, c22 and c33), although a significant non-linearity is also observed in c22 above
14 GPa. There has been some controversy regarding the behaviour of modulus c55 in San
Carlos olivine. Zha et al. (1998) determined a non-linear term to describe the pressure
dependence of c55 in San Carlos olivine as

c55 ¼ 77:7ð2:2Þ þ 1:65ð0:31ÞP 2 0:025ð009ÞP2 ðP in GPaÞ:

The aggregate compressional- (VP), bulk- (VB) and shear- (VS) wave velocities are
listed in Table 6.10 (Zha et al., 1998) together with Poisson’s ratio, s, which increases
from 0.24 to 0.32 with pressure from ambient pressure to 32 GPa.

TABLE 6.9
Axial linear compressibilities of (Mg12x, Fex)2SiO4 olivines at room pressure from compression measurements

d ln a/dP d ln b/dP d ln c/dP References

Fo 0.50 (10) 2.6 (4) 1.8 (4)

Fa17 0.38 (10) 2.6 (4) 1.6 (4) Andrault et al. (1995)
Fa66 0.37 (10) 3.6 (5) 2.2 (3)
Fa 0.43 (10) 4.3 (5) 0.78 (15)
Forsterite 0.70 2.8 1.5 Kudoh and Takéuchi (1985)
0.52 2.2 1.1 Will et al. (1986)
Fayalite 0.59 4.0 0.89 Kudoh and Takeda (1986)
0.45 2.4 1.0 Williams et al. (1990)
536 Chapter 6

Figure 6.12. Single-crystal elastic moduli of San Carlos olivine as a function of pressure. Filled symbols, present
data with 2s uncertainties; squares (Webb, 1989); circles (Chen et al., 1996); triangles, (Abramson et al., 1997);
dashed line, (Zhang et al., 1993). The solid lines are weighted least squares fits to the data of Zha et al. (1998a) and
the ambient-pressure data of Webb (1989). Pressure were determined by ruby fluorescence, and fitting curves
change only slightly when pressures from Brillouin data are used (Zha et al., 1998b).
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 537

TABLE 6.10
Aggregate elastic properties of San Carlos olivine (Zha et al., 1998)

P (GPa) KS (GPa) G (GPa) VP (km/s) VB (km/s) VS (km/s) s

2.5 140.5 (3.0) 81.4 (1.4) 8.55 (0.06) 6.42 (0.07) 4.89 (0.04) 0.257 (0.008)
5.0 153.5 (3.1) 85.5 (1.4) 8.79 (0.06) 6.65 (0.07) 4.97 (0.04) 0.265 (0.007)
8.1 166.7 (3.4) 91.6 (1.6) 9.03 (0.07) 6.87 (0.07) 5.09 (0.005) 0.267 (0.009)
14.1 185.5 (4.2) 94.8 (1.9) 9.22 (0.07) 7.12 (0.08) 5.09 (0.05) 0.279 (0.008)
18.8 199.7 (4.7) 99.4 (2.1) 9.40 (0.08) 7.29 (0.09) 5.14 (0.05) 0.287 (0.008)
24.6 216.8 (4.6) 101.2 (2.1) 9.54 (0.07) 7.49 (0.08) 5.12 (0.05) 0.298 (0.007)
32.3 247.0 (4.8) 104.5 (2.2) 9.83 (0.007) 7.86 (0.008) 5.11 (0.05) 0.315 (0.006)

Note: Bulk and shear moduli are averages of the Hashin–Shtrikman bounds. Uncertainties are two standard
deviations. Pressures are from integration of Brillouin and X-ray data.

6.3.3.2. Mode-Grüneisen parameter

Comparison of initial mode Grüneisen parameters of forsterite with those of fayalite
(Table 6, Hofmeister et al., 1989) shows that the Fe content affects the far-IR bonds more
than the mid-IR bands. This follows the assignments of far-IR bands to the metal ions and
of mid-IR bands to tetrahedral variations of SiO4. Four of the mid-IR bands (n4, n1 and 2n3)
have gi (0) values that are approximately the same (^ 10%) for forsterite and fayalite.
Initial values for the first mode-Grüneisen parameter range from 0.10 to 2.0, such
that far-IR modes (transitions or rotations) generally have values larger than unity, whereas
mid-IR modes (internal vibrations) all have values less than unity.
Among the far-IR bands, substitution of Fe for Mg exerts a strong to moderate
influence on gi(0), depending on the band. At pressures significantly above the stability
field of olivine, gi can become negative. The strong curvature of frequency with pressure
allows the determination of the first and also the second Grüneisen parameter, such that
qi (0) ranges from 0 to 9. The negative qi values are mandated by the linear dependence
of n on P.

6.3.4. (Mn, Fe, Co) olivines: compressibility

Zhang (1998) investigated the compression behaviour of synthetic Mg2SiO4,

Mn2SiO4, Fe2SiO4 and Co2SiO4 olivines up to pressures of 10 GPa. All four showed strong
anisotropic compression in the three crystallographic directions. The linear compressi-
bilities were determined as shown below.
Olivine ba ( £ 1023 GPa) bb ( £1023 GPa) bc ( £1023 GPa)
Mg2SiO4 1.53 2.90 2.32
Mn2SiO4 1.45 3.48 1.98
Fe2SiO4 1.35 3.29 1.76
Co2SiO4 1.25 2.82 2.01

Zhang (1998) measured the unit-cell parameters using single-crystal diffraction and
the eight-position centring procedure (King and Finger, 1979) that minimized systematic
errors and improved data precision. In the sequence of Mn – , Fe– and Co –olivine, the
538 Chapter 6

TABLE 6.11
Bulk moduli and their pressure derivatives of olivines (Zhang et al., 1998)

Olivine Sample KTO (GPa) K0TO V0 References

Mg2SiO4 SC 127 (4) 4.2 (8) 289.3 (1) Zhang et al. (1998)
Mn2SiO4 SC 129 (2) 3.0 (5) 324.3 (1) Zhang et al. (1998)
Fe2SiO4 SC 136 (3) 4.1 (7) 306.9 (1) Zhang et al. (1998)
Co2SiO4 SC 144 (2) 4.1 (5) 295.21 (7) Zhang et al. (1998)
Mg2SiO4 SC 128.1 (4) 4p Zhang et al. (1998)
Mn2SiO4 SC 125.2 (4) 4p Zhang et al. (1998)
Fe2SiO4 SC 136.1 (4) 4p Zhang et al. (1998)
Co2SiO4 SC 144.2 (3) 4p Zhang et al. (1998)
Mg2SiO4 SC 125 (2) 4.0 (4) Downs et al. (1996)
SC 122.6 4.3 Kudoh and Takeuchi (1985)
SC 129 (1) 4.2 (2) Duffy et al. (1995)
Poly 128 (8) 4p Andrault et al. (1995)
Fe2SiO4 Poly 125 (5) 4p Andrault et al. (1995)

SC, single crystal; poly, polycrystalline; p , assuming.

decrease in unit-cell volumes is the manifestation of increasing d electrons in the 3dx

configuration of these cations. However, Mg2SiO4 is more compressible than Co2SiO4,
Fe2SiO4 and slightly less compressible than Mn2SiO4. The bulk moduli and the pressure
derivative of olivines are presented in Table 6.11.
Like the compressibilities, the KTO of the transition-metal olivines increases with
decreasing molar volume, as is consistent with bulk modulus – volume systematics. Among
transition-metal olivines, such as (Mn, Fe, Co)2SiO4 the crystal with larger unit-cell
volume compresses more at high pressure than the crystal with smaller unit-cell volume.
In contrast, Mg2SiO4, with a smaller unit-cell volume compared with those of
Fe2SiO4 and Co2SiO4, shows greater compressibility at high pressures than the latter. This
is contrary to the bulk modulus –volume relationship (Fig. 6.13). The distinct electronic
configuration of Mg2þ compared with those of the transition-metal cations Mn2þ, Fe2þ and
Co2þ, results in different compression behaviours for Mg2SiO4 vis-à-vis (Mn, Fe,
Co)2SiO4.

6.3.5. Post-spinel transitions: phase-boundary study

Irifune et al. (1998) used a double-stage multi-anvil system, with eight sintered
diamond cubes as second-stage anvils, to observe the ilmenite – perovskite phase boundary
in MgSiO3 and the post-spinel phase boundary at pressures up to 28 GPa (Kato et al.,
1997). Using the new generation synchrotron radiation facility, Irifune et al. (1998)
developed a cell design to define the post-spinel phase boundary in Mg2SiO4 using an in
situ XRD technique. The sintered magnesia was used as the pressure medium and
pyrophyllite as the gasket material. Tungsten – carbide cubes with a truncated edge length
(TEL) of 3 mm were used as second-stage anvils.
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 539

Figure 6.13. Bulk modulus–volume relationship between (Mn, Fe, Co)2SiO4. The Mg2SiO4 falls off the line of
this relationship (from Zhang, 1998, q 1998 Springer-Verlag).

The phase boundary between spinel (g-phase) and MgSiO3 perovskite þ MgO
periclase was determined at temperatures between 1,400 and 1,8008C. The phase boundary
was seen to have a negative Clapeyron slope (dT/dP), estimated by quench experiments
and thermodynamics and located at ,21 GPa at temperatures (,1,6008C) corresponding to
the bottom of the mantle-transition region. This pressure seems to be lower than what was
earlier calculated for the 660-km (^ 30 km) seismic-discontinuity boundary, lying
between 22.5 and 24.5 GPa (Shearer and Masters, 1992). This discrepancy possibly arises
because the post-spinel phase boundary shifts toward higher pressures because of its
negative Clapeyron slope and also because of the presence of ferrous iron (about 10%)
replacing Mg in mantle olivine. Al2O3, CaO, Fe2O3 and also H2O might affect the post-
spinel transformations in the mantle.

6.3.6. OH2 ions

Young et al. (1993) showed that substantial amounts of OH can be incorporated in

the a- and b-phases of olivine at mantle pressures. The presence of even a small amount of
water can substantially influence the physical properties of silicate rocks. The creep
540 Chapter 6

strengths of olivine crystals and olivine-bearing rocks decrease by a factor of 2– 3 in the

presence of water (Mackwell et al., 1985; Karto et al., 1986).
Also in the presence of water, hydrogen ions may be introduced into the structure of
olivine, enhancing significantly the electrical conductivity of olivine (Karato, 1990). In
addition, point defects generated on the Si, O and metal sublattices, when hydrogen or
hydroxyl are incorporated into the olivine structure, should markedly influence the ionic
diffusivities and related kinetic properties.
Experiments at pressures between 50 and 300 MPa demonstrate that the
hydroxyl concentration increases systematically with increasing water fugacity (Bai
and Kohlstedt, 1992). Kohlstedt et al. (1996) estimated the solubility of water in the
a-, b- and g-phases of (Mg, Fe)2SiO4 under varying pressures at constant temperature
of 1,1008C. They observed that the hydroxyl stability increases systematically with
increasing confining pressure, reaching values as stated in the following stability
fields.
Stability field H/106 Si content Pressure (GPa)
a ,20,000 (<1,200 wt ppm H2O) 13
b ,400,000 (<24,000 wt ppm H2O) 14 –15
g ,450,000 (,27,000 wt ppm H2O) 19.5

All the observations about the dependence of hydroxyl solubility in a-olivine on

water fugacity have been summarized by Kohlstedt et al. (1996) as

COH ¼ AðTÞfHn2 O expð2PDV=RTÞ

where COH is the molar concentration of hydroxyl ion, AðTÞ ¼ 1:1 H=106 Si/MPa at
1,1008C, n ¼ 1 and DV ¼ 10.6 £ 1026 m3/mol.
These data demonstrate that the entire present-day water content of the upper
mantle could be interpreted as being present in olivine alone. They therefore concluded
that a free hydrous fluid phase cannot be stable in those regions of the upper mantle with a
normal concentration of hydrogen. Free hydrous fluids are restricted to special tectonic
environments, such as the mantle wedge above a subduction zone.

6.3.6.1. OH-bearing planar defects

Most of the olivine crystals obtained from kimberlite show dislocation
microstructures and no planar defects, whereas those parts of the crystals containing
planar defects are devoid of any dislocation. The defects may be surmised as having their
origin in (i) precipitation of OH-bearing monolayers from “hydrous olivine” or
(ii) hydration of olivine within the stability field of humite family minerals. These are
formed during the deformation process when in the presence of sufficient water glide, or
partial dislocation (generating planar defects) is energetically preferred as a mechanism of
deformation.
Green crystals of olivine obtained from kimberlite in Buell Park, Arizona, show
planar OH defects. The displacement sector is associated with the defects ðR ¼ 14 k011lÞ:
Infrared absorption of an OH-bearing monolayer within the olivine is seen, as noted in the
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 541

humite family of minerals (Kitamura et al., 1987). The 3,610, 3,598, 3,477 and 3,230 cm21
IR bands seen in olivine are presumed to be due to OH defects in the lattice. In the presence
of water, clinohumite and chondrodite, among the humite family minerals, are observed to
be stable under the pressure and temperature conditions of the upper mantle.

Hydroxyl concentration: Bi and values. To determine the concentration of hydroxyl

groups in a mineral, the IR spectra may be integrated from ,3,780 to 2,950 cm21, the
region dominated by the stretching vibrations due to O –H bonds. The integration can be
carried out using the calibration of Paterson (1982):
ð
Bi HðnÞ
COH ¼ ð6-8Þ
150z ð3;780 2 nÞ

where COH is the molar concentration of hydroxyl ion, z is an orientation factor and HðnÞ is
the absorption coefficient in mm21 at wave number n in cm21. For calculations of OH
concentrated in the case of a-, b- and g-olivines, the values of Bi and z are taken as
(Kohlstedt et al., 1996).
Bi values (H/106 Si)
a-olivine 4.39 £ 104
b-olivine 4.08 £ 104
g-olivine 3.97 £ 104
z values
alpha;-olivine 1/2 (Mackwell and Kohlstedt, 1990)
b- and g-olivine 1/3

In a-olivine (synthetic), the peak at 3,567 cm21 increases systematically with

increasing confining pressure and with water fugacity.

Mechanism for OH2 incorporation. In the olivine structure, a number of possible sites
have been proposed for the incorporation of hydroxyl ions (e.g., Bai and Kohlstedt, 1993;
Wright and Catlow, 1994). Wright and Catlow (1994) obtained a relatively low energy of
solution when OH is incorporated via a reaction between olivine and water in which Fe3þ
is reduced to Fe2þ ion:

2Fe0Me þ 2OXO þ H2 OðflÞ ! 2FeXMe þ 2ðOHÞzO þ 1=2O2 ðflÞ ð6-9Þ

where the Kröger – Vink defect notation has been used to specify species, site and charge of
defects in the olivine structure and the notation (fl) indicates a component in the hydrous
fluids phase.
At high temperature, the charge neutrality is affected and creep strength
significantly affected by the presence of water (Karato et al., 1986). Measurements on
the solubility of hydroxyl in olivine as a function of water fugacity indicated that the
0
interstitial defect (OH)j may also be an important site for the incorporation of hydroxyl into
the olivine structure (Bai and Kohlstedt, 1993). The defect incorporation involves
542 Chapter 6

the reaction:

OXO þ H2 O þ VXi ! ðOHÞzO þ ðOHÞ0j ð6-10Þ

For the charge neutrality, the two defects on the right-hand side of reaction (6-10) should
bear the equality
1=2
½ðOHÞzO  ¼ ½ðOHÞ0j  / fH2 O ð6-11Þ

Based on equation (6-11), the defect concentration in the reaction

ðOHÞzO þ ðOHÞ0j ! {ðOHÞzO þ ðOHÞ0j }x ð6-12Þ

will be dependent an the water fugacity as

1=2
{ðOHÞzO 2 ðOHÞ0j } / fH2 O ð6-13Þ

A water-fugacity exponent greater than 0.5 would suggest a significant fraction of

hydroxyl to be associated with other charged defects, as shown in equation (6-13).

6.3.7. Inter-diffusion and activation volume, V p

Misener (1973) studied the Mg – Fe inter-diffusion coefficient in olivine up to

3.5 GPa and measured the activation volume, V p ¼ 5.5 cm3/mol. Olivine at 2 GPa was
studied by Karato and Ogawa (1982), who determined the activation volume V p to be
14 cm3/mol.
The kinetics of dislocation recovery in olivine have been studied up to 10 GPa (and
1,5008C in a multi-anvil) by Karato et al. (1993). Dislocations in the (010) [100] slip
system appeared as high-temperature deformation (at room pressure). Karato et al.
statically annealed the specimens at high pressures and temperatures. The pressure
dependence of dislocation-recovery kinetics is shown later in Fig. 15.9 (Karato et al.,
1993) with the activation volume calculated assuming that it is independent of pressure.
The estimated value of the activation volume is small, V p ¼ 6 ^ 1 cm3/mol, possibly
arising from an interstitial mechanism of diffusion of the relevant transport limiting
species, such as oxygen or silicon. In general, the ionic diffusion for the olivine lattice has a
small activation volume and should not show much depth dependence in the upper mantle.
For further discussion, see Section 15.2.

6.3.8. Seismic and acoustic velocities: VP and VS

The velocity data for Fe-bearing wadsleyite are compared with those from Mg end-
members (Fig. 6.14). To date, acoustic measurements on Fe-bearing wadsleyite phases
have only been studied at ambient conditions using Brillouin scattering techniques
(Simogeikin et al., 1998).
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 543

Figure 6.14. Comparison of the pressure behaviour of the velocity data (filled squares) for Fe-bearing wadsleyite
with those for the Mg end-member (filled diamonds and lines from Li et al., 1996a) (from Sinogeikin et al., 1998,
q 1998 American Geophysical Union).

6.3.8.1. a – b system
Webb (1989) found that the P- and S-wave velocity contrasts between a- and
b-phases of Mg2SiO4 by ,13%. However, near 410 km, the velocity contrasts are
generally in the range of 4 –5% for compressional waves and 4 –4.6% for shear waves
(Mechie et al., 1993; Nolet et al., 1994). The velocity contrast between the a- and b-
phases of Mg2SiO4 is largely independent of pressure at ambient temperature (Duffy
et al., 1995).
Duffy et al. (1995) plotted sound velocities along 1,2508C adiabat in contrast to the
seismic-velocity profiles (Mechie et al., 1993) for the upper mantle. The acoustic velocity
jump in the a – b system corresponds to olivine fractions of 32 and 27% for compressional
(P) and shear (S) waves, respectively.
Recently, on both olivine (Zha et al., 1996) and wadsleyite (Gwanmesia and
Libermann, 1996), acoustic measurements have been made at temperatures up to 1,700 K
(/RP) and above 10 GPa pressures (RT ). Li et al. (1998) made acoustic-velocity
measurements and XRD studies on wadsleyite to 7 GPa and 873 K. The experimental
measurements of dK/dT and dG/dT for wadsleyite and other data for olivine (Mg, Fe)2SiO4
were used by Li et al. (1998) to calculate the elastic-wave velocities of olivine and
wadsleyite at the P, T conditions of the 410-km discontinuity. A calculation was
performed along a 1,673-km adiabat with the Eulerian finite-strain approach (Duffy and
Anderson, 1984).
544 Chapter 6

The ultrasonic measurements of elastic-wave velocities for a wadsleyite phase

with composition (Mg0.88FeO0.12)2SiO4 to 9.6 GPa at RT has been determined by Li
and Liebermann (2000) in a uniaxial split-cylinder apparatus. Bench-top velocity
measurements yielded the following values:

VP ¼ 9:33ð3Þ km=s; VS ¼ 5:43ð2Þ km=s; K ¼ 172ð2Þ GPa;

G ¼ 106ð1Þ GPa; K 050 ¼ 4:6ð1Þ; G00 ¼ 1:5ð1Þ:

P and S velocity contrast. The P- and S-wave velocities and the associated elastic moduli
were calculated for all pressure and temperature conditions. The velocity contrast between
the a and b polymorphs is 12.3 and 14.2% at ambient pressure for compressional (P) and
shear (S) waves, respectively and decreases to 9.8% for P-waves and 12.4% for S-waves
at the pressure of the 410-km discontinuity (,13.8 GPa). In comparison, the velocity
seismic contrast across the discontinuity is 4–5% although a recent study using stacks
of long period records obtained a S-velocity jump at the discontinuity of only 3–4%
(Shearer, 1996).
At 13.8 GPa, the aggregate compressional- and shear-wave velocities of the
forsterite (Mg2SiO4) are 2.7 ^ 0.7% lower than predicted from low-P data. The velocity
contrast between these a and b (ðVb 2 Va Þ=Va reaches 11.7 and 12.1% at 410-km depth
for P- and S-waves, respectively. To account for the seismic jumps of 4.6%, an olivine
content of 38 –39% for the upper mantle can be postulated. If dKs/dP and dG/dP of the
b-phase are assumed to be equivalent to a-phase values (4.2 and 1.4, respectively), the
allowed mantle olivine content increases to ,52% (for dG/dT ¼ 2 0.024 GPa/K).
Assuming a homogeneous mantle composition, the seismic velocity contrast/discontinuity
at 410-km depth can be satisfied only by a mantle containing , ,40% olivine, which is
well below the abundance assumed in peridotite-based upper-mantle models. These results
are consistent with an olivine fraction of 30– 50% in the upper mantle.
In the pyrolite composition, the jumps in the velocities of P- and S-waves,
associated with the a (Mg number ¼ 92.5) to b (Mg number ¼ 88.5) transformation can
be calculated from a mineral physics database (Duffy and Anderson, 1989). The values
calculated at ambient conditions (DVp ¼ 11.6% and DVs ¼ 12.1%) are found to be
smaller by about 10% than those expected in the isochemical transformation of olivine
with Mg number ,89 (DVp ¼ 12.8% and DVs ¼ 13.8%) (Irifune and Isshiki, 1998).
However, high-pressure studies on a- and b-phases have shown that the velocity jumps
decrease with increasing pressure. Mineral wave speed varies with composition; in isolated
Fo89, it is seen to increase across the a – b transition. This is discussed further in later
sections.
The calculated sound velocities in the (Mg, Fe)2SiO4 polymorphs at high
temperature and seismic velocity profiles for the upper mantle are compared in Fig. 6.15.
The high-temperature mineral velocities are calculated along a 1,600-K adiabat using the
method of Duffy et al. (1995). Figure 6.16 (read the caption) depicts that velocity increase
at 410 km under mantle conditions is consistent with olivine fractions of 30% (for shear
waves) and 40% (for compressional waves). For b-Mg2SiO4, the temperature derivative of
the shear modulus (dG/dT) is unknown.
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 545

Figure 6.15. Acoustic velocities in the a-, b-, and g-phase of (Mg0.9Fe0.1)2SiO4 along a 1,600 K adiabat
(solid lines) compared with seismic velocity profiles for the upper mantel (dashed lines). The lower solid curve
(at depths .410 km) is for ›G=›T ¼ 20:024GPa=K: The upper solid curve is for ›G=›T ¼ 20:014GPa=K:
The seismic models are from Grand and Helmberger (1984), Walck et al., (1984), and Mechie et al. (1993)
(Source: Zha et al., 1998c, q 1998 American Geophysical Union).

6.3.8.2. Fe/Mg in velocity relation

Since iron increases the density, the seismic velocities (P- and S-waves) of iron-rich
olivine should be lower than those of pure fosteritic olivine.
A strong decrease in G values with fayalite (Fe2SiO4) content has been observed. At
300 K, the shear moduli of forsterite (Mg2SiO4) and fayalite are 82 and 51 GPa,

Figure 6.16. Compressional and shear velocity (aggregate) of wadsleyite and forsterite. Open and filled symbols
are experimental data, solid lines are third-order finite strain fits to the data (Zha et al., 1998c).
546 Chapter 6

respectively. Indeed, the fayalite content affects the shear modulus of olivine much more
than the bulk modulus. This provides a useful constraint on the composition of the mantle.
The relations for the P- and S-wave velocities,
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffi
KS þ 4=3G G
VP ¼ and VS ¼
r r

can be used as an indicator of Fe2SiO4 variations.

Since VS depends only on the shear modulus and density, it follows that an
increase in Fe2SiO4 content (which decreases G but not KS) in the mantle will cause
both VP and VS to decrease but will cause an increase in VP/VS ratio. For example, if
olivine is pure Fe2SiO4 rather than pure Mg2SiO4, the VP/VS ratio becomes ,15%
higher.

6.3.9. Minor element partitioning in a ! b transformation

The partitioning of Fe, Mg, Ni, Cr, Al, Ti, Ca and Na, between co-existing olivine
and wadsleyite has been determined by Gudfinnsson and Wood (1998). Cation exchange is
generated by recrystallization. Other than Ca, all the elements are concentrated in
wadsleyite relative to olivine. These elements show greater compatibility in wadsleyite
relative to olivine in the order Ni , Na , Cr , Ti , Al. Weight-partition coefficients
Dwad-ol
tr (¼ wt% trace element in wadsleyite/wt% trace element in olivine) is ,2 for Ni, 3
for Na and between 5 and 8 for Cr, Ti and Al. A strong negative correlation has been
observed between the Si and Cr(þAl) contents of wadsleyite, indicating a coupled
substitution of 2Cr3þ for Mg2þ and Si4þ.
Any significant concentration of minor elements, such as of Ni, Cr, Al and H, may
possibly alter the stability of olivine (a) and wadsleyite (b) phases and thus affect the depth
and thickness of the 410-km discontinuity (Wood et al., 1996). Wood (1995) argued that a
trace element that is preferentially incorporated into wadsleyite would tend to stabilize this
phase to lower pressure and would broaden the olivine– wadsleyite transformation
interval. Conversely, if a trace element prefers olivine to wadsleyite, it would tend to
expand the stability field of olivine to a higher pressure (Gudfinnsson and Wood, 1998).
For example, high concentrations of FeTiO3 and chromite precipitates are seen in olivine
of the Alpe Arami peridotite massif, Switzerland, which possibly crystallized initially
wadsleyite phase with enough Ti and Cr in its lattice within the transition zone
(Dobrzhinetskaya et al., 1996).
In their experiments, Gudfinnsson and Wood (1998) assumed that the trace-element
content of olivine at 410 km is transferred progressively to wadsleyite through
the transition interval without the involvement of other phases, i.e., it is unbuffered.
But elements such as Al, Cr and Ti tend to concentrate in garnet, which tends to fix or
buffer the content of olivine and wadsleyite at constant values. Ti replaces Si at tetrahedral-
sites instead of Mg at octahedral-sites. The results of Gudfinnsson and Wood (1998)
demonstrate that the maximum concentrations of TiO2 and Cr2O3 that can dissolve
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 547

in wadsleyite are of the order of 0.6 and 2.0 wt%, respectively, at 14 GPa and 1,873 K
(Fig. 6.17).
The olivine of Alpe Arami is seen to contain about 1 vol% of exsolved FeTiO3
and 0.25% chromite lamellae. These, on the presumed homogenization in (Mg, Fe)2SiO4
solid solution, would suggest an original chemistry of about 0.5% TiO2 and 0.2% Cr2O3
in the solid solution. Olivine of this chemistry may be present even in fertile peridotite.
The relict porphyroclastic olivine in Alpe Arami massif might originally have been
wadsleyite.

6.3.9.1. Cr3þ and Al3þ in wadsleyite

The substitution mechanism of Cr3þ (or Al3þ) cannot always be determined
unequivocally from stoichiometry alone. The substitution of 2Cr3þ for 3Mg2þ can occur
with the creation of a magnesium vacancy:

3Mg2þ ¼ 2Cr3þ þ AHMg :

(Note: H enters wadsleyite structure where it is coupled with Mg-vacancy formation;

Inoue, 1994).
Because of the high CFSE of Cr3þ at the octahedral-site, Mg2þ replaces tetrahedral
4þ
Si in wadsleyite as

2½6 Mg2þ þ ½4 Si4þ ¼ ½4 Mg2þ þ 2½6 Cr3þ

This is analogous to that known in spinels. Gudfinnsson and Wood (1998) observed a
strong negative correlation between the Si and Cr(þ Al) contents of wadsleyite, indicating
a coupled substitution of 2Cr3þ for Mg2þ and Si4þ.

Figure 6.17. Maximum solubility of trace elements in olivine and wadsleyite observed in the experiments
by Gudfinnsson and Wood (1998), q 1998 Mineralogical Society of America.
548 Chapter 6

6.3.9.2. Ti4þ in olivine/wadsleyite

Olivines ((Mg, Fe)2SiO4 phases) in kimberlitic garnet (pyroxene solution in
garnet) lherzolite xenoliths, some of which originated as deep as 400 –500 km (Sautter,
1991), never show TiO2 contents more than 600 ppm. Trivalent cations are seen to be
highly soluble in wadsleyite (Dupas et al., 1994) and sufficient TiO2 can also be dissolved
in it (Green et al., 1998). Solubility of TiO2 may increase sufficiently at 12 GPa. The
phase which originally might have high solubility of TiO2 and trivalent cations, such as
Al3þ and Cr3þ, would possibly have been wadsleyite. Under pressure, wadsleyite and
ringwoodite show higher titanium solubility compared with olivine structure with low
solubility of Ti.
Ilmenite and olivine have the same basic hcp oxygen sublattice and thus many of
their corresponding d-spacings are similar. But on quenching, FeTiO3 of perovskite
structure develops LiNbO3 structure (Leinenweber et al., 1991).
At great depths with high pressure in a hydrous environment (at moderate
temperature), the solubility of TiO3 and trivalent cations in olivine may be significantly
greater than under dry conditions. Such samples could be formed by subduction to perhaps
150 km and could return to the surface.
Thus, in a mantle-wedge environment just above a downgoing slab, hydrous
fluids (melts?) emanating from the slab could metasomatize mantle peridotite and
perhaps significant TiO2 could be dissolved in olivine, along with elevated concentrations
of Al2O3 and Cr2O3. Subsequent dehydration of such material could precipitate FeTiO3 þ
chromite in the quantities observed in Alpe Arami massif (Dobrzinetskaya et al., 1996)
(see Section 5.4.3).

6.3.9.3. Cr2þ in olivine structure

Chromous ion (Cr2þ) has been reported from the diamonds exhumed from mantle-
derived magmatic bodies and from lunar rocks (e.g., Burns, 1975; Sutton et al., 1993). The
relationship of Cr2þ substitution within silicate phases can offer a useful monitor for
deciphering the lunar and terrestrial differential processes. Therefore, the distribution of
Cr2þ may serve as an important geochemical constraint for modelling the evolution of the
Earth’s mantle and its relationship to the Moon.
However, Cr2þ substitution in synthetic olivine and pyroxene is seen to be limited
to a maximum of 25 and 75 mol%, respectively (Li et al., 1995). At temperature below
1,4008C, attempts to synthesize Cr2þ-bearing orthosilicates resulted in the following
product only:

Cr2 O3 þ Cr metal þ SiO2 :

6.3.10. Partition coefficients: olivine –melt

Pressure effect on partition behaviour is important in the investigation of magma

genesis and the Earth’s evolution. Element partitioning between melt and high-pressure
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 549

magnesian silicates, such as olivine (Suzuki and Akaogi, 1995), majorite and silicate
perovskite (e.g., Kato, 1988, Drake et al., 1993), has been investigated. To determine the
equilibrium partition coefficients, the quantitative values of the cations of the crystals are
intrinsically important.
As early as 1968, the research of Onuma and his co-workers showed that, when
partition coefficients are plotted in a partition coefficient vs. ionic radius (PC – IR) diagram,
peaks corresponding to the most suitable size of cation (for the site of the crystal structure)
appear in the diagram. In olivine, the maximum of the PC is located near 70 pm,
corresponding to the ionic radius of Mg2þ (Matsui et al., 1977).
The partition coefficients between olivine and melt at upper-mantle conditions, for
3 –14 GPa, have been determined by Taura et al. (1998) for 27 trace elements: Li, Be, B,
Na, Mg, Al, Si, P, K, Ca, Sc, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Rb, Sr, Y, Zr, Cs, Ba, La and Ce
using secondary ion spectrometry (SIMS) and EPMA. The partition coefficients of
trivalent cations such as Cr, V, Sc, Y, etc., decrease with increase in pressure whereas those
of monovalent cations increase with pressure (Fig. 6.18).
The dependence of element partitioning on composition, temperature and
pressure between olivine and the co-existing melts has been studied for many divalent
cations such as Ni, Mg, Co, Fe, Mn and Ca (e.g., Jurewicz and Watson, 1988).
Partitioning of trivalent cations has also been determined by several workers, e.g.,
Beattie, 1994; Suzuki and Akaogi, 1995. Ozawa (1991) suggested that there is a
relatively large pressure dependence on elemental partitioning of monovalent and
trivalent cations between olivine and melt.
The partition coefficient of Al between olivine and silicate liquid, Dol/melt
Al , increases
with pressure up to 6 GPa (Agee and Walker, 1990) and that of Na also increases
with pressure up to 14.4 GPa, whereas Dol/melt Cr decreases with pressure (Ozawa, 1991)
ol=liq
(Fig. 6.18). The exchange partition coefficients, KDFe=Mg ; are in the range from 0.30 to 0.37
(Taura et al., 1998) (or 0.33 ^ 0.06; Zhang and Herzberg, 1996).
Note: Since the partition coefficient is represented by a concentration ratio of crystal
against melt, the coefficient for some major elements determined by SIMS might be
slightly smaller than those determined by EPMA because of the lower secondary ion yield
of olivine crystals as compared with quenched crystals.

6.3.10.1. Al3þ partitioning: Onuma diagram

The plots of partition coefficients against ionic radii, known as the “Onuma
diagram”, for the olivine –melt system at pressures between 3 and 14.4 GPa (Taura et al.,
1998) are shown in Fig. 6.19. The tetravalent cations, e.g., Si, Ti and Zr, make a smooth
curve for T-site occupation on Onuma diagrams. The curve drawn through the divalent
cations Ni, Mg, Co, Fe, Mn and Ca indicates an M-site parabola, although Ni, Co and Fe
deviate from the smooth curve for the M-site (Fig. 6.19). This deviation can be explained
as due to crystal-field effects (Matsui et al., 1977). The partition coefficients of Ni, Co and
Fe gradually decrease with pressure. The results of Taura et al. (1998) indicate that (Si4þ,
Mg2þ) ! (Al3þ, Cr3þ)-type substitutions are dominant in olivine at low pressure but
the substitution type changes to (Si4þ, Mg2þ) ! (Al3þ, Al3þ) and (Mg2þ, Mg2þ) !
(Naþ, Al3þ) with increase in pressure.
550 Chapter 6

Figure 6.18. Plots of olivine/melt partition coefficients vs. pressure. (a) monovalent cations and (b) trivalent cations
(after Taura et al., 1998, q 1998 Springer-Verlag).

When Al3þ occupies the T-site in olivine, Cr3þ may substitute to the adjacent
M2-site with a shared corner. Since the ionic radii of Al3þ . Si4þ and Mg2þ . Cr3þ, M2
octahedra with Cr3þ become compacted and T tetrahedron with Al3þ expands towards the
M2-site of Cr3þ. Thus, at lower pressure, the substitution (Si4þ, Mg2þ) ! (Al3þ, Cr3þ)
gets stabilized while, at higher pressure, (Si4þ, Mg2þ) ! (Naþ, Al3þ)-type substitution

Figure 6.19. Onuma diagrams of olivine/melt system at pressures from 3 to 14.4 GPa. Parabola-shaped lines are
connected among isovalent cations. Open square monovalent cations, solid circle divalent cations, solid triangle
trivalent cations, open diamond tetravalent cations, open triangle pentavalent cations (Taura et al., 1998, q 1998
Springer-Verlag).
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 551

dominates. Atomic proportions of Al, Cr and Na against (Al þ Cr þ Na) in olivine and in
melt are plotted in a ternary diagram. At high pressures, Al3þ tends to substitute in the
tetrahedral (T)-site; this substitution is accompanied by substitution of 3Al3þ and 2Naþ in
M-sites (see Figs. 6.20 and 6.21).
Under high pressure, the olivine structure with Al3þ shows a shortening of shared
edges. Positive correlation of the pressure dependence of Dol/melt
Al is due to incorporation of
Al3þ into the M-site of olivine.
The partition coefficient of Al increases with pressure while partition coefficients
of other trivalent cations such as Cr, V, Sc and Y decrease with pressure (Fig. 6.18).
The partition coefficients of Li and Na also increase with pressure.
Partitioning of Al, Na and Cr in olivine– melt is controlled by a substitution
mechanism into the olivine structure. The proportion of both Al and Na in olivine increases
with pressure whereas Cr decreases. The substitution of these cations can be expressed as

½AlM þ ½CrM þ ½NaM ¼ 2ð½Al3þ T þ ½Naþ Þ

where [x ] is the atomic fraction of x.

The abundance of AlT, AlM, Cr and Na as a function of pressure, shown in Fig. 6.21
(Taura et al., 1998), indicates that the amount of Al3þ in the T-site remains nearly constant
over the pressure range. The Al3þ content in M-sites increases significantly with pressure,
whereas the Cr3þ content in M-sites decreases. The Naþ content in M-sites also increases
with pressure.
For Chijiadian lherzolite, which experienced metamorphic pressures of at least
3.0 GPa (indicated by eclogite included within the lherzolite; Zhang et al., 1994), the
aluminium partitioning between enstatite and garnet indicates pressures of 4.1 ^ 0.6 GPa
and the distribution of calcium between olivine and diopside suggests equilibrium at
,4.4 GPa.

6.3.11. Compressibility and amorphization

Pressure-induced amorphization of olivine ((Mg, Fe)2 SiO4) has been

investigated at .50 GPa in a DAC through energy-dispersive XRD experiments
with synchrotron radiation (Andrault et al., 1995). The Birch –Muruagham EOS gives
the same room-P bulk modulus for all olivines, with KD ¼ 131 ^ 6 GPa (assuming
0
K 0 ¼ 4).
At higher P, the compression becomes nearly isotropic and the materials stiffen.
These changes could precede partial transformation of olivines to a high-P polymorph
related to the spinel structure. Instead of transforming to spinels, most of the material
undergoes P-induced amorphization, which takes place at considerably higher P for Mg
than for Fe-rich olivines.
However, no amorphous phase is observed in the high-temperature back-
transformation of b-Mg2SiO4 and g-Ni2SiO4 (Reynard et al., 1996). In such cases,
552 Chapter 6

Figure 6.20. Abundances of AlT, AlM, Cr and Na as a function of pressure. Site occupancy among Al3þ, Cr3þ, and
Naþ. (a) Projection on to the AlM –Cr–Na plane from AlT in the AlT –AlM –Cr–Na tetrahedron. Numbers in the
projection plot are ratios of AlT/(Na þ Al þ Cr). Solid line connects (AlT, AlM, Cr, Na) ¼ (0.5, 0, 0.5, 0.0) with
(AlT, AlM, Cr, Na) ¼ 0.17, 0.5, 0, 0.33). (b) Plot of Cr/(AlM þ Cr þ Na) vs. AlT/(Al þ Cr þ Na) (Taura et al.,
1998).

however, a two-step mechanism involving spinelloids (an olivine polymorph) is

inferred.
In olivine, the linear compressibilities decrease in the order b . c . a, thus
manifesting an anisotropy leading to a hexagonal symmetry at very high pressures. Beyond
this pressure, olivine amorphizes. Mg-olivine amorphizes at much higher pressures than
fayalite (Guyot and Reynard, 1992). However, evidence from Raman spectroscopy suggests
structural changes occur before amorphization (Reynard et al., 1994).
In Fig. 6.22a,b, the energy-dispersive X-ray diffraction patterns of the end-members
forsterite (Mg2SiO4) and fayalite (Fe2SiO4) show a decrease in the number of reflections,
an increase in linewidth of the peaks and a strong decrease in the intensity of the olivine
reflections at different pressures (Andrault et al., 1995). These features offer evidence for
progressive amorphization. TEM studies revealed formation of alternate lamellae of
crystalline and amorphous materials. The amorphization seems irreversible because of
very slow diffusion kinetics. No compositional changes have been noted during
amorphization.
The proportion of amorphous phase increases continuously with pressure. Fayalite,
however, directly amorphizes without going through an intermediate phase. TEM study of
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 553

Figure 6.21. Abundances of AlT, AlM, Cr and Na as a function of pressure (Taura et al., 1998, q Springer-
Verlag).

an olivine with 12% Fa has revealed that amorphization also takes place at the reported
pressure of 56 GPa (Jeanloz, 1977), but it is restricted to small zones of the sample.
Possibly the reason for this is that the temperature along the Hugoniot is high enough for
the kinetics of the formation of high-pressure crystalline phases to overcome the
competing amorphization.
In olivine, an increase in iron content leads to a compressibility decrease along the
c-axis and an increase along the b-axis. Kudoh and Takeuchi (1985) suggested that the
change with pressure of a, b and c unit-cell parameters of olivine would lead, at high
pressure, to a hexagonal close-packing lattice of oxygen. The a/b and a/c ratios of hcp
p p
lattices are 2/3 and 2̄/3̄. Deviations of (Mg, Fe)2SiO4 olivines from hcp symmetry
increase with Fe content.
The compressibility of polyhedra increases with their size. The differences in ionic
radii with rFe2þ ¼ 0.92 Å and rMg2þ ¼ 0.86 Å (Muller and Roy, 1974) would result in a
higher compressibility for iron-rich compounds. Since in olivines the octahedra are
interlinked along the c-axis, the compressibility along the c-axis is determined by that of
the (Mg, Fe)O6 octahedra, which are more compressible than SiO4 tetrahedra. However,
MgO6 octahedra show less compressibility than FeO6 octahedra. For (Mg, Fe)2SiO4
g-spinel (Hazen, 1993), the bulk modulus increases with the Fe content.
554 Chapter 6

Figure 6.22. Energy dispersive X-ray diffraction patterns of the end members (forsterite) (a) Mg2SiO4 and
(b) Fe2SiO4 (fayalite) at different pressures (D; during compression). Similar changes have been noted for the two
(Fe, Mg) olivines investigated (from Andrault et al., 1995, q 1995, Springer-Verlag).

6.4. b-Mg2SiO4 (wadsleyite)

6.4.1. Single-crystal elasticity

Both a- and b-phases of Mg2SiO4 are of orthorhombic symmetry and are thus
characterized by nine independent elastic moduli. Three of these, C11, C22 and C33, are the
longitudinal elastic moduli, C44, C55, C66 are the shear elastic moduli and the remaining
three (C12, C13 and C23) are the off-diagonal moduli.
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 555

The acoustic-velocity data are inverted by non-linear least squares using the
following equation (Rokhlin and Wang, 1992)
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 ffi
a 2p c þ 2p j
rVj2 ¼2 þ2 cos ; j ¼ 0; 1; 2; ð6-14Þ
3 3 3

described in detail in Section 4.4.3.7, to yield the elastic stiffness coefficients and three
Eularian angles, which relate the laboratory and crystallographic reference frames (Zha
et al., 1996).
Table 6.12 shows all the elastic constants at different pressures for wadsleyite.
Values for the aggregate moduli and their pressure derivatives are determined by
fitting the experimental data to finite-strain expressions for the bulk and shear moduli
0
(Davies and Dziewonski, 1975). For forsterite, the pressure derivatives of the bulk (K0) and
0
shear (G0) moduli are 4.2 ^ 0.2 and 1.4 ^ 0.1, respectively. For wadsleyite, the
corresponding pressure derivatives are 4.3 ^ 0.2 and 1.4 ^ 0.2.
The ambient pressure aggregate elastic moduli of wadsleyite obtained by Zha et al.
(1993) is comparable with those reported by Sawamoto et al. (1984) as K0 ¼ 174 GPa and
G0 ¼ 114 GPa.
For wadsleyite, Li et al. (1998) calculated the adiabatic bulk (K) and shear (G)
moduli and obtained the following values:

Kðat room conditionsÞ ¼ 172 ð2Þ GPa; dK=dP ¼ 4:2ð1Þ;

dK=dT ¼ 20:012 ð1Þ Gpa=K; GðRTÞ ¼ 113 ð1Þ GPa; dG=dP ¼ 1:5 ð1Þ;

dG=dT ¼ 20:017 ð1Þ GPa=K

The parameters for the thermal EOS of b-(Mg0.84Fe0.16)SiO4 are presented in

Table 6.13.
The high solubility of water in the b-phase suggests that water should be strongly
enriched in the transition zone (Young et al., 1993). Extrapolation of the thermodynamic
model for the solubility of water in olivine into the b-stability field yields a partition
coefficient for water between the b- and a-phases of D b/a ¼ 20 (Kohlstedt et al., 1996).
However, Young et al. (1993) obtained a value D b/a ¼ 40 from samples containing both
a and b. Considering the width of the transition zone to be constrained to less than 10 km,
the H2O in olivine near the 410-km discontinuity has been estimated by Wood (1995) to be
,3,400 H/106 Si (using D b/a ¼ 10). From these observations, it has been concluded by
Kohlstedt et al. (1996) that, away from regions of partial melting, the upper mantle is
,10 –20% saturated with water content.
Given such observations of high-water solubility in olivine and pyroxene phases,
added to the data for stishovite and perovskite (Lu et al., 1994; Meade et al., 1994), it can
be concluded that the mantle has a capability to store water several-fold the total of the
hydrosphere.
 556
TABLE 6.12
Elastic moduli of wadsleyite (Zha et al., 1998) (all in GPa)

P C11 C22 C33 C44 C55 C66 C12 C13 C23

0.0 370.47 ^ 7.84 367.74 ^ 6.50 272.43 ^ 5,82 111.20 ^ 3.58 122.48 ^ 4.00 103.05 ^ 3.86 65.59 ^ 4.54 95.20 ^ 5.18 105.14 ^ 4.36
3.1 379.28 ^ 11.46 382.01 ^ 10.40 292.21 ^ 4.46 111.24 ^ 2.16 122.71 ^ 1.84 100.70 ^ 6.10 85.39 ^ 6.62 105.53 ^ 5.44 112.39 ^ 5.28
6.3 393.35 ^ 7.90 399.90 ^ 12.30 316.69 ^ 6.06 126.59 ^ 3.68 117.53 ^ 2.58 116.96 ^ 5.76 83.91 ^ 6.14 118.23 ^ 4.62 119.10 ^ 4.98
8.1 404.22 ^ 11.22 418.33 ^ 16.78 323.99 ^ 5.42 121.37 ^ 6.36 123.58 ^ 6.74 118.56 ^ 6.98 94.42 ^ 6.82 122.31 ^ 5.48 123.37 ^ 6.70
9.4 414.42 ^ 11.16 439.78 ^ 13.04 333.08 ^ 5.67 127.13 ^ 5.32 122.77 ^ 3.80 118.43 ^ 6.40 103.89 ^ 5.90 123.78 ^ 6.89 131.81 ^ 5.92
10.1 428.83 ^ 11.30 415.97 ^ 9.86 333.91 ^ 7.90 117.55 ^ 2.22 128.19 ^ 3.08 121.09 ^ 2.40 104.48 ^ 4.10 130.21 ^ 6.90 110.77 ^ 5.80
10.5 421.89 ^ 6.72 425.66 ^ 14.22 330.73 ^ 7.70 121.65 ^ 6.76 127.21 ^ 4.32 116.16 ^ 3.32 107.78 ^ 4.38 127.90 ^ 5.42 125.39 ^ 5.98
14.2 444.47 ^ 15.86 464.89 ^ 16.13 386.83 ^ 11.30 130.94 ^ 3.72 121.67 ^ 2,76 129.71 ^ 3.30 123.93 ^ 4.56 141.95 ^ 2.18 151.85 ^ 19.60

Chapter 6
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 557

TABLE 6.13
Parameters of the thermal EOS of b-phase (Fei et al., 1992)

Parameters b-(Mg0.84Fe0.16)2SiO4
3
VP (cm /mol) 40.677
ba (TPa21) 2.24 (^0.05)
bb (TPa21) 1.74 (^0.06)
bg (TPa21) 1.81 (^0.06)
KTO (GPa) 174 (^3)
K0TO 4.0
(dKT/dT)P (GPa/K) 22.7 (^0.3) £ 1022
dT 5.1 (^0.8)

6.4.2. Hydrous wadsleyite, b-Mg22xSiH2xO4 (0.00 # x # 0.25)

Wadsleyite II is a hydrous spinelloid similar to wadsleyite, but with a b-axis 2.5

times greater. Some of the hydrous wadsleyite structure is monoclinic rather than
orthorhombic. Wadsleyite and wadsleyite II cannot be easily distinguished by powder
XRD and are similar optically. There is probably complete solid solution between hydrous
and anhydrous wadsleyite, at least in the composition range of 94.6– 100% Fo content. Si
vacancy in this structure is significant. The b-angle changes to 90.3978(9).
Inoue (1994) synthesized wadsleyite with a significant amount of H2O (3.1 wt%
determined by SIMS). The crystal structure of monoclinic hydrous wadsleyite is shown in
Fig. 6.24. Smyth et al. (1997) found a modification of hydrous wadsleyite with SG I2/m
(a sub-group of Imma). A continuous solid solution between hydrous and anhydrous
wadsleyites is suggested (Kudoh et al., 1996).
The deviation from orthorhombic symmetry is apparently caused by the ordering of
vacancy and divalent cations into two non-equivalent M3-sites (Smyth et al., 1997). The
presence of a curious extra Si-site is postulated to be the result of decompression of an
Mg –Si octahedron (Fig. 6.24), with the Si moving to an adjacent tetrahedral void on
release of pressure. The deviation from orthorhombic symmetry is the result of Si ordering
into one of the two M3-sites. This deviation should increase with pressure as Si moves
toward octahedral coordination with pressure.
From the comparison of structures, it is found that Mg-vacant structural modules
make the building units for the structure of hydrous wadsleyite. Three types of structural
modules are invoked in the hydrous wadsleyite structure (Fig. 6.25a –c) wherein OM
denotes a M3-site fully occupied by Mg, while VM2 ¼ two out of eight Mg3 positions are
vacant and VM1 ¼ one out of eight Mg3 position is vacant. A combination of these three
modules results in the variance of the hydrogen content. The dilution of symmetry from
orthorhombic to monoclinic in the hydrous wadsleyite structure possibly arises from the
lack of a mirror perpendicular to the a-axis of the module. The mode of these Mg-vacant
structural modules determines the relationship between hydrous wadsleyite and hydrous
ringwoodite (Kudoh and Inoue, 1999). The bond lengths and bond valences in hydrous
wadsleyite Mg1.75SiH0.5O4 are presented in Table 6.14.
558 Chapter 6

Figure 6.23. Schematic crystal structures for orthorhombic Mg wadsleyite (b phase) and cubic ringwoodite
(g phase). A single unit cell of the phase is drawn above a comparable block of g phase that has a complex
relationship to its unit cell edges. If the central half of the b phase unit cell (viz., that part within the indicator
markers above and below the drawing) were translated by 12 ½ac relative to the outer parts and the bonds were
reformed, the resulting structure would closely resemble that of the of g phase. On that basis, the b ! g transition
in (Mg, Fe)2SiO4 is thought likely to be a martensite-type shear (adapted from Morimoto et al., 1970).

Wadsleyite in the transition zone is shown to contain as much as 10,000 –

65,000 H/106 Si (Young et al., 1993) and is considered to be the major sink for protons in
the mantle. Smyth (1994) attempted to model the protonation with OH group formation in
the b-phase, wherein the O1-site is presumed to be energetically most favourable for
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 559

Figure 6.24. Crystal structure of monoclinic hydrous wadsleyite (b is horizontal), projected approximately down
[001]. The H position (labelled) is directly above O1, the oxygen shared by the four M3 cations. In the monoclinic
structure, there are two no-equivalent M3 sites labelled M3a and M3b. An extra Si position with about 3%
occupancy, shown with heavy lines as bonds to adjacent O positions, was observed adjacent to the M3 octahedra
and is postulated to be the result of partial Si occupancy of M3a at high pressure (Smyth et al., 1997, q 1997
Mineralogical Society of America).

protonation. Earlier, Downs (1989) found that, in addition to the minimum at the O1-site,
the electrostatic potential was at a minimum close to the O2-site, suggesting that
protonation could also occur at this second-site. Young et al. (1993) also observed two
(OH)-sites by spectral studies.
Wright and Catlow (1996), through calculations of the substitution energy of OH
for O at each of four-sites, found that the O1-site has the lowest energy, 15.77 eV, whilst

Figure 6.25. (a)–(c) Structural modules for hydrous wadsleyite. Occupied Mg3 site is shown by solid octahedron
and vacant Mg3 site by open octahedron. [OM ¼ Mg3 site is fully occupied by Mg, VM2 ¼ two out of eight
Mg3 positions are vacant. VM1 ¼ one out of eight Mg3 positions is vacant (Kudoh and Inoue, 1999, q 1999
Springer-Verlag).
560 Chapter 6

TABLE 6.14
Bond lengths (Å) and estimated bond valences (V ) in hydrous wadsleyite, Mg1.75SiH0.5O4

Anions Mg1 (97% Mg) Mg2 (94% Mg) Mg3 (78% Mg) Si (100% Si) V Anion chemistry
v
O1 2.14 (1) 2.065 (4)
0.28 0.28 1.4 OH
iii
O2 2.067 (9) 1.718 (5)
0.34 0.82 2.0 O
iv
O3 2.127 (8) 2.107 (6) 1.617 (7)
0.29 0.25 1.09 1.9 O
iv
O4 2.058 (5) 2.082 (5) 2.105 (4) 1.633 (4)
0.34 0.31 0.25 1.04 1.9 O
Average 2.081 (6) 2.089 (7) 2.092 (5) 1.650 (5)

the other three-sites have similar energies which are higher by ,1.5 eV. Thus, they agreed
with Smyth (1994) that the O1-site will complete most effectively with protons.

6.4.2.1. H2O in a – b transition

Wadsleyite can contain up to ,3 wt% of H2O (Inoue et al., 1995), which can induce
hydrolytic weakening of wadsleyite to cause it deform plastically. Kubo et al. (1998)
studied the effects of water on the a – b transformation kinetics in olivine. In their
experiments at 13.5 GPa/1,0308C, the growth of wadsleyite at the expense of olivine was
seen to be enhanced by the presence of water.
Wadsleyite has a lower dislocation density in wet runs than in dry-run phases. In
wadsleyite, water enhances the dislocation recovery and causes inelastic relaxation of the
localized pressure drop (due to transformation). This results in an increase in the growth
rate in wet runs.

OH2 solubility. When H2O is present, the stability field of high-pressure b-phase expands
into both the a-olivine and g-spinel fields (e.g., Lu et al., 1996). b-phase predominantly
occurs at depths between 400 and 520 km and is observed to take up as much as 3.1 wt%
H2O (e.g., Cynn and Hofmeister, 1994; Inoue et al., 1995).
b-(Mg, Fe)2SiO4 contains Si2O7 units, unlike the isolated SiO4 tetrahedra present in
olivine and g-spinel. The oxygen sites are “under-bonded” to allow protonation (Smyth,
1987). X-ray diffraction studies of hydrous b-(Mg, Fe)2SiO4 proved two structural types:
wadsleyite I and II (Smyth et al., 1997).
The difference in water solubility between b-phase and the anhydrous polymorph
could cause a release of water from material, convectively traversing the 410-km
discontinuity and potentially resulting in melting and plume-like upwellings (Young et al.,
1993). The dissolution of water in b-phase is strongly temperature dependent.
McMillan et al. (1991) synthesized a sample of b-Mg2SiO4 with ,9,000 H/106 Si
at 14.5 GPa/8508C. Later, Young et al. (1993) produced b-(Mg, Fe)2SiO4 with a greater
OH-content, ,65,000 H/106 Si.
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 561

Young et al. (1993) and Gasparik (1993) observed major amounts of OH in Fo92
wadsleyite with up to 1.6% of all O atoms protonated. Kohlstedt et al. (1996) reported up to
2.4 wt% H2O in wadsleyite (b-phase), 2.7 wt% in g-phase but only 0.12 wt% in a-phase
(olivine).
Prepared under hydrothermal conditions (Kohlstedt et al., 1996), wadsleyite shows
a group of large absorption peaks near 3,300 cm21 and a second group of smaller peaks
near 3,600 cm21 (Fig. 6.26). The OH content in the b-phase is ,400,000 H/106 Si. In the
experiments, high-pressure (15 GPa) samples show higher water fugacity.
The O – H stretching bands in micro-Raman (Fig. 6.27a) and micro-IR (Fig. 6.27b)
spectra are seen to occur in the regions 3,000 –4,000 and 4,400 – 2,800 cm21, respectively
(Kndoh et al., 1996).

Protonation-site. Smyth (1987) suggested that the O(1) structural-site is the most likely
oxygen position for protonation, while Downs (1989) favoured the O(2)- and also the O(1)-
site. Later, Smyth (1994) proposed a maximum OH content of 500,000 H/106 Si
(33,000 wt ppm) and a composition of Mg7Si4O14(OH)2 for fully hydrated b-phase.
Cynn and Hofmeister (1994) argued that the infrared absorption peak near 3,330 cm21 is
associated with protonation of the O(2)-site and the absorption bands near 3,600 cm21 are
due to protonation of the O(1)-site.
0
The positively charged (OH)O defect is charge compensated by two negatively
0
charged point defects, V00me and FeSi.

Figure 6.26. Unpolarized FTIR spectra from three samples of b phase produced under hydrothermal conditions.
The lower curve is from a sample prepared at 14 GPa, 1,1008C: the upper two curves are from samples synthesized
15 GPa, 1,1008C. The sinusoidal variations at wavenumbers below 3,200 cm21 interference fringes. Two distinct
groups of absorption bands are present. The wave numbers of some absorption peaks are labelled (Kohlstedt et al.,
1996, q Springer-Verlag).
562 Chapter 6

Figure 6.27. (a) Micro-Raman spectrum for a single crystal of hydrous wadsleyite in the region 3,000–
4,000 cm21 and showing the features due to O –H stretching. (b) Micro-IR spectrum for a single crystal of
hydrous wadsleyite in the regions 3,000–4,000 cm21 and 4,400–2,800 cm21, showing the features due to O–H
stretching (Kudoh et al., 1996, q 1996 Springer-Verlag).

Again, there is a reaction of water with an O vacancy to form the two OH groups:
V 000 þ H2 O þ Ox0 ! 2ðOHÞz0
where (OH)z denotes a hydroxyl group at an O-site. The energy for this is given by
DE2 ¼ 2EOH 2 EV0 þ EPT
where EOH is the energy needed to form an OH group at an O1-site and EPT is the energy of
the hypothetical gas-phase proton-transfer reaction (Wright et al., 1994).

6.4.2.2. Mg-vacant structural module

The mode of distribution of vacancies at the Mg3-site has been observed by Kudoh
and Inoue (1999) to correlate with the symmetry lowering (e.g., dilution of symmetry) of
the hydrous wadsleyite as a function of hydrogen content. (Note: The Si atom can occupy
the Si2-site only when the adjacent Mg3-site is vacant. This configuration avoids face
sharing between the SiO4 tetrahedron and MgO6 octahedron.) They found that the value of
occupancies of Si atoms at the Si2-site (11% for 0.3H-b and 7% for 0.5H-b) estimated
from peak heights in the different Fourier maps are consistent with the vacancy values at
the Mg3-site. One (in 0.3H-b) or two (in 0.5H-b) out of eight Mg3 positions in the unit cell
are seen as vacant. The existence of a Si atom at the Si2-site indicates new structural
modules (Fig. 6.25(a)– (c)). When there is no Mg –Si disorder or no Si vacancy in the
wadsleyite structure, maximum hydrogen content is constrained by the number of O1
oxygens in the unit cell and the formula with maximum hydrogen content is
Mg1.75SiH0.50O4 (Kudoh et al., 1996). A combination of structural modules gives the
variety of hydrogen content of hydrous wadsleyite. The commonly observed crystal
structures of hydrous wadsleyite are the average of structural modules randomly
distributed.
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 563

6.4.2.3. Fe in wadsleyite II
Experiments by Smyth and Kawamoto (1997) at higher Fe contents (Mg/(Mg þ
Fe) ¼ 0.822) at 17.5 ^ 1 and 18.0 ^ 1 GPa produced wadsleyite II with Mg/(Mg þ Fe)
ratios of 0.906 and 0.881, respectively. Finger et al. (1993) observed a preference of Fe for
M1 and M3 relative to M2 in Fe-bearing anhydrous wadsleyite. In the monoclinic
structure, there are two non-equivalent M3-sites that differ in their apparent scattering by
15s and at both M3-sites there is significant-site vacancy.
Smyth et al. (1997) have shown that cation vacancies at Si-sites as well as at M3 are
due to a major charge-compensation mechanism that allows wadsleyite to incorporate
variable amounts of H. Because the Mg/(Mg þ Fe) ratios in the transition zone are about
0.9, the hydrous wadsleyite is likely to have the wadsleyite II structure rather than the
monoclinic structure. Nevertheless, complex hydrous spinelloids appear to be the possible
constituents of the transition zone (Smyth et al., 1997).

6.4.2.4. Fe3þ in protonation

The solubility of water in wadsleyite is seen to be enhanced by the presence of ferric
iron. Wright and Catlow (1994) showed that low-energy redox reactions involving ferric
iron could affect water incorporation. They calculated an energy of 0.5 eV (47 kJ/mol) for
the following reactions:
1
2FezMg þ 2O0 þ H2 O ! 2FeMg þ 2ðOHÞz0 þ O ð6-15Þ
2 2
This reaction may break down as (defect notations from Kröger, 1972):
1
2FezMg ! 2FexMg þ Vzz0 þ O ð6-16Þ
2 2ðgÞ
where FezMg and FexMg are the ferric and ferrous ion, respectively, at Mg-sites and Vzz0
denotes the vacancy at oxygen-sites with two charges. The energy of the reactions is
determined by summing the energies of substitution and vacancy and half the dissociation
energy of an oxygen molecule. This gives E ¼ 1.64 eV.
Results of Mössbauer spectroscopic studies showed that Fe3þ partitions
preferentially into b-(Mg, Fe)2SiO4 with the minimum of (Fe3þ/Fetotal) ratio being 0.03
(O’Neill et al., 1993), which is much larger than for olivine. For this reason alone, more
water is expected in the b-phase. If Fe3þ is present in the b-phase, the dissociation of water
will be exothermic (Wright and Catlow, 1996).
Again, reactions involving the formation of Mg vacancies in b(Mg, Fe)2SiO4 have
an energy of 1.18 eV per H2O molecule when associated with the O1-site; this is much
lower than the same reaction in a-olivine (3.65 eV per H2O). This favours the dissolution
of water into the b-phase (Wright and Catlow, 1996).
Two factors control the b-phase in its incorporation of protons: first, the possible
0
presence of Fe3þ, whose reduction not only makes the protonation of OH -sites possible
but also of O-sites (e.g., O1) as well and, secondly, the unbonded oxygen can be easily
protonated and OH groups will be present as substitutional species of O-sites. Thus, the
presence of Fe3þ would promote the enhanced dissolution of water in the mantle phases.
564 Chapter 6

6.5. Olivine ! spinel transition: CFS

Pressure influences the olivine ! spinel transition in which DS and DV are

almost independent of composition and Pt directly varies as DU; which is again
linearly proportional to the ratio RM/RT (where RM ¼ ionic radii of divalent cations in
octahedral coordination and RT ¼ ionic radii of tetravalent cations in tetrahedral
coordination).
Olivine ! spinel transition may proceed through ion displacements such that
the (100) plane of olivine becomes the (111) plane of g-spinel (Poirier, 1981) (see
Fig. 6.9). Brearley et al. (1992) observed that, at 15 GPa, forsterite could transform to
b-phase either directly or through the formation of g-spinel, which is metastable at
high temperature.
Syon et al. (1971) suggested that the stability field of the spinel phase might be
diminished in favour of the b-phase at high pressures as the crystal-field stabilization of
Fe2þ in the g-phase is less than in the b-phase. The crystal-field splittings increase with
increasing Mg2SiO4 content due to the decreasing volumes of octahedra. The overall
crystal-field stabilization free energy, DGCFS, decreases with the decreasing Fe2SiO4
content. Olivines in the mantle contain ,10% Fe2SiO4, which means that crystal-field
energy contributes DPCFS (,5 kbar) in the lowering of the transition pressure to its real
value. Thus, the crystal-field stabilization may raise the olivine ! spinel boundary in the
mantle by ,15 km (Burns and Sung, 1978).
TEM and IR spectroscopic studies suggest the possibility of forming an
intermediate spinelloid phase between olivine and spinel at high pressure. New diffraction
peaks (Fig. 6.28) and additional Raman bands were also noted in this region resulting from
the condensation of SiO4 tetrahedra (Durben et al., 1993). MgGeO4 olivines (spinelloids)

Figure 6.28. Diffraction patters of Mg2SiO4 at 69 GPa with indexations for both forsterite and b spinel (Andrault
et al., 1995, q 1995 Springer-Verlag).
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 565

also show similar additional Raman bands (Reynard et al., 1994). The Raman spectra of
b-phase and spinel are shown in Fig. 6.29a,b.
Any attempt to describe the structural and physical properties of both Mg-olivine
and its spinel-polymorph (ringwoodite) and their response to increasing pressure have
centred on atomistic stimulations. The partially ionic models employing potentials, which
utilize fractional or partial charges coupled with short-range repulsion terms, satisfactorily
reproduce both structural and elastic properties (Price and Parker, 1984). The calculated
pressure derivatives of the elastic constants (d1ij/dP) are very similar to those measured by
Brillouin scattering (Bassett et al., 1982).

6.5.1. Oxygen sublattice transformation (hcp ! fcc): partial dislocations

In olivine structure, the oxygen sublattice is roughly hexagonal close packed (hcp)
and becomes face centred cubic (fcc) in the spinel structure. This transformation
hcp ! fcc can be achieved without diffusion but by the shear motion of partial (Shockley)
dislocations. The hcp ! fcc transformation occurs by the nucleation and glide of loops of
Shockley partials at the right levels in the stacking of dense O layers. Owing to their large
diffusion coefficients, reorganization of Fe and Mg cations would occur rapidly.
Dislocation-assisted transformation hcp ! fcc of the O sublattices has been
modelled by Raterron et al. (1998). The relevant partial dislocations can have Bergers
 1=2) and glide in the (100)ol
vectors of the type k0 1/6 1/2l (i.e., [0 1/6 1/2] and ½0 1=6
olivine plane, which is the dense plane of oxygens with the same structure as the {111}sp
plane of oxygens in spinel (Raterron et al., 1998) (see Fig. 6.9).

Figure 6.29. Raman spectra of (a) b-phase and (b) spinel (Fei, et al., 1991).
566 Chapter 6

Such a mechanism accounts for the epitaxial relationships between olivine and
spinel (100)olk{111}sp with [001]olkk110lsp. Such relationships have been characterized
for the olivine – ringwoodite polymorphic transformation and for the olivine –
(perovskite þ magneiowüstite) transformations (e.g., Kerschhofer et al., 1996; Wang
et al., 1997). One can produce two elementary layers of spinel with a twin in its midst with
only four partials (two with Burgers vectors [0 1/6 1/2] and two with ½0 1=6  1=2).
 The
surface energy associated with spinel twinning is quite low compared with the nucleation
energy. The partials lie in the inter-phase boundary. Pairs of partials with different Burgers
vectors can glide in the surrounding olivine without leaving a stacking fault in their wake.
There is a stacking fault only in the ribbon between the partials. The screw segments can
cross-slip and recombine, leading to a perfect c dislocation while the edge segments can
recombine only by climbing.
Under moderate differential stress and at high temperature, both a and c dislocation
systems are easily activated (e.g., Bai and Kohlstedt, 1992a,b).

6.5.2. Olivine –spinel compressibility

Point-group symmetries for the M1- and M2-sites in olivine and the B-site in spinel
are Ci, Cs and C3n, respectively. The volumes of the octahedra decrease in the order
M2 . M1 . B. The crystal-field splitting D 0 is, therefore, expected to increase in the
opposite order.
The average bond length (d) of Fe –O for M1 and M2 octahedra in olivine are 2.157
and 2.182 Å, respectively. Although the D 0 values for M1 and M2 octahedra are different,
the CFSE values of these octahedra are nearly the same (differing p 2%). This leads to
very little-site preference for Fe2þ and little cation ordering in the forsterite –fayalite
series. This disordering of divalent cations is responsible for making the Mg –Fe olivine
series a quasi-ideal solid solution.
Under increasing pressure, the volume and shape of the tetrahedra in Fe2SiO4
remain essentially unchanged and the compressibility differences between the M1 and M2
octahedra are negligible (Hazen, 1977). The K0 of the octahedra can be estimated from the
bulk modulus, K, as
K0 < K=ð1 þ VT =VM Þ
where VT and VM are the total volumes of the tetrahedra and octahedra, respectively. The
K0 value of Fe2SiO4 has been calculated as ,90 GPa, assuming VT/VM to be 0.33 and K to
be 119 GPa (Yagi et al., 1975).
In Fe2SiO4 spinel, the compression of octahedra and tetrahedra is expected to be
nearly uniform. But M octahedra in fayalite are more compressible than B octahedra in
spinel form; for this reason, under pressure D 0 of Fe2SiO4, fayalite increases more rapidly
that its spinel form.
For olivine (a) and spinel (g) solid solutions in the Mg2SiO4 –Fe2SiO4 system,
the mean thermal Gruneisen constants (gth) are seen to change with Fe content
(Fig. 6.30).
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 567

Figure 6.30. Mean thermal Gruneisen constants between 350 and 700 K, Yth versus mole percent of the Fe2SiO4
component for both olivine (a) and spinel (g) solid solutions in the system Mg2SiO4 –Fe2SiO4 (from Watanabe,
1987, q 1987 American Geophysical Union).

6.5.3. g-Mg2SiO4 (ringwoodite and inverse ringwoodite)

The general formula of ringwoodite can be written as

MgVI IV VI IV
22x Mgx Six Si12x O4 :

When x ¼ 0; it is called normal ringwoodite and when x ¼ 1; it is inverse ringwoodite.

Ringwoodite is considered to be the most abundant mineral in the lower part
(520 – 660 km) of the transition zone (e.g., Ita and Stixrude, 1992). At ambient
condition, it is cubic and has symmetry Fd 3m  with Z ¼ 8 Mg2SiO4 units in the
conventional unit cell. But its symmetry at transition zone P, T is not clearly known.
Inverse ringwoodite, with all Si in the octahedral-site and Mg split equally between
the tetrahedral and octahedral-sites, has not been found in laboratory experiments. The
other spinel (Mg2TiO4) structure is tetragonal, with space group P4122 and Z ¼ 4:
(Note: P4122 is a subgroup of Fd 3m:) 
The corner-sharing octahedra (the B-sites) provide the framework with three
oxygens of each triangular face of the octahedra forming a distorted cubo-octahedron
surrounding the A-site. Cations in the A-site thus have 12 near-neighbour oxygen atoms
but with M– O distances exhibiting a wide range of values. The A- and B-site cations are
only 279 nm apart; and rotation and tilting of the B-site octahedra produce distortion of the
cubo-octahedron forming the A-site. This is the cause for the wide range of M– O distances
and the ill-defined coordination number for the A-site.
Recent simulation (by molecular dynamics) shows that high densities can also
be reached without changes in Si coordination, essentially by the repacking of SiO4
tetrahedra (Kubicki and Lasaga, 1991). Kiefer et al. (1999) performed first-principles
calculations (variable-cell-shape molecular dynamics strategy) to investigate the influence
of cation exchange on the physical properties of ringwoodite. This technique combines
density functional theory (DFT) in the framework of the pseudopotential plane-wave
method (e.g., Pickett, 1989), which allows for efficient minimization of forces and stresses
through simultaneous relaxation of all structural parameters. This technique has
been used to study the pressure dependence of structural and elastic properties of several
mantle silicates.
568 Chapter 6

The hydrostatic compression of g-Mg2SiO4 to mantle pressures and 700 K was

determined by Meng et al. (1994). The P – V – T EOS for it were determined under
simultaneous high-P (#30 GPa) and high-T (#700 K) conditions. The measurements were
conducted in an internally heated DAC using synchrotron XRD with Ne pressure medium.
The isothermal bulk modulus and its pressure derivative, determined from the isothermal
compression data, are
Temperature (K) K0 (GPa) K00
300 182 (3) 4.2 (0.3)
700 171 (4) 4.4 (0.5)

Fitting all the P –V – T data to high-T Murnaghan equation of state yields:

KTo (GPa) K0To K (dKT/dT)o (d2KT/dPdT)o
22
182 (3.0) 4.0 (0.3) 22.7 (0.5) £ 10 GPa/K 5.5 (5.2) £ 1024 K21

The isothermal P –V relations at 300 and 700 K are presented in Fig. 6.31.
Below 670 km, a disproportionation reaction occurs and the spinel structure breaks
down to a mixture of silicate perovskite (Mg, Fe)SiO3 plus (Mg, Fe)O.

6.5.3.1. Under pressure

In normal ringwoodite, the compression is mainly taken up by the Mg octahedra,
which have a much lower bulk modulus than the Si tetrahedra. The Si tetrahedra remain
ideal over the whole pressure range, while the Mg octahedra become more ideal with
increasing pressure.

Figure 6.31. Isothermal P – V relations of g-Mg2SiO4 at 300 K (filled circles) and at 700 K (open circles). The
solid and dashed lines represent the least-squares fits of the compression data to a third order Birch–Murnaghan
equation of state (after Meng et al., 1994, q 1994 Springer-Verlag).
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 569

In contrast, the Mg-tetrahedra in inverse ringwoodite are the most compressible and
the Mg tetrahedra have the lowest bulk modulus (,45% lower than MgO6 octahedra and
,2.5 times lower than SiO6 octahedra). The Mg and Si octahedra get only weakly distorted
by pressure while the Mg tetrahedra distort more strongly. Put plainly, the Mg octahedra in
ringwoodite are much softer than the Si tetrahedra whereas, in inverse ringwoodite, the Mg
tetrahedra are the softest, much softer than the Mg and Si octahedral sites. The octahedra
form a continuous edge-sharing network whereas the Mg tetrahedra share only corners
with the surrounding polyhedra. Therefore, it is much easier to deform the Mg tetrahedra,
whose compressibility can also be seen in the quadratic elongation and the angular variance.
The results of Kieffer et al. (1999) show that, at low temperature, normal
ringwoodite is more stable than inverse ringwoodite over the whole pressure range up to
25 GPa, although inverse ringwoodite is more compressible than ringwoodite. The
compression of coordination polyhedra in normal and inverse Mg2SiO4 ringwoodite
obtained from LDA pseudopotential calculations by Kiefer et al. (1998) is presented
in Fig. 6.32.
Finger et al. (1977) have shown that the SiO4 tetrahedron in g-Ni2SiO4 has a
bulk modulus greater than 2.5 Mbar, whereas that of the NiO6 octahedron is 1.7 Mbar.
g-Mg2SiO4 is likely to be analogous to g-Ni2SiO4.
Deformation of spinels by dislocation creep has been discussed in Section 15.14.2.3.

Figure 6.32. Compression of coordination polyhedra in normal and inverse Mg2SiO4 ringwoodite from LDA
pseudopotential calculations (Kiefer et al., 1999, q Mineralogical Society of America).
570 Chapter 6

6.5.3.2. Thermodynamics
The free energy and enthalpy of ringwoodite are discussed below.

Free energy. The reaction Mg2SiO4(b) ¼ Mg2SiO4(g) in the mantle can be studied by
calculating the Gibbs free energies of b-spinel and g-spinel along the 1,600 K isotherm. At
lower pressure, b-spinel has a Gibbs free energy that is lower than that of the g-spinel.
With increasing pressure, the difference between the Gibbs free energies (DG) of these two
phases decreases. At pressure in the range 16 – 18 GPa, these two phases co-exist and are in
equilibrium (Akaogi et al., 1989).
The free energy of ringwoodite can be written in terms of x (disordering parameter):

Gðx; P; TÞ ¼ ð1 2 xÞGnormal þ xGorder-inverse 2 TSIV VI

conf ðxÞ 2 TSconf ðxÞ þ ixs ðx; P; TÞ ð6-17Þ

with

SIV
conf ðxÞ ¼ 2R½x ln x þ ð1 2 xÞ ln ð1 2 xÞ;

SVI
conf ðxÞ ¼ 22R½x=2 ln ðx=2Þ þ ð2 2 xÞ=2 ln ð2 2 x=2Þ

where Hxs (x, P, T ) describes the enthalpy change due to cation disorder.
Therefore, the energy difference between the normal and fully inverse ringwoodite
be written as

Ginverse 2 Gnormal ¼ Gorder-inverse 2 Gnormal 2 2RT ln 2 þ Hxs ð1; P; TÞ ð6-18Þ

The inverse content depends on Hxs, which is positive in spinels. The temperature at
which inverse ringwoodite becomes more stable than normal ringwoodite is determined
from the relation:

DH þ Hxs
T¼ ð6-19Þ
2R ln 2
where DH ¼ Horder-inverse 2 Hnormal and Hxs p DH (commonly).
Kiefer et al. (1999) concluded that the lower limit for the temperature at which
inverse ringwoodite becomes more stable than normal ringwoodite is 2,600 K.

Enthalpy. The enthalpy due to ordering is formulated as a second-order Taylor expansion

in the three cation distribution parameters and two compositional parameters that are
required to describe fully the cation distribution.
The enthalpy difference between normal ringwoodite and inverse ringwoodite
increases with pressure by 0.73 kJ/(mol GPa).

Phase equilibrium. Equilibrium conditions may be independently estimated from

thermodynamic data. Thus, the entropy S and heat capacity CV can be derived from
Raman spectroscopy measurements at high pressure if DH at 1 atm and DV at mantle
conditions are known.
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 571

Chopelas et al. (1994) used the Raman spectra of g-Mg2SiO4 at 20 GPa to

calculate the phase equilibria and heat capacity (Cv) at various P – T conditions. They
determined the slope for the b ! g transition as 950 ^ 4 bar/K and the slopes for the
b ! pV þ MgO and g ! pv þ MgO to be 2 7 ^ 3 and 25 ^ 4 bar/K, respectively.
These slopes result in a b– g –(MgO þ pv) triple point at 2 229 GPa and 2,200 K for
the iron-free system.

6.5.3.3. Symmetry analysis

g-Mg2SiO4 belongs to space group Fd3m (O7h) and has two formula units per
primitive cell. The 42 normal modes at the Brillouin zone centre may be divided among the
following symmetry species (Choplas et al., 1994):

T ¼ A1g ðRÞ þ Eg ðRÞ þ T1g þ 3T2g ðRÞ þ 2A2u þ 2Eu þ 4T1u ðIRÞ þ 2T2u ð6-20Þ

where R (Raman) and IR (infrared) denote spectral activity.

The observed modes in the spectra may be assigned to molecular motions by site-
group analysis (Farmer and Lazarev, 1974). This requires that the force constants of the
bonds in each of the two polyhedra must differ substantially.
In the site-group to factor-group analysis (Table 6.15a) by Chopelas et al. (1994),
the internal motions of the less compressible polyhedron (SiO4 tetrahedron) are assumed to
be separate entities. Such an approach was successful in enumerating the modes of olivine
(Chopelas, 1991) and garnet (Hofmeister and Chopelas, 1991).
Table 6.15b shows that only vibrational modes related to the SiO4 tetrahedron
in g-Mg2SiO4 should be present in the Raman spectrum. This suggests that the mode
frequencies should not vary significantly as other cations are substituted for Mg2þ.
Two of the high-energy modes, 600 and 796 cm21 (Fig. 6.33), assigned to internal
asymmetric modes (bending and stretching, respectively) of the SiO4 tetrahedron,
appear to be affected most strongly by the unit-cell size. In g-Ni2SiO4, g-Co2SiO4 and
in cubic garnets, the SiO4 internal-mode frequencies vary nearly with cell volume
(Hofmeister and Chopelas, 1991).
The Raman spectra (unpolarized) of g-Mg2SiO4 showed that the relative inten-
sities of the modes for 302, 372, 600 and 796 cm21 bands remained almost the same up to
,20 GPa (Fig. 6.33). The Grüneisen parameter for the Raman modes, gi, is calculated

TABLE 6.15a
Site-group analysis for g-Mg2SiO4 (Chopelas et al., 1994)

A1g A2g Eg T1g T2g A1u A2u Eu T1u T2u

SiO4int n1 0 n2 0 n3, n4 0 n1 n2 n3, n4 0

SiO4rot 0 0 0 1 0 0 0 0 0 1
SiO4trans 0 0 0 0 1 0 0 0 0a 0
Mgtrans 0 0 0 0 0 0 1 1 2 1
a
Acoustic modes subtracted to give total optic modes.
572 Chapter 6

TABLE 6.15b
Vibrational modes of g-Mg2SiO4 and assignment Raman modes in cm21

Mineral T2g Eg T2g T2g Ag aa (Å) Reference

b
g-Mg2SiO4 302 372 600 796 834 8.0649
c, d
g-Ni2SiO4 190 368 632 811 848 8.044
c
g-Co2SiO4 182 330 572 774 838 8.140
Assignment TSiO4 n2 n4 n3 n1
Infrared modes for g-Mg2SiO4 in cm21 (all T1u)
e
350 445 545 830
f
326–333 400–538 552–541 800–975
Assignment TMg TMg n4 n3
a
Lattice constants.
b
Chopelas et al. (1994).
c
Chopelas and Hofmesiter (1993, unpublished data).
d
The mode at 675 cm21 was reported earlier by Yamanaka and Ishii (1986); instead a mode at 632 cm21 was found.
e
From powder transmission spectra of Akaogi et al. (1984).
f
Unpublished reflectance data of Hofmeister (1993).

from
 
KT dn
gi ¼ ð6-21Þ
n0 dP T

where KT is the isothermal bulk modulus and n0 is the mode frequency at 1 atm and are
shown to vary between 0.66 and 1.54.

Figure 6.33. Unpolarized Raman spectra of single crystal g-Mg2SiO4 at 1 atm, 96, 143 and 193 kbar. The
expanded scale on the low energy modes is 10£ magnification (Chopelas et al., 1994, q 1994 Springer-Verlag).
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 573

6.5.4. Olivine-(enstatite)– spinel nucleation in subducting lithosphere

In the olivine –spinel transformation, the presence of clinoenstatite is seen to have a

strong effect on the kinetics of spinel nucleation (Sharp and Rubie, 1995). Nucleation can
be the rate-controlling step during metamorphic reactions in the Earth’s crust and can
result in the crystallization of non-equilibrium mineral assemblages (Wayle et al., 1989).
The experiments by Sharp and Rubie (1995) suggest that enstatite, the second most
abundant phase in the subducting lithosphere and in upper-mantle lithologies, is likely to
enhance nucleation of spinel and possibly b-phase within the Earth’s mantle. Sluggish
reaction kinetics can result in a deeply subducted wedge of metastable olivine and cause
deep-focus earthquakes through transformational faulting.
The similarity between the high-clinoenstatite (hcen) and spinel structures enables
enstatite to act as a template for the coherent nucleation of spinel through the formation of
a low-energy interface between the enstatite and the spinel nucleus. It is apparent that the
modified spinel (b) phase has a slightly more distorted close-packed structure than either
spinel or high clinoenstatite and therefore one would expect additional misfit-strain energy
in hcen –b-phase intergrowths.

6.5.4.1. Hydrous ringwoodite (g-Mg2SiO4)

(Mg, Fe)2SiO4 g-spinel has also been observed to retain significant quantities of
water at temperatures lower than those of the mantle: 2.7 wt% at temperatures of 1,1008C
and pressures corresponding to depths near 585 km (Kohlstedt et al., 1996; Inoue et al.,
1997).
In hydrous g-phase (ringwoodite), there is polymorphization in SiO4 tetrahedra
(Fig. 6.34). Since ringwoodite has a spinel structure, the probabilities of such arrangement
 [101], ½101;
are equal along the six directions, e.g., [110], ½110;  [011] and ½011:
Using the unit-cell parameter value as ag ¼ 8.0649 Å for anhydrous ringwoodite,
g-Mg2SiO4 (Sasaki et al., 1982) and bb ¼ 11.477 Å for hydrous wadsleyite, Mg1.86-
SiH0.28O4 (Kudoh and Inoune, 1999), the unit-cell edge length, ahy-g, for hydrous
ringwoodite can be calculated as

pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ahy2g ¼ ag þ 0:33ðbb = 
2 2 ag Þ 2 ag ¼ 8:082 A ð6-22Þ

This value is close to the value 8.079 Å observed by Inoue et al. (1998) for hydrous
ringwoodite.
The density difference between wadsleyite (D ¼ 3.474 g/cm3) and ringwoodite
(D ¼ 3.563 g/cm3) is mainly caused by the difference between the c-axis length of
wadsleyite (8.2566 Å) and the a-axis length of ringwoodite (8.0649 Å). The density of g is
2.6% greater than that of b and the a-axis length of g is 2– 3% shorter than the c-axis length
of b. This difference is ascribed to the tilting of the MgO6 octahedra and SiO4 tetrahedra
caused by the existence of the Si2O7 group in wadsleyite structure.
574 Chapter 6

Figure 6.34. Postulated structural modules for hydrous ringwoodite. The wadsleyite unit cell is outlined. Since
ringwoodite has a spinel structure, the probabilities of such arrangement are equal along the six directions, e.g.,
 [101], ½101;
[110], ½110;  [011] and ½011: Vacant tetrahedral Si2 site is shown by light shading and Si site by
heavy shading. Vacant Mg3 site is shown by open octahedron.

6.6. Fe2SiO4 systems

It is known that Fe2SiO4 melts incongruently into metallic iron plus liquid at low
pressures (Lindsley, 1967). At high pressures, a disproportionation (decomposition)
reaction of Fe2SiO4 produce wüstite, which is initially stoichiometric but, at higher
pressure (at 1,0008C), it becomes non-stoichiometric (Mao and Bell, 1977).

17:5 GPa 23 GPa

Fe2 SiO4 ! 2FeO þ SiO2 ! 2Fex O þ 2ð1 2 xÞFe þ SiO2 ð6-23Þ
spinel w€ust: stish: wust: 1-phase stish:

At ,1,0008C, the wüstite produced by the reaction above is non-stoichiometric in

the pressure range 10 –30 GPa.
At high pressure and temperature (.1,0008C), fayalite (FeSiO4), ferrosilite
(FeSiO3) and almandine should all disproportionate directly to an assemblage containing
FexO and g-iron (perhaps without passing through the stability of stoichiometric FeO).
This disproportionation of FeO into FexO þ (1 2 x)Fe is different from the disproportio-
nation of Fe2þ(?) into Fe0 þ Fe3þ(?) demonstrated in the basaltic glass by Bell and Mao
(1975). High pressure increases the non-stoichiometry of wüstite and results in an increase
in the yield of metallic iron (Liu, 1976). If the starting material is Fe2SiO4, the free iron
produced is only (1 2 x), which amounts to 0.03 or 1.6%.
Fayalite, which, like other iron-rich minerals is magnetically ordered at low
temperatures; is paramagnetic at room temperature and undergoes a phase transition
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 575

(Neel point) at 65 K to a collinear anti-ferromagnetic state. At 23 K, it has a second

transition point, below which a canted anti-ferromagnetic state is stable (Santoro et al.,
1966). Hayashi et al. (1987) measured as a function of pressure the Néel temperature, TN,
in the pressure range to 16 GPa and found the Néel temperature to increase linearly with
pressure at the rate of dTN/dP ¼ 2.2 ^ 0.2 K/GPa.
This result was explicable in terms of the classical Heisenberg exchange
Hamiltonian model on the assumption that the super-exchange interactions vary as the
210/3 power of the volume (see “TN: Bloch’s 10/3 law” section, Chapter 9) while the
volume dependence of the direct-exchange interactions is small and in the opposite
direction.

6.6.1. g-Fe2SiO4 spinel

By static compression measurements (at 4008C), the P – V – T behaviour of

g-Fe2SiO4 spinel was studied by Plymate and Stout (1994) under P # 24 GPa. Fitting the
data to the Murnaghan equation yields (at 4008C):

KðP0 Þ ¼ 177:3 ^ 17:4 GPa and K 0 ðT0 Þ ¼ 5:4 ^ 2:5:

All available data in the T interval 258C , T , 4008C give the average value of
(dK/dT)P0 ¼ 2 4.1(^ 6.2)1022 GPa/8C.
A five-parameter VðP; TÞ equation for g-Fe2SiO4 based on simultaneous
regression of these data, combined with the elevated P – T data of Yagi et al. (1987)
and the extrapolated thermal expansion values from Sukuki et al. (1979), yields
isochores which have very little curvature ((d 2T/dP 2)V ¼ 0). This is in marked
contrast to the isochores for fayalite (Plymate and Stout, 1990), which exhibit
pronounced n9 ((d2T/dP 2)V , 0).
Along the fayalite – g-Fe2SiO4 reaction boundary, DVR varies from a minimum
of ,8.3% at ,4508C to ,8.9% at 1,2008C. Extrapolation of the fayalite and
g-Fe2SiO4 VðP; TÞ relationships to the T and P of the 400-km discontinuity suggests a
DV of approximately 8.4% at that depth (,10% less than the 9.3% DV at ambient
conditions).

6.6.1.1. Fe2SiO4 – Fe3O4 system

Fe2SiO4 has olivine structure (a-Fe2SiO4) at ambient condition and transforms
directly into spinel (g-Fe2SiO4) under high pressure (Akimoto et al., 1967), unlike
Mn2SiO4 and Co2SiO4, for which an intervening modified spinel phase is reported.
In the Fe3O4 – Fe2SiO4 system, the high-pressure phase equilibria are significant in
gaining an understanding of the Earth’s mantle, especially its oxidation state, electrical
conductivity, etc. Cation distribution and electron conductivity in the spinel solid solution
have been elucidated by X-ray crystal-structure refinement and AC-impedance
measurement.
576 Chapter 6

In the system Fe3O4 –Fe2SiO4, a spinelloid phase has been reported by Canil et al.
(1990). A spinelloid phase is stable for intermediate compositions in the pressure range
from 3 to 9 GPa; the synthesized spinelloid phase is indexed assuming a nickel
aluminosilicate V-type structure (Ohtaka et al., 1997). Ti4þ substitution in Fe3O4 yields a
continuous solid solution between magnetite Fe3O4 and ulvospinel Fe2TiO4 but Si4þ
substitution in Fe3O4 at ambient pressure is not known.
A complete solid solution of the system Fe2SiO4 – Fe3O4 with spinel structure
occurs above 9 GPa at 1,2008C (Ohtaka et al., 1997). A spinelloid phase is stable for
intermediate compositions in the pressure range 3 – 4 GPa.

6.6.2. Cr2SiO4 : Cr21 orthosilicates

Cr2þ has been reported from lunar minerals and from olivine inclusions in mantle-
derived diamonds (e.g. Sutton et al., 1993). The distribution of Cr2þ provides an important
geochemical constraint for the models of the evolution of the Earth’s mantle and its
relationship to the Moon. Cr2þ substitution is found to be limited to a maximum of ,25
and 75 mol%, respectively, in the olivine and pyroxene solid-solution series (Li et al.,
1995).
In synthesized Cr2SiO4 crystals, the Cr2þ ion occupies the octahedral site. Since
2þ
Cr ion is too small for the six-coordinated octahedral site, the site gets strongly distorted
and Cr2þ is displaced from the centre and the coordination approximates a planar
arrangement of four oxygen ligands with two additional, very weak bonds to one side of
this distorted planar Cr2þO4 group. Stabilization of such a square-planar geometry is
expected for Cr2þ cations because the 3d4 configuration normally leads to a strong Jahn –
Teller distortion in 6-fold coordination. The Cr– Cr distance between two adjacent
polyhedra reduces to 2.75 Å. This reduced distance is indicative that interaction between
the Cr atoms exists and that a weak metal – metal bonding is present. This is also supported
by DFT calculations.

6.6.2.1. XRD and electronic spectroscopy

Polarized electronic-absorption spectra (Furche and Langer, 1998) have been
interpreted in terms of both localized d – d and d – dp transitions, based on the suggestion
that a dynamic exchange process occurs between mono- and di-nuclear species on a short
time scale. (Note: Such electronic coupling has not, however, been observed in Cd2SiO4
(Miletich et al., 1998).)
The electronic absorption band at ,18,300 cm21 for Ekc (parallel to the shortest
Cr –Cr vector in the structure), which has a very large (5,000 cm21) half-width, indicative
of electronic interaction between metal centres.
The high-pressure form of Cr2SiO4 remains orthorhombic up to 9.22 GPa. A third-
order Birch – Murnaghan EOS fitted to the V – P data yields (Miletich et al., 1999):

K ¼ 94:7 ð4Þ GPa; K 0 ¼ 8:32 ð14Þ and V0 ¼ 610:10 ð3Þ A:


MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 577

6.6.2.2. M– M bonding and Cr dimerization

Evidence has been observed for the changes of subtle stereochemical features with
pressure, such as the increasing planarity of the 4-fold square-planar oxygen configuration.
The electronic absorption bands arise from orbital overlap. The diamagnetic response
would suggest electron pairing from orbital overlap due to metal – metal bonding. An
increase in orbital overlap occurs due to reduction in inter-atomic distances under pressure,
while temperature enhances the probability for electronic transfer between the two Cr
atoms. All these offer evidence for a weak metal –metal bonding interaction and a silicate-
bridged chromium dimer. Multiple metal – metal bonding and dimerization might as well
stabilize other transition-metal species in silicate-melt species in silicate-melt structures at
high P – T and thus play an important role for geochemical fractionation of these elements
during the evolution of the Earth (Miletich et al., 1999).

6.6.2.3. Compressional anisotropy

Compressional anisotropy of the thenardite-type AB2O4 structures are basically
controlled by cation repulsion through shared polyhedral edges and consequently through
the polyhedral geometries.
The high compressional anisotropy observed in Cr2SiO4 arises because the Cr atom
is a square-planar CrO4 configuration which lies parallel to (001). In this structure type,
edge sharing is the most characteristic feature in which the SiO4 tetrahedra share edges
with neighbouring CrO6 octahedra. The strong uniaxial elongation of the SiO4 tetrahedra
along the diad parallel to the c-axis is a result of the topological connectivity (see Fig. 2 in
Miletich et al., 1998), which is also responsible for the distortion of the MO6 polyhedron
(Miletich et al., 1999).

6.6.3. Ni2SiO4: deformation

Nickel orthosilicate (Ni2SiO4) samples were studied from 3 to 20 MPa and fO2 from
1021 to 105 Pa by Wolfenstine and Kohlstedt (1994).
At T # 1,773 K, the dominant deformation mechanism is cobble creep, whose rate
is limited by grain-boundary diffusion of silicon. At T between 1,803 and 1,813 K,
enhanced creep rates and increased ductility occur. These are associated with fast diffusion
through a thin intergranular film produced by kinetic decomposition of Ni2SiO4 under the
imposed stress gradient. Wolfenstine and Kohlstedt (1994) deformed samples of nickel
orthosilicate grains by differential stresses of 3– 20 MPa (at 1,573 – 1,813 K) and under fO2
from 1021 to 105 Pa.

6.6.4. Mg2GeO4 olivine

Reynard et al. (1994) studied the structural modifications in Mg2GeO4 olivines

under pressure and the Raman spectra of the samples were taken at P # 34 GPa at RT.
Near 11 GPa, two sharp extra bands are seen to appear in the 600 – 700 cm21
frequency range. Above 22 GPa, along with these two, the position and relative intensity of
578 Chapter 6

the other vibrational bands change drastically. The intensity of the sharp bands decreases
progressively above 25 GPa.
During decompression to atmospheric P, the high-pressure phase partially reverts to
olivine structure. From this, it can be concluded that GeO4 tetrahedra are polymerized on
metastable transformation from olivine structure. The metastable high-P phase is a
structurally disordered spinelloid close to the hypothetical W- or Ep-phase.

6.7. Pyroxenes

Pyroxenes have composition: ASiO3, where A consists of divalent cations, Mg,

Fe, Ca or mixtures of cations of valencies þ1, þ2 and þ3. Pyroxene structures consist
of alternate layers of MO6 octahedra and SiO4 tetrahedra parallel to the (100) plane
(where M ¼ divalent cations). The single tetrahedral (formally (SiO3)22) chains occur
in layers parallel to (100). SiO4 tetrahedra share two corners with adjacent tetrahedra
forming infinite chains running parallel to the c-axis. There are two non-equivalent
octahedral M-sites: M1 and M2 (see Fig. 6.35a,b). The former is regular, while M2 is
distorted and accommodates larger cations than the M1-site can. The local
coordination of M2 may be nearly 8-fold when the site is occupied by larger cations
such as Na or Ca. The size of the cation occupying the octahedral M2-site may
determine the rotation of the tetrahedral SiO4 chains that are linked with the MO6
octahedra. The different symmetries among pyroxene structural types result from
different stacking sequences of the octahedral layers and form symmetrically distinct
tetrahedral chains. In high-pressure pyroxenes, Si and Mg are known to form an
ordered edge-sharing octahedral chain in this structure.
In these silicate minerals, the Si – O bonds are linked in array to form a skeleton,
whose articulation under the influence of pressure and temperature offer valuable informa-
tion for understanding the adaptations and transformations to different phases. Ab initio
quantum-mechanical calculations have been developed to extend the reliable predictions
of the behaviour to mineral phases under pressure (e.g., Stixrude and Cohen, 1993).
Five Ca-poor pyroxene polymorphs are known: orthopyroxene (Pbca), protopy-
roxene (Pbcn), low-clinopyroxene (P2I/c), high-clinopyroxene (C2/c) and high-P
clinopyroxene (C2/c) (see Fig. 6.35c). The five polymorphs differ, mainly in the
configurations of the tetrahedral chains relative to the orientation of the octahedral bands.
A first-order displacive-phase transformation from the Pbcn space group to P2Icn was
observed between 2.03 and 2.50 GPa, which is characterized by a discontinuous decrease in
a, c and V by 1.1, 2.4 and 2.6%, respectively (Yang et al., 1999). The prominent structural
changes associated with the Pbcn to PIcn transformation involve an abrupt splitting of one
type of O-rotated silicate chain in low-pressure protopyroxene into S-rotated A and O-
rotated B chains in high-pressure protopyroxene. Both O- and S-rotated silicate chains
should occur in the P2Ica structure. The discovery of the P2Icn protopyroxene structure
(Yang et al., 1999) implies the possible existence of P2Ica orthopyroxene at high pressures.
The transition Pbcn ! P2Icn is the result of differential compression between SiO4
tetrahedra and MO6 octahedra. The two MO6 octahedra, e.g., M1 and M2, became less
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 579

Figure 6.35. (a –b) Schematic representation of the structure of high-pressure C2/c clinoenstatite viewed down
(001). (c) The structure of high-pressure C2/c clinoenstatite viewed down (100). Si are located at the centre of
oxygen tetrahedra; the M1 cation sites are represented by octahedra, while the M2 sites are represented by spheres
(Wentzcovitch et al., 1995, q 1995 Springer-Verlag).

distorted at higher pressures and their volumes decreased linearly with pressure (Fig. 6.36)
(Yang et al., 1999).
Earlier, it was thought that Ca-rich diopside and hedenbergite and jadeite crystallize
in monoclinic C2/c structure while the rest are either P21/c (low clinopyroxene) or Pbca
(orthopyroxene). But Ca-free clinoenstatite (MgSiO3) phase with C2/c symmetry has been
observed at high pressures (e.g., Kanzaki, 1991). Similarly, a high-pressure C2/c structure
type for FeSiO3 has also been established (Hugh-Jones et al., 1994).
The Pbca ! C2/c transformation in MgSiO3 starts at 7 GPa and 7008C, with the
transition pressure increasing with temperature. This pressure corresponds to an estimated
depth of 220 km. (Note: The Lehman discontinuity occurs between the depths of 180 and
580 Chapter 6

Figure 6.36. M1 and M2 octahedral volumes as a function of pressure (Yang et al., 1999, q 1999 Mineralogical
Society of America)

280 km.) Hugh-Jones et al. (1994) showed that the Pbca ! C2/c transition may start at the
pressure 7 GPa but at a slightly higher temperature of 8208C (see Fig. 6.37 from Woodland
et al. (1997); read the caption).
In wadeite structure (silica-rich K[vi] [iv]
2 Si3 Si1O9), silicon occurs in all network-
forming positions, in both 4- and 6-fold coordination. Each bridging O1 atom is

Figure 6.37. PT-path for the synthesis of (Mg, Fe)SiO3 clinopyroxenes (from Woodland et al., 1997). Step 1:
pressurization at room temperature; Step 2: heating to achieve the desired pressure–temperature conditions for the
synthesis; Step 3: termination of experiment by shutting off power while at high pressure; Step 4: decompression
to ambient conditions. The phase boundaries for MgSiO3 pyroxenes are shown for reference, indicating that
the experiments were within the stability field of C2/c clinopyroxene (q 1997 Mineralogical Society of America).
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 581

[vi] [vi] [iv]

coordinated to two Si and two K, while O2 atoms are linked to one Si, one Si
and two K.

6.7.1. Structural chains and angles

Silicon and oxygen atoms occupy generally equivalent positions in the C2/c
structures. In orthopyroxenes, the tetrahedral chains are of two symmetrically distinct
types: the one showing a more extended chain of smaller tetrahedra is called the “A” chain
and the other more linked chain with larger tetrahedra is referred to as the “B” chain. Thus,
two symmetry-distinct silicate chains (“A” and “B”) are created in the P21/c structure upon
inversion from C2/c phase, which contains a single chain (see Fig. 6.38a –c). The “A” chain
of the P21/c structure is an “S”-type chain (3 – 3 –3 angle ø 167.1 (9)8) which is very
similar to the “S”-type chain in the structure of the high-temperature C2/c phase (3 – 3– 3
angle ø 169.5 (8)8). The high-pressure phase is, however, quite distinct, containing “K”-
type chains (3 – 3 –3 ø 138.4 (9)8, which is very similar to the “B” chains of P21/c
structure (3 – 3– 3 ø 144.1 (1)8) (Christy and Angels, 1995).
The chain-extension angle (3 – 3– 3) in diopside changes by 2.8– 5.38/GPa (Levier
and Prewitt, 1981). The structural changes between the high pressure and the high-T
structure of C2/c clinoferrosilite are probably restricted to a narrow interval, which would
comprise either a cross-over or a phase transition (Christy and Angels, 1995).

6.7.2. MgSiO3 –FeSiO3 system

MgSiO3 forms a solid solution with FeSiO3 in the series MgSiO3 –FeSiO3. The
low-pressure phase of MgSiO3 is enstatite, an orthorhombic pyroxene with two formula
units per unit cell (Mg2Si2O6). Its upper-mantle phase, brought as xenoliths in peridotites
and ophiolites, has the composition Mg(12x)FexSi2O6 (with x ø 0.1). Above 2,000 K and
pressures of 17– 19 GPa, pyroxene transforms into a phase with garnet structure (isolated
tetrahedra) and four formula units per unit cell (Mg4Si4)O12.

Figure 6.38. (a) –(c) Polyhedral representations of the clinopyroxene structures of FeSiO3. M1 sites are shown as
octahedra, M2 sites circles. Note that the tetrahedral chains in the high-temperature C2/c structure (a) are straight
(S), those in the high-pressure C2/c structure (b) are kinked (K), and that the structure of P21/c clinopyroxene shown
in (c) contains chains of both types in alternative layers (from Christy and Angels, 1995, q Springer-Verlag).
582 Chapter 6

At upper-mantle pressures, aluminous silicate garnet, named majorite, predomi-

nates. This mineral was first identified in shocked meteorites. Below 2,000 K and above
20 GPa, MgSiO3 transforms from garnet structure to ilmenite structure (an Al2O3
corundum structure with Mg2þ and Si4þ located in an ordered alternate fashion in the
six-coordinated Al3þ-sites). The P – T phase diagram for MgSiO3 has been presented in
Fig. 6.4, along with the P-wave velocities (in parentheses) corresponding to the phases.
Liu (1976) studied the system and obtained solid solutions of perovskite at
P . 20 GPa. Yagi et al. (1979) studied this at 1,2738C and P up to 30 GPa.
Kato (1986) studied this system at P ¼ 20 GPa and T # 2,473 K and obtained the
stability field of garnet phase with cubic and tetragonal structures. Ohtani et al. (1991)
investigated at T ¼ 2,073 K and P # 25 GPa and only obtained tetragonal garnet. The
univariant reaction Px þ g þ St ¼ Mj is found to be stable at 16 GPa.
Diamond-anvil studies have determined the density of silicate perovskite at room
temperature for the entire pressure range of the lower mantle (Ito and Takahashi, 1989).
However, at 7 GPa pressure and utilizing synchrotron radiation, it has been seen that the
magnesium end-member perovskite remained metastable up to 1,0008C and then
transformed, first into an amorphous phase and then to the low-pressure enstatite phase
(Wang et al., 1991).
(Mg, Fe)SiO3 perovskite was known to be stable to at least 60 GPa (Mao et al.,
1977). Hence, it was considered to be the most abundant mineral in the Earth. Moreover,
later experiment (Knittle and Jeanloz, 1987a) proved its stability extended beyond
100 GPa.
At lower-mantle pressure, silicate ilmenite transforms to perovskite structure
having corner-sharing SiO6 octahedra with Mg2þ in dodecahedral sites. The lower mantle
is presumed to be dominated by silicate perovskites (cf. Jeanloz and Thompson, 1983).
Because of their high-temperature superconducting properties, much interest has been
generated in the crystal chemistry of perovskite-structure phases.
(Mg, Fe)SiO3 perovskite constitutes possibly more than 80 vol% of the lower
mantle. If so, it is the most abundant mineral in the Earth. Up to 1 Mbar at RT, it is
orthorhombic. High-pressure perovskite structures have also been seen in the case of
CaSiO3, CaMgSi2O6 (diopside) and CaMgSi2O6 – NaAlSi2O6 (diopside –jadeite) solid-
solution phases. However, at high pressure, FeSiO3 decomposes to FeO and SiO2
(stishovite).

6.7.2.1. MgSiO3 orthopyroxene

The space group of MgSiO3, orthoenstatite, is Pbca. An X-ray diffraction study of it
at 2.05 and 5.17 GPa has revealed no change in the symmetry (i.e., Pbca space group). It is
observed that the difference in volume between that observed at pressures in excess of
4 GPa and that calculated by extrapolation appears to evolve continuously from zero at
,4 GPa (Angel, 1996). In addition, the individual components of the spontaneous strain
tensor (calculated as, e.g., 111 ¼ a=a0 2 1; where a is the high-pressure cell parameter and
a0 is that extrapolated from the low-pressure regime) also evolve continuously from zero,
as does the scalar strain (Fig. 6.39) (Angel, 1997).
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 583

Figure 6.39. The axial strains (a) and the scalar strains (b) arising from the change in compression mechanism in
MgSiO3 orthopyroxene. These are calculated from the cell parameter measurements (Hugh-Jones and Angel,
1994), and using second-order polynomials for the extrapolation of the low-pressure cell parameters to pressures
in excess of 4 GPa (from Angel, 1996, q 1996 Overseas Publishers Association).

At high pressures, the unit-cell parameters a, b and c show a significant break in

slope at 4 GPa. This is also reflected in the plot of the unit-cell volume (Fig. 6.40). Raman
frequencies are also seen to change at a pressure above 4 GPa but, at pressures below
4 GPa, the SiO4 tetrahedra are incompressible. However, the compression is accom-
modated by rotations and tilts of these tetrahedra, which in turn result in compression of
the octahedral sites. Above 4 GPa, the volumes of both the symmetrically distinct
tetrahedra show significant compression.
As stated earlier, the compression is initially accommodated by the rotation of the
tetrahedral, which results in shortening of the M2 –O3 bond lengths. When M-sites are
occupied by small cations like Mg and Fe, the two M2 –O3 bonds become significantly
longer and can compress rapidly. Thus, it quickly becomes harder to show unusually high
values of K 0 (i.e., dKT/dP). At a critical point, this compression mechanism may become
energetically unfavourable and the tetrahedra in turn start to manifest compression.
However, when larger size Ca-cation substitutes at Mg- or Fe-sites, the tetrahedral
compression is not manifested so easily, even up to ,10 GPa.

Intra-crystalline Mg – Fe ordering. Mg – Fe ordering between two octahedral sites in

orthopyroxene [(Mg, Fe)2Si2O6] provides a sensitive indicator for thermal histories of
mantle xenoliths (Dyar et al., 1992), meteorites (Molin et al., 1991), granulites,
charnockites (Saxena and Dal Negro, 1983) and ophiolites (Skogby, 1992), etc.
Ordering between M1- and M2-sites of orthopyroxene can be expressed by the
distribution coefficient, KD:

KD ¼ ðXFeM2 =XMgM2 Þ=ðXFeM1 =XMgM1 Þ

Note: KD ¼ 1 signifies complete disordering.

584 Chapter 6

Figure 6.40. Top: The variation of the unit-cell volume of MgSiO3 orthopyroxene with pressure. Symbols are for
data measured by Hugh-Jones and Angel (1994), lines are the equations of state to the data. The excess volume
arising from the change in compression mechanism at 4 GPa is indicated. Bottom: Variation with pressure of the
volumes of the two symmetrically distinct tetrahedra in MgSiO3 orthopyroxene. Error bars are the maximum
uncertainties (Angel, 1996, q 1996 Overseas Publishers Association).

For slowly cooled natural samples, KD . 0.50 have been reported (Tribandino and
Talarico, 1992).
A small pressure effect on cation ordering, as reported by Hazen and Navrotsky
(1996), shows a small positive volume of disordering for Mg – Fe orthopyroxene:
DVdisorder ¼ Vdisorder 2 Vordered
causing KD to increase with increasing pressure. DVdisorder is a function of composition
with a maximum at XFe ¼ 0.5 (Domeneghetti et al., 1995).
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 585

Extrapolation of the thermodynamic model of Kroll et al. (1997) for ordering in

Mg –Fe orthopyroxene to 1,2758C yields DG0exch ¼ 213.3 kJ/mol, when DG0rec is assumed
to equal zero and W M1 ¼ 3.0 kJ/mol and W M2 ¼ 1.2 kJ/mol.
Hazen et al. (1993) synthesized a single crystal of intermediate orthopyroxene,
(Mg0.56 Fe0.44)2Si2O6, at 11 GPa and 1,6008C. This rapidly quenched crystal
displayed a high degree of disorder, KD ¼ 3.9. Comparison with low-pressure
orthopyroxene quenched from similar temperatures indicates that pressures corres-
ponding to the transition zone offer little effect on Mg – Fe ordering between M1 and
M2, unlike intra-crystalline ordering in several other dense magnesian iron silicates
(Hazen et al., 1993).

6.7.2.2. Orthoenstitite – clinoenstatite: LCLEN ! HCLEN

In the MgSiO3 system, the position of the orthoenstatite – high clinoenstatite – low
clinoenstatite triple point has not been fully determined experimentally. A linear
extrapolation from lower pressures of the low clinoenstatite , orthoenstatite boundary
(Grover, 1972) to orthoenstatite , high clinoenstatite boundary, determined by Pacalo
and Gasparik (1990), is seen to lie at about 7.9 GPa/9208C.
The phase-equilibrium behaviour of the pure end-member composition (MgSiO3) is
often used as a zeroth-order model for the phase-transition behaviour of the orthopyroxene
component of peridotite assemblages within the upper mantle.
The EOS of low clinoenstatite and orthoenstatite has been measured by Angel and
Hugh-Jones (1994) by single-crystal XRD to pressures in excess of 8.5 GPa. The volume
variation with pressure with low-clinoenstatite (LCLEN, space group P21/c) is described
by a third-order Birch –Murnaghan EOS with the parameters V0 ¼ 31.292 (8) cm3/mol,
0
KTO ¼ 111.1 (3.3) GPa and KTO ¼ 6.6 (1.1). LCLEN is seen to undergo a reversible
transition to a high clinoenstatite (HCLEN, space group C2/c) between 5.3 and 8.0 GPa
(Angel et al., 1992). This transformation is reversible but with significant hysteresis at RT.
The low clinoenstatite is recovered on decompression to below 5.3 GPa. The structure of
high-pressure clinoenstatite was shown earlier in Fig. 6.35c.

Orthoenstatite: thermoelasticity. Thermoelastic properties of (Mg, Fe)SiO3 orthopyrox-

ene are of great geophysical importance for understanding the composition and dynamics
of the Earth’s crust and upper mantle. Orthoenstatite (OREN) is the stable phase of
MgSiO3 at modest temperatures and pressures. Zhao et al. (1995) measured the P – V – T
data of MgSiO3 OREN by single-crystal X-ray diffraction at simultaneous high pressures
(4.6 GPa) and temperatures (up to 1,000 K). The P – V – T data are combined with
compressional data at room temperature (Hugh-Jones and Angel, 1979) and thermal-
expansion data at atmospheric pressure (Yang and Ghose, 1994). These are fitted by a
modified Birch– Murnaghan EOS for diverse temperatures, such that thermoelastic
properties of MgSiO3 OREN can be completely derived.
The final fitted thermoelastic parameters for MgSiO3 OREN derived by Zhao et al.
(1995) are as follows:
Thermal expansion, avo ¼ a þ bT, where a ¼ 2.86 (29) £ 1025 K21 and
b ¼ 0.72 (1) £ 1028 K22.
586 Chapter 6

Isothermal bulk modulus, K ¼ 102.8 (2) GPa, K0 ¼ dK/dP ¼ 10.2 (1.2), K0 ¼ dK/dT ¼
2 0.037 (5) GPa/K ¼ 2 0.0296 GPa/K (Frisillo and Barsch, 1972).
Adiabatic bulk modulus, Ks ¼ 107:8 GPa (Weidner et al., 1978).
The large value of (see also Yang and Ghose, 1994) indicates that this phase is also
very sensitive to changes in pressure and temperature.
The change of thermal expansion with pressure is directly related to the change of
bulk modulus with temperature. This thermodynamic relation is
   
da 1 dKT
¼
dP T KT2 dT P
Using the bulk modulus KTO ¼ 102:8 GPa and temperature derivative of bulk
modulus dK=dT ¼ 20:037 GPa=K obtained from the P – V – T EOS, the pressure
derivative of thermal expansion has been estimated by Zhao et al. (1995) as da=dP ¼
3:5 £ 1026 K=GPa:
Using the same data, Angel (Zhao et al., 1995) derived a slightly different set of
thermoelastic parameters by performing a fit in which each point is inversely weighted by
its uncertainty. The results obtained are

a ¼ 1:95 £ 1025 K21 ; b ¼ 2:23 £ 1028 K22 and K ¼ 96:6 ð2:4Þ GPa; K 0 ¼ 13:9 ð1:9Þ;

K 0 ¼ 20:022 GPa=K

Thermal Grüneisen parameter. As stated earlier, the thermal Grüneisen parameter (g th)
at ambient conditions can be derived from the P – V – T experimental data with a formula:
av KT
g th ¼
rCv
For MgSiO3 OREN, Zhao et al. (1995) obtained av ¼ 3:08 £ 1025 K21 ; KT ¼ 102:8 GPa
and r ¼ 3:196 g=cm3 : The heat capacity, CV, obtained by Frisillo and Barsch (1972) and
Yang and Ghose (1994) are 94.16 and 94.50 J/mol/K, respectively. Both values were
derived from acoustic modes, while the contributions from the optic branch were
neglected.
The values of av and Cv for OREN are not well resolved experimentally and hence
the value of gth is not well documented.

Anderson – Grüneisen parameter. The Anderson– Grüneisen parameter, a widely applied

parameter in Earth-mantle modelling, is directly related to the temperature derivative of
bulk modulus as
 
1 dKT
dT ¼ ð6-24Þ
av KT dT P
At ambient conditions, av ¼ 3:08 £ 1025 K21 ; KT ¼ 102:8 GPa and
dK=dT ¼ 2 0:037 GPa=K: The Anderson– Grüneisen parameter dT < 11.6, estimated by
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 587

Zhao et al. (1995) for MgSiO3 OREN, is large compared with other silicates and oxides but is
typical for pyroxene minerals (Frisillo and Barsch, 1972).

Anomalous volume behaviour. The volume variation of MgSiO3 OREN (space group
Pbca) displays an anomalous discontinuity at ,4 GPa. The EOS with reference to this
pressure is stated below (Angel and Hugh-Jones, 1994):
P ,4 GPa .4 GPa
V0 (cm3/mol) 31.354 (8) 31.23 (10) cm3
KTo (GPa) 95.8 (3.0) 122.8 (16.5)
K 0To 14.9 (2.0) 5.6 (2.9)

There is evidence for similar changes under compression in other pyroxene

structures (Zhang and Hafner, 1992). This is probably a manifestation of a general
phenomenon in complex silicates, which have enough structural degrees of freedom to
allow changes in the compression mechanism.

Vibrational modes. OREN (Pbca, D2h) has eight formula units (Mg2Si2O6) per unit cell.
The 80 atoms in a unit cell require 240 vibrations (3n, where n is the number of atoms per
unit cell) at any one point in the Brillouin zone. All atoms in the formula unit are unique
and are of Ci symmetry. The vibrations are distributed evenly among all the symmetry
species. The irreducible representation for OREN is

G ¼ 30A1g ðRÞ þ 30B1g ðRÞ þ 30B2g ðRÞ þ 30B3g ðRÞ þ 30A1u

þ 30B1u ðIRÞþ30B2u ðIRÞ þ 30B3u ðIRÞ

where R and IR denote Raman and infrared activity, respectively, of the various vibrational
symmetries. For a total of 120 Raman active modes, there are 30 vibrations for each of the
four Raman symmetries.
Chopelas (1999) has obtained the nearly complete 1-atm polarized spectra of
MgSiO3 OREN and high-pressure Raman data on OREN (Pbca) to 24.5 GPa and majorite
to 33.6 GPa. He found the slopes for the following transitions:
MgSiO3 phase transitions Slope (bar/K)
Clinoenstatite (C2/c) to majorite 212
Majorite to ilmenite 46
Ilmenite to perovskite 246
Majorite to perovskite 26

The Raman spectra of OREN at various pressures are shown in Fig. 6.41. Above
8.7 GPa, the prominent OREN peaks weaken in intensity while new peaks start to grow.
However, at 12 GPa, the OREN peaks vanish. At the highest experimental pressure
588 Chapter 6

Figure 6.41. Raman spectrum of synthetic MgSiO3 orthoenstatite 140– 1,200 cm21 at various pressures to
24.5 GPa at room temperature. Disappearance of some prominent peaks are seen at low pressures (e.g., that
marked with an asterisk at about 1,020 cm21), and some new prominent peaks appear above ,9 GPa (e.g., that
denoted by an arrow at about 680 cm21) (Chopelas, 1999, q 1999, Mineralogical Society of America).

(24.5 GPa), the final spectrum appears to be similar to C2/c clinoenstatite (or diopside)
(Chopelas and Boehler, 1992a).

6.7.2.3. Aluminous orthopyroxene: elasticity and velocities

From the Al2O3 content of enstatite co-existing with spinel (MgAl2O4) or pyrope
(Mg3Al2Si3O12), the geobarometry is derived. If the equilibrium temperature is known, the
equilibrium pressure can be calculated on the basis of Al2O3 content in enstatite. The Al
content increases with temperature and decreases with pressure. Exsolution of spinel from
enstatite is observed in xenoliths from kimberlites.

Elasticity and velocities. To a pressure of 12.5 GPa, the elastic constants, average bulk,
shear moduli and lattice parameters of a natural orthopyroxene from Kilbourne Hole have
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 589

been determined by Chai and Brown (1997). It contains 5 wt% Al2O3 and 1 wt% CaO.
The relation between its bulk modulus and P (GPa) is observed to be
K ¼ 115:4 þ 7:82P 2 0:18P2 ð6-25Þ
The isotropic shear modulus (G) shows a linear relationship with P (with
experimental uncertainty):
G ¼ 77:9 2 1:44P
At 1 bar, the isotropic longitudinal velocity (5.15 km/s) and shear velocity
(4.86 km/s) are ,3% faster than those found in non-aluminous orthopyroxene. At
P . 4 GPa, the average compressional wave velocities in the orthopyroxene exceed those
in olivine (Chai et al., 1996).
These results call into question the assumption, common in models created to
interpret mantle seismic data, that high velocities in garnet are balanced by low velocities
in orthopyroxene.

6.7.2.4. MD simulation
STOP simulation of the MgSiO3 polymorphs have been performed by using
molecular dynamics (MD) based on empirical model potential (Matsui and Price, 1992;
Winkler and Dove, 1992). Calculations on MgSiO3 perovskites were performed using the
electron gas model (Cohen et al., 1989) and two-body potentials derived from the modified
electron gas model (e.g., Bukowinski and Wolf, 1988).
Stixrude and Cohen (1993), using the LAPW method within the LDA, have
calculated the energetics of the rotations of the SiO6 octahedra responsible for the
cubic ! orthorhombic phase transition.
Very interesting phase transitions from OREN are simulated at high pressures at
,2,000 K. The orthorhombic phase is characterized by shared units of MgO6 octahedra
and SiO4 tetrahedra. These transform first at 8 GPa to a higher symmetry structure
involving SiO5(?) units and then at 11 GPa to a still higher symmetry structure involving
SiO6 and MgO8 units, similar to that present in the orthorhombic perovskite structure. Each
of the transition steps involves a volume collapse of ,10%. In the first step, the
compression is along the c-axis and in the second the compression is along the a-axis and is
large.
Based on the assumption of isotropic compression and no internal relaxation, the
EOS of the orthorhombic modification seems to be stable throughout the lower-mantle
pressure (D’Arco et al., 1993). In the MD simulation by Chapelot et al. (1996), the
orthorhombic phase remains stable beyond 100 GPa at ambient temperatures. However,
when the temperature is increased (at 70 GPa or so) the orthorhombic distortion decreases
and, at about 5,400 K, transition to a tetragonal phase occurs with a much smaller
distortion than for the cubic symmetry (at low temperatures). The tetragonal phase shows
many interesting properties. The tetragonal distortion is very small but the distortion
fluctuates locally and changes its direction among the a, b and c directions at a time scale of
a few picoseconds. Thus, the time-averaged symmetry may even be observed as cubic.
590 Chapter 6

The orthorhombic to tetragonal transition is also accompanied by a significant

increase in the simulated diffusion of the oxygen atoms, like a partial melting of the
oxygen sublattices, thus forming a solid ionic conductor. In an investigation on the full
melting of the lattice by Chapolet et al. (1996), it is found that, starting from a perfect
crystal, all the sublattices melt at ,7,000 K and 70 GPa at a time scale of 10 ps. But this
is somewhat large compared with the other measurements by Zerr and Boehler (1993)
at 5,000 K.

6.7.2.5. Ab initio simulation: Hartree –Fock

The relative stability of MgSiO3 – ilmenite, MgSiO3 –perovskite and (periclase þ
stishovite) assemblage phases as a function of pressure was investigated by D’Arco et al.
(1994) with the periodic quantum-mechanical ab initio Hartree – Fock program
CRYSTAL. The study showed that small basis sets reproduce correctly experimental
geometries. However, larger basis sets (“triple zeta” quality, plus polarization d functions)
are needed to yield significant thermodynamical results.
On the basis of the calculations, it appeared that in the investigated pressure
range (0 , P , 60 GPa) the mineralogical assemblage periclase þ stishovite has
higher enthalpy than MgSiO3 –ilmenite or MgSiO3 – perovskite and that ilmenite
transforms to orthorhombic perovskite at ,29.4 GPa, in good agreement with
experimental data.

6.7.3. Clinopyroxene

In all the polymorphs of monoclinic pyroxene, the structure consists of the same
stacking sequence of octahedral bands but differs in the magnitude of the rotation of the
tetrahedral chains. A straight chain (“S”) would have an angle of 1808 (^ up to 208) and
would result in the M2 cation-sites being eight-coordinated by oxygen atoms. A kinked
chain (“K”) with angles of less than 1458 reduces the coordination of the M2-sites to 6-fold
(Fig. 6.38a –c).
Pyroxenes of compositions on the MgSiO3 –FeSiO3 join show a structure with
P21/c symmetry, which is most stable at low temperatures and pressures. This structure
inverts to C2/c symmetry at high temperatures (Shimobayashi and Kitamura, 1991).
Generalized phase boundaries of clinopyroxenes can be seen in Fig. 6.37. At low pressure,
all FeSiO3 pyroxene polymorphs are metastable with respect to orthopyroxene (Angel and
Hugh-Jones, 1994).
Fe atoms in C2/c polymorphs occupy special positions (symmetry-constrained M1-
and M2-sites) within the unit cell. A symmetry reduction to P21/c results in the relaxation
of the symmetry constraints but this does not lead to a splitting of either the M1 or M2
positions into pairs of sites in the P21/c structure. In pyroxene structures, the geometry of
the M1-site is invariant with compositional changes, while that of the M2-site is
determined by the rotations of the tetrahedral chains.
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 591

Monoclinic clinopyroxene with C2/c structure is stable at high pressure and

modest temperatures for both the MgSiO3 and FeSiO3 end-members (e.g., Hugh-Jones
et al., 1994). For a typical mantle pyroxene composition of XEn ¼ 0.9, the monoclinic
C2/c polymorph becomes stable at depths of about 300 km. Clinopyroxene may be limited
to ,200-km thick zone, straddling the 400-km discontinuity (Woodland and Angel, 1977).
Hence, a knowledge of the thermodynamics becomes important for the construction
of viable mineralogical models of the upper mantle and transition zone to a depth of
,660 km.
In clinopyroxene, the bonding states of Fe2þ and Mg2þ at ambient states are
different. The substitution of Mg2þ for Fe2þ reduces the pseudo-axial crystal-field splitting
of Fe2þ, resulting in a decrease of distortion. Again, different electronic structures of Mg2þ
and Fe2þ dictate the intrinsic effective bond lengths of Fe– O and Mg –O, as well as the
different bonding characters in clinopyroxenes.

6.7.3.1. C2/c clinoenstatite

Wentzcovitch et al. (1995) applied the ab initio variable cell shape molecular
dynamics (VS-MD) algorithm to study MgSiO3-enstatite and HP clinoenstatite, C2/c.
The C2/c clinopyroxenes (Fig. 6.23.3a) contain 40 atoms per unit cell and 18
structural parameters (“free”). The Bravais lattice is monoclinic with b , 1068 (hence
called “clino”). High-pressure C2/c structures are also associated with monoclinic
b-angles that are in the range 100 –1028 (Hugh-Jones et al., 1994), which are significantly
smaller than other monoclinic pyroxenes ðb ¼ 1068Þ: The overall effect of pressure is to
decrease distortions away from the ideal packing of oxygens. In C2/c enstatite, the chain-
extension angle seems to go asymptotically towards 1348. Heating causes the cation
polyhedra to expand more than the strongly bonded tetrahedra. The kM1 – Ol and kM2 – Ol
bond lengths in orthopyroxene would increase at most by 0.5% upon increasing
temperature by 300 K, while the kSi – Ol bonds might actually contract slightly
(Wentzcovitch et al., 1995).
The chains are all equivalent and symmetrically related to one another. The
tetrahedral chains of (SiO3)2n n stretch along the c-axis and have a periodicity of two
tetrahedra. The basic topology of the structure, when viewed along the c-axis, was shown
in Fig. 6.35a. The M1-site lies between the apices of the opposing tetrahedra and the
M2-site lies between their bases. There is a structural transition to the low-pressure P21/c
phase between 5 and 7.9 GPa. This involves a tetrahedral chain instability. Thus C2/c
enstatite is thermodynamically stable above 7.9 GPa at room temperature. At lower
pressures, the structure distorts into another form of clinoenstatite with P21/c symmetry, in
which the tetrahedral chains are differentiated into “O-rotated” and “S-rotated”.
At 8-GPa pressure, a large fluctuation in the structural parameters causes a
change in tetrahedral tilting and chain-extension angles. Below and above 8 GPa, the
extension angles decrease steadily. The a lattice parameter also behaves anomalously
at this point and the polyhedral compressibilities change. Evidently, the fluctuations
occur at pressures close to the P21/c (low-pressure) to C2/c (high-pressure)
592 Chapter 6

transformation (7 – 8 GPa), which involves tetrahedral chain rotations (see Fig. 6.37)
(see also Wentzcovitch et al., 1995).
However, there are mineralogical cases displaying anomalous structural disconti-
nuities not involving symmetry changes, as is most notably seen in MgSiO3. In OREN, the
excessive chain flexibility accommodates heat, pressure and cation substitution and,
apparently, OREN can do so in a discontinuous manner.

6.7.3.2. Diopside – hedenbergite join

Pyroxenes with composition near Ca(Fe, Mg)Si2O6 on the diopside – hedenbergite
join have been investigated (at high temperature and high pressure) on isothermal com-
pression (Vaidya et al., 1973), for elasticity (Kendelin and Weidner, 1988), crystal structure
(Levien and Prewitt, 1981) and their phase transformations (Tamai and Yagi, 1989).
Mössbauer parameters, such as isomer shift, d and the quadrupole splitting, DEQ, of
57
Fe at M1 in Ca(Fe, Mg)Si2O6 clinopyroxenes, are observed to decrease with increasing
pressure between the ambient pressure and 10 GPa (Fig. 6.42; Zhang and Hafner, 1992).
Significantly, the study revealed abrupt changes in d and DEQ between 3.8 and 4.3 GPa.
The volume coefficients of d and DEQ are shown in Table 6.16.
The relative invariance of the isomer shift, d, to the substitution of Mg for Fe at
constant pressure is remarkable. It implies relatively invariable Fe – O bonds at M1-sites
occupied by Fe compared with the corresponding average (Fe, Mg) – O bonds at all
M1-sites, at least in the Fe-rich part of the Ca(Fe, Mg)Si2O6 system. With increasing
Mg substitution at M1, a decrease of the pseudoaxial crystal-field splitting D of Fe2þ is
seen (Zhang and Hafner, 1992).
The abrupt discontinuities in d (Fig. 6.42) and DE0 (see Fig. 6.43) found between
3.8 and 4.3 GPa are characteristic of a first-order phase transition. The hysteresis in DEQ
found in Hd80Di20 provides additional evidence that the phase transition is thermo-
dynamically of first order. The phase transition is reversible and could be of displacive
type. Prewitt et al. (1971) reported a thermal hysteresis for the P21/c– C2/c transition in
natural pigeonite.

Hedenbergite CaFeSi2O6: NFS spectra. Zhang et al. (1999) studied hedenbergite using
57
Fe nuclear forward scattering (NFS) spectra of synchrotron radiation at pressures up to
68 GPa. They observed a reversible phase transition between 53 and 68 GPa at room
temperature, which is probably a transition from the paramagnetic phase at low pressures
to a magnetic phase at high pressures.
The DEQ values determined from the time spectra of hedenbergite at low pressure
are in good agreement with the linear dependence of isomer shift (d) and quadruple
splitting DEQ on pressure between 0 and 4 GPa. These were consistent with the
discontinuity at ,4 GPa observed earlier by Zhang and Hafner (1992) (Fig. 6.42 and 6.43).
The time spectra recorded at 53.3 and 68 GPa showed changes in shape, signifying a new
phase transition.
STOP spectrum was fitted by use of the Hamiltonian for the general case of
the mixed electrostatic and magnetic interaction. The final hyperfine values obtained
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 593

Figure 6.42. Dependence of d of hedenbergite, Hd80Di20 and Hd60Di40 on pressure. There is a discontinuity at
approximately 4 GPa (cf. dashed line) (Zhang and Hafner, 1992, q 1992 Mineralogical Society of America).

in this way were

H0 ¼ 66 ^ 0:5 kOe; DEQ ¼ 1:2 ^ 2 mm=s
 
Vxx
h¼ ¼ 0:6 ^ 0:2; u ðH0 ^Vzz Þ ¼ 30^108
Vyy 2Vzz
594 Chapter 6

TABLE 6.16
Volume coefficients of d and DEQ

Sample Pressure range (GPa) (›d/› ln V) (›DE0/› lnV )

Hd 0.0–4.2 1.03 3.67

4.4–9.9 2.23 7.19
Hd60Di40 0.0–3.6 0.67 1.54
3.9–7.9 0.96 3.82

Note: V calculated from the lattice constants of Zhang et al. (1999).

and the angle w between the projection of H0 on the XY plane was constrained to be
equal to 0.
Hedenbergite is magnetically ordered below TN ¼ 41K: If the transition between
53 and 68 GPa is a magnetic transition similar to the one observed at 41 K, TN must depend
critically on pressure. The paramagnetic time spectrum obtained at ambient pressure after
pressure release is indicative of a reversible transition between 53 and 68 GPa. However, it
excludes a reconstructive transition (Zhang et al., 1999).

Compressibilities. For interpreting the ›d/›P and ›DEQ/›P, information on the

compression of the local volume around the position of the resonantly absorbing nucleus
is needed over the pressure range.

Figure 6.43. DEQ values. Solid squares refer to conventional Mössbauer velocity spectra. Open circles to NFS
time spectra at 0, 3.7, 5.9 and 13.7 GPa. The estimated total errors are indicated (Zhang et al., 1999, q 1999
Mineralogical Society of America).
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 595

TABLE 6.17
Axial and volume compressibilities of clinopyroxenes Ca(Fe, Mg)Si2O6 (bx £ 1023 GPa21)

ba £ 1023 bb £ 1023 bc £ 1023 bv £ 1023 ba : bb : bc

GPa21 GPa21 GPa21 GPa21 ( £ 1023 GPa21)

Hda 2.22 (8) 3.3 (1) 2.6 (1) 7.62 (7) 0.65 (4) : 1.00(6) : 0.7(6)
Hd60Di40b 1.9 (2) 3.1 (2) 2.2 (2) 6.5 (2) 0.6 (1) : 1.0(1) : 0.7(1)
Dic 2.58 (3) 3.274 (7) 2.597 (3) 7.85 (1) 0.78 (1) : 1.00(4) : 0.793(2)

Note: ba, bb, bc are the axial compressibilities, bv is the volume compressibility; errors calculated using error
propagation procedure.
a
0.0–3.7 GPa (Zhang et al., 1999).
b
0.0 –4.9 GPa (Zhang et al., 1999).
c
0.0–5.3 GPa (Levien and Prewitt, 1981).

Hedenbergite and Hd60Di30 exhibit anisotropic compression; b being the most

compressible direction. The compressibilities ba, bb and bc for the system Ca(Fe,
Mg)Si2O6 are listed in Table 6.17.
The bulk moduli K0 were calculated by Zhang and Hafner (1992) using a Birch–
Murnaghan equation of state (assuming ð›K0 =›PT Þ ¼ 4). The data were similar to those
obtained from Brillouin spectroscopy (Table 6.18).
The bond distances of the M1 polyhedra in Ca(Fe, Mg)Si2O6 clinopyroxenes
decrease with increasing pressure at constant Fe/Mg ratio; they also decrease with
decreasing Fe/Mg ratio at ambient pressure. For example, in hedenbergite the shortest
bond is dFe – O ¼ 2.087 Å and the average bond at M1 is kldllFe – O ¼ 2.130 Å and kldll
dMg – O ¼ 2.050 Å; and in diopside at ambient conditions kldllMg – O ¼ 2.077 Å at M1
(Cameron et al., 1973). The relative reduction of 2.5% of kldll from hedenbergite to
diopside at ambient pressure corresponds to 1.7% reduction in diopside from ambient
pressure to 5 GPa (Levien and Prewitt, 1981).

Tetrahedral compression. The high-pressure structure determinations of an intermediate

synthetic (Mg, Fe)SiO3 orthopyroxene by Hugh-Jones et al. (1997) also confirmed their

TABLE 6.18
Bulk moduli (Mbar) of clinopyroxenes

Samples K0a References

Hdb Synthetic 1.19 (2) Zhang and Hafner (1992)

Hd60Di40c Synthetic 0.827 (1) Zhang and Hafner (1992)
Hedenbergite Natural 0.71 (2) Vaidya et al. (1973)
Hedenbergite Synthetic 1.20 Kandelin and Weidner (1988)
Fassaite Natural 0.991 (15) Hazen and Finger (1977)
Diopside Natural 1.146 (4) Levein and Prewitt (1981)
Diopside Natural 1.22 (2) McCormick et al. (1989)
a
Fitted by use of a Birch–Murnaghan equation of state, assuming K 00 ¼ 4:0:
b
Fitted to data #3.7 GPa.
c
Fitted to data #3.1 GPa.
596 Chapter 6

earlier observation that the tetrahedra within the structure undergo significant compression
by ,8 GPa. At pressures in excess of ,4 GPa significant structural change occurs in
orthopyroxenes.
Raman data also indicate a non-symmetry breaking structural change between ,3.5
and ,6 GPa indicated by the change in the rate of change of the Raman frequencies with
pressure.
In natural orthopyroxene the presence of Ca2þ and/or Al3þ may significantly reduce
the compressibility. The presence of one Ca2þ per ,50 M2 cation-sites might be sufficient
to reduce the amount of M2 compression, thus suppressing the appearance of a high
pressure regime of tetrahedral compression. Study on synthetic sample by Hugh-Jones et al.
(1997) showed that when the 03A – 03A distance compresses beyond ,2.60 Å and the
03B – 03B distance below ,2.75 Å, significant tetrahedral compression is observed. These
O3 – O3 bond lengths are the most important structural elements in determining the
mechanism for compression of a Ca2þ-poor (Mg, Fe)SiO3 orthopyroxene.

Shocked diopside. In shocked diopside the defect microstructures consist of a large

variety of shock-induced defects, a high density of glide dislocations, (100)[101],
(110)[001] and (100)[010], mechanical twin lamellae (mostly parallel to (100)),
amorphous lamellae parallel to a few planes with low crystallographic indices and
heterogeneously distributed molten zones (3 – 20 mm size). The latter after cooling appear
as a glass with a chemical composition very close to that of the original diopside (Leroux
et al., 1994).
These features may serve as a diagnostic tool for recognizing impact phenomena on
planetary bodies of the solar system.

Clinoenstatite in diopside: “Alpe Arami”. The mineralogists of the University of

California while studying the garnet lherzolite from Swiss Alps (Alpe Arami) by TEM
found clinoenstatite lamellae exsolved in diopside grains. These lamellae probably
precipitated as high pressure c-centered monoclinic polymorph of enstatite, which is stable
only at depth . 250 km at upper mantle temperature. These diopsides also contain
numerous chromite exsolution lamellae and rare OREN lamellae.
This exsolution of clino-enstatite from diopside may have occurred at very high
pressure, P . 8 GPa, prevalent for a normal mantle geotherm (Pacalo and Gasparik,
1990). The clinoenstatite platelets, tens of nanometer wide, show crystal domains of subtly
differing orientation. High pressure laboratory studies have shown that such “antiphase
boundaries” form when the high pressure structure of a mineral relaxes to the low-pressure
form. The antiphase domains offer good evidence that high pressure clinoenstatite gave
rise to the present phase (Bozilov et al., 1999).
The clinoenstatite lamellae (SG P21/c) and diopside (SG C2/c) share the b axis, with
their c axes subparallel. These results imply that the rocks were exhumed from a minimum
depth of 250 km before continental collision (Bozhilov et al., 1999). The exsolution of
pigeonite lamellse in augite (or diopside) is quite common. There is a considerable
literature on the C2/c ! P21/c transformation in basic volcanic rocks in which subcalcic
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 597

augite (SG C2/c) first crystallized from the magma and then inverted to pigeonite (SG
P21/c).
Geophysicists consider that Alpe Arami, as a chunk of rock, could have risen
from hundreds of kilometres (,250 km) down. The mechanism may be postulated as
the two colliding continents push buoyant continental crust to mantle depths, where it
breaks free and bobs back to the surface, collecting a bit of the inherently dense
mantle as it moves up.

Polymorphism of low Ca-pyroxene. A similar transformation has also been seen to

take place in multianvil experiments upon decompression of HP clinoenstatite (e.g., Hugh-
Jones et al., 1996). HPclen also has the C2/c space group but has a different structure and
topology from augite. It has not been possible to retain this phase in low Fe pyroxenes
during decompression. Room temperature decompression induces C2/c ! P21/c trans-
formation with an over-stepping of the phase boundary by ,1 GPa. The most striking
differences between Htclen and Hpclen lie in density change (Dr , 3%) and in the different
degree of kinking of the silicate tetrahedral chains, expressed macroscopically as different
angles of monoclinity (b), which are ,1108C in Htclen and ,1018C in Hpclen (Bozhilov
et al., 1999).
Pyroxenes with low Ca contents are known to crystallize in five different
polymorphs: protoenstatite (Pbcn), OREN (Oren, Pbca), low clinoenstatite (Lclen, P21/c)
and high pressure clinoenstatite (Hpclen, C2/c) (e.g., Hugh-Jones et al., 1996).
There are four possibilities for the occurrence of the Lclen lamellae in
diopside: (i) direct exsolution of Lclen, (ii) exsolution of Oren with later
transformation to Lclen by shearing or reconstructive growth, (iii) exsolution of
Htclen with later transformation to Lclen, or (iv) exsolution of Hpclen with later
transformation to Lclen.
Transformation of Oren to Lclen is a well-documented process (Hugh-Jones et al.,
1996). Oren ! Lclen transformation is reconstructive, involving breaking of strong bonds
and rearranging of the structure. The presence of antiphase domains within the lamellae
indicates that a precursor phase transformed to Lclen structure by a displacive
transformation.

Mg – Fe distribution (a) Intracrystalline distribution: The intersite distribution of Mg and

Fe2þ in the M1 and M2 octahedral-sites in high-P C2/c (Mg, Fe)SiO3 clinopyroxene was
determined on quenched samples by Woodland et al. (1997). The high pressure C2/c
polymorph is unquenchable and the recovered sample usually shows the low pressure
P21/c structure. Fe acts to lower the pressure for the orthorhombic to high pressure
monoclinic transition. The schematic P – T path for the synthesis of (Mg, Fe)SiO3
clinopyroxene adopted by Woodland et al. (1997) is shown in Fig. 6.37. Mössbauer spectra
at 81 K indicate that Fe2þ is strongly ordered in the M2-site.
The state of ordering in the high pressure clinopyroxene is similar to
orthopyroxene, especially if cation ordering in the high pressure clinopyroxene has a
small pressure dependence.
598 Chapter 6

(b) Intercrystalline distribution: For the Mg – Fe partitioning reactions between

co-existing ferromagnesian oxides and silicates the free energies involved are small (a few
kilojoules per mole of (Mg þ Fe2þ)).
O’Neill and Wall (1987) reviewed a variety of data on activity-
composition relations in Mg – Fe olivines and derived a virtual model with the
regular solution interaction parameter, Wol GMg – Fe ¼ 5,000 J/mol (one atom basis).
Wiser and Wood (1991) determined the activity-composition relations in Mg – Fe
olivines from the exchange of Mg and Fe between olivine and co-existing magnesio-
wüstite solid solution (with excess Fe metal) at 1,400 K and found WGMg –
ol
Fe ¼ 3,700 ^ 800 J/mol.
(i) olivine/pyroxene: The (Mg – Fe) distribution coefficient, KD, . between co-
existing olivine and orthopyroxene is defined as
ol opx ol opx
KD ¼ XFe XMg =XMg XFe

This depends on composition and on temperature, but near Fe/(Mg þ Fe) ¼ 0.1
(i.e., mantle composition) these effects cancel out and KD becomes insensitive to
temperature. The thermodynamic mixing properties of Mg –Fe2þ olivine solid solutions
show small and near-symmetric deviations from ideality, with Wol GMg – Fe ranging between
2,000 and 8,000 J/mol.
The olivine-garnet Mg – Fe2þ distribution studies by Hackler and Wood (1989) at
0.9 GPa (1,273 K) gave Wol GMg – Fe ¼ 3,700 ^ 108 kJ/mol.
An analysis of the data on activity coefficients in binary Mg – Fe2þ ferrite, chromite
and aluminate spinels from olivine-spinel Mg –Fe2þ partitioning (O’Neill and Wall, 1987)
sp
shows that these solutions are close to ideal, with WGMg – Fe ¼ 7,000 ^ 1,000 J/mol. The
2þ
olivine– ilmenite Mg –Fe partitioning data of Anderson and Lindsley (1979) at 1 bar
give Wilm opx
GMg – Fe ¼ 6,200 ^ 1,500 J/mol. For orthopyroxene, WGMg – Fe ¼ 2,800 ^ 900
J/mol is apparently independent of temperature.
For the mixing of divalent cations in a variety of structures, Davies and Navrotsky
(1983) found

W1 – 2 ¼ 100:8DV1 – 2 2 0:4ðkJ=molÞ

where DV1 – 2 ¼ V2 2 V1 =0:5ðV2 2 V1 Þ

(ii) pyroxene/garnet: With increasing pressure (10 – 15 GPa) all phases such as,
garnet, orthopyroxene and clinopyroxene become more Mg-rich. However, the relatively
Mg-poor majorite garnet becomes more abundant at these pressures. Due to the formation
of majorite garnet at pressures between 10 and 15 GPa, the composition of mantle olivine
in its polymorphic forms undergo change. The equilibrium partitioning relations between
these co-existing phases require that this gradual dissolution of Mg-rich pyroxene into the
garnet-majorite solid solution be accompanied by transfer of iron from olivine and its
polymorphs into the garnet phase, resulting in net Mg-enrichment of all polymorphs of
olivine.
(iii) Perovskite/magnesio-wüstite: With high iron content perovskite becomes
unstable and disproportionates into a mixture of magnesio-wüstite and stishovite. But iron
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 599

is not disproportionated equally between perovskite and magnesio-wüstite, it preferentially

goes into magnesio-wüstite. The distribution coefficient, KD, can be written as

KD ¼ ðXFe =XMg Þmw =ðXFe =XMg Þpv

where X is the wt% of Fe or Mg.

The value of KD is found to decrease with increase in pressure and above 40 GPa its
value (KD ø 3.5) remains constant.
(c) Thermobarometry: Alpe Arami. The Fe – Mg partitioning among garnet,
enstatite and diopside in the Alpe Arami massif indicates a temperature of
,875 ^ 508C. Thermobarometric measurements indicate maximum pressures to range
between 3 and 5 GPa. (Note: Temperature estimates are from the garnet –clinopyroxene
thermometer of Krog (1988) and the clinopyroxene –orthopyroxene thermometer of Brey
and Koehler (1990) and derived from the composition of mineral cores.) Brenker and Brey
(1998), however, show that the minimum conditions to satisfy for the origin of Arami
massif are 5 GPa (160 km) and 1,400 K, during 35 – 40 million years ago.

6.7.3.3. Enstatite– diopside – jadeite join: garnet

At pressures greater than 10 GPa the composition of mantle pyroxenes can be
closely approximated by the ternary system enstatite – diopside – jadeite (Gasparik, 1989).
With increasing pressure pyroxene transforms to garnet (e.g., Ito and Takahashi, 1987).
The garnet co-existing with compositions in the ternary system Mg2Si2O6(En) –
CaMgSi2O6(Di)– NaAlSi2O6(Jd) should be rich in calcium. Garnet was seen to reach its
maximum Ca content at 17 GPa. It exsolved CaSiO3 perovskite at higher pressures.
The maximum Na content of garnet limited by the co-existing pyroxene did not
exceed 51 mol% jadeite at 22 GPa and 2,1008C. At 22 GPa, pyroxene was replaced with
NaAlSiO4 (calcium ferrite structure) and stishovite under anhydrous conditions, while in
the presence of H2O a new hydrous Na-bearing phase with the ideal composition Na7(Ca,
Mg)3AlSi5O9(OH)18 was synthesized instead.
Garnet co-existing with CaSiO3 perovskite and MgSiO3 ilmenite at 22 GPa and
1,4008C was En51Di9Jd40, much similar to the first garnet appearing in the ternary system
at 13 GPa.
The new data (Gasparik, 1996) are applicable to the Earth’s transition zone (400 –
670 km depths) and suggest that the transformation from the eclogite to garnetite would
occur primarily over a limited depth interval from 400 to 550 km. Gaps in the observed
garnet compositions suggest immiscibility, which could potentially cause a sharp 400 km
discontinuity in an eclogite mantle.
The stability field of garnet expands with increasing pressure from 13 to 17 GPa by
the dissolution of pyroxene in garnet. But the field contracts between 17 and 25 GPa
through the exsolution of CaSiO3 perovskite and MgSiO3 ilmenite. At 17 GPa the
maximum solubility that has been noted is 50 mol% jadeite and 40% diopside. The
minimum solution composition is noted at 13 GPa as En51Di9Jd40; this is the same
chemistry obtained as the exsolution product at 22 GPa.
600 Chapter 6

Enstatite – jadeite join. Gasparik (1992) determined experimentally the phase relations on
the enstatite-jadeite join at solidus temperatures and 9 – 15.2 GPa and at 1,400 – 2,0508C at
17.5 –21.9 GPa, with a split-sphere anvil apparatus (USSA-2000). An immiscibility in
garnet was found and the stability of NaAlSiO4 (calcium ferrite structure) with stishovite
was determined.
To calculate a complete phase diagram for the enstatite – jadeite join at 500–
2,5008C up to 27 GPa a thermodynamic model was developed. The results indicate that the
400 and 670 km discontinuities could correspond respectively to the formation and the
breakdown of garnet with a pyroxene composition. Solidification was completed at 400 km
depth by crystallization of Na-enriched residual melts, which produced a pyroxene layer at
300 –400 km depth.
Mixing properties of the enstatite-jadeite garnet are based on the compositions of
garnet co-existing with pyroxene or ilmenite at 13.5 –21.9 GPa. The activities of garnet,
which is a solution of Mg-majorite (Mg4Si4O12) and Na-majorite (Na2Al2Si4O12), were
approximated by the Redlich –Kister equation identical to the one used for high
clinopyroxene.
Several univariant curves were calculated by Gasparik (1992) from the intersection
of divariant surfaces. These include the transition of orthopyroxene to high-P
clinopyroxene formation of garnet from two pyroxenes and the breakdown of garnet to
perovskite, NaAlSiO4 and stishovite. The univariant curve indicating the formation of
garnet from two pyroxenes has a slightly positive slope and occurs at pressures
corresponding to the 400 km discontinuity.
Although the stability field of two co-existing garnets on the enstatite –jadeite join
is limited, the two-garnet field could expand in more complex systems by stabilization of
the sodium-poor garnet to higher pressures. This happens on the enstatite – ferrosilite join,
as shown by Ohtani et al. (1991). Thus, in more complex mantle compositions, two
co-existing garnets can be stable at 670 km depth. The 670 km discontinuity could thus
correspond to the break down of almost pure (Mg, Fe)4Si4O12 garnet to perovskite,
magnesio-wüstite, stishovite and minor sodium-rich garnet limited in composition to the
pyrope – NaPx join (þ Fe).
At low temperatures, magnesium and sodium are restricted to the cubic-site while
the octahedral-site is occupied only by aluminium and silicon. Expansion of the garnet
stability to enstatite-rich compositions with increasing temperature occurs primarily due to
disorder of magnesium over the two-sites. Aluminium remains restricted to the octahedral-
site; however, the slight excess of jadeite over En50Jd50 composition at very high pressures
reveals the presence of aluminium in the cubic-site and thus the possibility for a limited
aluminium disorder at such pressures.

Diopside – jadeite join (omphacite). Recent seismic tomography strongly suggests that
most of the subducted oceanic lithosphere terminates and stagnates in the transition zone
(e.g., Fukao et al., 1992). Since the basaltic component of the oceanic lithosphere is rich in
Na-and Ca-compositions under high P – T conditions pyroxene phases of diopside – jadeite
join become significantly important. Pyroxenes occurring in the blueschist and eclogite
facies metamorphic rocks have compositions which are usually very close to the diopside
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 601

(Di-CaMgSi2O6)- jadeite (Jd-NaAlSi2O6) binary system. Such pyroxenes are referred to as

omphacites. Diopside – jadeite join draws much of petrologists attention because
omphacites are major constituents of blue-schists and eclogites, which are the
metamorphic products of basalts. Omphacite-based geobarometry is used for blue-schists
and eclogites.
The results from an experimental study of the simple system Di-Jd-SiO2 can be
directly applied to natural occurrences. However, phase relations at low temperatures of the
blueschist facies are complicated by ordering of the intermediate Di-Jd pyroxene,
accompanied by a change in symmetry from C2/c to P2/n. Liu (1980) reported that the
diopside – jadeite pyroxene (omphacite) was stable up to 20 GPa at 1,0008C. The pyroxene
ranging in composition from diopside to 40 mol% jadeite transformed to a non-quenchable
phase named diopside II at 23 – 25 GPa, while the more jadeitic omphacite was stable to
greater pressures and transformed to NaAlSiO4 þ stishovite þ diopside II at 28 GPa.
Using the split-sphere anvil apparatus (USSA-2000), Gasparik (1989) determined
the compositions of co-existing pyroxene and garnet at 1,6508C and 15.2 –16.5 GPa with
starting composition close to the diopside –jadeite join. Later (1996) using the same initial
conditions he experimentally determined the phase relations on the diopside – jadeite join
at 16– 22 GPa at ,1,5008C under hydrostatic conditions and at 2,1008C under anhydrous
conditions.
In his experiments on diopside –jadeite join Gasparik (1996) used mixtures of high
purity wollastonite (CaSiO3), nepheline (NaAlSiO4), sodium disilicate (Na2Si2O5),
periclase (MgO), brucite (Mg(OH)2) and amorphous SiO2.
In the system enstatite – diopside jadeite (Mg2Si2O6 –CaMgSi2O6 – NaAlSi2O6) the
garnet that formed reached its maximum Ca content at 17 GPa and exsolved CaSiO3
perovskite at higher pressures. At 22 GPa pyroxene was replaced with NaAlSiO4 (calcium
ferrite structure) and stishovite under anhydrous conditions. In the presence of H2O,
however, a hydrous phase of composition Na7(Ca, Mg)3AlSi5O9(OH)18 was synthesized.
Activity-composition relations for Di-Jd solution illustrate the apparent pseudo-
ideality of the solution above 1,0008C (Wood et al., 1980). The increased stability of the
intermediate Di-Hd pyroxene, apparently is caused by ordering of cations in both
octahedral-sites.
The experimentally determined transition between the disordered and short-range
ordered C2/c pyroxene at 6008C should closely approximate the C2/c ! P2/n transition
observed in nature.
Disorder in the Di-Jd solution was found by Gasparik (1985) to be significantly
smaller than complete disorder implied by the ionic two-site model, which was used by
Holland (1983) for extrapolating the 6008C phase relations to higher temperatures. Using
the available data Gasparik (1985) determined the following mixing properties of the
diopside (Di)-jadeite (Jd) solid solution

GXS ¼XJd XDi ½12;60029:45T þð12;60027:6TÞðXJd 2XDi Þ2ð21;400216:2TÞðXJd 2XDi Þ2 

The Di-Jd solution is close to ideal above 1,0008C but miscible below 5658C.
The Di-Jd solvus is slightly asymmetric, with the crest at composition Di42.4Jd57.6.
602 Chapter 6

Garnet co-existing with CaSiO3 perovskite and MgSiO3 ilmenite at 22 GPa and
1,4008C was En51Di9Jd40, coincidentally identical to the first garnet forming on the
En-diop-jd ternary. These data suggest that the transformation from eclogite to garnetite
would occur at the pressure range for 400 – 500 km depth.

Enstatite – diopside join: melting relations. Melting relations on the Mg2Si2O6(En) –

CaMgSi2O6(Di) join were determined by Gasparik (1996) at 7 to 22.4 GPa pressure in a
split-sphere apparatus (USSA-2000). Melting is peritectic in enstatite-rich compositions at
7 –12.4 GPa (1,840 –2,1008C) and eutectic at higher pressures; while the diopside rich
clinopyroxene melts azeotropically at 7 – 16.5 GPa and up to 3008C lower temperatures
than the eutectic. Orthopyroxene is replaced with enstatite-rich clinopyroxene at 12 GPa
and 2,0908C.
Through melting experiments Gasparik (1996) identified the following two-phase
assemblages along the enstatite – diopside solidus:
(i) at 7 –12 GPa: orthopyroxene þ clinopyroxene
(ii) at 13 –15.8 GPa: two-clinopyroxene
(iii) at 15.8 GPa: garnet þ clinopyroxene
(iv) at 17 and 18 GPa: garnet þ CM phase
(v) at 19 –22 GPa: perovskite
(vi) .22 GPa: two perovskites.

6.7.3.4. Clinopyroxene and anorthite

Though the atoms in the high P – T structures of (Mg, Fe) clinopyroxene and
anorthite occupy the same Wyckoff-sites in structural conformation, they are quite distinct.
The P1̄ symmetry of anorthite (at room P – T) transforms at both high P and T (,2408C) to
I1̄ symmetry phase. Christy and Angel (1995) described the energetic of the high-P, T
isosymmetric crossover or phase transitions in terms of an appropriate order parameter and
showed that the stability of the low-symmetry phases arises naturally as a consequence of
simple inhomogeneity in the order parameter.

6.7.3.5. Potassium in clinopyroxene

Cation vacancies may provide localized strain relief to facilitate substitution of
potassium into the pyroxene structure. Harlow and Veblen (1991) suggested that K would
more readily substitute into clinopyroxenes with larger unit-cell volumes. They further
proposed that substitution of larger Cr3þ or Fe3þ for Al3þ in the M1-site helps expand the
M2-site to accommodate K. Harlow (1996) suggested that substitution of Mg into M2-site
would facilitate K substitution. Because Mg is smaller that Ca, small M2 polyhedra
occupied by Mg could compensate for larger M2 polyhedra containing K. Similarly in
clinopyroxene K would be accommodated in large M2-site through coupled substitution of
K and smaller cations in M2 and a larger cation Cr3þ in M1. Again, the polyhedral
compressibilities of Na and K are large in comparison with Ca and hence Na and K
decrease in size more with increasing pressure. This explains the observed increase in K
content of clinopyroxene with increasing pressure.
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 603

An aluminous garnet M3Al2Si3O12 (M ¼ Mg, Fe, Ca, etc) can be seen as

intermediate between MSiO3 and Al2O3. Mg-pyroxenes dissolve in garnet with increasing
pressure, leading to the formation of majoritic garnet following the reaction
M3 Al2 Si3 O12 þ 4xMSiO3 ! ð1 þ xÞM3 Al2=ð1þxÞ ðM0:5 Si0:5 Þ2x=ð1þxÞ Si3 O12
with 0 , x , 3 (see Fig. 6.44). In majorite, two Al3þ cations are replaced by one divalent
and one tetravalent cations.

6.7.3.6. Pyroxene– garnet transition: Martian mantle

A pyroxene garnet solid solution (at 1,4008C) and its phase transition are shown in
Fig. 6.44. In the enstatite – diopside system two clinopyroxenes appear at pressures
. 11 GPa, but with higher contents of CaSiO3 (Gasparik, 1990; Bertka and Fei, 1997). At
pressures between 16 and 17 GPa and temperatures . 1,6008C majorite replaces
clinopyroxene.
Dissolution of pyroxene in garnet has very little effect on thermal expansion, but
substitution of iron and magnesium in the pyrope – almandine join seems to have a large
effect.

Figure 6.44. Phase relations in the system M3(M0.5Si0.5)2Si3O12 –M3Al2Si3O12 at 1,4008C. The upper most
phase’s boundaries represent complex transitions (modified from Madon, 1992, q 1992 Academic Press, Inc).
604 Chapter 6

When the d0 is calculated, using the (dK/dT)P values as reported by Soga (1967), its
values become 5.3 for pyrope and 6.4 for almandine. For the pyroxene-garnet solid
solution it is observed (Yagi et al., 1987) that with increasing pyroxene component the bulk
modulus seems to decrease. And also that the thermal expansion of majorite is expected to
be similar to that of garnet with the same Fe/Mg ratio.
A study on iron-rich mantle composition (Mg #75) at ,3.0 GPa produced
subsolidus clinopyroxene composition whose CaSiO3 was lower than expected for a
Mg #90 mantle at similar P, T conditions (Bertka and Holloway, 1993). Compared with the
Earth’s mantle an increased iron content in the Martian mantle will lead to clinopyroxene
with lower CaSiO3 in the Martian mantle. The complete transition of clinopyroxene to
majorite will occur at a similar pressure (,16 GPa) in both the planets’ mantles, if the
corresponding aerotherm or geotherm temperature is high, ,1,6008C (Bertka and Fei,
1997).
A change in Al and Si atomic concentrations of majorite with pressure is
observed. Majorite first appears at 13 GPa and continues growing in proportion up to
17 GPa. At higher pressures when pyroxene is no longer stable and Mg – Fe perovskite
appears the concentrations of Al and Si remain nearly constant. In the Martian mantle
assemblage, determined by Morgan and Anders (1979) (designated briefly as MA), a large
increase in majorite abundance and the appearance of magnesio-wüstite occur at the
expense of b phase and g-spinel.

Clinopyroxene-garnet at the transition zone. In subducting oceanic lithosphere, peridotite

and basalt constitute the major rock types, which contain diopside (CaMgSi2O6) as a major
mineral. Subsolidus phase equilibrium studies on these rocks in the P – T conditions of the
transition region are of fundamental importance in interpreting the seismic structure of the
transition zone and in predicting the likely phase transitions experienced by the subducting
oceanic lithosphere (Takahashi and Ito, 1987; Gasparik, 1990). For this objective in view,
one may examine the breakdown of diopside in the thermal range prevalent at the
transition zone. At temperatures above 1,6008C diopside transforms according to the
breakdown reaction

CaMgSi2 O6 ¼ CaSiO3 þ ðCa; MgÞSiO3

Di Ca-Pv Gt

whereas at temperature below ,1,3008C diopside decomposes as

CaMgSi2 O6 ¼ CaSiO3 þ MgSiO3

Di Ca-Pv Ilm

Gasparik (1989) reported the appearance of (Ca, Mg)SiO3 garnet along the join
diopside – enstatite at 15.7 GPa and 1,6508C. Later, Canil (1994) found that diopside
transforms to CaSiO3-perovskite and (Ca, Mg)SiO3-garnet or ilmenite bearing
assemblages at pressures of 17– 18 GPa. The Clapeyron slopes of the diopside breakdown
reactions are found to be steep and negative (22 MPa/K), consistent with most other
perovskite-forming reactions in simple systems related to the Earth’s mantle (e.g.,
Navrotsky, 1989).
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 605

Canil (1994) experimented on a natural fertile peridotite composition (KLB 1) at

1,5008C and under pressure of 17 and 18 GPa. These conditions were selected to
bracket the intersection of transition region geotherm with the 520 km discontinuity in
mid-transition zone (see also Fig. 6.45). At 17 GPa an assemblage of b-(Mg,
Fe)2SiO4 þ g-(Mg, Fe)2SiO4 þ majorite-rich garnet was produced. But at 18 GPa,
CaSiO3-perovskite was observed to have exsolved from garnet. When coupled with a
re-examination of clinopyroxene stability in other more complex peridotite systems, the
experimental data of Canil (1994) support the proposal that the 520 km discontinuity in
transition region is caused by the exsolution of CaSiO3-rich perovskite from majorite-
garnet. The physical properties of some mid-transition zone phases are presented in
Table 6.19.
A portion of the KLB-1 peridotite P – T diagram is shown in Fig. 6.45 (Zhang and
Herzberg, 1994).

Supersilicic clinopyroxene. Pressure stabilizes supersilic clinopyroxene. Occurrences

of silica rods in clinopyroxenes and experimental results have amply demonstrated that
topotaxial growth of quartz lamellae in eclogitic Na – Ca clinopyroxenes results from
exsolution from a former supersilicic clinopyroxene stabilized at VHP conditions.

Figure 6.45. Portion of KLB-1 peridotite P – T phase diagram after Zhang and Herzberg (1994). The width of
the b-phase g-spinel is adopted from Kito and Ito (1989) (Source: Agee, 1998, q 1998 Mineralogical Society
of America).
606 Chapter 6

A good amount of experimental studies on the pressure solubilities of SiO22

clinopyroxenes have been carried out. Of these, some selected ones are cited below.
Workers Clinopyroxene Conditions Comments
P (GPa) T (8C)
Mao (1971) Jd–Ca Tsch ,4.0 1,100–1,700 7.5% excess SiO2
Wood and Henderson (1978) Di–Ca Tsch (CaAl2SiO6) 2.5 –3.5 1,200–1,450
Zharikov et al. (1984) Di–Ca-eskola (Ca0.5AlSi2O6) 3.5 –5.7 1,2008C
Angel et al. (1988) Na–Mg 10 –15 1,600 Silicon as
(IVSi þ VISi)

Oriented needles of quartz (,2 –20 mm wide and ,5 –200 mm long) have been
seen to occur within Ca –Na clinopyroxene (Cpx II) in the eclogite facies rocks of North
Dabie complex (NDC), China (see Section 11.6.3).
These SiO2 needles are interpreted as exsolution products from a precursor, non-
stoichiometric omphacitic clinopyroxene, which contained excess silica at the
peak metamorphic (pressure) conditions. The supporting evidences include (Tsai and
Liou, 2000):
(a) quartz needles occur only in matrix clinopyroxene (Cpx II), but not within any other
phases nor as an individual phase in the matrix;
(b) omphacitic clinopyroxene included in garnet does not contain quartz needles;
(c) needle-bearing matrix Cpx II shows systematically higher Si contents compared with
that of needle-free matrix clinopyroxene (Cpx III);
(d) “broad-beam” EPM analyses of the needle-bearing Cpx-II indicate a supersilicic
nature of the precursor, according to definition: (Si þ Ti) . (Ca þ Mg þ Fe2þ þ
Mn þ Ni – 2Na);
(e) the needle-bearing core of the matrix clinopyroxene has a higher Na-content
(Jd10 – 17), whereas the needle-free rim has lower Na content (Jd , 10);
(f) the quartz-needle microstructure resembles quartz exsolution in omphacite or
Ca –Na-clinopyroxene reported from coesite-bearing eclogites elsewhere.

6.7.3.7. FeSiO3: clinoferrosilite

The ortho-clino FeSiO3 phase boundaries of Akimoto et al. (1965) are shown in
Fig. 6.46. In situ single crystal XRD experiments on FeSiO3 by Hugh-Jones et al. (1994)
have revealed a transition with 3% decrease in unit cell volume and considerable structural
rearrangement of the tetrahedral chains. All through the transition the M1 and M2-sites
occupied by Fe2þ ions behave differently. The size and distortion of M2-site change
significantly compared with those of M1-site, which is least altered. In the P21/c structure
the M1-site shows no change in covalency or orbital occupation and a small increase in
ground state splitting. The M2-site, on the other hand, undergoes change in bonding or a
change in ground state splitting. For the M1-site the change in electronic structure is small.
Only a slight increase in covalency of the Fe –O bonds occurs along with a small increase
in the ground state splitting. The latter is possibly arising from the changes in the nnn
environment due to M2 atoms. For the M2-site the changes become greater due to
 MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes
TABLE 6.19
Physical properties of some mid-transition zone phases (at 1 bar and 298K where applicable).

Garnet (Mg3Al2Si3O12 þ Fe3Al2Si3O12 Majorite (MgSiO3 þ FeSiO3) Clinopyroxene (CaMgSi2O6 Ca-perovskite

þ Ca3Al2Si3O12) þ CaFeSi2O6) CaSiO3

Density (g/cm3) 3.561(XPy) þ 4.318(XAl) þ 3.617(XCa)a 3.522b þ 0.973(XFeSiO3) 3.277 þ 1.73(XCaFeSi2O6)c 4.23d
Thermal expansion (K21) £ 1025 2.38 2.38 2.55 2.5
Bulk modulus KS (GPa) 175(XPy) þ 176(XAl) þ 169(XGr) 160b 113 286d
Shear modulus G (GPa) 90(XPy) þ 98(XAl) þ 104(XGr) 90b 67 159e
dKs/dP 4.9 5.8d 4.5 4.0
dG/dP 1.4 2.9d 1.7 1.9
2dKS/dT (GPa/8C) 0.021 0.021 0.013 0.027
2dG/dT (GPa/8C) 0.010 0.010 0.010 0.023
Gruneisen parameter, g 1.24(XPy) þ 1.06 (XAl) þ 1.05(XCr) 1.24 1.06 2 0.11(XCaFeSi2O6) 1.96
Debye temperature, u (K) 981(XPy) þ 909 (XAl) þ 904(XGr) 949 2 129(XFeSiO3) 941 2 96(XCaFeSi2O6) 917

Values are from Duffy and Anderson (1989) and Anderson (1989) and from the following references:
a
Skinner (1966) and Akimoto and Akaogi (1977); b Pacalo and Weidner (1997); c Cameron et al. (1973); d Rigden et al. (1994); eMao et al. (1989).

607
608 Chapter 6

Figure 6.46. The ortho-clino FeSiO3 phase boundaries (using Akimoto et al., 1965 and Lindsley, 1965). The
pressures and temperatures of Lindsley’s reversals are represented by squares (Hugh-Jones et al., 1994, q 1994
Mineralogical Society of America).

contribution from the orbital occupation changes by transfer of electron density from 4s to
3d orbitals. In addition, with a probable increase in covalency of Fe –O bond the ground
state splitting is likely to decrease.
High pressure 57Fe Mössbauer effect experiment on synthetic FeSiO3 was
performed by McCammon and Tennant (1996) using DCA up to 3.8 GPa. A transition
to the C2/c structure was observed between 1.33 and 1.74 GPa and is characterized by a
large increase in quadrupole splitting (DEQ) for the M1-site and a small decrease in DEQ
for the M2-site.
Under pressure the changes in the isomer-shift (d0) and quadrupole splittings (DEQ)
of M1 and M2-sites in pyroxenes of P21/c and C2/c structures that have been noted by
them are shown in Table 6.20.
Evidently, in the C2/c structure the M1 and M2-sites behave in a more similar
manner as a function of pressure. Further, the distortion of the M2 (more distorted than
M1)-site decreases with increasing pressure up to 6.5 GPa (Hugh-Jones et al., 1995).
The experiments by Hugh-Jones et al. (1994) have established that the high pressure
phases of MgSiO3 (Angel et al., 1992) and FeSiO3 having space group C2/c, show a large
stability region. The transition from the low-pressure form with space group P21/c to the
C2/c modification can be observed at room temperature and high pressures; and this
transition is reversible. Through the phase transition from P21/c to the C2/c structure
 MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes
TABLE 6.20
Pressure dependence of d and DEQ for clinopyroxenes (cfs, clinoferrosilite; hd, hedenbergite)

Phase Site P21/c C2/c References

dd/dP (mm/s/GPa) cfs M1 28(4) £ 1023 27(28) £ 1024 McCammon and Tennant (1996)
cfs M2 6(4) £ 1023 21(3) £ 1023 McCammon and Tennant (1996)
hd M1 28 £ 1023 Zhang and Hafner (1992)
ddSOD /dP (mm/s/GPa) cfs 22(1) £ 1023 22(1) £ 1023 McCammon and Tennant (1996)
hd 21 £ 1023 Zhang and Hafner (1992)
dd0 /dP (mm/s/GPa; calculated) cfs M1 26(4) £ 1023 1(3) £ 1023 McCammon and Tennant (1996)
cfs M2 8(4) £ 1023 1(3) £ 1023 McCammon and Tennant (1996)
hd M1 27 £ 1023 Zhang and Hafner (1992)
dDEQ /dP (mm/s/GPa) cfs M1 6.3(9) £ 1022 6(6) £ 1023 McCammon and Tennant (1996)
cfs M2 4.0(9) £ 1022 3.6(6) £ 1022 McCammon and Tennant (1996)
hd M1 23 £ 1022 Zhang and Hafner (1992)

609
610 Chapter 6

the M1 and M2-sites behave differently. The variation with pressure of unit cell volume
(Å3)and b angle of clinoferrosilite is shown in Figs. 6.47a,b.

FeSiO3 solubility. The maximum solubility of FeSiO3 in perovskite structure is important

for understanding the nature of 660 km discontinuity (Fei et al., 1991).
The solubility of FeSiO3 component into perovskite is limited to (Mg0.8Fe0.2)SiO3
composition and an assemblage of Mg-wüstite (Mg0.24Fe0.76)O plus stishovite co-exists
with perovskite in the compositions with iron content higher than (Mg0.8Fe0.2)SiO3 (Yagi
et al., 1979).
The solubility of FeSiO3 component into ilmenite and perovskite is found to be
limited to nearly 10 mol%. At pressure above ,25 GPa, the system MgSiO3 – FeSiO3
is divided into three stability fields according to FeSiO3 content (XFe): perovskite
for 0 , X Fe , 0.1; perovskite (XFe , 0.1) þ (Mg0.5Fe 0.5)O þ stishovite for
0.1 , XFe , 0.5; and (Mg, Fe)O þ stishovite for 0.5 , XFe , 1 (Ito and Yamda, 1982).
This suggests that some amount of oxide mixture of stishovite and relatively iron rich
Mg-wüstite would be present with Mg-perovskite in the lower mantle.

CaSiO3: wollastonite II. The high pressure polymorph of CaSiO3, wollastonite II

(or wahlstromite) is stable at pressure above 3.5 GPa at 8008C and 3.0 GPa at 1,0008C
(Essene, 1974; Huang and Wyllie, 1975b). This phase may occur in some VHP
impure marble.

6.7.3.8. Na-pyroxene
Na(Mg0.5Si0.5)Si2O6: [6]Si. Single crystal of Na-pyroxene [Na(Mg0.5Si0.5)Si2O6], were
synthesized from a stoichiometric mix of high-purity SiO2, MgO and Na2Si2O5 at 1,6008C
and 15 GPa using a split-sphere anvil apparatus (USSA-2000) (Gasparik, 1988; Angel
et al., 1988).

Figure 6.47. (a) The variation of b with pressure of the clinoferrosilite (FeSiO3), showing a distinct discontinuity
of some 58 at the transition from P21/c to C2/c. The uncertainties in the measurements of b and the pressure are
considerably less than the size of the symbols used. (b) The variation of unit-cell volume of the clinoferrosilite
(FeSiO3) with pressure, showing a discontinuous jump of about 12 Å3 at the position of the P21/c to C2/c
transition. The uncertainties in the measurements of both the volume of the unit cell and the pressure are
considerably less than the size of the symbols used (Hugh-Jones et al., 1994, q 1994 Mineralogical Society
of America).
MgO – FeO –SiO2 (MFS) System: Olivines and Pyroxenes 611

This Na-pyroxene contains excess silicon, some of which are in octahedral

coordination. The difference in size and charge between Si and Mg results in ordering of
these two cations within the M1-sites, while M2 remains occupied by Na. But the size
difference is so large that it results in total ordering in M1 and M1(1)-sites. The M1(1)– O
mean distance found as 1.811 Å is similar to that in octahedrally coordinated Si, seen in
stishovite, MgSiO3 ilmenite and perovskite.
The stability of this silica-rich sodium pyroxene is conditioned by excess silicon
over aluminium. The incorporation of excess Si could stabilize the pyroxene and expand its
stability field to higher pressures.

Na2 – MgO –Al2O3 – SiO2 system: forsterite– jadeite join. The Fo – Jd join belongs to the
quaternary system Na2O – MgO –Al2O3 –SiO2. Such a system can have up to four co-
existing phases: forsterite, clinopyroxene, garnet and Na2Mg2Si2O7 (NMS), with NMS
being the solidus phase. The melting curves of NMS and Na2MgSiO4 (N2MS) were
determined from 1 atm to 22 GPa. These phases are expected to form in the Earth’s mantle
instead of nephelines at high pressures (higher than the stability of nepheline).
NMS bearing assemblages are expected to form in the Earth’s mantle from
nepheline normative compositions at the pressures at which nepheline is no longer stable.
Partial melting of the NMS-bearing assemblages at 4 –6 GPa produces melts which lie to
the N2MS-rich side of the Fo– Jd join. These melts are unusual by their Na/Al ratios much
higher than unity.
The partial melts generated at even greater depths than the alkali basaltic magmas
will again have a unique signature reflecting the melting of NMS in the source region. It is
possible that such melts are responsible for the alkaline volcanism associated with refitting
of thick continental lithosphere.
Melting experiments on the forsterite –jadeite (Fo – Jd) join (Mg2SiO4 – NaAlSi2O6)
were investigated by Gasparik and Litvin (1997) using USSA-2000 (see “Na(Mg0.5Si0.5)-
Si2O6: [6]Si”) at 4– 22 GPa. This experiment was done to determine the character of the
partial melts originating at depths greater than those for alkali basaltic melts.
The resulting Fo –Jd solidus can be approximated by the following Simon equation:

PðGPaÞ ¼ 4 þ {½TðKÞ=1; 4736:2 2 1}

Fosterite and jadeite were seen to start reacting in the vicinity of the solidus.

(Na, K)0.9 (Mg, Fe)2 (Mg, Fe, Al, Si)6O12: a lower mantle Al-phase. A new Al-rich
phase, [K0.56Na0.32][Ca0.04Mg1.66Fe2þ 3þ
0.3 ] [Mg0.98Fe0.3 Al2.62Si2.1]O12 was synthesized at
24 GPa, 1,7008C by Gasparik et al. (2000). Single crystal XRD revealed the unit cell to be
hexagonal (SG P63/m), the other parameters are
 c ¼ 2:779ð1ÞA;
a ¼ 8:830ð1ÞA;  3 and Z ¼ 1; r ¼ 3:97g=cm3
 V ¼ 187:65ð5ÞA

Compositional variations and high density suggest complex solid solution behaviour of
this phase at the lower mantle pressures, similar to pyroxenes at the upper mantle and
majorite garnet at the transition zone pressures.
612 Chapter 6

The Al-rich phase synthesized at 22 GPa has a substantially higher Al content

than observed at 24 GPa and may represent a composition at the very limit of its stability.
The stoichiometry requires Ca to be present at M3-site and the substitution CaAl ¼ NaSi to
occur. This phase may be a host for Na in the lower mantle. The other likely phase is
NaAlSiO4 in calcium ferrite structure. Nevertheless, the Al content of an average mantle is
too low to allow for the widespread occurrence of such Al-rich phase.

6.7.4. Akermanite, CaMgSi2O7: incommensurate to normal phase transition

Akermanite, Ca2MgSi2O7, is an end member of the melilite group of minerals and

usually occurs in igneous and metamorphic rocks, meteorites and blast furnace slag.
It is recognized that akermanite and iron-akermanite Ca2(Fe, Mg)Si2O7, exhibit an
incommensurate (1C) structure at room temperature with modulations along [110] and
 and that they undergo a reversible phase transition to a high-temperature normal (N)
½110
structure at 352– 523 K, depending on the composition (e.g., Seifert et al., 1987). A review
of the macroscopic and microscopic changes at the 1C –N phase transition in compounds
of similar behaviour, such as melilites can be seen from Seifert and Röthlisberger (1993).
The crystal structures of both the 1C and N phases of akermanite are tetragonal
(SG ¼ P421m), characterized by tetrahedral sheets consisting of [Si2O7] dimers
interconnected by tetrahedrally coordinated Mg cations in the form of five membered
rings. The sheets lie parallel to the (001) plane and are linked together along c by large Ca
cations in distorted eight-coordinated-sites (Yang et al., 1997).
Seifert et al. (1987) found two distinct [(Fe, Mg)O4] tetrahedral-sites in modulated
Fe –akermanites. These two-sites become less distinguishable at critical temperature,
where phase transition occurs and merge into one-site in the normal structure.
Yang et al. (1997) investigated the structural change associated with the
incommensurate (1C) to normal (N) phase transition in akermanite under high-pressure
using single crystal XRD up to 3.79 GPa. At ,1.33 GPa the 1C phase is seen to transform
to the N phase.
The structural transformation is marked by a small but discernable change in the
slopes of all unit cell parameters as a function of pressure. It is reversible with an apparent
hysteresis and is classified as tricritical phase transition.
From room pressure to 3.79 GPa the volume of the [SiO4] tetrahedron is unchanged
(2.16 Å3), whereas the volumes of the [MgO4] and [CaO8] polyhedra decrease from 3.61 to
3.55(1) Å3 and 32.8 to 30.9(2) Å3, respectively.
At ambient condition there is a mismatch between the size of Ca atoms and the
configuration of tetrahedral sheets, which appears to be responsible for the formation of
tetrahedral structure. With pressure the misfit diminishes through the relative rotation and
distortion of the [MgO4] and [SiO4] tetrahedra.
 613

Chapter 7
(K2O, Na2O, CaO)– Al2O3 – SiO2 System

7.1. KAlSi3O8 – NaAlSi3O8 – CaAlSi3O8 felspars

The felspar group (KAlSi3O8 – NaAlSi3O8 – CaAl2Si2O8 system) amongst the

tetrahedral framework alumino-silicates comprise ,60% of the total volume of the
Earth’s crust. Felspars are kinetically stable at normal condition. Their thermodynamic
properties such as bulk moduli, vibrational entropy and heat capacities can be determined
from spectroscopic studies under pressure. The variations in bulk moduli can be
understood from the pressure variation of unit-cell parameters.
The low symmetry of ordered felspars maximizes the degrees of freedom associated
with compression mechanism. Felspars compress mainly by bond-angle bending, while the
high-symmetry closest-packed materials (occurring in deep Earth) compress mainly by
bond shortening. The T – O –T angle bending appears to control the compression of these
structures but the chemistry of the tetrahedral cations and bonding of the extra framework
cations constrain the energetics of angle bending. The tetrahedral cations can affect
significant changes in bulk modulus whereas the extra-framework cations affect
compression mechanisms as well.
The felspars show an effectively non-collapsible framework structure where
changes in volume can be associated with changes in individual T – O –T angles but the
overall angle does not vary. In felspars, the T –O –T angle changes from being Si – O –Si
alone (as in silica) to being a mix of Si – O – Si and Al –O – Si (or B – O –Si). Al – O –Si
appears to be twice as soft as Si – O – Si whereas B – O – Si is twice as stiff
(Nicholas et al., 1992).
The felspar bulk moduli are controlled by the stiffness of the T –O – T angles and the
strength of the M– O bonds. For details of the effects of pressure on the felspar group of
minerals, readers are advised to see Angel (1994). The variation of the T –Oc –T angle vs.
the modulus of some felspars is shown in Fig. 7.1. The bending energy of the T– O – T
angle is primarily affected by the charge of the M cation and the M– O bond length. Also,
the energy required to bend the T – O –T angle is increased significantly if the bridging O
is bonded to three or more atoms (Geisinger et al., 1985).
Substitution of different tetrahedral cations affects the compressibility of the
felspar structure. Compared with anorthite, the sodic and potassic felspars are seen to be
more compressible. This happens because, in anorthite, the electrons from closer and
more highly charged Ca atoms exert more influence on the bridging bond than what
614 Chapter 7

Figure 7.1. The variation of the average T–Oc – T angle vs. bulk modulus for some felspars. Anorthite does not
follow the trend of alkali felspars, possibly for the effect of the relatively strong Ca–O bonds affecting the strain
direction in the structure (Downs et al., 1994; q 1994 Mineralogical Society of America).

happens in the K and Na felspars. Sodium felspars are less compressible than potassium
felspars and the molar volume of sodium felspars is smaller. This is because the Na – O
bond is shorter than the K – O bond. The electron density map shows that K is 7-fold
coordinated whereas Na is 5-fold (Downs et al., 1996). The K atom is bonded to all
bridging O atoms involved with Al –O – Si linkages (including Oco) (see Fig. 7.1).
Changing the bonding of the K atom would change the bending energetics of all
associated T –O – T linkages, which would then change the compression mechanism of
the structure. Microcline compresses by sharing the four-membered rings, which in turn
compresses the K-bearing channels. At ,4 GPa, a new bond between K and Obm appears
that alters the compression mechanism and this explains the discontinuity in the pressure
variation of crystallographic parameters observed by Allan and Angel (1997). Thus, the
compression behaviour of the alkali felspars is dominated by the compression of alkali-
containing channels. The observed variety of compression pathways results from T– O – T
angle-bending energetics that are coupled with the effects of alkali – cation bonding
(Downs et al., 1999).
Although in albite the Al –Oco – Si angle changes most under pressure, it shows the
least change in microcline structure (Fig. 7.2). However, in the latter, the Al – Obo –Si and
Si– Obm – Si angles display the greatest decrease with pressure. This corresponds to the
shearing of the four-membered rings.
With increasing P and T, the felspar structure with AlO4 tetrahedra becomes
unstable and new structures with AlO6 are formed. However, much below 4 GPa pressure,
the tetrahedra are not expected to change significantly.
(K2O, Na2O, CaO) – Al2O3 –SiO2 System 615

Figure 7.2. The structure of albite and microcline viewed down b p with 0:0 , y , 0:5: Differences in compression
mechanism between the two structures are illustrated. The T– Oco –T angle in albite bends in such a way that the
chains effectively slide over each other as demonstrated by the arrows in the bottom of the figure. In microcline the
major angle bending occurs at Obo and Obm, effectively shearing the four-membered rings as again demonstrated
with the arrows at the bottom of the figure (Downs et al., 1999, q 1999 Mineralogical Society of America).

7.1.1. Bulk moduli

In the 1920s, using pressure techniques, Adams and Gibson (1929) determined the
value for the bulk modulus of labradorite (Ab52An48) as 68 GPa However, the earliest
work dates back to 1923. A list of early workers on felspar studies is presented below:

Workers Material
Adams and Williamson (1923) Oligoclase (Ab78Ab22)
Labradorite (Ab48An52)
Microcline (Or91Ab9)
Bridgeman (1928) Adularia
Adams and Gibson (1929) Labradorite (Ab48An52)
Bridgeman (1948) Orthoclase (x2)
Labradorite
Yoder and Weir (1951) Albite
Oligoclase (Ab78Ab22)
Vaidya et al. (1973) Labradorite

In general, felspars are stiffer than framework silicates (e.g., quartz: KT ¼ 41:4 GPa;
Glinnemann et al., 1992) but softer than most other low-pressure minerals (e.g.,
pyroxenes: KT ¼ 100 – 120GPa; forsterite: KT ¼ 123GPa; Kudoh and Takeuchi, 1985).
In plagioclase felspars, the bulk modulus is seen to increase with anorthite content,
anorthite being about 45% stiffer than albite. The alkali felspars have very similar bulk
moduli (radius of Na ¼ 1:18 Å, K ¼ 1:51 Å) but anorthites containing high Ca (ionic
radius 1.12 Å) are much stiffer. The stereograms of Fig. 7.3 show that there is great
similarity between the directional distribution of the M –O bonds in both alkali and
plagioclase structures.
616 Chapter 7

Figure 7.3. The orientation of interatomic vectors in sanidine and anorthite. All vectors are plotted in the upper
hemisphere. The angular distribution of bonds in triclinic alkali felspars is similar to sanidine, except that each
single vector in sanidine is replaced by a closely spaced pair of vectors (Angel, 1994, q 1194 Kulwer, Dordrecht).

However, the T – T vectors across shared oxygen atoms and the T –O bonds do show
significant differences, which suggest that the tetrahedral framework is controlling the
compressibility of felspars (Angel, 1994). Under pressure, the rigid SiO4 and AlO4
tetrahedra in felspars remain almost unchanged but the T– O – T linkages bend (Fig. 7.4), as
happens in other silicate minerals. The rigid tetrahedra tilt around their shared apices and
the oxygen atoms are pushed out into the cages containing the extra-framework cations
(M), thereby reducing the M– O cation distances. The stiffness of the T –O – T bond angles
and M– O bonds control the bulk moduli of felspars.
Increasing disorder of Al and Si results in a small decrease in bulk modulus, which
occurs in response to the increase in the number of Al– O – Al and Si –O –Si linkages in
anorthite at the expense of Si – O –Al linkages that originate from disordering. MO
calculations suggest that Al –O –Al and Si –O –Si linkages are slightly stiffer than
Al –O – Si linkages (Geisinger et al., 1985). That is why the replacement of Al in albite to
produce reedmergnerite, with Si –O – B linkages, results in a 19% increase in the bulk
modulus (see Fig. 7.1).
High-pressure P1 ! I 1 phase transition (change in framework conformation) in
anorthite involves a large decrease in the bulk moduli (Angel, 1988). The bulk moduli of
alkali felspars in both ordered (i.e., albite and microcline) and disordered (i.e., sanidine)
structural states increase slightly with increase in K content. The bulk modulus is seen to
(K2O, Na2O, CaO) – Al2O3 –SiO2 System 617

Figure 7.4. The variation of the T–O –T angles with pressure. Error bars indicate 1 sd. Best-fit lines are
superimposed: solid lines represent Si–O –Si angles and dashed lines represent Al–O–Si angles. The O atom
involved in the angle is indicated. Note that only the Al–O –Si angles decrease with increasing pressure (Downs,
et al., 1994, q 1994 Mineralogical Society of America).

increase with increased Si, Al order from sanidine to microcline. The bulk modulus of a
fully disordered anorthite is ,60 GPa, akin to alkali felspars.

7.1.2. Compressibilities: M – O and kT – O – Tl

The felspars are generally stiffer than SiO2 polymorphs as revealed from
their bulk moduli stated below. The bulk moduli, M– O bond length and T– O – T
angle changes (Downs et al., 1999) in some felspars with pressure are presented in
Table 7.1.
Albite is the most compressible, followed by microcline, then reedmernerite and
anorthite (Fig. 7.1). The molar volume of sodium felspars is smaller because the Na –O
bond is shorter than the K –O bond.

TABLE 7.1
K0, M– O length and T–O –T angles of some felspars under pressures (Downs et al., 1999)

Species (composition) angles K0 (GPa) M– O bond length (Å/GPa) T– O– T (degree/GPa)

Microcline (KAlSi3O6) 60 20.021(1) 20.39(4)

Albite (NaAlSi3O8) (low) 55 20.031(2) 20.23(6)
Reedmergnerite (NaBSi3O8) 69 20.023(1) 20.35(4)
Anorthite (CaAl2Si2O8) 80 20.29(11)
618 Chapter 7

The K0 values observed by Angel (1997) and his co-workers for some felspar
structures are as follows:

Low albite (NaAlSi3O8) 70 GPa

High sanidine (KAlSi3O8) 67 GPa
Anorthite (CaAl2Si2O8) 94 GPa
Anorthite (CaAl2Si2O8) 83 GPa (when K0 is taken as 4)
Reedmergnerite (NaBSi3O8) 69 GPa
Danburite (CaB2Si2O8) 114 GPa
Microcline (KAlSiO6) 63 GPa

The different compressibilities appear to be due to differences between the stronger

Ca2þ –O and the weaker Naþ – O and Kþ –O bonds as well as between the stiffer B – O –Si
and the softer Al –O –Si and Si – O –Si bridging angles. However, little difference of K0
values for low albite (,70 GPa) and high sanidine (67 GPa), with complete Al – Si disorder
over the tetrahedral sites, suggests almost no difference between Si –O – Si and Al – O –Si
linkages (Downs et al., 1994).
These data may allow comparison of the effects of M site (M ¼ Na, K, Ca) diversity
in the interstitial cavities of the negatively charged tetrahedral framework and for
comparing different T –O – T linkages (T ¼ Si, Al, B).
The volume of the NaO7 polyhedron varies linearly with the volume of the unit
cell (Fig. 7.5) with the bulk modulus of NaO7 polyhedron calculated as K0 ¼ 26ð2ÞGPa:
Low albite contains eight non-equivalent O atoms, each associated with bridging
T –O –T angles. The observed decrease in Al– O – Si angles with pressure (Fig. 7.4) in low
albite appears to be consistent with the molecular orbital calculations carried out by
Nicholas et al. (1992). They showed that the Al– O – Si angle with a quadratic force

Figure 7.5. The variation of the volume of the Na polyhedron with the volume of the unit cell. The best-fit linear
equation is represented by the line (Downs et al., 1996, q 1996 Mineralogical Society of America).
(K2O, Na2O, CaO) – Al2O3 –SiO2 System 619

Figure 7.6. The unit-cell volume of low albite as a function of pressure. The solid circles represent the
experimental data. The solid line represents the nonlinear best-fit curve for a third-order Birch Murnaghan
equation of state (Downs, et al., 1994, q 1994 Mineralogical Society of America).

constant of 8.95 kcal/(mol rad2) is twice as soft in albite (low) molecules as the Si – O –Si
angle with a quadratic force constant of 17.34 kcal/(mol rad2). The variation of the unit-
cell volume of low albite as a function of pressure is shown in Fig. 7.6.
The average Al –O – Si angle in low albite varies with pressure as: kAl – O – Sil ¼
136:4ð2Þ 2 0:70ð11ÞP: The average T –O – T angle decreases at a rate of 2 0.23(6)8/GPa
(Table 7.1). This is a higher rate than that observed in anorthite, 2 0.29(11)8/GPa (Angel,
1988). The shorter but stronger Ca2þ – Obr bonds in the latter tend to stiffen the Al – O –Si
angles compared with the longer but weaker Naþ – Obr bonds in the former. The phase
relations of albite NaAlSi3O8 are presented in Fig. 7.7.

7.1.2.1. Unit strains in felspars

The narrower the T –O –T angle, the more energy is needed to compress the angle
even further. This is consistent with the stiffening of the major axis of the unit-strain
ellipsoid with pressure. The strain ellipsoid for anorthite is significantly different from
those for the alkali felspars. The principal unit strains of felspars at 4 –5 GPa is presented in
Table 7.2 (Angel, 1994).
The magnitudes of the principal unit-strain coefficients between room pressure and
P, as a function of pressure, have been calculated with the Strain program of Ohashi by
Hazen and Finger (1982). The best linear fits of the give data: l11 l ¼ 0:0107ð1Þ 2
0:040ð5ÞP; l12 l ¼ 0:0041ð1Þ 2 0:018ð5ÞP; l13 l ¼ 0:0035ð1Þ 2 0:017ð2ÞP: The major axes
620 Chapter 7

Figure 7.7. The P –T phase relations in NaAlSi3O8 (From Yagi et al., 1994a,b, q 1994 AIP,.N.Y.).

of the strain ellipsoid at 3.78 GPa are oriented at 26, 82 and 918 with respect to a, b and
c: 113, 109 and 158, and 103, 21 and 758 for 11 ; 12 and 13 ; respectively.

7.2. KAlSi3O8 system

The solidus temperature of the system KAlSi3O8 in dry peridotite is shown in

Fig. 7.8. The solidus of peridotite is that of KLB-1 reported by Takahashi (1986). Potassic
phases can reduce the melting temperature of the mantle.
At pressures between 4 and 10 Gpa, KAlSi3O8 (K-felspar) breaks down as
(Kinomura et al., 1975) (Fig. 7.8):
2KAlSi3 O8 ¼ K2 Si4 O9 þ Al2 SiO5 þ SiO2
K-felspar wadeite-type kyanite coesite

In the wadeite-type structure, SiO4 tetrahedra and SiO6 octahedra are present and
are linked.
TABLE 7.2
Principal unit strains of feldspars at 4–5 GPa (Angel, 1994)

Sample P (GPa) 11 12 13 11/(11 þ 12 þ 13) (%)

An100 5.0 71 24 7 70
An100 (QOD ¼ 7.8) 4.7 71 31 12 62
An98 4.5 80 30 12 66
An89 4.6 77 29 14 64
An72 4.2 68 40 27 50
Albite 4.9 86 32 25 60
Reedmergnerite 4.6 77 29 19 62
Sanidine (Or98) 4.9 92 39 16 63
Microcline (Or87) 4.7 91 39 14 63

Note: All principal strains are in units of 104 GPa21.

(K2O, Na2O, CaO) – Al2O3 –SiO2 System 621

Figure 7.8. Solidus temperature of the system KAlSi3O8 (solid line) and the dry peridotite (dotted line). The
solidus of peridotite (dotted line) is that of KLB-1 reported by Takahashi, (1986). Potassic phases can reduce
the melting temperature of mantle (Urakawa et al., 1994, q 1994 Springer-Verlag).

Beyond 10 Gpa, a hollandite-type phase is formed with KAlSi3O8 composition. The

tetragonal symmetry of this involves disordering of Al and Si ions, which occupy a single
octahedral site. The EOS of this phase has been determined by Zhang et al. (1993) as
KT ¼ 180ð3Þ GPa (presuming K 0 ¼ 4).
Sanidine, the low-pressure phase of KAlSi3O8, is a predominant potassic phase in
the Earth’s crust and the uppermost mantle. The two well-known high-pressure K-bearing
alumino-silicates are: wadeite-type K2Si4O9 and hollandite-type KAlSi3O8.
In microcline, the Al – Oco –Si angle undergoes least change under pressure while in
albite it changes the most. In the former, the Al –Obo – Si and Si – Obm –Si angles display
the greatest decreases with pressure. This corresponds to a shearing of the four-membered
rings, which in turn compresses the K-bearing channel (Downs et al., 1999). At about
4 GPa, a new bond between K and Obm alters the compression mechanism, leading
to a discontinuity in the pressure variation of crystallographic parameters (Allan and
Angel, 1997).
Kinomura et al. (1975) reported wet-condition synthesis of cymrite-structured
KAlSi3O8·H2O, which was seen to be more stable than sanidine. The breakdown pressure
of KAlSi3O8·H2O must be different from that of sanidine.

7.2.1. Stability

In the presence of water, the stability of K-felspar is limited to lower pressures and,
like albite, it breaks down to sheet silicates. At high pressures, K-felspar reacts with
orthopyroxene (MgSiO3) to produce biotite, whereas reaction with pyrope-almandine
622 Chapter 7

garnets produce siderophyllite. Often a breakdown of these micas during metamorphism

leads to formation of felspars.
K-felspars are seen to be stable at very high pressure. Smyth and Hatton (1977)
reported a sanidine– grospydite from a kimberlite containing high sanidine and free silica
(in the form of coesite), plus clinopyroxene, kyanite and garnet. For these phases,
equilibrium conditions were estimated to be . 9008C at 3 GPa.
With increasing pressure, orthoclase is seen to break down successively as:
orthoclase ! sanidine þ jadeite þ quartz=coesite
ðin hydrous environmentÞ ! KAlSi3 O8 ·H2 O þ jadeite þ coesite
Ringwood and his co-workers (1967) investigated felspars under pressure to
transform to hollandite structure. Sanidine, transformed into hollandite structure at
12 GPa/9008C, is seen to have both Al and Si at 6-fold coordination. Hollandite structure
with [6]Si and [6]Al seems to be the most abundant phase in the transition zone but ordering
of cations in the octahedral B site has never been observed in any hollandite structure.
High-pressure Si and Ba alumino-silicates with hollandite structure were
investigated by Reid and Ringwood (1969). The Si –O bond length is usually 1.775 Å
(in stishovite; Ross et al., 1990) and the (Al0.25Si0.75)– O bond length is 1.816 Å in
KAlSi3O8 hollandite (Zhang et al., 1993). The ratio of rB =rA is noted as 0.35 (see
Section 7.3).

7.2.2. Phase relations

Ringwood et al. (1967) synthesized the hollandite-structured KAlSi3O8 from

sanidine at 12 GPa and 9008C. KAlSi3O8 sanidine transforms with increasing pressure as
(Kinomura et al., 1975; Yagi and Akaogi, 1991):

6 , 8 Gpa wadeite-type K2Si4O9 þ kyanite þ coesite (stishovite?)

9 , 11 Gpa hollandite-type KAlSi3O8.

The first phase boundary between sanidine and the three-phase assemblage of
wadeite-type K2Si4O9 þ kyanite þ coesite (Fig. 7.8) is approximately determined by the
relation:

PðGPaÞ ¼ 4:5 þ 1:7 £ 1023 T=8C:

The second phase boundary between the three-phase assemblage and hollandite-type
KAlSi3O8 seems to be located around 9.0 , 9.5 GPa at 1,2008C (Urakawa et al., 1994). The
melting point of hollandite-type KAlSi3O8 is between 1,700 and 1,8008C at 11 GPa.
Urakawa et al. (1994) carried out an in situ XRD study on KAlSi3O8 using the
cubic-type high-P apparatus, MAX90, combined with synchrotron radiation.
At # 11 GPa pressure, the phase relations of sanidine, with the wadeite-type
K2Si4O9 þ kyanite (Al2SiO5) þ coesite (SiO2) assemblage, and KAlSi3O8 hollandite were
studied (Fig. 7.9). The melting T of potassic phases has also been determined.
(K2O, Na2O, CaO) – Al2O3 –SiO2 System 623

Figure 7.9. Phase relations in KAlSi3O8. Circles represent the experimental runs. Areas of different symbols
in one circle correspond roughly to the volumes of the different phases (or assemblages). Stishovite was found to
co-exist with coasite, kyanite, and wadeite at pressures within 0.5 GPa below the upper boundary (From Yagi
et al., 1994, q 1994 Springer-Verlag).

The breakdown of sanidine into three phases reduces the melting temperatures: K2Si4O9
melts first at ,1,5008C in a three-phase region. The solidus temperature of the system
KAlSiO3 in dry peridotite is shown in Fig. 7.8.
If these potassic phases host K in the Earth’s mantle, the mantle-solidus
temperatures will be much lower than the reported dry-solidus temperature of peridotite.

7.2.3. Displacive-phase transition

Using powder XRD, Harrison et al. (1994) have measured the temperature of
evolution of the displacive order parameter of hypersolvus in Al – Si disordered alkali
felspars with compositions Or31 and Or20.
The monoclinic – triclinic transition shows second-order behaviour and bi-linear
order parameter strain coupling. The transition temperatures are 443 K (Or31) and 750 K
(Or30). Temperature-evolution of the peak width (G ) of the (132) reflection was found to
depend on the grain size of the sample, with an anomalous increase of G at Tc in fine-
grained material. This effect has been rationalized by Harrison et al. (1994) in terms of
surface relaxation occurring as T approaches Tc. No anomalous line broadening is observed
in coarse-grained material.

7.3. Hollandite-type compounds

Hollandite-type compounds have a general formula, AxB8O16, where A represents

large mono-to-divalent cations with x # 2; and B represents smaller, 2 – 5-valence cations.
A large variety of chemical species is accommodated in the structure: A may be
represented by large cations such as Ag, Ba, Cl, Cs, K, Na, Pb, Rb, Sr and Ti; whereas
624 Chapter 7

B-type cations may be Al, Co, Cr, Cu, Fe, Ga, Ge, In, Mg, Ni, Sb, Sc, Si, Sn, Ti and Zn.
These form some 30 different end-member compositions (Pentighaus, 1978).
Hollandite structure consists of edge-sharing octahedra that form a four-sided,
eight-membered channel that is capable of containing low-valence large-radius cations.
The O atoms and the large cations together form a hexagonal closest-packed structure
(a sort of analogue to the cubic closest-packed perovskite). The similarity of the structure
to stishovite should remind one of the similarities of coesite and the felspars.
The type mineral hollandite, BaMn8O16, has a crystal structure with tetragonal
symmetry, 14/m. The large A-type cations occupy the 2 £ 2 channels and are ideally
located at the intersection of the mirror and the 4-fold axis of rotation, i.e., at the Wyckoff
position 2b, (0,0, 1/2). They are coordinated with eight O atoms, all at equal bond lengths,
located at the corners of a rectangular prism.
The symmetry depends on the ratio of the average ionic radius of B cation to that of
A. When rB =rA . 0:48; the tunnel cations are too small for the tunnel and so the octahedra
tilt, the tunnels walls distort and the symmetry is reduced to monoclinic, 12/m. The unit-
cell volume is dependent on the B – O bond length.
Large alkali or alkaline earth cations occupy positions along large channels that run
parallel to c. This hollandite structure, therefore, becomes the repository of alkalis in the
Earth’s mantle. Natural hollandites are typically monoclinic (pseudotetragonal), owing to
ordering of Mn3þ and Mn4þ or other cations at octahedral sites. Synthetic hollandite
(K1.33Mn4þ 3þ
6.67Mn1.33O16) (Liu, 1978b) displays diffuse X-ray diffraction, characteristic of
some short-range order, but long-range disorder of Mn4þ and Mn3þ.
Hollandite bears a close resemblance to calcium ferrite (CaFe2O4) structure
(Yamada et al., 1983). Both these structures consist of double octahedral chains which are
joined to form tunnels parallel to c that accommodate alkali or alkaline-earth cations.
Ringwood et al. (1967) transformed sanidine at 9008C and 12 GPa into the
hollandite structure with Al and Si randomly occupying the octahedral sites. Considering
the year of reporting, this was the first oxide structure identified with both Al and Si
displaying 6-fold coordination and only the second after stishovite with [6]Si. Reid and
Ringwood (1969) synthesized SrxAl2xSi422xO8 and BaxAl22xSi422xO8, x , 0:75; with the
hollandite structure but failed in transforming felspars rich in Na, Ca and Rb.
Because it accommodates many chemical species, the hollandite structure has been
considered as a host for high-level radioactive wastes, along with the perovskite and
zirconolite in the synthetic titanate ceramic synroc (e.g., Smith and Lumpkin, 1993).
Hollandite structure easily immobilizes the Ba, Cs and Rb isotopes. Also, because of its
channel vacancies, the hollandite structure offers a classic example of a one-dimensional
electrochemical super-ionic conductor (Beyeler, 1976).

7.3.1. Pb –hollandite

Single crystals of lead alumino-silicate hollandite, with composition

Pb0.81Al0.19Si24O72, have been synthesized at 16.5 GPa and 1,4508C by Downs et al.
(1995). The crystals are tetragonal (14, Z ¼ 2) and in these Si and Al are disordered on
the octahedral site. From a mixture of orthopyroxene, orthoenstatite and jadeite at 1,4508C
(K2O, Na2O, CaO) – Al2O3 –SiO2 System 625

and 16.5 GPa (with a Pb flux,) a phase was synthesized of composition Pb0.8Al1.6Si2.4O8
with hollandite structure (Downs et al., 1995). This shows that alumino-silicate hollandite
is capable of holding large cations of low valence at high P– T region, such as the transition
zone.
The stability and structure of a new high-pressure Na1.8Ca0.2Si6O14 have been
determined by Gasparik et al. (1995) employing pressures from 8 to 16 GPa (950 –
2,3008C) on a system Na2O – CaO –5SiO2.

7.4. Anorthoclase (KAlSi3O8 – NaAlSi3O8)

Alkali felspars (KAlSi3O8 – NaAlSi3O8 system) are one of the most abundant
minerals in the Earth’s crust. High-pressure transitions of these felspars have been
investigated by many workers.
KAlSi3O8 felspar transforms to hollandite structure at about 12 GPa (Ringwood
et al., 1967). Between K-felspar and hollandite structure, there exists a field of three-phase
assemblage: wadeite-type K2Si4O9, kyanite Al2SiO5 and coesite SiO2. At much lower
pressure (2 –3 GPa), NaAlSiO6 jadeite plus quartz occur. At about 23 Gpa, jadeite is seen
to dissociate into calcium-ferrite-type NaAlSi2O4 plus stishovite, while NaAlSi3O8
hollandite is stable near 20 GPa (Liu, 1978b). Hollandite structure in the solid-solution
series KAlSi3O8 – NaAlSi3O8 may be important as the K- and Na-bearing mineral phase in
the deep mantle.

7.4.1. Phase relations

Yagi et al. (1994) examined the system at 5– 23 GPa and 700 –1,2008C. KAlSi3O8
sanidine first dissociates into a mixture of wadeite-type K2Si4O9, kyanite and coesite at
,7 GPa, which further recombines into KAlSi3O8 hollandite at 9 –10 GP (Fig. 7.9). In
contrast, NaAlSi3O8 hollandite is stable at 800– 1,2008C near 23 GPa, where the mixture
of jadeite plus stishovite changes directly into the assemblage of calcium ferrite-type
NaAlSiO4 plus stishovite. Kinomura et al. (1975) placed the boundary between wadeite,
kyanite plus stishovite and hollandite at ,11– 12 GPa at 700 – 9008C.
Phase relations in the system KAlSi3O8 –NaAlSi3O8 at 1,0008C show that the
NaAlSi3O8 component gradually dissolves into hollandite with increasing pressure. The
maximum solubility of NaAlSi3O8 in hollandite at 1,0008C is ,40 mol% at 22.5 GPa,
above which it decreases with pressure. The hollandite solid solution in this system may be
an important candidate as a host mineral for K and Na in the uppermost lower mantle.
In the NaAlSi3O8 albite system, the high-pressure phase relation shows that at about
22.5 Gpa, NaAlSi2O6 jadeite plus stishovite changes into a mixture of calcium ferrite-type
NaAlSiO4 and stishovite (Fig. 7.7).

ðaÞ NaAlSi2 O6 þ SiO2 ¼ NaAlSi3 O8

Jadeite stishovite hollandite

and
626 Chapter 7

ðbÞ NaAlSi3 O8 ¼ NaAlSiO4 þ 2SiO2

Hollandite Ca-ferrite stishovite

The density increases for the above two reactions are calculated to be 4.7 and
10.5%, respectively. From the Hugoniot data, Ahrens and Liu (1973) concluded that albite
transforms to hollandite structure in the pressure range of 40– 80 GPa (in shock-
compression experiments).
Liu (1978) reported the synthesis of NaAlSi3O8 hollandite at 21 –24 GPa from
jadeite and stishovite. Above 24 Gpa, this transforms to calcium ferrite structure. In the
system KAlSi3O8 (Or) –NaAlSi3O8 (Ab), the albite component dissolves gradually into
KAlSi3O8 hollandite with increasing pressure above 14 GPa. However, this solubility in
hollandite is rather limited to 1,0008C temperature. Liu (1978) also confirmed that
KAlSi3O8 hollandite is stable at . 30 GPa (1,0008C). Hence, it is likely to be the most
K-bearing stable phase in the lower mantle. Since the abundance of K in the mantle is
much smaller than Na, the amount of hollandite in the lower mantle would possibly depend
on the solubility of K into calcium ferrite-type NaAlSiO4.

7.5. Plagioclase felspars (NaAlSi3O8 –CaAl2Si2O8)

It is difficult to determine whether or not the M-site cation is coordinated to a given

O atom in the alkali felspars. In many structures, the M –O bond lengths display a broad
range of values with no apparent distinguishable gaps. If M ¼ K or Rb, then the
coordination number is usually chosen as nine but, for M ¼ Li, the coordination appears to
be five (Smith and Brown, 1988). When M ¼ Na, the coordination appears to depend, in
part, on the framework cations. For NaAlGe3O8 and NaGaGe3O8, there are four O atoms
with the short M– O bond lengths (< 2.4 Å) and, for reedmergnerite, NaBSi3O8, there
are five.

7.5.1. Albite, NaAlSi3O8

The P – T phase relations of NaAlSi3O8 are shown in Fig. 7.7. In albite, the Na –O
bond length decreases smoothly with pressure. The unit-cell volume change with pressure
is shown in Fig. 7.6. It has already been discussed in Section 7.1 that the Oco atom in albite
is underbonded in an Al – Oco –Si linkage that should be soft compared with any other
angle in the structure. The Al– Oco – Si angle is the most compressible in albite structure
and also the Al– O – Si angle is weaker than the Si – O –Si angle. In albite under pressure,
only Al– O – Si angle decreases while Si – O – Si angles show little change (or even an
increase).
Traditionally, mineralogists have chosen the coordination number of Na in albite
as five, seven or nine. The stability boundary of albite has been defined by Holland
(1980) as:
P ¼ 0:00265Tð8CÞ þ 0:035GPað^0:05GPaÞ
(K2O, Na2O, CaO) – Al2O3 –SiO2 System 627

Figure 7.10. Breakdown reactions of albite overlain upon the kyanite– andalusite –sillimainte relations. Reaction
(1) is paragonite ¼ albite þ corundum þ H2 O and reaction (2) is paragonite ¼ albite þ AlSiO5 þ H2 O (Angel,
1994, q 1994 Kluwer Pub).

Above this field, it transforms to jadeite þ quartz (over the temperature range 600–
1,2008C) (Fig. 7.10) and at very high pressures (14 –28 GPa) it amorphizes (to “glass”).
These have been studied through infrared (Williams et al., 1993) and Raman (Daniel et al.,
1993) spectroscopy.
At higher pressure, quartz transforms to coesite and then to stishovite. When the
pressure reaches 21 –24 GPa and the temperature is around 1,0008C, jadeite reacts with
stishovite to form a single-phase hollandite (Liu, 1978). Above 24 Gpa, calcium ferrite
structure is attained along with stishovite (Fig. 7.7). However, Gasparik (1992) reported
this to occur at lower pressure (21 GPa) with jadeite composition.
Water is seen to catalyze the breakdown reaction to jadeite þ quartz down to
3.5 GPa. An addition of Fe3þ to albite produces a jadeite – acmite solid solution, while an
addition of MgSiO3 displaces the equilibrium breakdown of albite to lower pressure as a
result of the lowering of jadeite activity by enstatite.

7.5.1.1. Al – Si order –disorder

The Al– Si order – disorder in felspars can be determined from XRD analyses using

the diffraction peak positions (e.g., D131 ¼ 2uð131Þ 2 2uð131ÞÞ; cell parameters, or T –O
bond lengths (Kroll and Ribbe, 1980). For the Al – Si disordering, the covalent argument
suggests that because of p bonding the Si – O – Si angles should be larger on average than
the Al –O –Si angles.
Again, if the large anisotropic displacements of the Na atoms indicate static
disorder, then the low albite structure might transform from C 1 to P1 symmetry upon
decrease in volume as a result of Na atoms ordering into two different sites (analogous to
the transformation observed in anorthite) but such transformation is not observed.
628 Chapter 7

Low-temperature albite displays complete ordering of Al atoms into one of the four
symmetrically non-equivalent, tetrahedrally coordinated cation sites. At higher tempera-
tures, disordering leads to the occupancy of all the four tetrahedral sites by Al atoms. A
specific degree of order (or disorder) may be achieved by controlling the pressure and
temperature (Goldsmith and Jenkins, 1985).
For disordered albite, the energy term, PDV; will be smaller than for ordered albite.
Therefore, with enough heat to aid the kinetics and with the application of pressure, Al –Si
ordering is encouraged. Hydrogen atom plays an important role in promoting Al –Si inter-
diffusion in albite at high pressures. Coupled with such effects of H, its bond
accompanying strain may weaken the structure, leading to bond breaking, which could
easily be followed by cation diffusion.

7.5.1.2. Low albite: kAl– O – Sil change

XRD studies of albite (e.g., Armbruster et al., 1990) have indicated that Na atom is
not statistically disordered but instead vibrates with large anisotropic motion about a single
centre (Smith et al., 1986). Na occupies an interstitial cavity in the negatively charged
alumino-silicate framework.
Low albite contains eight non-equivalent O atoms, each associated with a bridging
T –O –T angle. Al – O –Si angles are seen to decrease with pressure. The observation
that only Al– O – Si angles decrease with pressure in low albite appears to be consistent
with the MO calculations (Nicholas et al., 1992). However, the average T– O – T angle
studied by Downs et al. (1994) barely decreases with pressure with a rate of
2 0.23(6)8/GPa. This behaviour is in contrast with that noted for quartz, cristobalite
and coesite, wherein the V=V0 ratio seems to vary linearly with the normalized average
Si– O – Si angle.
The average Al – O –Si angle in low albite has been seen to vary with pres-
sure as kAl – O – Sil ¼ 136:4ð2Þ 2 0:70ð11ÞP: Compared with the longer and weaker
Naþ – Obr bonds, the shorter and stronger Ca2þ –Obr bonds tend to stiffen the Al – O –Si
angles.

EOS. Downs et al. (1994) used their experimental pressure and weighted-volume data
on low albite to a non-linear third-order Birch – Murnaghan EOS that gave a zero-pressure
volume, V0 ¼ 664:04ð9Þ Å3, a zero-pressure bulk modulus, K0 ¼ 54ð1Þ GPa and
K0 ¼ 6ð1Þ (see Fig. 7.6). Earlier, Hackwell and Angel (1995) obtained the bulk modulus,
K0 as 57(2) GPa (presuming K0 ¼ 4).
The inter-axial angles were seen (Downs et al., 1994) to vary systematically as
da=dP ¼ 20:0228=GPa
db=dP ¼ 0:1148=GPa
and
dg=dP ¼ 0:0848=GPa:
In the pressure range of ,4 Gpa, the average T – O bond lengths for the four non-
equivalent tetrahedra show little change. Such a rigidity of T– O bonds is comparable with
(K2O, Na2O, CaO) – Al2O3 –SiO2 System 629

those observed for framework structures of SiO2 such as quartz and cristobalite (e.g.,
Downs and Palmer, 1994) and for the AlO4 and SiO4 tetrahedra in the anorthite structure
(Angel, 1988). Some other workers (e.g., Levien and Prewitt, 1981), however, noted that
these bond lengths change only insignificantly.

7.5.2. Anorthite, CaAl2Si2O8

The structure of CaAl2Si2O8 consists of a three-dimensional framework of corner-

linked AlO4 and SiO4 tetrahedra, with Ca atoms occupying larger cavities within the
framework. Anorthite has a three-dimensional framework structure of corner-linked SiO4
and AlO4 tetrahedra, with Ca in the larger cavities in the framework. There are two
symmetrically distinct chains of tetrahedral, the conformations of which differ by up to 408
(Si – O –Al bond angles). The structure responds to intensive variables by tilting the
tetrahedral about their corner oxygens. The bridging oxygen atoms can be pushed out
into the cages containing the Ca cation. The Al and Si cations are almost completely ordered
over the tetrahedral sites, but small changes in Al., Si order can be introduced by heating
anorthite at temperatures in excess of 9008C (Angel et al., 1990; Carpenter et al., 1990). The
P1 ¼ I 1 phase-transition boundary in pure natural anorthite (An100) is shown in Fig. 7.11
(Hackwell and Angel, 1995).
The high-pressure breakdown of anorthite to grossular þ kyanite þ quartz offers
the simplest expression for the important boundary between granulites and eclogites.
Below 7258C, anorthite reacts with water as:

anorthite þ H2 O ! zoisite þ kyanite þ quartz þ vapour:

The phase boundary for anorthite reaction is defined by:

P ¼ 0:00204Tð8CÞ 2 0:459 GPa:

Figure 7.11. Reversals of the P1 $ I 1 phase boundary in natural An100 anorthite as determined by in situ high-
pressure, high-temperature single-crystal x-ray diffraction (Hackwell and Angel, 1995, q 1995 Mineralogical
Society of America).
630 Chapter 7

This boundary is valid up to a triple point at ,1.02 GPa and 7258C, at which point
melt appears. Anorthite-rich plagioclase under pressure first disproportionates as:
plagioclase ! Na-plagioclase þ garnet þ kyanite þ quartz
At higher pressures (3 GPa, 1,2008C), the plagioclase gives way to pyroxene.
Angel (1988) experimented on the structural change of anorthite between pressures
of 2.5 and 3.1 GPa and studied the phase transition at 2.6 GPa. The data were refined,
assuming all to be I 1 symmetry. It was observed that the tetrahedral bond lengths and
volumes remained constant over the pressure interval. However, the average T– O – T
angle (kT –O –Tl) seemed to vary as 136.8(3)–0.29(11) P, GPa (see Fig. 7.4). Angel
(1988) concluded that the compression is accommodated by tilting of the rigid tetrahedra
and bending of T – O –T angles into interstitial cavities occupied by Ca cations.
High-pressure experiments on synthetic anorthite glass revealed that anorthite
remains stable up to 18 GPa when it inverts to produce hollandite þ Ca –Al perovskite.

7.5.2.1. Al – Si order –disorder

The Al and Si cations are almost completely ordered over the tetrahedral sites but
small changes in Al, Si order can be introduced by heating anorthite at temperatures in
excess of 9008C (Angel et al., 1990).
Al, Si ordering in anorthite, CaAl2Si2O6, has been studied in both equilibrium and
non-equilibrium states (e.g., Salje et al., 1993, Phillips and Kirkpatrick, 1995). However,
exact order parameter measurements in anorthite are difficult to achieve using laboratory
X-ray sources because of the similar scattering power of Al and Si. Measurements usually
rely on secondary effects, such as those arising from spontaneous strain and variation of
average tetrahedral bond lengths due to size differences of the ordering cations.
Al – Si order – disorder can be quantified by an order parameter, QOD. For
completely ordered crystal, QOD ¼ 1; while it is zero in a completely disordered case.
In an anorthite sample with the least amount of induced Al, Si disorder ðQOD ¼ 0:87Þ; the
evidence for the P1 $ I 1 transition is provided by rapid changes in the unit-cell angles
between 2.5 and 3.0 GPa. In this pressure range, the unit-cell angles are smeared out.
Around 2.6 Gpa, the unit-cell angles show a sharp change. The results of a study by Angel
(1994) on anorthite-rich felspars are presented in Table 7.3.

TABLE 7.3
Parameters of the P1 to I 1 phase transition in anorthite-rich feldspars (Angel, 1994)

Sample Composition QOD Transition bracket (GPa) 1s

VP An100 0.92 2.46–2.67 0.0106

MS An97 –An100 0.91 2.51–2.66 0.0095
115,082 An94 –An98 0.88 2.97–3.12 0.0073
84,332 An90 –An92 n.d. 3.32–3.42 0.0049
87,975 An88 –An90 0.85 3.69–3.90 0.0030

Note: Sample numbers refer to the Harker collection of the University of Cambridge. Composition shows the
range for these samples reported by Carpenter et al. (1990). Individual crystals used in the high-pressure
experiments were not probed.
(K2O, Na2O, CaO) – Al2O3 –SiO2 System 631

It is seen that the transition pressure increases with increasing albite content and the
linear relationship is obtained as (Angel, 1994):

Ptr ¼ 1:04Nab þ 2:53ðGPaÞ

where Nab is the mol% of albite in solid solution.

The experiments by Angel (1988) demonstrated that for the P1 $ I 1 transition at
high temperature (Redfern, 1992), the state of Al, Si order within the anorthite has a large
effect on both the position of the transition and its character.
In fully ordered anorthite, the P1 ! I 1 transition is tri-critical (Redfern and Salje,
1987) but the addition of Na and/or Si, Al disorder decreases the transition temperature and
causes a change to second-order behaviour (e.g., Redfern, 1992). At elevated pressures, the
P1 – I 1 transition for Si, Al ordered anorthite is first order and occurs near 2.5 GPa at room
temperature (Angel et al., 1989).
Coupled substitution. The influence of coupled substitution of (Na þ Si) for
(Ca þ Al) (i.e., by albite component) on the P1 $ I 1 transition has been studied
extensively by Angel et al. (1990) and Carpenter et al. (1990).

7.5.2.2. Structure
The tetrahedral framework of anorthite can be considered as comprising chains of
four rings of tetrahedra that run parallel to the c-axis of the conventional cell setting. The
tetrahedra in the four rings share Ob- and Od-type oxygen atoms (see Smith and Brown,
1988, for felspar site notations) and the chains are formed by the sharing of Oa oxygen
atoms between the T1o and T1m tetrahedra from successive four rings along the c-axis
(Fig. 7.12). The three-dimensional framework of anorthite is formed by the cross-linking
of these chains through the Oc oxygen atoms.
In the P1 structure, there are two symmetrically distinct chains, one containing only
atoms distinguished by a final “o” in their level (e.g., Obmzo) and one containing atoms
distinguished by a final “i” (e.g., Obmzi). Christy and Angel (1995) referred to these two
distinct chains as the o-chain and the i-chain (Fig. 7.12). In the I 1 structures, these two
chains become symmetrically equivalent so each atomic site in the I 1 structures
corresponds to two distinct sites in the P1 structure.
The three anorthite structures, P1 (o-chain), P1 (i-chain) and I 1;
 differ in the tilting
or rotation of the SiO4 and AlO4 tetrahedra about their common oxygen atoms, which
can be quantified as the Al –O – Si bond angles. The bond angles corresponding to the
O positions (at 3.1 GPa and 1,4308C) are presented in Table 7.4.
The pattern of distortions of the four rings in the i chain of the P1 structure and the
corresponding differences in bond angles in o and i chains (in the P1 structure) are shown
in Fig. 7.12.
A study of Table 7.4 reveals that the high-pressure I 1 structure has chains showing
the same pattern of distortion as the i chain of the P1 structure whereas the chains in the
high-temperature I 1 are distorted in the same way as the o chain in the P1 structure. In their
detailed study on anorthite structure, Christy and Angles (1995) concluded that the high-
pressure and high-temperature I 1 phases of anorthite have distinct structural configurations
632 Chapter 7

Figure 7.12. A polyhedral representation of the i chain in P1 anorthite at room pressure and temperature. The
distortion of this chain from that of the average structure is clear from the deviation of the rings from a square.
The same pattern of distortion occurs in the chain in the high-pressure phase, whereas the opposite pattern
of distortions occurs in the o chain of the P1 structure and in the structure of the high temperature I 1 phase
(Christy and Angel, 1995; q 1995 Springer-Verlag).

of their aluminosilicate tetrahedral frameworks and that the P1 structure can be described
as being composed of two distinct tetrahedral chains, one that has a structure similar to that
found in the high-pressure phase and the other with the same pattern of distortion as that in
the high-temperature phase.

7.5.2.3. Phase diagram

p
The P1 I 1 phase boundary in a schematic P– T –X phase diagram for natural,
well-ordered anorthite-rich plagioclases is shown in Fig. 7.13. Both the high-temperature
and high-pressure phases show I 1 structure but the two regimes are separated by a cross-
over region, causing sharp curvature in P – T space. At high pressures and low
temperatures, the phase boundary is marked by significant discontinuities in the unit-
cell angles of anorthite and a complete disappearance of the c and d reflections within the
I 1 phase field. Room-temperature experiments also revealed the same characteristics
(Angel, 1988, 1996).
(K2O, Na2O, CaO) – Al2O3 –SiO2 System 633

TABLE 7.4
Si–O –Al bond angles in anorthite

O position P1 o chain P1 i chain I 1 3.1 GPa I 1 1,4308C

Oa1o 137.0 138.4 138.1 140.5

Oa1z 136.9 134.5 136.0 137.8
Oa2o 127.0 122.2 121.3 125.4
Oa2z 123.4 125.1 127.2 125.7
Oboo 129.3 137.5 136.0 136.5
Oboz 138.9 126.7 122.8 140.3
Obmo 170.2 145.7 142.2 153.9
Obmz 143.3 166.7 170.7 148.6
Ocoo 133.9 130.3 129.6 128.6
Ocoz 131.9 131.4 130.6 134.4
Ocmo 130.5 130.6 128.0 131.3
Ocmz 127.2 130.6 129.1 129.3
Odoo 136.0 125.6 125.0 131.2
Odoz 123.9 134.6 129.7 133.4
Odmo 139.3 164.6 169.2 146.7
Odmz 163.8 135.9 135.1 148.9

Figure 7.13. A schematic P–T –XAb phase diagram for natural, well-ordered anorthite-rich plagioclases. The
phase boundary surface separates a P1 stability field at low pressures, temperatures and albite content from the I 1
phase field which has two distinct regimes separated by a “crossover” transition, indicated at An100 is unknown,
and the point of curvature may move to lower temperatures or disappear altogether. The phase boundary surface
must, however, be crossed by a line of tri-critical points (dotted line) separating first-order from second-order
character, but its phase trajectory is unknown (Angel, 1996 q 1996 Kulwer).
634 Chapter 7

7.5.2.4. P1 $ I 1 transition: non-ferroic displacive

As discussed earlier, at high pressure, anorthite undergoes phase transition from
tri-clinic P1 symmetry to I 1 symmetry but this does not involve any change in point-
group symmetry (hence, “non-ferroic”; Wadhawan, 1983). Also, there are no symmetry-
breaking components of the spontaneous strain associated with the transition. The scalar
strain provides a measure of the magnitude of the distortion (since 1s / Q2o ) accompanying
the I 1 to P1 symmetry change.
Pure anorthite (An100) was investigated by Angel (1988) using diffraction
reflections below the phase-transition pressure (at 2.5 GPa) and above the phase transition
(at 3.1 GPa). No significant changes in tetrahedral bond lengths and internal (O –T – O)
angles between 0.1 MPa and 2.5 GPa were noted by Angel, confirming that the tetrahedra
are essentially rigid towards compression. However, compression within anorthite
structure in pressure up to 2.5 GPa is seen to be accommodated by the reduction in the
longer Ca– O bond distances arising from small flexing of the T– O – T bond angles
between essentially rigid tetrahedra.
At 3.1 Gpa, a substantial change in the relative intensities of the reflection occur
with Ia . Ib and the cand d reflections become absent. All these suggest that the high-
pressure phase has true I 1 symmetry. The scattering density at the Ca sites within the
structure at 3.1 GPa shows a single maximum at each site (Fig. 7.14) in contrast to the split
sites observed in the structures at 0.1 MPa and 2.5 GPa. Significant changes are noted in
the T –O – T bond angles. The largest changes (up to 178) occur at Obmo, Obmz, Odoo and
Odmz, which are the sites showing the largest deviations from I 1 symmetry in the low-
pressure P1 phase.

Figure 7.14. The scattering density at the Ca sites in the anorthite structure at various pressures. Maps on the left
are sections through the Mooo site at (a) 2.5 GPa, average P1 structure (b) 3.1 GPa, I 1 structure. Parts (c) and (d)
are the Mzoo sites under the same conditions. Note the distinct change from split sites in the P1 average structure
to single sites in the I 1 structure (Angel, 1994, q 1994 Kluwer).
(K2O, Na2O, CaO) – Al2O3 –SiO2 System 635

Ca atoms are seen to occupy single sites, manifested by single minima in a more
sharply defined potential well (Angel, 1994). As discussed earlier, the high-pressure I 1
structure has single Ca sites, whereas the high-temperature structure appears to have split
sites between which the Ca atoms hop. At high pressure, the transition is strongly first order
in character, while it is continuous at high temperature (Redfern and Salje, 1987). This
observation would suggest that the high-pressure and high-temperature phases are distinct.
This is marked by the disappearance of the c and d reflections.
Hackwell and Angel (1995) used an externally heated diamond-anvil pressure cell
to reverse the P1 ! I 1 phase boundary in P –T space (Fig. 7.11, see the caption). From
2.5 GPa at room temperature, the phase boundary has a negative slope,
dP=dT , 20:0025 GPa K21, showing up to a temperature of ,1658C. From 165 to
,2308C, the boundary becomes steeper, passing through a well-constrained bracket
centred at 2308C and 1.7 GPa. From this pressure to the ambient pressure, the boundary is
almost vertical in P – T space with dP=dT , 20:2 GPa K21. Along the high-pressure limb
of the transition (first order) boundary, the transition is marked by the disappearance of the
c and d reflections.
A new high-pressure phase transition in an anorthite single crystal has been
revealed by Raman spectroscopic studies (Daniel et al., 1995) to 11.4 GPa (at RT). A first-
order phase transition is observed on increasing P at 2.6 GPa. The P1 ! I 1 phase transition
in anorthite has been reversed in situ at 2.6 GPa and ,2558C in an internally heated DAC
using single-crystal XRD (Hackwell and Angel, 1995).
Another example of displacive transition in a crystalline silicate is offered by the
calcium silicate, CaSi2O5. At P ¼ 0:17 – 0:205GPa; triclinic CaSi2O5 is seen to transform
to monoclinic phase (Angel, 1997).

7.5.2.5. Amorphization
High-pressure amorphization of anorthite has been observed by high-energy
dispersive XRD of powdered samples of highly ordered anorthite (from Val Pasmeda) held
under static pressure in a DAC (Redfern, 1996). The onset of amorphization is observed at
10 –14 GPa and at 14– 20 GPa, anorthite becomes completely X-ray amorphous. A
partially amorphized anorthite can be regarded as a spatially heterogeneous anti-glass,
with long-range order maintained but the translational disorder dominating at shorter
correlation lengths.
However, Daniel et al. (1996) observed in their studies that above 16 GPa, anorthite
is fully amorphous but only the samples pressurized to . 22 GPa remain amorphous on
recovery at ambient conditions. The high-P behaviour of anorthite is highly sensitive to
deviatoric stresses. Daniel et al.’s (1996) calculations showed that the free energies of the
crystalline and amorphous materials are equal at ,30 GPa.

7.5.2.6. Shock transition to glass: Raman results

Structural modification induced by dynamic compression (by shock wave) up to
40 GPa in anorthite glass has been investigated by Raman spectroscopy (Reynard et al.,
1999). It is observed that densification increases with increasing shock pressure and a
maximum density increase ðDr=r0 Þ of 2.2% is obtained for a shock pressure of 24.0 GPa.
636 Chapter 7

This is associated with an increase of the three-membered ring concentration

(observed at intermediate pressure). The products of the highest shock pressures are
seen to be less densified than those at intermediate pressure (Okuno et al., 1999; Reynard
et al., 1999). In these glasses, the structures consist of fully polymerized tetrahedral
networks where alkali (Na, K) or calcalkaline earth (Ca) cations act as “charge
compensators” for Al replacing Si as a network former when compared with pure silica
(Taylor and Brown, 1979). It is observed that the bulk moduli of melts increase along the
charge-compensated SiO2 – CaAl2O4 join (Webb and Cowitial., 1996) with increasing Ca
content. Under the influence of non-bonding interactions, glasses along the SiO2 – NaAlO2
join should show a behaviour intermediate between SiO2 and CaAl2Si2O8 under shock
compression.
The Raman spectrum of the anorthite glass is characterized by a group of high-
frequency bands in the 900– 1,200 cm21 region, which correspond to the stretching
vibrations of T –O – T (T ¼ Al, Si) linkages within the tetrahedral network. Bands in the
700 –800 cm21 have been attributed, by analogy with vitreous silica, to deformation
modes involving in-cage motion of tetrahedral cations in the highly polymerized network
and to AlO4 stretching vibrations (e.g., Sharma et al., 1984; McMillan et al., 1994). Finally,
the bands in the 300– 600 cm21 region are attributed to the bending vibrations of T– O – T
linkages and tetrahedra (e.g., Sharma et al., 1981).
With increasing shock pressure, the prominent evolution in the Raman spectra
occurs in the relative intensity of the peak near 580 cm21 while other peaks show only
minor broadening and frequency changes. This change is similar to that observed on
statically compressed anorthite glass to 17 GPa and ambient temperature (Daniel et al.,
1997). The intensity variation of the 580 cm21 peak is proportional to that of the density
increase. The 580 cm21 peak is the likely result of a densification mechanism in anorthite
glass subjected to shock waves (Reynard et al., 1999).
High-temperature Raman spectra show the density decrease to be related to the
decrease in intensity of the 580 cm21 band, which starts at ,900 K.
For a minimum time of 1026 s (i.e., shock-duration time), a corresponding viscosity
of ,105 Pa s is required for the melt to relax fully. This corresponds for anorthite to a
temperature of ,1,450 K (Urbain et al., 1982).

7.5.3. P1 ! I 1 : 29Si MAS-NMR spectroscopic study

In anorthite (CaAl2Si2O8), when the alumino-silicate framework contains ordered

Si, Al., the structure contains 16 inequivalent tetrahedral sites in which eight Si sites
alternate with eight Al –Sites such that each Si atom is linked to four tetrahedral Al
neighbours and vice versa (Angel et al., 1990).
Anorthite shows a phase transition from a low-temperature stable tri-clinic form
 to a high-temperature phase ðI 1Þ
ðP1Þ  at 2418C. The primitive unit cell of the I 1 form is
half as large as that of the P1 phase.
The P1 – I 1 transition corresponds to the addition of a body-centring sub-cell
transition, 1/2[111], which halves the number of crystallographically inequivalent sites.
Thus, in the high-temperature phase, the Ca atoms appear to occupy split sites
(K2O, Na2O, CaO) – Al2O3 –SiO2 System 637

corresponding approximately to a superposition of the pseudosymmetric P1 Ca positions,

i.e., those separated by a translation of ,1/2[111]. The I 1 symmetry corresponds to an
average of antiphase-related P1 domains. The correlated motion of the Ca atoms may
relate to the relatively slow fluctuations of the anti-phase domain boundaries observed at or
near Tc by electron microscopy (Van Tendeloo et al., 1989).
29
Si MAS-NMR spectroscopy is sensitive to short-range structure and relatively
slow dynamical processes (Phillips et al., 1992). Hence, it can provide information about
the structural and dynamical characteristics of structural-phase transitions in silicates,
particularly in structures like felspars, which display short-range variations in the state of
order affecting phase stability. In plagioclases (anorthite, in particular), the short-range Si,
Al order parameter can be obtained from a moment analysis of the 29Si MAS-NMR
spectrum (Phillips et al., 1992; Phillips and Kirkpatrick, 1995).
The 27Al NMR spectra eight resonances at low temperatures corresponding to the
eight inequivalent Al site of P1 anorthite. With increasing temperature, a gradual pair-wise
convergence is noted and only four resonances are seen to occur above the transition. The
hopping rate of Ca atoms between the split positions must be . ,6 kHz to give spectra
consistent with I 1 symmetry.

7.6. Reedmergnerite, NaBSiO8

In albite and reedmergnerite, NaBSi3O8, the kT –O – Tl angles are similar and there
is only 0.007 Å difference in kR(Na –O)l. Hackwell and Angel (1992) undertook a powder
XRD study of cell parameters for albite, reedmergnerite, anorthite and danburite
(CaB2Si2O8) and showed that the B analogue is more compressible than the Al-containing
structure.
Downs et al. (1999) obtained the structural and volume compressibility data for
reedmergnerite by single-crystal XRD at P # 4:7GPa: They obtained K0 ¼ 69:8ð5Þ GPa
(with K0 ¼ 4). Unit-strain tensors indicated the unit-cell compression to be anisotropic.
Na – O bonds decreased systematically while T– O – T angles showed wide behaviour.
The compression of reedmergnerite is similar to that of low albite, wherein bending
of the (Al, B) –Oco – Si angle compresses the Na-bearing zigzag channels.
The differences in compressibility between these B-bearing phases and other
plagioclases are due to differences in bending-force constants of B – O – Si vs. Al – O –Si.
Downs et al. (1999) determined the zero-pressure volume of reedmergnerite as
V0 ¼ 589:17ð3Þ Å3 and the zero-pressure bulk modulus, K0 ¼ 69:8ð5Þ GPa.
This page is intentionally left blank
 639

Chapter 8
Al2O3 – SiO2 and (CaO– MgO)–Al2O3 – SiO2 Systems

8.1. Al2O3 – SiO2 system

In the Al2O3 – SiO2 system, the polymorphs of Al2SiO3 — kyanite, sillimanite

and andalusite — are very important minerals to be learnt about in metamorphic and
experimental petrology because of their abundance in pelitic rocks and their simple
chemistry. A knowledge of the crystal structures of these polymorphs as a function of
pressure is of paramount importance for understanding the stability relationships
within the Al2SiO5 system. Small changes in thermodynamic properties greatly
influence the position of Al2SiO5 phase transition. These are considered as primary
thermobarometers for metamorphic grade studies of rocks such as metamorphosed
pelitic sediments.
These polymorphs provide an interesting crystal-chemical system. On the one
hand, all these structures have Si exclusively in tetrahedral coordination and one-half
of the total Al in octahedral coordination. The crystal structures of the three Al2SiO5
polymorphs have been summarized by Papike and Cameron (1976) and Kerrick
(1990). In these polymorphs, aluminium occurs in three different coordinations, in
addition to one-half of the total Al that is octahedrally coordinated in each structure.
These AlO6 octahedra share edges to form chains running parallel to the c-axis. This
is illustrated with the structure of sillimanite (see Fig. 8.1a and b). On the other hand,
the remaining Al is four-, five- and six-coordinated in sillimanite andalusite and
kyanite, respectively.
Relative to sillimanite and andalusite, kyanite is the high-pressure phase in the
Al2SiO5 system. The pressure effects on the crystal structures of andalusite
and kyanite have been studied by Ralph et al. (1984) and Yang et al. (1997a),
respectively.
For andalusite and sillimanite, the bulk moduli of 166 and 175 GPa (Voigt bound
values) were obtained by Vaughan and Weidner (1978); the corresponding Reuss bound
values they obtained were 158 and 166 GPa. They measured the elastic constants of
andalusite and sillimanite using the Brillouin scattering method. Their results show that
c33, the incompressibility along the c-axis, in both crystals is much greater than either c11
or c22 (c11 ¼ 233.4, c22 ¼ 289.0 and c33 ¼ 380.1 GPa for andalusite and c11 ¼ 287.3,
c22 ¼ 231.9 and c33 ¼ 388.4 GPa for sillimanite), suggesting a strong anisotropy of
compressibility for the two crystals.
640 Chapter 8

Figure 8.1. Crystal structure of sillimanite projected (a) down c and (b) down a (Yang et al., 1997c).
Al2O3 –SiO2 and (CaO – MgO) – Al2O3 –SiO2 Systems 641

8.1.1. Sillimanite and andalusite

Sillimanite under pressure (up to 5.29 GPa) shows a linear increase in a- and b-
axes, whereas c decreases nonlinearly. It exhibits the least compressibility along c but the
least thermal expansivity along a (Winter and Ghose, 1979). The bulk moduli of
aluminosilicates of this group and their polyhedra are determined as:

Workers K (GPa)

Sillimanite
Yang et al. (1997) 171(1)
Matsui (1996) 175
Vaughan and Weidner (1978) 175 (Voight bound values); 166 (Reuss bound values)
Andalusite
Yang et al. (1997) 151(3)
Ralph et al. (1984) 135(10)
Vaughan and Weidner (1978) 166 (Voight bound values); 158 (Reuss bound values)
Kyanite
Yang et al. (1997) 193
[Al1O6] 162(8)
[Al2O4] 269(33)
[SiO4] 367(89)

In all these polymorphic structures, [SiO4] tetrahedra are the most rigid units,
whereas the [AlO6] octahedra are the most compressible.
A decrease in compressibilities of the [AlO6] octahedra is noted as kyanite !
sillimanite ! andalusite in the order of their bulk moduli. Indeed, [AlO6] octahedra
control the compression of Al2SiO5 polymorphs. The compression anisotropy of
sillimanite is primarily a consequence of its topological anisotropy coupled with the
compression anisotropy of the Al– O bonds within the [Al1O6] octahedron (Fig. 8.2).
The edge-shared [Al1O6] octahedral chain is a common structural feature shared by all
three Al2SiO5 polymorphs. The octahedral chains are cross-linked by the relatively rigid
tetrahedra through corner sharing, leaving open structural tunnels running parallel to c as
well. Winter and Ghose (1979) noted that, with increasing temperature, both the relatively
rigid [Al2O4] and [SiO4] tetrahedra rotate in the same direction (clockwise). However, with
increased pressure, the two different tetrahedra rotate in the opposite direction, with the
[Al2O4] tetrahedra rotating anti-clockwise and the [Si4] tetrahedron rotating clockwise.
This is manifested by the changes in their out-of-plane tilting angles. A clockwise rotation
of the tetrahedra facilitates the expansion along b and works against the expansion along a
(Winter and Ghose, 1979).
The bulk moduli of the Al2SiO5 polymorphs appear to decrease linearly as the
molar volumes, Vm, increase with K0 ¼ 4.53– 5.8 Vm. Figure 8.3a and b shows the
relationship between bulk moduli and molar volumes for andalusite, sillimanite and
kyanite (Yang et al., 1997a,b).
642 Chapter 8

Figure 8.2. Variations of the Al – O bond distances within the [AlO4 ] tetrahedron with pressure
(Yang et al., 1997c).

In andalusite and sillimanite, there are two types of polyhedral chains running
parallel to c. In addition to chains of edge-shared AlO6 octahedra in both structures, there
are also fully extended double chains consisting of corner-shared SiO4 tetrahedra and AlO5
hexahedra in andalusite or SiO4 and AlO4 tetrahedra in sillimanite (see Figs. 4 and 5 in
Vaughan and Weidner, 1978). The structures of andalusite and sillimanite are less
compressible along c than along a or b because the polyhedral chains become the load-
bearing framework along c.
The relative volume compressibilities of andalusite (Ralph et al., 1984), kyanite
(Yang et al., 1996) and sillimanite as a function of pressure are shown in Fig. 8.4
(Yang et al., 1997).

8.1.2. Kyanite

Relative to sillimanite and andalusite, kyanite is the higher pressure phase in the
Al2SiO3 system. Kyanite is a common accessory mineral in eclogite-facies rocks for both
meta-basaltic and meta-sedimentary bulk compositions. When such crustal rocks are
subducted, kyanite participates in several breakdown reactions. In mafic rocks, such
reactions involve paragonite, zoisite, lawsonite and pumpellyite while, in metagreywackes
or metapelites, staurolite and chloritoid are involved.
Neither kyanite nor corundum has been reported so far from typical mantle
compositions, such as lherzolites or harzburgites but, near or below the transition zone,
fertilization of the peridotite is expected by subduction (Ringwood, 1991) and Al phases
other than garnet probably play a major role in mantle evolution. As an aluminium-
carrying phase, garnet has a structure which is stable down to a depth corresponding to a
pressure of , 20 GPa. A high-pressure phase of Al2SiO5, having a structure similar to
Al2O3 –SiO2 and (CaO – MgO) – Al2O3 –SiO2 Systems 643

Figure 8.3. (a) The relationship between the bulk moduli of the three Al2SiO5 polymorphs and their molar
volumes. Molar volumes are 44.158, 50.052 and 51.572 cm3/mol for kyanite, sillimanite, and andalusite,
respectively. The bulk moduli for three polymorphs are taken from Vaughan and Weidner (1978), Ralph et al.
(1984), and Yang et al.(1997). (b) The bulk modulus– volume relationship for AlO6 octahedra in kyanite (Yang
et al., 1997a, q1997 Mineralogical Society of America).

V3O5, is proposed to serve as a host of aluminium in the lower mantle (Ahmed-Zaid and
Madon, 1991).
Its structure is considered as a distorted cubic close-packed arrangement of
O atoms, with 10% of the tetrahedral sites filled with Si and 40% of the octahedral sites
filled with Al. There are four crystallographically distinct Al sites (Al1, Al2, Al3 and Al4)
and two Si sites (Si1 and Si2) (Fig. 8.5; Yang et al., 1997). Kyanite is 12– 14% denser than
sillimanite, which can be explained by its distorted cubic close-packing arrangement
of O atoms and its four AlO6 octahedra share several edges in a quite complex manner
644 Chapter 8

Figure 8.4. Relative volume compressibilities ðV=V0 Þ of andalusite (Ralph et al., 1984), kyanite (Yang et al.,
1997a), and sillimanite (Yang et al., 1997c), as a function of pressure q1997 Springer-Verlag.

(five each for the Al1 and Al3 octahedra and four each for the Al2 and Al4 octahedra). The
distortion in the packing is due to mismatched sizes of SiO4 and AlO6 groups, as well as the
narrow O – O contacts required for edge-sharing polyhedra. The AlO6 octahedra appear to
become more regular with increasing pressure, indicating that distortion in the closest
packing may decrease with pressure. The closest-packing distortion parameter (Ucp) and

Figure 8.5. Crystal structure of kyanite (read text). The strain ellipsoid was calculated using the unit-cell
parameters at 4.56 GPa relative to those at room pressure (Yang et al., 1997a, q 1997 Mineralogical Society
of America).
Al2O3 –SiO2 and (CaO – MgO) – Al2O3 –SiO2 Systems 645

TABLE 8.1
The closest-packing distortion parameter (Ucp) and the ideal closest-packed O atom radius in kyanite at various
pressures and temperatures (Yang et al., 1997)

P or T Ucp Radius (Å)

0.00 GPa 0.0658 1.372

1.35 GPa 0.0635 1.369
2.54 GPa 0.0631 1.367
3.73 GPa 0.0614 1.364
4.56 GPa 0.0607 1.362
258C 0.0652 1.373
4008C 0.0692 1.374
6008C 0.0689 1.379

Note: high-temperature data are taken from Winter and Ghose (1979).

the ideal closest-packed O atom radius in kyanite at various temperatures and pressures
are shown in Table 8.1.
Comodi et al. (1997) investigated the compressibilities of a kyanite crystal in a
DAC and found the mean axial compressibilities as:

a-axis : 2:00ð8Þ £ 1024 bar21

b-axis : 1:90ð4Þ £ 1024 bar21
c-axis : 2:00ð4Þ £ 1024 bar21
The principal axes of the strain ellipsoid were calculated with the STRAIN program
(Ohashi, 1982) and an anisotropic strain ratios were observed as:
b1 : b2 : b3 ¼ 1:4 : 1:2 : 1:
The greatest compressibility and smallest hardness are found along [001]. This
explains why the hardness along [001] in kyanite is less than that measured along normal to
[001] by about one unit on the Moh’s scale.

8.1.2.1. Bulk modulus

In comparison with andalusite and sillimanite, kyanite shows a much less
anisotropic compressibility and a large bulk modulus, which may contribute to its stability
at high pressures. For kyanite, Matsui (1996) obtained a bulk modulus of 197 GPa. Kyanite
compression appears to be complicated by the triclinic symmetry and the variety of
polyhedral units.
The isothermal bulk modulus, K0, and its pressure derivative as determined by
Comodo et al. (1997) appear as:
K0 ¼ 156ð10ÞGPa
K 00 ¼ 5:6ð55Þ
3
V0 ¼ 293:92ð9Þ A
646 Chapter 8

When K00 is set at 4, K0 becomes 160(3) GPa. In comparison, pyrope (another

compact phase) shows K0 ¼ 143 – 175 GPa (Hazen and Finger, 1978).
The a angle increases with P whereas it decreases with T. The angles b and g
remain almost unchanged with P as well as with T whereas, under pressure, AlO6
octahedra tend to become more regular. However, with temperature, their distortion in
terms of both longitudinal and shear strain increases.

8.1.2.2. dP/dT slope and stability

The disproportionation reaction
kyanite ¼ corundum þ stishovite
Al2 SiO5 Al2 O3 SiO2

is reported by Schmidt et al. (1997) to occur at 17.5 GPa and 10008C with a gently positive
dP/dT slope and with kyanite lying on the high-T, low-P side of the reaction. Earlier,
however, Harlov and Newton (1993) reversed the metastable equilibrium: kyanite ¼
corundum þ quartz. This again showed a positive slope, with kyanite appearing on the
low-T, high-P side of the reaction. Such a metastable reaction must be of significance in
natural occurrences for high-grade metamorphic rocks. As long as quartz or coesite is
present in association with corundum, kyanite should be favoured on the high-P side (with
a lower volume) of decomposition reactions.
The relatively high entropy of kyanite counterbalances its relatively large volume.
Kyanite is stable as its free-energy function is lower than that of corundum þ stishovite.
The geometrical structural invariance, expressed by the bv/av ratio and obtained
from the average compressibility and average thermal expansion of the cell volume
(Winter and Ghose, 1979), has been determined as 238C/kbar by Comodi et al. (1997). The
EOS, which applies in crystal P – T conditions, can be defined as:

V=V0 ¼ 1 þ 3:00ð7Þ £ 125 T 2 5:8ð1Þ £ 1024 P

where T is in 8C and P in kbar.

8.2. CaO –MgO – Al2O3 – SiO2 (CMAS) system

The CMAS system accounts for over 90% of the mantle composition and thus the
CMAS phase relations represent a suitable starting point for estimating mantle mineralogy.
For constructing phase diagrams for the system, the phase relations on the enstatite –
diopside join are essential. The first P –T phase diagram for the CMAS system, valid for
the whole range of P – T conditions in the upper mantle, was reported by Gasparik (1990).
This system helps the development of thermodynamic models relevant for mantle studies.
It is also employed in determining the thermobarometry of mantle xenoliths and
metamorphic rocks.
The CMAS system includes three ternary solid solutions: orthopyroxene,
clinopyroxene and garnet, and thus a divariant assemblage could have up to six
independent compositional variables.
Al2O3 –SiO2 and (CaO – MgO) – Al2O3 –SiO2 Systems 647

An internally consistent thermodynamic model for the CMAS system has been
developed by Gasparik (2000). This can be used as a simple tool for calculating the
phase relations in the CMAS system at pressures up to 10 GPa. The thermo-
dynamic properties of the end-member phases in the CMAS system are presented by
Gasparik (2000).
The CMAS system offers a very satisfactory approximation of the complex natural
compositions found in mantle xenoliths in terms of predicting the stability of mineral
phases and the correct phase relations. It is likely that the CMAS system is a good analogue
for the whole mantle and is thus essential for estimating its mineral and chemical
composition (Gasparik, 1990).
The compositions of the two co-existing pyroxenes on the CMAS solidus are
important in the processes that led to the differentiation of the upper mantle. Herzberg
et al. (1990) determined the compositions of the co-existing orthopyroxene,
clinopyroxene and garnet at 20808C and 10 GPa. From the compositions, they were
able to deduce that the melting of a chondritic mantle is peritectic and the composition of
the melt is outside the triangle formed by the compositions of the co-existing
orthopyroxene, clinopyroxene and garnet. Hence, the melting of the three-phase
assemblage produces peritectic melt and additional orthopyroxene. The peritectic melt
could evolve into the peridotitic compositions observed in mantle xenoliths only by
separation from the orthopyroxene residue and subsequent fractional crystallization of
clinopyroxene and garnet (Gasparik, 1990).
High-pressure garnets containing octahedrally coordinated silicon possibly
comprise almost half of the Earth’s transition zone (Ita and Strixrude, 1992). These
garnets mostly show tetragonal symmetry and one such, the MgSiO3 majorite garnet,
contains more than three silicons per 12 oxygens. Gasparik (1989, 1990, 1992) showed a
broad stability range and unexpected compositional complexity in high-pressure garnets in
the Na2O –MgO – Al2O3 – CaO – SiO3 (NMCAS) system.
Majorite (Mg3(MgSi)Si3O12 tetragonal space group I41 =a) has been synthesized at
around 17 GPa and 17008C (Hazen et al., 1994). Remarkably, this shows a pervasive
twinning by reflection on (110). In calcium – majorite (synthesized by Hazen et al., 1994), a
complete Mg – Si ordering on two octahedral sites, as well as Ca – Mg ordering on two
dodecahedral sites, has been observed along with the characteristic majorite reflection on
(110). Na –majorite, however, is cubic (space group Ia3d), optically isotropic and is
synthesized at 16.5 GPa and 16008C.
Gasparik (1990) outlined a simplified approach in modelling the CMAS system.
Garnet is treated as a pseudobinary solution of (pyrope, grossular) and (Mj, CM). Calcium
expands the stability of garnet causing displacement of phase boundaries, as discussed
below.
(a) In the CMAS system, the spinel lherzolite –garnet lherzolite boundary is displaced to
lower pressures with respect to the spinel peridotite– garnet peridotite boundary in the
MAS system (e.g., Gasparik, 1984).
(b) Formation of garnet on the enstatite – diopside join occurs at lower pressures than
the formation of MgSiO3 garnet from high-P clinoenstatite (Herzberg and
Gasparik, 1991).
648 Chapter 8

(c) Breakdown of enstatite– diopside garnet to ilmenite and CaSiO3 perovskite is

displaced to higher pressures than the corresponding boundary in the MgSiO3 system.
(d) The role of calcium to cause the expansion of the stability field can be explained by the
activity of calcium (aCa) in the dodecahedral site of garnet with an empirical
expression:

RT ln aCa ¼ 3RT ln½0:15 2 ð2:3P £ 1026 2 0:27Þ2  ðP in barsÞ

CMAS System (Gasparik, 1990). The phase relations in the CMAS system can be
calculated using the thermodynamic models for the constituent simple systems: MgO –
SiO2, MgO – Al2O3-SiO2 and CaO – MgO – SiO2.
At the lowest pressures, the assemblage consists of enstatite-rich orthopyroxene,
diopside-rich clinopyroxene, forsterite and anorthite (An-CaAl2Si2O8) corresponding to
olivine-gabbro. At 7– 9 kbar, anorthite is replaced with MgAl2O4 spinel, producing spinel
lherzolite. At pressures between 1.8 and 2.0 GPa, spinel is replaced with garnet, producing
garnet lherzolite. The garnet lherzolite assemblage is often used as a thermobarometer.
Garnet in the CMAS system is a reciprocal solution of pyrope, grossular, Mg –
majorite and Ca – majorite. Garnet may be treated as a pseudobinary solution of (Py, Gr) –
(MgMj, CaMj).
Phase relations in the system MgO –Al2O3 – SiO2 and CaO – MgO – SiO2 have been
experimentally investigated at 16.6 – 22.6 G.Pa pressures and 1350 – 25008C temperatures
by Gasparik (1990) using a split-sphere anvil apparatus (USSA 2000).
MgSiO3 ilmenite is stable to lower pressures (17.5 GPa) and CaSiO3 perovskite
forms at P (bar) ¼ 15T(8C) þ 151,000.
MgO –Al2O3- – SiO2. The three-component chemical system MgO – Al2O32 – SiO2
(MAS) accounts for 90% of Earth’s mantle. In the MAS system, the spinel peridotite
assemblage consists of orthopyroxene, forsterite and MgAl2O4 spinel. The stability is
limited to pressures below 2 –3 GPa. At very low pressures, spinel is replaced with
cordierite. The bulk composition of the mantle falls within the subsystem enstatite –
pyrope – forsterite (Mg2Si2O6 –Mg3Al2Si3O12 –Mg2SiO4), which is a close analogue for
spinel (Sp-MgAl2O4)- and garnet – peridotites.
The compositions of co-existing pyroxene and garnet on the enstatite –pyrope
join are primarily pressure dependent and are thus suitable for the geobarometry of
mantle xenoliths. Aluminious pyroxenes are fundamentally important for thermo-
barometry, especially of spinel peridotites and garnet peridotites. Gasparik (1992)
determined the melting temperatures on the join at different pressures between 8 and
15.2 GPa with a split-sphere anvil apparatus (USSA 2000). His work revealed a large
temperature dependence of the compositions of garnet co-existing with pyroxene. The
variation in garnet composition is likely to be caused by disordering.

8.2.1. Thermodynamic equilibria parameters of CMAS system

Clinopyroxene in the CMAS system can have up to four components in solution:

diopside (CaMgSi2O3), Ca-tschermak pyroxene (CaAl2SiO6), Ca-Eskola pyroxene
Al2O3 –SiO2 and (CaO – MgO) – Al2O3 –SiO2 Systems 649

(Ca0.5AlSi2O6) and enstatite (Mg2Si2O6). Diopside and Ca-Tschermak pyroxene are the
only end-members that are stable and can form a complete solid solution.
Gasparik (1986) studied the SiO 2-saturated CMAS system in which
the clinopyroxene in equilibrium with anorthite (An), kyanite (Ky) or a silica polymorph
incorporates variable amounts of the Ca-Eskola component. In the CMAS system,
he experimentally determined the equilibrium composition of clinopyroxene co-existing
with anorthite and quartz under varying P – T conditions (1200 – 14008C, 1.4 – 3.04 GPa).
The data were fitted to the Redlich –Kister equation for a ternary solution, providing
the following mixing properties of the CaMgSi 2O6- – CaAl2SiO6- – Ca0.5AlSi2O6
(i.e., Di-CaTs-CaEs) pyroxene:

Gexcess
Px ¼ XCaTa XDi ½10; 000 2 10T 2 0:07Pð50; 000 2 31:19T þ 0:04PÞðXCaTs 2 XDi Þ

2 7000ðXCaTs 2 XDi Þ2  þ XCaTs XCaEs ½220160 þ 14760ðXCaTs 2 XCaEs Þ

þ XDi XCaES ½25950

ðunits : joules; kelvins; barsÞ

The equilibria in the CMAS system and the corresponding parameters are shown in
Table 8.2. The listed parameters are internally consistent and hence it is possible to
calculate from these the thermodynamic parameters of all participating phases in the
CMAS system. These are listed in Table 8.3 (Gasparik, 1996b).

8.2.2. Garnet structure

Garnets of general chemistry, A2B2Si3O12, are among the few major minerals
which constitute the Earth’s upper mantle and the transition zone Garnets have the
general formula [viii]A2þ[vi]
3 B3þ[iv]
2 Si4þ
3 O12, where eight-coordinated A is commonly Mg,
Fe, Mn or Ca and six-coordinated B is usually Fe, Al or Cr. The garnet structure consists
of a corner-linked framework of alternating Si4þ tetrahedra and trivalent octahedra; it is
formed of a relatively rigid, corner-linked framework of alternating silicate tetrahedra
and octahedra with Mg2þ, Al3þ or Si4þ. The interstices in the framework define larger
dodecahedral cation sites containing Mg2þ, Ca2þ or Na1þ. This framework defines eight-
coordinated sites occupied by divalent cations. The structure is cubic with space group
Ia3d, with only four symmetry-independent atoms: A, B, Si and O. Garnets may be
visualized formed of a framework structure with a three-dimensional network of corner-
linked silicate tetrahedra and Fe3þ octahedra. The inter-polyhedral angle between
tetrahedra and octahedra (the Si –O –Fe angle) undergoes the largest change — a
decrease of 1.58 — between room pressure and 5 GPa. At high pressures, however, this
angle appears to become constant at ,1328. High-pressure garnets incorporate Si in both
four- and six-fold coordination.
Cubic andradite and pyrope garnets adopt space group Ia3d: Three cation types —
Si at (3/8, 0, 1/4), the octahedral cation at (0, 0, 0) and the eight-coordinated cations
at (1/8, 0, 1/4) — are in fixed special positions, whereas oxygen is in the general position.
650 Chapter 8

TABLE 8.2
Equilibria in the system CaO–MgO–Al2O3 –SiO2 and the corresponding parameters (from Gasparik, 1986b)

No. Equilibrium DGT;P ¼ DHT0 2 DS0P 2 cT d þ PDVT0 2 bP2

1 OEn ¼ hEn DGð1Þ ¼ 3457 2 1:95T þ 0:042P 2 1027 P2

2 OEn ¼ CEn DGð2Þ ¼ 3300 þ 2:5T 2 0:08P
3 OEn ¼ IEn DGð3Þ ¼ 21921 þ 2:29T 2 0:011P þ 1027 P2
4 PEn ¼ OEn Dið4Þ ¼ 217932 þ 35:55T 2 0:6T 1:5 ð1 2 633 £ 1028 PÞ 2 0:32P
5 2CEn ¼ Mj DGð5Þ ¼ 135900 þ 157:6T 2 30T 1:2 2 0:88P 2 1026 P2
6 Mj ¼ 2Bt þ 2 St DGð6Þ ¼ 19100 2 132T þ 30T 1:2 2 0:32P þ 1026 P2
7 Mj ¼ 4MgI1 DGð7Þ ¼ 90200 2 134:4T þ 30T 1:2 2 0:7P þ 1026 P2
8 Mj ¼ 4MgPv DGð8Þ ¼ 310200 2 152T þ 30T 1:2 2 1:54P þ 1026 P2
9 Fo ¼ Bt DGð9Þ ¼ 30000 þ 6:7T 2 0:284P
10 Bt ¼ sSp DGð10Þ ¼ 10700 þ 7:6T 2 0:12P
11 CEn ¼ Bt þ St DGð11Þ ¼ 77500 þ 12:8T 2 0:6P
12 Bt þ St ¼ 2MgI1 DGð12Þ ¼ 35550 2 1:2T 2 0:19P
13 sSp þ St ¼ 2MgI1 DGð13Þ ¼ 24850 2 8:8T 2 0:07P
14 MgI1 ¼ MgPv DGð14Þ ¼ 55000 2 4:4T 2 0:21P
15 Bt ¼ MgPv þ Pc DGð15Þ ¼ 112025 2 0:5P
16 sSp ¼ MgPv þ Pc DGð16Þ ¼ 101325 2 7:6T 2 0:38P
17 ODi ¼ Di DGð17Þ ¼ 232245 þ 12T 2 0:13P
18 ODi ¼ CDi DGð18Þ ¼ 222000 þ T þ 0:2P
19 PDi ¼ Di DGð19Þ ¼ 211920 2 7T
20 2Di ¼ CM DGð20Þ ¼ 113000 þ 5T 2 0:7P
21 2CM ¼ 4CaPv þ Mj DGð21Þ ¼ 364000 þ 45T 2 2:6P
22 OEn þ aSp ¼ MgTs þ Fo DGð22Þ ¼ 23400 2 2T
23 OEn þ MgTs ¼ Py DGð23Þ ¼ 8600 2 92:5T þ 19T 1:2 2 ð0:85 2 35 £ 1027 PÞP
24 Py ¼ 3MgI1 þ AIll DGð24Þ ¼ 180000 þ 40T 2 0:7P
25 A1I1 ¼ A1Pv DGð25Þ ¼ 58000 2 4:4T 2 0:21P

Units: Joules, Kelvins and bars; OEn or thoenstatite; hEn high clinoenstatite. CEn high-P clienoenstatite; lEn low
clinoenstatite; PEn protoenstatite; Mj majorite. Bt beta phase; St stishovite; MgIl ilmenite; MgPvMgSiO3
perovskite; Fo forsterite; sSp Mg2SiO4 spinel; Pc periclase Odi orthodiopside; Di diopside; CDi high-P
clinodiopside; PDi protodiopside; CM CM phase; CaPv CaSiO3 perovskite; aSp MgAl2O4 spinel; MgTs
Mg-tschermak pyroxene; Py pyrope. AlIl Al2O3 ilmenite; AlPv Al2O3 perovskite (from Gasparik, 1986b).

In crustal silicate garnets, the cation valences for octahedral and dodecahedral sites are
usually þ 3 and þ 2, respectively. Any increase in octahedral cation valence is
compensated by a decrease in average dodecahedral cation valence.
Haggerty and Sautter (1990) reported silica-rich garnet in mantle-derived nodules
having more than four Si atoms per 12 O atoms, as opposed to three Si in crustal garnets.
The depth postulated from this Si enrichment goes deeper than 300 km — the deepest
mantle sample so far recorded.
The compressibility of garnet depends almost entirely on the extent to which the
tetrahedral – octahedral framework changes with pressure. All known silicate garnets have
essentially full occupancy by silicon of tetrahedra, giving an average tetrahedral cation
valence of þ 4. High-pressure silicate garnets may incorporate six-coordinated silicon and
thus approaches a pyroxene composition, such as synthetic garnet MgSiO3, which
Al2O3 –SiO2 and (CaO – MgO) – Al2O3 –SiO2 Systems 651

TABLE 8.3
Thermodynamics properties of phase in the system MgO–Al2O3 –SiO2

Phase DHfQ;970 (kJ mol21) S 0970 (J mol21 K21) V 0298 (J bar21) a £ 105 KT (kbar) K0

Beta phase (Mg2SiO4) 232.0 268.3 4.052a 3.2b 1720c 5

Corundum 0 176.8 2.558b 2.6d 2500e 4
Enstatite (Opx) 269.5 382.5 6.264f 3.2d 1070g 4
Enstatite (Cpx) 261.2 376.3 6.034 3.2 1280 4
Frosterite 262.0 275.0 4.360h 3.5h 1280e 5
Ilmenite (MgSiO3) 24.1 186.2 2.635a 2.5i 2100 j 4
Ilmenite (Al2O3) 95.8 165.7 2.635 2.5 2100 4
Majorite 52.3 756.6 11.431k 1.4 1180 4
Mg- Tschermak Px 27.0 380.5 5.888l 3.2 1070 4
Periclase 0 79.0 1.124m 4.2m 1610a 4
Perovskite (MgSiO3) 65.0 189.7 2.446a 4.4 2440n 4
Perovskite (Al2O3) 139.7 169.2 2.446 4.4 2440 4
Pyrope 281.9 764.3 11.320o 2.7p 1740e 4
Spinel (MgAl2O4) 222.5 270.7 3.971q 2.7d 1950e 5
Spinel (Mg2SiO4) 221.3 260.7 3.965a 2.6b 1820c 5
Stishovite 45.0 103.0 1.401a 2.3r 3060s 5
a
Jeanloz and Thompson (1983); bAkaogi et al. (1984); cWeidner et al. (1984); dSkinner (1966); eSumino and
Anderson (1984); fChatterjee and Schreyer (1972); gSumino and Anderson (1984); hHazen (1976a); iAshida et al.
(1988); jWeidner and Ito (1985); kAngel et al. (1989); lGasparik and Newton (1984); mHazen (1976b); nYeganeh-
Haeri et al. (1989); oNewton et al. (1977); pHazen and Finger (1978); qRobie et al. (1978); rIto et al. (1974);
s
Weidner et al. (1982).

corresponds structurally to Mn3(MnSi) Si3O12 (Fujino et al., 1986). This garnet shows
tetragonal symmetry because Mn and Si order in two symmetrically distinct octahedral
sites (Ringwood and Major, 1967).
Garnets constitute a major component of the Earth’s mantle. But the P – T
conditions at which they transform to high pressure polymorphs, especially to the
perovskite structure, are highly uncertain as compared to the other mantle minerals
(Mg,Fe)2SiO4-olivine and (Mg, Fe)SiO3-pyroxene.
Calcium content: thermobarometer. The mole fraction Ca in the dodecahedral site
of garnet (XCa) is related to pressure as
XCa ¼ 0:15 2 3 £ 1027 P ðP in kbarÞ
The activity of the Mg – majorite component in the garnet solution corresponds to

RT ln aMgMj ¼ 3RT lnð0:85 þ 3 £ 1027 PÞ þ 2RT lnð1 2 XAl Þ þ ð15; 000 þ 63TÞ
 ð0:15 2 3 £ 1027 PÞ2

where XAl is the mole fraction of Al in the octahedral site.

The Ca content in the M2 site of clinopyroxene is unaffected by the presence of Al
in the M1 site, because mixing in the two dodecahedral sites of clinopyroxene is mutually
independent (Gasparik, 1984a).
652 Chapter 8

The Ca content of orthopyroxene is very low at all pressures. Orthopyroxene is

treated as an ideal binary solution of the enstatite and Mg-Tschermak components.

8.2.2.1. Andradite
In andradite, the Ca-containing eight-coordinated polyhedron, Fe3þ-containing
octahedron and Si-containing tetrahedron — all show significant compression. Between
ambinent pressure and 19 GPa the total compression of average Si – O, Fe– O and Ca –O
bonds amounts to 3.1, 1.9 and 4.0%, corresponding to polyhedral bulk moduli of 200, 330
and 160 GPa (each ^ 10%), respectively (Hazen and Finger, 1989).
At pressures up to 12.5 GPa, the O –Si –O and O – Fe –O angles are essentially
unchanged and the O – Ca –O angles vary by only ,0.58; also no significant Si –O
compression is ever observed.
The high pressure behaviour of andradite up to 5 GPa is similar to that of
framework silicates, in which a corner-linked network of smaller and more rigid polyhedra
“collapses” about the larger and more compressible polyhedra.

8.2.2.2. Pyrope Mg3 Al2(SiO4)3

Magnesium-rich garnet (i.e., pyrope) is considered to be one of the dominant
minerals in the upper mantle and transition zone. Various experimental techniques have
been employed for determining the bulk modulus of pyrope (see Table 8.4). The results
have been tabulated in Table 8.3. Along with other phases of MgO-Al2O3-SiO2 system.
Zhang et al. (1998) determined by X-ray diffraction the unit cell parameters of
pyrope single crystal at 33 GPa. The volume data from ambient pressure to 33 GPa were
fitted to a third-order Birch-Murnaghan EOS
"

5=3 #( "
 #)
3 V0 7=3 V0 3 0 V0 2=3
P ¼ KTO 2 1 þ ð4 2 K TO Þ 21 ð8:1Þ
2 V V 4 V
0
where KTO and KTO are the isothermal bulk modulus and its first pressure derivative, V0 the
unit-cell volume at ambient pressure.
The average Al– O bond length 1.889(2) Å in pyrope is 6.3% smaller than the value
of Fe3þ – O 2.016(2) Å in andradite at ambient pressure, whereas the polyhedral bulk
modulus of AlO6 is ,36% smaller than that of Fe3þ – O6 in andradite with a value of
330 ^ 33 GPa (Hazen and Finger, 1989).
From ambient pressure to 19 GPa the Si – O bond in pyrope experiences
compression ,1.1%, whereas that of andradite is 3.1%. The bulk moduli are
580(24) GPa for pyrope and 200(20) GPa for andradite. In garnets, with increasing
pressure, the SiO4 tetrahedra and MgO8 polyhedra become more regular, whereas the AlO6
octahedra become distorted. The polyhedral volumes of MgO8, AlO6 and SiO4 together
account for ,44.2% of the unit cell volume. It is possible to obtain a bulk modulus from
those of the constituent polyhedra in a very close packed structure like garnet. However
MgO8 contributes primarily to the compression of the whole structure (Zhang et al., 1998).
The bulk moduli of AlO6 and SiO4 are 211 (11) and 580 (24) GPa respectively, with
pressure derivative fixed to 4.
Al2O3 –SiO2 and (CaO – MgO) – Al2O3 –SiO2 Systems 653

The linkages between AlO6 and SiO4 parallel to the crystallographic axes are
similar to chains which are again connected to each other and form a three dimensional
framework. The compression can be induced by kinking the Si – O –Al angle between
the two polyhedra. In pyrope this kinking is manifested by a continuous decrease of the
Si – O – Al angle from 131.0(2) to 126.9(2) degrees between ambient pressure and
33 GPa (Fig. 8.6).
A linear relationship between normalized distortion parameters for MgO8 and AlO6
polyhedra and the normalized Si –O – Al kinking angle (Fig. 8.7) strongly suggests that the
kinking (bending) between AlO6 and SiO4 is the governing factor in the compression of the
pyrope structure.
The side bands of pyrope strongly resemble that of YAG and the frequencies of
most of the modes correspond well to the positions of the available infrared and
Raman data (Hofmeister and Chopelas, 1991). The intensity variation of the band
around 1100 cm21 with pyrope content shows a linear trend Fig. 8.7 (Hofmeister and
chopelas, 1991).
At about 28 GPa the slope of the optical mode side band at 852 cm21 decreases
by about 45%. A similar decrease in resolution of side band fluorescence spectra is
observed for MgSiO3 majorite at 38 GPa (Chopelas, 1991), where the Raman spectrum
exhibited dramatic changes with new major peaks appearing as are characteristic of
cubic garnets. The peak, characteristic of the tetragonal structure, on the other hand,
disappears.
A plot of velocity versus volume reveals a break in slope of the velocity – volume
function at approximately 28 GPa. At this pressure an increase in the slope of VP versus
V/V0 by over 60% is seen.

Figure 8.6. The change of the kinking angle Si–O –Al to 33 GPa (Source: Zhang et al., 1998a).
654 Chapter 8

Figure 8.7. The linear intensity variation of the band around 1100 cm21 with the pyrope content (Chopelas, 1991).

8.2.3. Pyrope ! ilmenite ! perovskite transformation: Al-content

For pyrope garnet Liu (1977) observed a transformation to the ilmenite structure at
,24 GPa and to the perovskite structure at ,30 GPa (/,10008C). Pyrope has been
observed to transform to an aluminious perovskite and a corundum (Al2O3)/ilmenite
(MgSiO3) solid solution (Irifune et al., 1996). All aluminium is observed to be
accommodated in a single perovskite phase above 43 GPa (Serghiou et al., 1998).
Sherghiou et al. (1998) observed that with increasing pressure pyrope transforms to
an ilmenite phase above ,21.5 GPa, to perovskite plus ilmenite above ,24 GPa and
to perovskite above 29 GPa. A small amount of Al2O3 co-exists with perovskite up to
43 GPa. But only above 43 GPa the entire Al2O3 content of pyrope is accommodated in the
perovskite structure.

8.2.4. Almandine (Fe3Al2Si3O12) break-down

For the results of their high P – T synthesize experiments on almandine in a laser-

heated DAC, Kesson et al. (1995) offered two possibilities: one was the formation of a
Ca-ferrite structured FeAl2O4 plus stishovite and the other, decomposition of almandine
to constituent oxides. Later Conrad (1998) demonstrated that almandine (Fe3Al2Si3O12)
decomposes to oxides FeO, SiO2 and Al2O3, rather than transforming to a perovskite-
structured phase at lower mantle pressures. The complete transformation of almandine to
oxides was found at a pressure of only 22 GPa. The reaction proceeded as:

Fe3 Al2 Si3 O12 ! 3Fe þ 3SiO2 þ Al2 O3

This behaviour differs from that of pyrope (Mg2Al2Si2O12), which transforms first
to an ilmenite (or corundum) structure and then to perovskite at lower mantle conditions
see the earlier section).
Al2O3 –SiO2 and (CaO – MgO) – Al2O3 –SiO2 Systems 655

However, Conrad (1998) observed no other silicate products, such as a proposed

Ca-ferrite structured FeAl2O3, produced from almandine.

8.2.5. Factors for garnet compression

Grossular, uvarovite andradite and spessartine have relatively large unit-cell

volume and display an inverse bulk-modulus– volume trend (Olijnyk et al., 1991).
However, pyrope and almandine, with relatively small compressible cations (Mg2þ and
Fe2þ), are strangely much more compressible than expected (Bass, 1989). This increase in
compressibility of Mg –Fe garnets may arise from a lack of structural support provided by
relatively small dodecahedral cations for the octahedral-tetrahedral framework. Indeed,
Mg –Fe –garnets have a bulk modulus close to about 176 GPa (see Table 8.3).
The Mg2þ octahedra are more compressible than those of Al3þ and Si4þ and thus
contribute to a lower garnet bulk modulus. On the other-hand, Na –majorite (NAMAJ),
with relatively incompressible Si4þ in all octahedral sites, has the largest bulk modulus
reported for any silicate garnet value. It also shows a small unit-cell volume
(1485.5 ^ 0.3 Å3) (Fig. 8.8). The two factors, molar volume and octahedral cation
valence, appear to control garnet compression.

Figure 8.8. A plot of garnet bulk modulus (in GPa) versus unit-cell volume (in Å3) reveals bulk modulus–volume
systematics. Solid circles represent the five specimens from this study, while open circles are: “and”, andradite
(Bass, 1986); “uv”, uvarovite (Bass 1986); “gr”, grossnlar (Bass, 1989); “sp”, spessartine (Bass, 1989); “al”,
almandine (Bass, 1989); “pyr”, pyrope (O’Neill et al., 1991). The dashed line represents a bulk modulus–volume
trend defined by andradite, uvarovite, grossular, and spessartine. Note that NAMAJ a garnet with Si4 filling both
octahedral cation sites, is significantly less compressible than the other garnets recorded. MAJ and CAMAJ with
Mg2þ and Si4þ in octahedral sites lie below the general bulk modulus– volume trends delined by other garnets
(Hazen et al., 1994a,b, q1994 Mineralogical Society of America).
656 Chapter 8

8.2.6. Bulk moduli

The P – V resolution of garnet crystals may fit to a second-order Birch-Murnaghan

EOS for which the pressure derivative of bulk modulus, (K 0 ), is fixed to a value of 4; while
bulk modulus, K and room-pressure cell-volume, V0, are varied. The total compressibility
of garnet between room pressure and 5 GPa is only about 2% — not enough of a change to
derive a meaningful value of K 0 .
For isoelectronic silicate garnets the bulk moduli are observed to vary from about 160
to 180 GPa. Substitution of magnesium with iron slightly increases the bulk modulus and K0
increases approximately 1.4% from pyrope to almandine (Yagi et al., 1987). Ultrasonic
measurements by Bakuska et al. (1978) also show similar values. However, they reported
that the dissolution of the pyroxene component in garnet decreases the bulk modulus, which
varies linearly with composition. The bulk modulus of garnet-structured MgSiO3 becomes
160 GPa. Two factors such as molar volume and octahedral cation valence, appear to control
garnet compression (Hazen et al., 1994). Using single crystal X-ray diffraction analysis of
pyrope up to 5 GPa Levien et al. (1979) obtained K0 ¼ 175 GPa and K00 ¼ 4.5. Leitner et al.
(1980) measured acoustic wave velocity using the Brillouin scattering technique and
calculated the isothermal bulk modulus to be 175 GPa, the bulk modules and its pressure
derivative for pyrope determined by various workers are presented in Table 8.4.
Pressure –volume data (Table 8.4) of andradite obtained by Hazen and Finger
(1989) were fitted to a Birch-Murnaghan EOS, with an assumed value of 4 for dK/dT.
Those yield a bulk modulus of 159 ^ 2, while Bass (1986) obtained this value as 158 ^ 2
(for andradite) using Brillouin scattering. Later, Hazen et al. (1994) through an experiment
on a high-pressure mount with several crystals, found that the bulk moduli of the garnets
range from 164.8 ^ 2.3 GPa for Ca – majorite to 191.5 ^ 2.5 GPa for Na –majorite
(assuming K0 ¼ 4) (see Fig. 8.8).
Hazen and Finger (1989) documented high pressure crystal structures of pyrope
and andradite (Ca3Fe2Si3O12) and observed that garnet bulk compression approximately

TABLE 8.4
Bulk modulus and its pressure derivative for pyrope, Mg3Al2(SiO4)3 (SC single crystal, PC polycrystalline)
(Source: Zhang et al., 1998a)

K (GPa) K0 K00 P (GPa) Sample Method References

168.2(4) 4.74(16) – 1 SC Ultrasonic Bonczar and Graham (1997)

173.0(9) – – 0 SC Ultrasonic Babuska et al. (1978)
173.6(4) 4.93(6) 20.28(4) 3 SC Ultrasonic Webb (1989)
175(1) – – 0 SC Brillouin scattering Leitner et al. (1980)
172.8(3) – – 0 SC Brillouin scattering O’Neill et al. (1989)
190(6) 5.45 – 35 PC X-ray diffraction Takahashi and Liu (1970)
171(3) 1.8(7) – 10 PC X-ray diffraction Sato et al. (1978)
175(1) 4.5(5) – 5 SC X-ray diffraction Levien et al. (1979)
172.8 3.8(1.0) – 25 PC X-ray diffraction Leger et al. (1990)
174(3) 4 – 4.7 SC X-ray diffraction Hazen et al. (1994)
171(2) 4.4(2) – 32.5 SC X-ray diffraction Zhang et al. (1998)
Al2O3 –SiO2 and (CaO – MgO) – Al2O3 –SiO2 Systems 657

TABLE 8.5
Bulk modulus K0 versus dK0 /dP of garnets (Yagi et al., 1987)

dK0 /dP 1 2 3 4

Material
Pyrope 173.5(8) 171.1(7) 168.7(8) 166.4(8)
58% En, 48% Py 166.5(5) 164.2(6) 162.0(6) 159.8(6)
almandine 176.0(22) 173.4(22) 170.8(22) 168.3(23)
18% Fs, 82% Alm 172.7(11) 169.9(1) 167.2(11) 164.6(11)

Unit of bulk modulus is 1 GPa.

equals that of the dodecahedral sites. In andradite, e.g., bulk moduli for the dodecahedral
site and the crystal are 160 and 159 GPa, respectively. It appeared to Hazen et al. (1994)
that the dodecahedral site merely scales (under compression) the much more rigid
tetrahedral – octahedral framework. Furthermore, the compressibility of the framework is
determined primarily by the valence of the octahedral cations.
For pyrope (Mg3Al2Si3O12) the bulk modulus was obtained as 177 GPa (Leitner
et al., 1980). Refinement of eight andradite P – V data yields K ¼ 146 ^ 15 GPa and
dK/dP ¼ 6.1 ^ 2.3. Six pyrope P – V data points give K ¼ 163 ^ 3 GPa and
dK/dP ¼ 12 ^ 2. Presuming dK/dP ¼ 1, 2, 3 and 4, the corresponding bulk moduli of
pyrope, (En58Py48) and almandine (Fs18Aln82) are presented in Table 8.5.
The bulk modulus –volume relationship of natural silicate garnets provides some
rationale for the observed variations in bulk moduli from 155 onward (Fig. 8.8 see the
caption). Garnets having trivalent octahedral cations and relatively large dodecahedral
cations (Ca2þ and Mn2þ), including grossular, uvarovite andradite and spessartine, have
relatively large unit-cell volumes and display an inverse bulk-modulus – volume trend
(e.g., Olijnyk et al., 1991). This increased compressibility of Mg – and Fe –garnets may
result from the lack of structural support for the octahedral-tetrahedral framework,
provided by relatively small dodecahedral cations. Most Mg – or Fe –garnets with unit cell
volumes less than about 1600 Å3 have bulk modulus close to about 176 GPa; and thus do
not follow the bulk modulus– volume trend of garnets with greater unit-cell volumes.

8.2.7. Thermal expansion

Dissolution of pyroxene in garnet has very little effect on thermal expansion, but
substitution of iron and magnesium in the pyrope-almandine join seems to have a large
effect.
For the pyroxene-garnet solid solution it is observed (Yagi et al., 1987) that with
increasing pyroxene component the bulk modulus seems to decrease; and also that the
thermal expansion of majorite is expected to be similar to that of garnet with the same
Fe /Mg ratio.
When do is calculated, using the (dK/dT)P data reported by Soga (1967), its values
become 5.3 for pyrope and 6.4 for almandine.
658 Chapter 8

8.2.8. YAG

Yttirium aluminium garnet, Y3Al2(AlO4)3, under high pressure shows spectra,

wherein the signal quality or resolution does not degrade with pressure. The acoustic
modes are seen at 190 and 319 cm21 (Chopelas et al., 1996, Fig. 8.9a and b). These
frequencies yielded high-pressure velocities consistent with the available ultrasonic data
(Yogurtchu et al., 1980). As in MgAl2O4, the shear velocity of YAG changes very little
with pressure. The fluorescence spectra of Cr-doped YAG-at various pressures are shown
in Fig. 8.10 (see caption).

8.2.9. Mg – Cr – garnet

Garnet in diamonds of peridotite suite is essentially the solid solution of

pyrope – knorringite system; and the garnet inclusions in the diamond of kimberlite are
seen to contain up to 50% of knorringite molecule, Mg3Cr2Si3O12 (Sobolev, 1974;
Tsai et al., 1979).
Garnet containing up to 22% of knorringite molecule was synthesized at 5 GPa and
12008C (Maleniovsky et al., 1975). Knorringite end member is reported (Ringwood, 1977)
to have been synthesized at about 7 GPa and 14008C. By experiment Irifune et al. (1982)
observed that at 12008C a single phase of Mg3AlCrSi3O12 garnet was stable at pressures
above 8 GPa. Below 8 GPa a mixture of garnet solid solution and enstatite þ corundum
solid solution was obtained (Fig. 8.11a).

Mg3 AlCrSi3 O12 $ 3MgSiO3 þ ðAl; CrÞ2 O3

garnetsss enstatite corundumss

Figure 8.9. (a) The transverse and longitudinal acoustic modes of YAG at 190 and 319 cm21, respectively, versus
pressure. The uncertainties indicated by the error bars are about 0.5%. (b) The shear and compressional velocities
of YAG are calculated. The lattice constant at each pressure was calculated with the third order finite strain
equation of state (Birch, 1978) and a bulk modulus of 180.3 GPa and its pressure derivative of 5.2 (from Cholpelas
et al., 1996).
Al2O3 –SiO2 and (CaO – MgO) – Al2O3 –SiO2 Systems 659

Figure 8.10. Fluorescence spectrum of chromium-doped YAG under pressure. The R1 line is set to 0 cm21 in
order to emphasize the pressure effect on the sidebands. The band at 688.8 nm shifts at a rate of 0.0385 nm/GPa.
The acoustic modes are indicated at each pressure. Note that the quality intensity and resolution of the sidebands
continue even to the highest pressures (from Chopelas et al., 1996, q 1996 The Geochemical Society).

The lower limit of the pressure where garnet is stable increases sharply with the
increase of the knorringite content. At 12008C and at 3 GPa the content of knorringite
molecule in pyrope is 8 mol%, whereas at 5 GPa the content increases to 22 mol%. The
solubility limit of knorringite molecule in pyrope is seen to vary linearly as a function of
temperature and pressure. In contrast to the garnets of other regions the garnets of
peridotite suite, especially those included in diamond, have a compositional feature of low
Ca and high Mg and Cr contents.
Garnet is the major phase, next to spinels and clino-pyroxenes, capable of
accommodating Al and Cr. The Cr/(Cr þ Al) value of garnet solid solution may serve as an
useful indicator of the minimum pressure necessary for crystallization and the Cr/(Cr þ
Al) value of garnet (0.2 –0.4) corresponds approximately to the depths of the upper mantle
where diamonds form. Garnets in diamonds are reported to possess the ratio of
Cr/(Cr þ Al) around 0.5 and are regarded to be equilibrated at pressure higher than 8 GPa,
i.e., at depths below 240 km in the upper mantle. The deep mantle xenoliths originate at
such depths (Mercier and Carter, 1975).
660 Chapter 8

Figure 8.11. (a) Phase diagram of pyrope Mg3Al2Si3O12 knorringite Mg3Cr2Si3O12 system at 1200 and 14008C.
(I) garnet solid solution þ enstatite þ sapphirine þ kyanite solid solution, (II) enstatite þ sapphirine þ kyanite
solid solution, and (III) enstatite þ sapphirine þ kyanite solid solution þ corundum solid solution (Irifune and
Hiraya, 1983). (b) Solubility limit of knorringite molecule in pyrope as function of pressure and temperature
(Irifune and Hiraya,1983; Irifune et al., 1982, q 1982 Elsevier Scientific Publishing Co.).

8.2.9.1. Cr, Al fractionation in garnets

The undepleted lherzolite has the ratio Cr/(Cr þ Al) as , 0.07 (Maaløe and Aoki,
1977), while that of basalt is 1023. The process of fractionation is presumed to give rise to
an enrichment of Al in the shallower upper mantle and an enrichment of Cr in the deeper
Al2O3 –SiO2 and (CaO – MgO) – Al2O3 –SiO2 Systems 661

part of the upper mantle. Garnet peridotites with diamonds, having inclusions of garnets
with high amount of knorringite molecule (Cr/(Cr þ Al) ¼ 0.2 –0.4), possibly originate at
the deeper part of the upper mantle extending down to 240 km depth.
Analysis of the experimental data on pyrope – knorringite solid solution leads to the
following observations:
(1) Garnetss at low pressure dissociates to enstatite þ corundum.
(2) Enstatite shows very little solubility for Al3þ and Cr3þ (limited to a few wt%,
Malinovsky et al., 1975) under high pressure.
(3) Garnet and corundum solid solutions are represented by symmetric simple solution of
Guggenheim (1959).
(4) The molar volume of each component does not change with temperature and pressure.
Chromium concentration in subsolidus lherzolite and liquidus komatiitic garnets
are not affected by pressure, in contrast to the garnets in spinel harzburgite. In the latter at
constant temperature, Cr partitions more strongly into garnet with increased pressure
(Canil and Wei, 1992). This observation is consistent with phase equilibria for garnets
on the join pyrope – knorringite (Fig. 8.11b see caption) which show that with increasing
pressure at constant temperature, garnets richer in knorringite component are stable
along the join (Irifune and Hiraya, 1983). Similar observations were also made by Brey
et al. (1991).

8.2.10. Tetragonal garnets

At crustal pressure most garnets have cubic symmetry with the general formula A2þ
3
B3þ
2 Si3O12, where eight-fold coordinated A is usually Mg, Fe, Mn or Ca and six-fold
coordinated B is Al, Fe, or Cr. High pressure garnets are able to incorporate cations of
valence 1þ , 2þ , 3þ and 4þ and cations of wide range of size and electronic structure.
Silicate garnets synthesized at high pressure commonly incorporate [6]Si, as in
Mn3(Mn,Si)Si3O12 (equivalent to MnSiO3), which has tetragonal symmetry because of
the ordering of Mn and Si in octahedral sites (Prewitt and Sleight, 1969). Tetragonal
garnets have three symmetry-independent SiO4 tetrahedra, all of which are nearly regular,
with typical orthosilicate mean T –O distance close to 1.64 Å.
In majorite tetragonal garnet the following substitutions occur in the crystal-
lographic sites as:
tetrahedral site: Si4þ and Al3þ
octahedral: Si4þ, Al3þ and Fe3þ
dodecahedral (12)-coordination site: Mg2þ, Fe2þ and Ca2þ
IR spectroscopy has confirmed the octahedral coordination of Si in several high
pressure phases (e.g., Weng et al., 1983). Under increasing pressure pyroxene þ garnet
phases transform first to majorite garnet, then ilmenite and eventually to perovskite
structures. When garnet forms a solid solution with orthopyroxene under pressure, 2Al ,
Mg; Si replacement occurs successively in an octahedral site and the total amount of Si
exceed value of 3 (12 oxygen basis).
662 Chapter 8

8.2.10.1. Majorite garnet

Garnets from Coorara meteorite with more than three Si atoms per 12 O atoms were
named as majorite by Smith and Mason (1970). Majorite is a garnet structured pyroxene-
garnet solid solution and is an important component in the region between the lower
portion of the upper mantle and the transition zone. The solid solubility of pyroxene in
garnet increases drastically with pressure (Akaogi and Akimoto, 1977).
Majorite-rich garnets (end-member majorite composition is Mg4Si4O12) are
obtained from three somces: (1) in shocked meteorites, (2) HP-HT experiments on
complex compositions expectedly available in the transition zone and (3) inclusions in
majorite-bearing garnets in diamonds, believed to have come from transition zone
(Haggerty and Sauter, 1990).
Commonly, silica atoms per 12 oxygen (p.f.u) act as the prime parameter indicating
the proportion of majorite end member component. The critical Si content for the cubic to
tetragonal phase transition for majorite garnets investigated by Heinemann et al. (1997) is
observed to be Si ¼ 3.8 p.f.u. The measurements by different workers give a range of
values as shown in the following:

Critical Si content p.f.u Source Workers

3.8 Synthetic Heinemann et al. (1997)
3.55– 3.99 Shocked chondrites Jeanloz, 1981; McMillan
et al., 1989; Chen
et al., 1996
3.91– 4.01 Low Al–majorite Jeanloz, 1981; Chen
et al., 1996
3.24 (13 GPa, 16508C ; 400 Km) to 3.69
(20 GPa, 16008C ; 660 Km)
3.27 (13.5 GPa, 18508C) to 3.64 (17.5 GPa, 17958C) KLB-I anhydrous peridotite Ohtani et al. (1995)
3.58 (23 GPa, 15008) to 3.41 (25 GPa, 15008) pyrolite composition Irifune, 1994
3.55 (16 GPa, 19008C) to 3.63 (20 GPa, 20008C) Chondritic bulk composition Yurimoto and Ohtani,
1992

MgSiO3-rich garnets are the major constituents of the Earth’s transition zone (Ito and
Takahashi, 1987). Garnets in the upper mantle contain little amounts of MgSiO3, while in the
lower mantle MgSiO3 enriched garnets decompose to perovskite under increasing pressure.
Pure MgSiO3 majorite is tetragonally distorted (Angel et al., 1989). In this
structure, one half of the octahedral sites is occupied by Si and the other half by Mg,
according to the formula [8]Mg[6] [6] [4]
3 Mg Si Si3O12. Because of the ordering of Mg and Si in
the octahedral sites, the symmetry of the garnet structure is reduced from cubic to
tetragonal. In addition, each of the dodecahedral (or distorted cubic) and octahedral sites
occur in two different forms. One of the dodecahedral sites in majorite is much more
distorted than in cubic garnets; the crystal field stabilization of Fe2þ in such a site could be
significantly different from that in other forms of garnets.
Synthetic majorite shows polysynthetic twinning, which characteristically arises
due to a cubic to tetragonal phase transition upon quenching from high temperature.
Synthesizing untwinned majorite appears impossible. Typically, two twin laws prevail
with sizes of untwinned domains in the order of 1 mm (or even less).
Al2O3 –SiO2 and (CaO – MgO) – Al2O3 –SiO2 Systems 663

The chemical composition of majorite in the Earth’s mantle can be approximated

by the solid solution Mg3Al2Si3O12-MgSiO3 and the structural formula can be expressed as
Mg3(MgxSixAl222x)Si3O12. Majorite solid solution in this system undergoes the series of
symmetry changes, Ia3d  ! I41 =acd ! I41 =a; with increasing MgSiO3 component
(see also Fig. 8.16). Strong electrostatic interaction between the dodecahedral (Mg2þ)
and tetrahedral (Si4þ) cations is observed from atomic thermal motion and electron density
distribution. The number of Al and Si atoms for 12 oxygens in the formula of garnet
(or majorite) versus pressure show a distinct change around 15 GPa (Fig. 8.12, Bertka and
Fei, 1997; read the caprion). The Brillouin spectrum of tetragonal majorite (En100), show
the frequency shift from the elastically scattered central Rayleigh peak (Fig. 8.13). Peaks at
^ 39 GHz are also due to Rayleigh scattering. Longitudinal and transverse acoustic
modes are indicated by P and S, respectively. No peak splitting or significant broadening
are seen.
For composition 0 # x , 0:8 the system is cubic ðIa3dÞ  and stable. In the range
0:8 # x , 1:0 it is tetragonal ðI41 =aÞ; wherein the cations Mg2þ and Si4þ order on the two
nonequivalent octahedral sites (e.g., Parise et al., 1996; Heinenmann et al., 1997). Phase
 to I41 =a in MgSiO3 majorite is observed (Wang et al., 1993) to occur
transition from Ia3d
around 22 GPa and 2300 –23508C. Tetrahedral site in I41 =acd structure loses the center of
symmetry with the symmetry reduction from Ia3d  to I41 =acd: The symmetry reduction
may be caused by the electronic polarization of the cations due to neighbouring cation –
cation interaction (Nakatsuka et al., 1999). The 57Fe Mössbauer spectrum (80K) of
majorite garnet (Mg/(Mg þ Fe ¼ 0.9)) synthesized at 18 GPa and 18008C in a Re capsule
is shown in Fig. 8.14 (O’Neill et al., 1993).

Figure 8.12. Numbers of Al and Si atoms for 12 oxygens in the formula of garnet (or majorite) versus pressure.
Solid symbols denote data from Akaogi and Akimoto (1979) (circles), Herzberg and Zhang (1993) (squares) and
Canil (1991) (triangles). Open symbols denote the study by Bertka and Fei (1997) and show oxides as starting
material(circles), glass(squares), Oxides with flux (diamond), and mineral mix. The dashed line shows the change
in Al and Si atomic concentrations of majorite with increasing pressure (Bertka and Fei, 1997).
664 Chapter 8

Figure 8.13. Brillouin spectrum of tetragonal majorite (En100), showing the frequency shift from the elastically
scattered central Rayleigh peak. Peaks at ^ 39 GHz are also due to Rayleigh scattering. Longitudinal and
transverse acoustic modes are indicated by P and S, respectively. No peak splitting or significant broadening are
seen (Sinogeikin et al., 1997, q1997 Springer-Verlag).

Enstatite – pyrope (Mg4Si4O12 –Mg3Al2Si3O12) join: elasticity. The elastic properties

of high-pressure phases of the enstatite – pyrope (Mg4Si4O12 –Mg3Al2Si3O12) join
have been investigated. Brillouin scattering technique has been employed at ambient
conditions for determining the elastic properties of pure synthetic pyrope (e.g., O’Neill
et al., 1991). Elastic properties of intermediate-pyrope solid solutions have also been
studied by Brillouin spectroscopy (Bass and Kanzaki, 1990; Yeganeh-Haeri et al.,
1990), static compression experiments (Yagi et al., 1987) and ultrasonic interfero-
metry (Ringden et al., 1994). The elastic properties of end-member Mg4Si4O12
majorite (denoted by En100) were investigated by volume compression (Yagi et al.,
1992) and Brillouin scattering (Pacalo and Weidner, 1997; Sinogeikin et al., 1997;
Fig. 8.13). The bulk modulus versus dK0/dPfor pyrope and En – Py s.s are shown
in Fig. 8.15.
Sinogeikin et al. (1997) measured the longitudinal and elastic wave velocities of
three compositions: En100 majorite, En80Py20 and En50Py50. From the measured velocities
the bulk and shear moduli were calculated using standard formula. The compressibility of
garnet may involve appreciable angle bending strains and relatively compliant polyhedra
may accommodate most of the compression (as opposed to the stiff SiO4 tetrahedra).
A decrease in both K and m are seen with increasing En content and the bulk modulus K of
pure majorite is substantially lower (,6%) than that of pyrope, whereas m for En100 is
slightly lower than, or perhaps equal to, m of pyrope.
Al2O3 –SiO2 and (CaO – MgO) – Al2O3 –SiO2 Systems 665

Figure 8.14. 57Fe Mössbauer spectrum (80K) of majorite garnet synthesized at 18008C, 18 GPa in a Re capsule,
fitted to three symmetrical Lorentzian doublets that corresponds to Fe2þ in the octahedral site, Fe2þ in the
dodecahedral site, and Fe3þ in tetrahedra site (O’Neill et al., 1993, q 1993 American Geophysical Union).

K appears to be constant at the value of < 87 GPa at the majorite-rich end; however,
a sharp step-like change in moduli over a narrow compositional interval in the range En70 –
En80 have been noted (Sinogeikin et al., 1997). These two compositional regimes with
differing elastic properties are related to the ordering of Mg, Si and Al on the octahedral
sites in the garnet structure, which is observed only for majorite-rich samples (Angel et al.,
1989; Hatch and Ghose, 1989).
A linear behaviour of K and XEn relationship can be characterized by a form:
K ¼ KPy þ aXEn
where KPy is the bulk modulus of pure pyrope and XEn is the mole percentage of majorite.
The values of a considered by different workers are:

a value Authors
, 20.05 Yeganeh-Haeri et al., 1990
20.11 Yagi et al., 1992
20.13 Pacalo and Weidner, 1997
20.22 Bass and Kanzaki, 1990
666 Chapter 8

Figure 8.15. Bulk modulus versus dK0/dP for pyrope and En-Py s.s, with comparisons in pyrope and other
measurements. Solid squares represent isothermal compressions on pyrope using the powder diffraction and
single-crystal analysis (LPW) (Yagi et al., 1987, q 1993 American Geophysical Union).

Using the data of Pacalo and Weidner (1997) and Sinogeikin et al. (1997) one can
obtain the relation
K ¼ 173 2 0:1 XEn and m ¼ 92 2 0:03 XEn
The contrast in moduli between Mg – majorite and garnet is less substantial: 6% for
K and 3% for m; whereas the contrast in these values for mantle pyroxene and garnet is
, 45% for K and 27% for m. Thus, the transformation of pyroxene to majorite garnet will
cause a overwhelming change in elasticity; and this change consequently will have a
dominant influence on the seismic velocity gradient in the transition zone. The elastic
property of majorite thus assumes significance in the seismic models of the earth.
The sudden change in elasticity at 70– 80% majorite may be caused by a change in
symmetry of the garnet structure from cubic to tetragonal. Jeanloz (1981) studied the bulk
modulus of majorite obtained from a meteorite by means of high-pressure X-ray diffraction
using a diamond-anvil cell.

Majorite in TZ: tetragonal versus cubic. Garnets rich in majorite component is expected
at the base of transition zone For most part of the transition zone the values between 3.3 and
3.6 p.f.u seem to be reasonable estimates of Si p.f.u.
Al2O3 –SiO2 and (CaO – MgO) – Al2O3 –SiO2 Systems 667

Majorite garnet breaks down to garnet þ clinopyroxene, which may appear as

exsolutions. This assemblage of majorite as inclusions in diamonds may indicate a source
at the transition zone. These garnets show a range of 3.34, i.e, to a maximum of 34 mol%
majorite end member.
Heinemann et al. (1997) concluded that garnets in the transition zone should have
cubic symmetry rather than tetragonal symmetry, because the pyrope (end member
Mg3Al2Si3O12) content of such garnets is greater than the critical value of 20 mol%. The
elastic strain and the intensity of (222) reflection as a function of composition indicate a
second-order phase transition near 20 mol% pyrope. The cubic structure in the transition
zone may further be stabilized by cations like Fe3þ, Na etc. This cubic form was earlier
stipulated by Hatch and Ghose (1989) but this contrasts with Wang et al. (1993) who
suggested that majorites in the transition zone must be tetragonal. Cubic to tetragonal
phase transition in Mg4Si4O12 garnet occurs below 20008C at 19 GPa. A composition
dependent Landau model is consistent with a direct transformation from Ia3d to I41 =a:
Heinemann et al. (1997) have shown that for majorite with 3.8 p.f.u. the transition
temperature is ,1000 –13008C at 19 GPa and decreases with increasing pyrope content
(Fig. 8.16 read the caption).

Elastic softening in Ia3d ! I14/a : seismic velocity in TZ. In ferroelastic transitions

elastic softening is common Ferroelastic phase transitions as well as other phase transitions
commonly result in a softening of elastic properties at temperatures well above and below
the transition temperature (Poirier, 1982; Salje, 1990). The softening effect is generally

Figure 8.16. Derived schematic isobaric T – X phase diagram for Al- poor majorites at 19 GPa. The shaded area
represents the estimated cut-off temperature at which intra-crystalline ordering processes are frozen in for the
given quench rate (Heinemann et al.,1997, q 1997 Springer and Verlag).
668 Chapter 8

greater on the low symmetry side of the transition. Softening of the Youngs modulus is
very strong for the Ia3d ! I41 =a transition. Such a softening effect could also play a role in
majorite garnets within the transition zone. Softening of elastic constants, which
commonly accompanies ferroelastic phase transitions, may affect the seismic velocities
of garnets in the deeper transition zone where majorite contents are highest.
Elastic softening of majorite garnets under transition zone ðP; TÞ conditions may
provide a solution to the observed discrepancy of 2% lowering of seismic velocity within
the transition zone than what is expected for a homogeneous mantle composition.

8.2.10.2. Compressibility
Garnet compressibility is apparently controlled by molar volume and octahedral
cation valence. Hazen et al. (1994) measured by single crystal XRD the compressibilities
of five synthetic majorite [Mg3(MgSi)Si2O12], Ca-majorite [(Ca0.49Mg2.51)(MgSi)Si3O12],
Na – majorite [(Na1.88Mg0.12)(Mg0.06Si1.94)Si3O12], an intermediate composition [(Na0.37
Mg2.48)(Mg0.13Al1.07Si0.80)Si3O12], along with one natural pyrope [(Mg2.84Fe0.10Ca0.06)
Al2Si3O12]. At the same pressure these samples showed small differences in
compressibilities.
High pressure garnets containing octahedrally coordinated silicon may constitute
about half of the Earth’s transition zone (Ita and Stixrude, 1992). Gasparik (1990, 1992)
investigated garnets in the composition range Na2O – MgO – CaO – Al2O3 – SiO2
(NMCAS). High silicon (.3 Si per 12 O) garnets have been reported from such end
member composition as (Na2Mg)Si3O12 [Namaj], intermediate composition in the
Pyrope – Majorite – Namaj ternary and phases from the majorite – CaSiO3 join
(Hazen et al., 1994).
8.2.10.3. Bulk modulus
Majorite and Ca –majorite with divalent Mg cations in half of the octahedral
sites, are more compressible than garnets of similar volume with all trivalent octahedral
cations, even though the average cations valence is þ 3 for these tetragonal garnets.
Hazen et al. (1994) suggested that the Mg2þ octahedra are significantly more
compressible than those with Al3þ or Si4þ and thus contribute to lower garnet bulk
modulus. On the other hand, Na – majorite, with relatively incompressible Si4þ in all
octahedral sites, has the largest bulk modulus reported for any silicate garnet (Fig. 8.8;
Hazen et al., 1994).
The tetragonal majorite –garnet so displays a nearly isotropic compression. The
value of majorite bulk modulus is reported to range beyond 162 GPa. For Na –majorite, the
bulk modulus, determined through Brillouin spectroscopy, is reported to be 173.5 GPa
(Pacalo et al., 1992).
Bulk moduli of the garnets range from 164.8 ^ 2.3 GPa for calcium majorite to
191.5 ^ 2.5 GPa for sodium majorite, assuming K0 ¼ 4.
Based on systematic ultrasonic measurements on various garnets, the adiabatic bulk
modulus of majorite has been estimated by several authors (Wang and Simmons, 1974;
Jeanloz, 1981).
Al2O3 –SiO2 and (CaO – MgO) – Al2O3 –SiO2 Systems 669

8.2.10.4. Vibrational modes: I41/a

Garnets in the tetragonal distortion ðI41 =aÞ have many more active Raman and
infrared modes than garnets in the cubic structure. For tetragonal garnets, the irreducible
representation is:

G ¼ 25Ag ðRÞ þ 27Bg ðRÞ þ 28Eg ðRÞ þ 32Au ðIRÞ þ =31Bu þ 33Eu ðIRÞ ð8:2Þ

Thus, nearly all the vibrational modes of this garnet can be seen by infrared and
Raman spectroscopy.
Chopelas (1999) has obtained Raman spectra of MgSiO3 majorite at 1 atm and
between 11 to 33.6 GPa. The frequency is seen to increase linearly with pressure from
1 atm to , 26.6 GPa. Above 28.3 GPa, two new peaks begin to appear in the spectrum and
several of the minor peaks disappear. Upon decompression the new prominent peaks
disappear below P ¼ 26.6 GPa.

8.2.11. Ca – garnets

In majorite-type garnets the tendency to order octahedral and dodecahedral cations

appears to be a driving force behind the reduction of symmetry from the more typical cubic
garnet symmetry. Similar symmetry-breaking might operate in perovskites of intermediate
composition of CaMgSi2O6.
Calcium ion in the eight-fold coordination has the radius of 1.12 Å, whereas
the radii of Fe and Mg are 0.92 and 0.89 Å, respectively (Shannon, 1976). For the
transition zone condition the system CaO – MgO –Al2O3 –SiO2 is of special interest
(Gasparik, 1990).
Of the seven symmetrically distinct cation sites in tetragonal garnets only three Si
tetrahedra display similar crystal chemical behaviour. Of the remaining four sites two
octahedra and two dodecahedra behave differently and incorporate varieties of elements.
These lead to element partitioning. Furthermore, the presence of extensive garnet solid
solutions may expand its stability into pressure and temperature ranges for the upper
portion of the lower mantle (O’Neill and Jeanloz, 1994). Mantle-derived garnets with more
than three Si atoms per 12 O atoms underscore the chemical complexity of these high
pressure minerals (Haggerty and Sautter, 1990).
Single crystals of Ca-bearing majorite, a garnet with composition (Ca0.49Mg2.51)
(Mg, Si)Si3O12, have been synthesized by Hazen et al. (1994) at 18.2 GPa and 20508C (see
Fig. 8.17). This silicate garnet displays ordering on both octahedral and dodecahedral sites.
This may affect element partitioning at high pressure and stabilize the garnet structure in
the transition zone and upper portion of the lower mantle.
A nearly complete octahedral ordering in Ca-bearing majorite may be coupled with
the unanticipated ordering of Ca and Mg on the two symmetrically distinct octahedral
sites, D1 and D2. Ca is concentrated in the D2 site. One possibly significant difference
between D1 and D2 is the distribution of second-nearest neighbour cations. Ordering of Ca
onto D2 minimizes Ca –Ca interactions, which may be significant at the high pressure and
temperature of Ca-bearing majorite synthesiz (Hazen et al., 1994).
670 Chapter 8

Figure 8.17. The structure of tetragonal Ca-bearing majorite (shown with one a axis vertical and the c axis to the
left and diagonally out from the page) features a corner-linked framework of SiO4 tetrahedra, SiO6 octahedra
(smaller), and MgO6 octahedra (larger), represented here as polygedra. This framework defines two symmetrically
distinct dodecahedral cation sites (spheres), which contain Ca and Mg. Ca concentrates in the D2 site, which is
represented by the larger spheres (from Hazen et al., 1994, q1994 Mineralogical Society of America).

The distribution KD measured as (XCaD2/XMgD2)=(XCaD1/XMgD1) is calculated to be

5.4, which is unusually large for a silicate quenched from high temperature. Most crustal
silicates quenched from high temperature have KD , 2 (completely disordered; ordered
sites have KD ¼ 1), especially for large cation sites that share edges and thus would seem
to disorder easily (Hazen et al., 1993).This phenomenon underscores the point that a
marked tendency for cation ordering may work even at high temperatures (near 20008C).

8.2.12. Andradite –skiagite solid solution

Woodland and O’Neill (1995) investigated the stability of Ca bearing garnets on

the join andradite –“skiagite” as a function of P at 11008C. The experiments were
performed using a series of glasses and slags with fixed fixed Fe3þ/Fetotal in a piston
cylinder apparatus and a multi-anvil press. The maximum solubility of “skiagite” in
andradite increases from 12 mol% at 1.7 GPa to 48 and 64 mol% at 6.0 and 7.0 GPa,
respectively. This is greater than that found for the solubility of “skiagite” in almandine
(Fe2þ
3 Al2Si3O12) at the same P – T conditions. The equilibrium garnet solid solutions at
11008C co-exist with spinel þ clinopyroxene þ quartz or coesite at 0.7 , P , 7.0 GPa.
Above 7.0 GPa garnet may co-exist with either spinel þ clinopyroxene þ coesite,
or, with Fe2SiO4-rich spinel þ spinel þ coesite, depending on the bulk Fe3þ/ Fetotal.
The co-existing clinopyroxene becomes progressively more Fe-rich with increasing P
until a miscibility gap is intersected at 7.0 GPa. Above this P, the pyroxenes are Ca-poor
and monoclinic (space group P21 =c), but probably have the C2=c space group at the
conditions of the experiments. There is a highly asymmetric miscibility gap between this
Ca-poor high-P form and the hedenbergite-clinoferrosilite solid solution, similar to that is
Al2O3 –SiO2 and (CaO – MgO) – Al2O3 –SiO2 Systems 671

found between Ca-poor orthopyroxene and hedenbergite solid solutions at lower P.

Determination of the mixing properties of the garnet solid solutions and the fG81373K:lbar for
“skiagite” can potentially be determined from the Ca –Fe2þ exchange between garnet and
pyroxene. However, this analysis is constrained by problems in modelling the properties of
the co-existing clinopyroxene.

8.2.13. Calderite garnet, Mn3Fe31

2 Si3O12

Calderite garnet (Mn3Fe3þ2 Si3O12), is stable only at pressure above 3.0 GPa
(Schreyer and Baller, 1981), has been described in VHP meta-manganiferous quartzite
from the Western Alps by Reinecke et al. (1992).
This page is intentionally left blank
 673

Chapter 9
AB2X4 Structure
This page is intentionally left blank
 675

(A) Oxide Spinels

This page is intentionally left blank
AB2X4 Structure 677

9.1. Introduction

Spinels have the general formula AB2O4, where A is a divalent cation and B is a
trivalent cation; examples are: MgAl2O4 spinel, Fe3O4 magnetite, FeCr2O4 chromite and
Mg2SiO4 ringwoodite. AB2O4 oxides of spinel structure are associated with many
geological environments — from aluminates, ferrites and chromites in metamorphic,
sedimentary and igneous rocks of the Earth’s crust, to high-pressure forms of ortho-


silicates in the Earth’s mantle. The 2-3 class of oxide spinels (space group Fd3m) is
represented by the general formula unit AB2O4. The oxygen anions form an
approximately cubic close-packed arrangement and the cations distribute themselves
over one tetrahedral site and two octahedral sites per formula unit according to the
general scheme:

A12n Bn ðAn=2 B12n=2 Þ2 O4

where parentheses represent cations on octahedral sites. The variable n is referred to

as the inversion parameter. Two ordered configurations of the spinel structure can be
adopted at low temperatures: the normal configuration with n ¼ 0 and the inverse
configuration with n ¼ 1: At elevated temperatures, the cations became increasingly
randomly distributed over tetrahedral and octahedral site. A value of n ¼ 2=3
corresponded to a completely random distribution of A and B over the three cation
sites per formula unit. Because the order –disorder (OD) process in spinel is of the
non-convergent type (there is no symmetry difference between an ordered and a
totally disordered spinel), such a completely random distribution would only be
anticipated at infinite temperature and is approached asymptotically on increasing
temperature. In cubic spinels, AB2O4, where A and B are both transition-metal ions,
the Jahn – Teller (JT) effect can arise from A or B sites or both. The lowest electronic
state can be a doublet (E spinel) or a triplet (T spinel).
The end-members magnetite (Fe3O4) and spinel (MgAl2O4) adopt the inverse and
normal cation distributions, respectively, at low temperature. Cation ordering in their solid
solutions is expected to be a complex function of composition owing to the conflict of site
preference displayed by Mg and Fe2þ cations in the end-members (O’Neill and Navrotsky,
1984). Experimental determination of the ordering is made especially difficult in spinels
containing Fe2þ and Fe3þ cations because these can exchange with each other relatively
rapidly by the transfer of an electron also, because of the many independent variables
needed to describe the cation distribution, a combination of several independent
experimental observations is required to obtain unique values for all the cation
occupancies.
The difficulty in determining cation distributions experimentally prompted
several theoretical studies of cation ordering in this system (Nell and Wood, 1989; Sack
and Ghiorso, 1991). These models allow calculations of the cation distribution as a
function of temperature and composition and are commonly used to ascertain oxygen
fugacities in the upper mantle (Wood, 1990). To date, only one in situ study exists with
678 Chapter 9

which to test these theoretical models: Nell et al. (1989) measured in situ cation
distributions using the electrical conductivity/Seebeck-effect technique (Mason, 1987).
The cation ordering in the (Fe3O4)12x(MgAl2O4)x solid solution has been
investigated through in situ structure refinements using time-of-flight neutron powder
diffraction (Redfern et al., 1996). Time-of-flight neutron scattering is an ideal probe of
cation ordering in this mineral owing to the good contrast between the neutron scattering
lengths of Fe, Mg and Al (Table 9.1), the ability to perform the experiment in situ under high
vacuum and the ability to record the entire diffraction pattern rapidly, which lowers the
probability of exsolution occurring at temperatures below 1,0008C. The data did not allow
the specific distribution of Mg, Al, Fe2þ and Fe3þ to be calculated uniquely but permitted the
evaluation of various thermodynamic theories of the cation-ordering process.
Samples of the (Fe3O4)12x(MgAl2O4)x solid solution x ¼ 0:4; 0.5 and 0.75 were
synthesized from the oxides using the technique described by Harrison and Putnsi (1995).
The starting materials used by Harrison et al. (1999) for all syntheses were 99.9% pure
Fe2O3, MgO and Al2O3.
The spinels show a wide range of composition and geographical distribution.
Their varied compositions show wide magnetic behaviour which provoked consider-
able interest in geophysical studies. They also show a wide range of physical
properties such as thermal and electrical conductivity. Inter- and intra-crystalline
spinel equilibria are sensitive to T, P, f O2, bulk rock and fluid composition. Such
properties call for their use as petrogenetic indicators, for example as geothermo-
meters, geobarometers and oxygen-fugacity sensors (e.g., Sack and Ghiorso, 1991).
The versatile spinel structure may accommodate cations of þ1, þ2, þ3 or þ4 valence
in either tetrahedral or octahedral coordination and thereby forms stable compounds
over a wide range of temperature, pressure and composition.
AB2O4 spinels are cubic ternary oxides, space group Fd3mðO
7
h Þ and Z ¼ 8,
whose elementary cell contains 32 O atoms in cubic close-packing arrangement.
Sixteen octahedrally (M) and 13 tetrahedrally (T) coordinated sites can host various
cations (e.g., Fe, Al, Mg and Mn). Mg –Al –Fe-rich spinels are considered to be important
constituents of the upper mantle of the Earth. Natural Mg – Al– Fe spinels tend to have
normal structures (e.g., Della Giusta et al., 1996) with Mg and Al cations ordered in T and
M sites, respectively.

TABLE 9.1
Individual scattering lengths and room-temperature tetrahedral-oxygen bond lengths

Cation b (fm)a Rtet (Å)b

Mg 5.375 1.965
Al 3.449 1.77
Fe2þ 9.45 1.995
Fe3þ 9.45 1.865
a
Sears (1992).
b
From O’Neill and Navrotsky (1983).
AB2X4 Structure 679

Studies on the high-pressure crystal chemistry of isomorphous compounds, such as

ABO4 scheelites (Hazen et al., 1985) and A2BO4 silicates, such as olivine and chrysoberyl,
demonstrated the extreme variability of physical properties that may result from the
mixed-valence substitutions of the type Ca2þW6þ , La3þNb5þ or Mg2þSi4þ , 2Al3þ in
a given structure type.

9.1.1. Normal/inverse spinels

Spinels have the general formula A2þB3þ 2 O4, where A and B represent divalent and
trivalent cations, respectively (Fig. 9.1). However, all the variety of spinels can be
represented as (A12nBn) [AnB22n]O4: parentheses () and brackets [] stand for tetrahedral
and octahedral sites, respectively. The values of n ¼ 0; 2/3 and 1 correspond to spinels
which are called normal, random and inverse, respectively (Navrotsky and Kleppa, 1967).
In a unit cell of a spinel, eight of the total of 64 tetrahedral sites and 16 of the 32 octahedral
sites are occupied by cations.
Thus, when all these tetrahedral and octahedral sites are occupied exclusively by
divalent and trivalent cations, respectively, they make a normal (N) spinel, (A)[B2]O4.
From the radius ratio concept, this should be termed “abnormal” since, in contrast, the
smaller tetrahedral site should favour the smaller trivalent cations while the octahedral site
should be the host of larger divalent cations.
However, when the trivalent ions fill the eight tetrahedral sites and the remaining
eight trivalent and eight divalent cations occupy the octahedral sites, the phase is called an
inverse (I) spinel. A random or statistical cation distribution would be formulated as
(A1/3B2/3)[A2/3B4/3]O4. In an intermediate spinel when n ¼ 1=8; it is called 1/8 inverse
(1/8 I). The spinel types corresponding to the divalent and trivalent cations are presented in
Table 9.2 (compiled by Burns, 1993, p. 248).

Figure 9.1. The spinel structure. The characteristic A and B sites containing the tetrahedral and octahedral
site cations in the structure are shown on the left. The fcc lattice of A site ions in a unit cell of spinel is shown on
the right.
680 Chapter 9

TABLE 9.2
Experimental and theoretical cation distributions in spinels with the corresponding type (Burns, 1993)

A B
Al3þ V3þ Cr3þ Mn3þ Fe3þ Co3þ Ga3þ
Exp Th Exp Th Exp Th Exp Th Exp Th Exp Th Exp Th

Mg2þ 7/8 I O N N N N – Np I O N N I O
Zn2þ N O N N N N Np Np N O N N N O
Cd2þ N O N N N N N Np N O N N N O
Mn2þ N O N N N N Np Np I O N N N O
Fe2þ N I N NþI N N – Np I I N N N I
Co2þ N I N NþI N N IþN Np I I N N N I
Ni2þ 3/4 I I – I N N 7/8 I I þ Np I I 7/8 I N I I
Cu2þ I Ip – Ip N N 3/4 I Np 7/8 Ip I N N N I

Exp, experimental; Th, theoretical; N, normal spinel; I, inverse spinel; O, no prediction; –, no data;
p , tetragonally distorted.

9.1.2. CFSE in spinels

The crystal-field stabilization energies (CFSE) of transition-metal ions in

tetrahedral and octahedral co-ordinations are determined from the measurements of the
absorption spectra of oxides, silicates and glasses. The CFSE as measured in oxygen
coordinated structures are summarized in Table 9.3.
Trivalent cations with high OSPE, such as Cr3þ, Mn3þ, V3þ and Co3þ (low spin),
form normal spinels. Ni2þ and Cu2þ have a strong tendency to from inverse spinels.

TABLE 9.3
Octahedral site preference energies of transition-metal ions in structure

Cation Electrons in Electronic Octahedral Tetrahedral Octahedral site no =nt

d orbital configuration CFSE (Eo) CFSE (Et) preference energy
(kJ/mol) (kJ/mol) (OSPE)

Ca2þ, Sc3þ, Ti4þ d0 0 0 0 0 0

Ti3þ d1 (t2g)1 287.4 258.6 228.8 15
V3þ d2 (t2g)2 2160.2 2160.7 253.5 158
Cr3þ d3 (t2g)3 2224.7 266.9 2157.8 2.9 £ 106
Cr2þ d4 (t2g)2(eg)1 2100.4 229.3 271.1 829
Mn3þ d4 (t2g)3(eg)1 2135.6 240.2 295.4 8.208
Mn2þ, Fe3þ d5 (t2g)3(eg)2 0 0 0 0
Fe2þ d6 (t2g)4(eg)2 249.8 233.1 216.7 5
Co3þ d6 (t2g)4(eg)2 2188.3 2108.8 279.5 1.827
Co2þ d7 (t2g)5(eg)2 292.9 261.9 231.0 19
Ni2þ d8 (t2g)5(eg)3 2122.2 236.0 286.2 3.440
Cu2þ d9 (t2g)6(eg)3 290.4 226.8 263.7 407
Zn2þ, Ga3þ, Ge4þ d10 (t2g)6(eg)4 0 0 0 0

no =nt ¼ exp 2 ½ðEo 2 Et Þ=RT at T ¼ 1; 0008C.

AB2X4 Structure 681

Against this observed role of CFSE, however, it must be noted that CFSE
contributes only less than 10% to the lattice energy. CFSE values are small in comparison
with the absolute heat of hydration or lattice energies. Thus, they make up only a small
fraction of the total energy of transition-metal compounds. For example: CFSE of Cr3þ
contributes less than 10% to its heat of hydration.

9.1.3. JT effect

In cubic spinels, AB2O4, where A and B are both transition-metal ions, the JT effect
can arise from A or B sites or both. The lowest electronic state can be a doublet (E spinel)
or a triplet (T spinel). Since the electrons responsible for the JT effect are on the outside
of the ions, they interact strongly with the lattice, giving rise to structural-phase transitions
(at high temperature).

9.1.4. Crystal structure

The spinel-type structure consists of a cubic close-packed arrangement of oxygen in

which 1/2 of the octahedral sites and 1/8th of the tetrahedral 10 sites are occupied by
cations. In spinel, the tetrahedral and octahedral cations are Mg2þ and Al3þ, respectively.
In magnetite, the tetrahedral position is fully occupied by Fe3þ, whereas the disordered
octahedral site contains 1/2 Fe2þ and 1/2 Fe3þ. In silicate spinels, on the other hand, the
tetrahedral and octahedral cations are Si4þ and M2þ, respectively.
The oxygen anions are in a cubic closest-packed arrangement with the B cation in
octahedral sites and the A cation in tetrahedral sites. The octahedra form edge-sharing
chains that are linked together by the isolated tetrahedra. The changes in the x position of
the oxygen atom cause changes in the cubic cell edge, which is determined from the
octahedral, do, and tetrahedral, dT, bond lengths as:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
40dT þ 8 33do2 2 8dT2
a¼ pffiffiffi ð9-1Þ
11 3
8
a¼ pffiffiffi ð5dT þ AÞ ð9-2Þ
11 3
where
p
A ¼ ð33do2 2 8dT2 Þ ð9-3Þ

The ideal spinel structure is cubic with space group Fd3m and Z ¼ 8 (e.g., Hill et al.,
1979). In the commonly used setting with an inversion centre at the origin,
the configuration of the tetrahedral cation at (1/8, 1/8, 1/8) and the octahedral cation at
(1/2, 1/2, 1/2) are employed. The oxygen parameter in this setting is designated as u, u, u
(u , 0.25). The octahedral and tetrahedral cation positions may be defined as (3/8, 3/8,
3/8) and (0, 0, 0), respectively, in which case the oxygen positional coordinate is
u0 ¼ 1/2 2 u. Furthermore, the origin may be shifted by (1/4, 1/4, 1/4) to a point with 43m


682 Chapter 9

symmetry. It is necessary to identify the choice of both origin and fixed cation coordination
prior to any comparison of spinel structures.
Hill et al. (1979), in their analysis of 149 AB2O4 oxide spinels, noted a close
relationship between the oxygen u parameter (the only variable atomic position coordinate
in the spinel structure) and the relative sizes of A and B cations. In particular, the ratio R of
the octahedral to tetrahedral cation – oxygen bond lengths is related to u by the exact
expression:
 0:5
R2 2 11R2 1
2 þ 2
4 3 48 18
u¼ 2
ð9-4Þ
2R 2 2

They concluded that the standard ionic radii account for most of the observed range
of the oxygen u parameter of spinels and other factors such as electronic structure are not
significant in cubic spinels.

9.1.4.1. Compressibility
Hazen and Finger (1979) demonstrated that the compressibilities of cation
coordination polyhedra (both octahedra and tetrahedra) are proportional to
polyhedral volume and are inversely proportional to formal valance. A difference in
octahedral and tetrahedral compressions may lead to systematic variation of u with
pressure.
Most spinel-type compounds show bulk moduli of about 2 Mbar (see Table 9.5)
although individual cation polyhedral moduli may differ greatly. Divalent Mg2þ
tetrahedra in spinel are twice as compressible as tetravalent Si4þ in SiO4 tetrahedra
(measured in g-Ni2SiO4), whereas trivalent Fe3þ tetrahedra in magnetite have
intermediate compressibility. Divalent Ni2þ octahedra in the silicate spinel are 50%
more compressible than Al3þ octahedra in Mg-aluminate and the mixed Fe2þ – Fe3þ
octahedra of magnetite (Fe valence ø 2.5) are of intermediate compressibility.
The corner-sharing octahedra (the B sites) provide the framework with three
oxygens of each triangular face of the octahedra forming a distorted cuboctahedron
surrounding the A site; cations in the A site thus have 12 near-neighbour oxygen atoms but
with M –O distances exhibiting a wide range of values. The A- and B site cations are only
279 nm apart and rotation and tilting of the B site octahedra produce distortion of the
cuboctahedron forming the A site. This is the cause of the wide range of M– O distances
and the ill-defined coordination number for the A site.
Recent simulation (molecular dynamics) shows that high densities noted in silicate
spinels can also be reached without changes in Si coordination but by repacking of SiO4
tetrahedra (Kubicki and Lasaga, 1991).
If spinel octahedra and tetrahedra contribute equally to crystal compression then it is
probable that all oxides spinels have a similar bulk modulus of about 2 Mbar. Any increase in
octahedral compressibility that results from cations of lower volume will be offset by the
decrease in tetrahedral compressibility of higher valence cations. This behaviour would,
AB2X4 Structure 683

however, contrast sharply with that of scheelite-type ABO4 oxides, which vary in com-
pressibility by more than a factor of four as a result of mixed valence cation substitutions.
Silicate spinels in the Earth’s interior may be subjected to considerable cation
disorder, which may alter the stability and transport properties of these important mantle
minerals. However, the 2-Mbar bulk moduli of these phases should be little affected by the
degree of ordering.
Below 670 km, a disproportionation reaction occurs and the spinel structure breaks
down to a mixture of silicate perovskite [(Mg,Fe)SiO3] plus [(Mg,Fe)O].

9.1.5. High-pressure studies

High-T and high-P annealing of A2þB3þ 2 O4-type spinels at 12 GPa and 1,273 K
resulted in four types of behaviour (Ringwood and Reid, 1969):
(a) MnAl2O4, FeAl2O4 and NiAl2O4: These decompose to AO (rocksalt) þ B2O3
(corundum) mixtures.
(b) ZnAl2O4, MgAl2O4, MgCr2O4, CoGa2O4, NiGa2O4 and ZnGa2O4: These did not
decompose at this P, T.
(c) CdCr2O4 and CdFe2O4: These transform to denser AB2O4 structures.
(d) Fe3O4 and ZnFe2O4: Their behaviour is inconclusive because of the reduction of
Fe3þ to Fe2þ at this P, T.
The compressibility of the tetrahedral (dT) and octahedral (do) bonds determines the
compressibility of the spinel structure. The compressibilities of these bonds may be of
three types (Finger et al., 1986): (i) the compressibilities of do and dT are equal, (ii) that of
dT is softer than do and (iii) do is softer than dT.
Under pressure, there is a blue shift of Fe2þ and Cr3þ CF bands in oxide spinel
(MgAl2O4) and magnesiochromites (Mao and Bell, 1975). In silicate spinels such as
g-phase or ringwoodite (g-Fe2SiO4), the Fe2þ CF band at 11,430 cm blue shifts to the
visible region (Mao and Bell, 1972). Similar blue shifts of CF bands in octahedral Ni2þ are
noted in Ni2SiO4 (Yagi and Mao, 1977).
The Fe2þ CF band in g-Fe2SiO4 shows a strong pressure-induced red-shift of the
oxygen ! Fe absorption edge and vanishes beyond 20-GPa pressure. A similar effect
operating on mantle minerals, like olivine and magnesiowüstite, influences the radiative
heat transfer and electrical conductivity in the Earth’s interior.

9.1.5.1. Compressibility
Differentiating equation (9-1) with respect to pressure (P) (or temperature), an
expression for the dimensional change in cell edge (a) can be obtained as (Hazen and
Yang,1999):
 
da da 0 33do d0o 2 8dT d0T
¼ pffiffiffi 5dT þ ð9-5Þ
dP 11 3 A
where d0 ¼ dd=dP or dd=dT as appropriate.
684 Chapter 9

Dividing equation (9-5) by equation (9-2) yields an exact expression for the linear
compressibility (or thermal expansion) of a in terms of bond distances dT and do and their P
(or T) derivatives:

da 5dT 33do d 0o 2 8dT d 0T da

2b ¼ ¼ þ a¼ ð9-6Þ
adP 5dT þ A ð5dT þ AÞA adT

Variations of cation –oxygen bond distances (d) with P and T are available from
Hazen and Finger (1982, Tables 6-3 and 7-2). The bond compressibilities, b, and bond
thermal expansivities, a, are determined as follows:
   
1 dd 1 dd
2b ¼ ; a¼
d dP d dT

The values of these parameters for metal –oxygen distances of six common cations
in natural spinels, such as Mg2þ, Fe2þ, Al3þ, Fe3þ, Si4þ and Ti4þ, are given in Table 9.4
below.
The thermal expansivities of tetrahedral R2þ – O bonds are much greater than
that of tetrahedral R3þ –O bonds (whereas octahedral bond expansivities are similar for
R2þ – O and R3þ –O bonds) but the tetrahedral and octahedral compressibilities of divalent
and trivalent cations are similar.
The effects of OD reactions on EOS differ systematically for A2þB3þ 2 O4 (2-3)
spinels compared with A4þB2þ 2 O4 (4-2) spinels. Normal 2-3 spinels are less compressible
than inverse variants. The 4-2 titanate and silicate spinels show significantly greater
compressibilities and expansivities than their normal spinel forms.
Compressibilities for normal vs. inverse variants of A2þB3þ 4þ 2þ
2 O4 and A B2 O4
spinels are predicted to differ by as much as 17% — and thermal expansivities by as much
as 15% — as a result of the differential compressibilities or thermal expansivities of
divalent, trivalent and tetravalent cations in tetravalent vs. octahedral coordination (Hazen
and Yang, 1999).

TABLE 9.4
Tetrahedral (dT) and octahedral (do) bond distances and their pressure and temperature derivatives, for cubic
spinels

Bond dT do ›dT =›dP ›do =›dP ›dT =›T ›do =›T

(Å) (Å) (104 Å/GPa) (104 Å/GPa) (106 Å/8C) (106 Å/8C)

Mg2þ –O 1.96 2.08 46 47 19 29

Fe2þ –O 2.01 2.13 49 47 19 28
Al3þ –O 1.77 1.91 27 25 2 17
Fe3þ –O 1.89 2.00 33 29 3 18
Si4þ –O 1.65 1.79 13 19 0 10
Ti4þ – O 1.80 1.98 14 25 0 14

Source: Hazen and Yang (1999).

AB2X4 Structure 685

9.1.5.2. Polyhedral bulk moduli, K

The isothermal bulk modulus at zero pressure may be evaluated by means of a least
squares fit of the P –V data to the Birch –Murnaghan equation:

P ¼ ð3=2ÞK0 ½ðV=V0 Þ27=3 2 ðV=V0 Þ25=3 {1 2 ð3=4Þð4 2 K 00 ÞðV=V0 Þ22=3 2 1Þ}

where V, Kand K 0 are volume, bulk modulus and ðdK=dPÞ; respectively, and the subscript
zero indicates the quantities evaluated at zero pressure. The available experimental values
for K 0 0 for the spinel group of crystals are listed in Table 9.5.
Anderson and Anderson (1970) calculated K 0 0 values using power or
exponential functions for the repulsive potential in the Born –Mayer-type central
force approximation of the inter-atomic potential. By means of the power function
repulsive potential, the K 0 0 value of MgO was determined as 3.85 (Anderson and
Anderson, 1970; Spetzler, 1970). For NiFeO4 spinel, the K 0 0 value was determined as
4.41 (presuming K 00 ¼ 4 ^ 0:4). Waldron (1955), however, obtained a value for K0 of
1.944 Mbar, which compares well with Lieberman’s (1969) value of 1.823 Mbar
determined by the ultrasonic pulse superposition method. For magnetite, the K0 values
have been obtained as 1.70 ^ 0.05 Mbar by Bridgman (1949) and as
1.83 ^ 0.10 Mbar by Mao et al. (1974) (see Table 9.6). The measured bulk modulus
of spinel (assuming K 0 ¼ 4) is 1.94 ^ 0.06 Mbar, determined ultrasonically (Finger
et al., 1985) (Table 9.5).

9.2. MgAl2O4 spinel

Mg –Al –spinel (MgAl2O4) is a common constituent of a low-pressure zone of

shallow upper mantle and is seen in peridotite xenoliths. MgAl2O4 ideally has a normal
spinel structure. The A site (Mg2þ) has tetrahedral coordination with full Td symmetry,
while the B site (Al3þ) has 6-fold distorted octahedral coordination belonging to the D3d
point group. The B site, where Cr3þ substitutes in the lattice, has a centre of inversion.
In mineral spinel, the Mg –O tetrahedral bonds are soft while Al– O octahedral
bonds are stronger. Its structure has recently been determined by Pavese et al. (1998) to
3.8 GPa in their study of cations partitioning as a function of pressure. Because of this, the
softer-bonded tetrahedra compress more rapidly than the octahedra. The latter undergo
distortion, increasing the length of the shared octahedral edge and bringing about the

TABLE 9.5
Values of K0 and K 00 of spinels

K0 (GPa) K 00 Authors

190 ^ 2 4 Pavese et al. (1999b)

195 ^ 7 2.2 ^ 2 Pavese et al. (1999b)
195 Wang and Simmons (1972)
194 ^ 6 Finger et al. (1986)
198 ^ 8 Askarpour et al. (1993)
686 Chapter 9

TABLE 9.6
Bulk modulus of magnetite

Reference KTO K 0TO Pmax Sample Method

Mao et al. (1974) 183(10) 4(0.4) 23 Powder XRD

Wilburn and Bassett (1977) 155(12) 4 6.5 Powder XRD
Hazen et al. (1981) 189(14) 4 4.5 Single crystal XRD
Finger et al. (1986) 186(5) 4(0.4) 4.5 Single crystal XRD
Nakagiri et al. (1986) 181(2) 5.5(15) 4.5 Single crystal XRD
Staun-Olsen et al. (1994) 200(20) – 5.5 Powder XRD
Gerward and Staun Olsen (1995) 215(25) 7.5(40) 25 Powder XRD
Haavik et al. (2000) 217(2) 4 27 Selected datasets

Note: K 0 TO values of “4” are assumed.

consequent destabilization of the structure. Above 15 GPa, MgAl2O4 decomposes to a

mixture of MgO (periclase) and Al2O3 (corundum) (e.g., Liu, 1980).
However, with T . 1,0008C and P . 25 GPa, MgAl2O4 is seen to adopt the
CaFe2O4 (calcium ferrite) structure (Irifune et al., 1991). This CaFe2O4 structure is akin to
the high-pressure structure of magnetite (FeFe2O4), wherein Fe3þ is in octahedral
coordination and Fe2þ is in 8-fold coordination. This structure has a zero-pressure density
of 3.937(3) g cm23, which is , 2% denser than the lower pressure assemblage of
periclase þ corundum.
The bulk modulus of spinel was determined as 190– 194 GPa with K 00 ¼ 4 by the
diffraction technique and as198 GPa by Brillouin spectroscopy (Askarpour et al., 1993). A
recent study by Funamori et al. (1998), employing synchrotron XRD with a DAC up to
70 GPa, showed that MgAl2O4 spinel transform as:

MgAl2 O4 ðspinel structureÞ ! Al2 O3 þ MgO ! ðCaFe2 O4 structureÞ

,40 GPa
! ðCaTi2 O4 structureÞ

High-pressure studies on this were carried out by Liu (1978, 1980) and Finger
et al. (1986). Under pressure, the oxygen u parameter (the only variable
atomic position coordinate in spinel structure) of magnesium aluminate spinel
decreases, thus indicating that the magnesium tetrahedron compresses more than the
aluminium octahedron.
It is seen that the crystal bulk modulus is often simply the average of the
tetrahedral and octahedral bulk moduli; in MgAl2 O4 spinel, it is
(2.6 þ 1.2)/2 ¼ 1.9 Mbar. This is in sharp contrast to other binary oxide compounds,
including olivine-type and scheelite-type minerals in which tetrahedral bulk moduli
have little effect on the macroscopic crystal compression whereas the crystal bulk
modulus is controlled by the larger cation polyhedra. All oxide spinels, however, seem
to have a similar bulk modulus (of about 2 Mbar).
AB2X4 Structure 687

Yamanaka and Takeuchi (1983) described the high-temperature structure of

MgAl2O4 to 1,7008C. At temperature up to , 7008C, the structural changes are opposite to
those observed with increasing pressure. In this spinel, a 1% volume decrease with either
compression or cooling is accompanied by a 0.0003 unit increase in u. This spinel, like
Ni2SiO4 spinel, conforms to the “inverse relationship” structural response to temperature
and pressure. In Mg-aluminate spinel, however, the inverse relationship fails dramatically
above , 7008C owing to the disordering of Mg and Al. As Al enters the tetrahedral
positions and Mg the octahedral positions, the u parameter decreases sharply. This
decrease corresponds to the increase in octahedral M– O distance and the decrease in
tetrahedral M –O distance.
The contrast between Al and Mg neutron-scattering length makes an unambiguous
separation of the contribution of diffraction signals from these two chemical species.
Pavese et al. (1999a) employed a minimization technique to extract the partitioning of
Mg, Al and vacancies in a synthetic spinel. They used a combination of the simplex
and Newton –Raphson methods and implemented an extension of the MINUIT code
(James and Roos, 1975).
T-cation sites are often affected by local clustering effects (as a consequence of
different second-neighbour shells surrounding the tetrahedrally and octahedrally
coordinated cations) that make M and/or T sites energetically non-equivalent,
as demonstrated by Mössbauer spectroscopy (Carbonim et al., 1996). These effects may
cause the anomalous behaviour of the T- and M site atomic displacement parameters
(Pavese et al., 1999a,b).
Using the Birch –Murnaghan EOS, the values for K0 and K 0 0 were obtained by
different authors as given in Table 9.5.

9.2.1. Spectral models

For MgAl2O4, spectral measurements were made to 21 GPa by Chopelas and

Hofmeister (1991). A change in compressional mechanism was observed as a break in the
slope of some of the optical modes (in the Raman, infrared and side-band spectra) and as a
decrease in frequencies of the acoustic modes. However, the shear velocity was not
observed to change with pressure (Chopelas, 1996).
In MgAl2O4, the 42 normal modes at the Brillouin zone centre may be divided into
three acoustic modes of T1u symmetry and 39 optic modes distributed among the following
symmetry species:

G ¼ A1g ðRÞ þ Eg ðRÞ þ T1g þ 3T2g ðRÞ þ 2A2u þ 2Eu þ 4T1u ðIRÞ þ 2T2u

where R (Raman) and IR (infrared) denote spectral activity (Chopelas and Hofmeister,
1991; Chopelas, 1996). The following modes are observed:
Raman fundamental modes: 312 cm21 ðT2g Þ; 407 cm21 ðEg Þ; 490 cm21 ðT2g Þ;
666 cm21 ðT2g Þ; 767 cm21 ðA1g Þ
Infrared modes (all T2u ; transverse optic components): 304, 476, 578, 676 cm21.
688 Chapter 9

9.2.1.1. Cr3þ in MgAl2O4

Cr3þ fluorescence in MgAl2O4 (as in MgO) shows vibrational modes at the same
energy as seen for Raman and IR modes. The side-band modes represent vibrational modes
of the unperturbed lattice (Chopelas and Hofmeister, 1991). However, in the side bands,
the six modes at 150, 184, 257, 326 and 444 cm21 do not correspond to Raman or IR
vibrational energies. The assignment of these modes are as follows:

Modes Assignments
21
218, 257 cm Transverse acoustic modes
326 cm21 Longitudinal acoustic modes from inelastic neutron scattering (Thompson and Grimes, 1978)
444 cm21 Optic mode, its IR/Raman inactive (Chopelas and Hofmeister, 1991)

9.2.2. (MgAl2O4)x (Fe3O4)12x solid solution

In the cation distribution in (Fe3O4)12x (MgAl2O4)x, three parameters, p, q and r,

describe the solid solution:

Alp Fe3þ 2þ 3þ 2þ
q Mgr Fe12p2q2r ½Al2x2p Fe222x2q Mgx2r Fepþqþr2x O4

The solid solution can be reduced to a binary on the grounds that there is no partitioning of
Mg relative to Fe2þ between tetrahedral and octahedral sites, as has been observed in the
cation ordering in the end-members MgAl2O4, FeAl2O4, MgFe2O4 and Fe3O4 (e.g.,
O’Neill et al., 1992).
The Mg/Fe2þ distributions are determined by the values of x and ðp þ qÞ:

Alp Fe3þ 2þ 3þ 2þ
q Mgxð12p2qÞ Feð12xÞð12p2qÞ ½Al2x2p Fe222x2q MgxðpþqÞ Feð12xÞðpþqÞ O4

For x ¼ 0:75; Nell et al. (1989) predict that Fe3þ orders into octahedral sites. However, in
their experiment, Harrison et al. (1999) observed that Fe3þ ordered into the tetrahedral
site during isothermal annealing of their sample below Tr (the relaxation temperature at
which the cation distribution rapidly approaches equilibrium). Equilibrium is defined by
the global minimum of the free energy with respect to the cation order parameters. This
yields three equations that are solved simultaneously to give p, q and r. Al occurs
predominantly at the octahedral sites with a slowly increasing quantity on tetrahedral sites
at high temperatures.
MgAl2O4 shows an ordering of Mg (and Fe2þ) with a strong tetrahedral site
preference (relative to Al). In inversely ordered Fe3O4, an octahedral preference is shown
by Fe2þ (and Mg) relative to Fe3þ. In solid solution of the two end-members, Fe2þ and Mg
are randomly distributed between sites that are not already occupied by Al.

9.2.3. Order –disorder (OD): cation partitioning

OD reactions involving tetrahedral and octahedral cation diffusion induced

by temperature have been widely studied by spectroscopic and diffraction techniques.
AB2X4 Structure 689

Hazen and Navrotsky (1996) speculated on how the pressure affects OD reactions in
condensed matter and pointed out that the basic component provides an important
contribution in driving such phenomena.
Employing the powder neutron diffraction (POLARIS) diffractometer, Pavese
et al. (1999b) observed that the octahedral and tetrahedral site-scattering lengths (bM
and bT) show abrupt changes with increasing pressure from 0.68 to 1.82 GPa,
suggesting an OD reaction takes place. They also noted a partially disordered Al– Mg
distribution, whereas the vacancies are fully segregated into the T sites. Increasing
pressure causes Mg atoms to order into the T site, whereas Al atoms exhibit preference
for the 6-fold coordination. Vacancies are limited to the tetrahedral site at low pressure
and, at high pressure, they equipartition over both M and T sites. Since vacancies
occupy the T sites and soften them, under pressure kT –Ol undergoes compression
whereas Mg and Al cations fully occupy the M sites and make the kM –Ol bond length
more rigid.
Above the OD transition pressure, the T sites become more Mg rich, at the
expense of vacancies and Al, and become less compressible. The kM – Ol bond is stiff
because of the site being occupied by Mg and Al but it collapses between 0.6 and
1.8 GPa. At T sites, the vacancies are twice as likely as at M sites owing to the site
multiplicity of the former.
Results of all studies indicate that the tetrahedron is softer than the octahedron. The
significant increase of the stiffness of the octahedron with respect to the tetrahedron is
effected by the migration of tetrahedrally coordinated Al to the M sites.
The octahedral volume VM vs. P shows distinct changes at the OD reaction
pressure, with a shrinkage taking place between 3 and 4 GPa (Finger et al., 1986). The rate
of octahedral volume variation (VM) with pressure is noted as:

Authors Sample VM/kbar VM/VT (low P to high P)

3
Finger et al. (1986) Natural 20.0030 Å /kbar 2.6 –2.71
Pavese et al. (1999b) Synthetic 20.0015 Å3/kbar (MRV ¼ 160) 2.49–2.54

Cation partitioning in spinels is closely related to pressure and temperature as

both the energetics and the configurational entropy are dependent on the atomic
distributions at T and M sites. Pressure is generally regarded as an obstacle to
processes based on thermally activated intra-crystalline diffusion because the
compressional forces reduce the inter-atomic voids. The lattice parameter of spinel
can be expressed in terms of T – O and M –O bond lengths. Based on ionic radii, the
calculation of Menegazzo et al. (1997) shows that the inverse spinel volume is 1.3%
smaller than its normal spinel form.

9.2.4. Magnetic behaviour: MS

In the majority of spinels, the magnetic moments associated with the Fe2þ and
3þ
Fe cations adopt a co-linear arrangement below the Curie temperature, with
690 Chapter 9

moments on tetrahedral sites aligned anti-parallel to those on octahedral sites. These

cations in spinels exchange rapidly with each other simply by the transfer of an
electron, and also a collinear spin structure is adopted in all compositions in the solid
solution.
The net saturation magnetization is given simply by the difference between the
octahedral and tetrahedral sublattice magnetizations:

oct oct tet tet

MS ¼ 2ðmFe3þ XFe 3þ þ mFe2þ XFe2þ Þ 2 mFe3þ XFe3þ 2 mFe2þ XFe2þ ð9-7Þ

where mFe3þ and mFe2þ are the 0 K magnetic moments of the Fe3þ and Fe2þ cations (5 and
4mB, respectively; 1mB ¼ 9.27 £ 10224 Am2).
For the solid-solution (MgAl2O4)x(Fe3O4)12x, measurements of MS provide a
precise and very sensitive constraint on the cation distribution. The values of MS ¼ 1:32;
0.88 and 0.025mB were determined by Harrison et al. (1977) for samples with x ¼ 0:4; 0.5
and 0.75, respectively.
From the data of Nell et al. (1989), Harrison et al. (1999) calculated average
saturation magnetization for x ¼ 0:5 and 0.75 as MS ¼ 1:38 and 2 2.65mB.
This discrepancy from the earlier observed values of MS ¼ 0:88 and 0.025mB
(Harrison, 1977) may indicate that the method of Nell et al. (1989) overestimates the
amount of Fe3þtet .

9.3. Magnetite, Fe3O4

In magnetite structure, the bond lengths (do and dT) are equal in strength. The
crystal structure of magnetite has been determined to be 4.5 GPa (Finger et al., 1986) and
30 GPa (Haavik et al., 1998). The bulk modulus has been determined as 220 GPa. The
phase boundary shows a slope of 2 688C/GPa (Huang and Bassett, 1986).
Magnetite (Fe3O4) is a mixed valence iron oxide with some Fe2þ ions occupying
the tetrahedral sites and the Fe2þ and the remaining Fe3þ ions occupying the octahedral

(e.g., Fleet, 1981). It is the best known
sites in the spinel structure (space group, Fd 3m)
ferrimagnetic inverse spinel.
Magnetite (Fe3O4) at room temperature and pressure has the inverse spinel
structure. Using the centro-symmetric description of the space group, the tetrahedral
cations are located at (1/8, 1/8, 1/8), the octahedral cations at (1/2, 1/2, 1/2) and the
oxygen atoms at (u, u, u), where u ¼ 0.25. According to Fleet (1981), u ¼ 0.2549(1) at
RT. One Fe3þ per formula unit is at the octahedral site (1/8, 1/8, 1/8) with equipoint
16(d). The oxygen u coordinate, the only variable atomic position coordinate for the
spinel structure, does not vary with pressure for P , 4:5 GPa (e.g., Nakagiri et al.,
1986). With pressure, the oxygen position parameter (u) does not change (within 2s). No
cation-distribution change is indicated under pressure up to 43 GPa (Haavik et al., 2000)
and, hence, inverse spinel modification seems stable in this pressure range.
AB2X4 Structure 691

When it is presumed that the unit cell contains two molecules, the density
(6.24 g cm23) of the high-pressure phase becomes consistent with the density predicted for a
phase of the magnetite composition, having all its cations in 6-fold coordination.
Magnetite shows two structural modifications: (a) low temperature and (b) high
pressure. The former has the space group Imma and has the transition temperature at
,120 K. The high-pressure phase is monoclinic and the transition occurs at 25.0(1) GPa
(Mao et al., 1974).
The high-pressure behaviour of magnetite has been studied by several authors
employing the methods as tabled below:
Methods Authors

X-ray diffraction Mao et al. (1974)

Mössbauer spectroscopy
Electrical resistivity measurements
Huang and Bassett (1986)
Nakagiri et al. (1986)
Finger et al. (1986)
Pasternak et al. (1994)
Haavik et al. (2000)
Mao et al. (1977)
Pasternak et al. (1994)
Ramasesha et al. (1994)
Rozenberg et al. (1996)
Morris and Willams et al. (1997)

As stated, magnetite transforms sluggishly to the high-pressure monoclinic phase

at , 25 GPa (Mao et al., 1974). On releasing the pressure, it reverts back to the original
phase at 5 GPa. The phase transition, confirmed by Huang and Bassett (1986) (Fig.
9.2a), is seen to be accompanied by changes in magnetic properties (e.g., Pasternak et al.,
1994) and electrical resistivity (Morris and Williams, 1997). Mössbauer
measurements by some workers (Mao et al., 1977; Pasternak et al., 1994) indicated
that the orthorhombic high-pressure phase is not magnetically ordered. High-
pressure electrical measurements (Morris and Williams, 1997) showed that the
high-P phase is not metallic. Hence, the structural transformation is accompanied
by a change from the ambient ferrimagnetic state to the paramagnetic state
(Mao et al., 1977).
Three transformations are observed with temperature. The Verwey transition is
related to a change in the degree of electron localization of the iron atoms (Verwey, 1939)
and, above the transition temperature, fast electron exchange is observed between the
octahedral Fe2þ and Fe3þ ions. With increasing temperature, a gradual disordering is
observed of the inverse spinel to transform to a random one (Wu and Mason, 1981). At
Tc ¼ 848.5 K, magnetite becomes paramagnetic (Grønvold and Sveen, 1974).
Nakagiri et al. (1986) reported that tetrahedra and octahedra have different
polyhedral bulk moduli values (for P , 4.5 GPa) of 170(10) and 200(8) GPa, respectively.
The values of the bulk modulus of magnetite determined by various workers are
presented in Table 9.6.
692 Chapter 9
AB2X4 Structure 693

9.3.1. h-Fe3O4

The high-P modification of Fe3O4, designated as h-Fe3O4, appears above ,25 GPa
and is paramagnetic (Pasternak et al., 1994). The calculated density of h-Fe3O4 indicates
that all iron atoms in it are placed in 6-fold coordination (Mao et al., 1974) through a
massive reconstructive transition from the low-P spinel type, wherein iron occurs in both
4- and 6-fold coordination.
Magnetite and h-Fe3O4 appear to co-exist as a two-phase mixture over several GPa
(e.g., Pasternak et al., 1994) and hence this transition appears to be of first order. For
pressures up to 66 GPa, the fast electron exchange between Fe2þ and Fe3þ, characteristic
of Verwey transition, is still present. This is also reflected by the change in resistivity value
(Morris and Williams, 1997). Under pressure, the XRD reflections for magnetite are seen
to persist till 30.3 GPa, manifesting the slow kinetics of the transformation. On
decompression, h-Fe3O4 does not completely revert to magnetite even down to 6 GPa.
The transformation involves atomic motion and/or large strain, which reduces the
long-range order.
Fei et al. (1999) determined the crystal structure of the h-Fe3O4 phase using an
imaging plate detector and monochromatic synchrotron X-ray radiation. The high-P phase
shows the Pbcm space group (CaMn2O4-type structure) with the following cell parameters
(at 23.96 GPa, 823 K):
a ¼ 2:7992ð3Þ Å
b ¼ 9:4097ð15Þ Å
c ¼ 9:4832ð9Þ Å
Fe3þ occupies an octahedral site and Fe2þ is in an 8-fold coordinated site described
as a bicapped trigonal prism. However, it is also consistent with a more symmetric
CaTi2O4-type structure.
The high-P CaMn2O4-type Fe3O4 phase is ,6.5% more dense than the spinel form
at 24 GPa (Fei et al., 1999). The CaMn2O4-type structure, closely related to the CaFe2O4-
type structure, is one of the densest AB2O4 structures (see Section 9.3.4). The trivalent
ions occupy the octahedral sites while the divalent ions occupy the 8-fold coordination site.
MgAl 2O 4 spinel also transforms to the CaFe 2O 4-type structure at ,25 GPa
(Irifune et al., 1991).
The Fe3þ – O22 bond lengths for the octahedral sites and the Fe2þ – O22 bond
lengths for the 8-fold coordinated sites at 24 GPa range from 1.715 to 2.589 Å and from
1.775 to 2.719 Å, respectively (Table 9.7) (Fei et al., 1999).

Figure 9.2. Phase transformation in Fe3O4 at high pressures and temperatures. The solid and open circles
indicate P– T conditions under which the high-pressure Fe3O4 phase and magnetite were observed,
respectively. The arrows indicate the experimental P –T path. Thick line represents the phase boundary of the
transformation from magnetite to its high-pressure phase, determined by Huang and Bassett (1986) (Fei et al.,
1999, q1999 Mineralogical Society of America). (b) The variation of the oxygen u parameter in spinel,
magnetite and nickel silicate spinel. Increasing u with pressure indicates octahedron is more compressible
than the tetrahedron, while decreasing u with pressure, indicates that the tetrahedron is more compressible
(Finger et al., 1986).
694 Chapter 9

TABLE 9.7
Refined unit-cell and atomic positional parameters and selected inter-atomic distances (Å) for the high-pressure
phase of Fe3O4 at 24 GPa and 823 K (Fei et al., 1999)

Ions Index x y z
2þ
Fe 4d 0.724(6) 0.3757(5) 0.25
Fe3þ 8e 0.246(4) 0.1107(4) 0.0879(4)
O22 4c 0.506(5) 0.25 0
O22 4d 0.180(9) 0.2447(19) 0.25
O22 8e 0.296(11) 0.4899(13) 0.0980(15)

Bond distance Bond distance

Fe2þ –O2 1.775(12) £ 1 Fe3þ –O1 1.715(12) £ 1

Fe2þ –O2 1.959(14) £ 1 Fe3þ –O3 1.716(25) £ 1
Fe2þ –O3 2.159(25) £ 2 Fe3þ –O3 1.899(29) £ 1
Fe2þ –O3 2.409(21) £ 2 Fe3þ –O2 1.997(5) £ 1
Fe2þ –O1 2.718(12) £ 2 Fe3þ –O3 2.006(14) £ 1
Fe2þ –Fe2þ 2.7992(3) Fe3þ –O1 2.589(12) £ 1
Fe3þ –Fe3þ 2.7992(3)
Fe2þ –Fe3þ 2.695(5)

Notes: The final discrepancy indices (Larson and Von Dreele, 1986) are Rwp ¼ 0.019; Rp ¼ 0.013; R(F 2) ¼ 0.17;
and reduced X 2 ¼ 9.4. Space group: Pbcm; Z ¼ 4, a ¼ 2.7992(3) Å, b ¼ 9.4097(15) Å and c ¼ 9.4832(9) Å.

9.3.1.1. EOS and molar volume

Haavik et al. (2000) obtained the XRD pattern for h-Fe3O4, which, along
with the data for hematite (a-Fe2O3) and wüstite around , 40 GPa, led them to
suggest that the HP transition of Fe3O4 is related to structural rearrangement
rather than decomposition to Fe2O3 þ FeO. As already mentioned, magnetite and
h-Fe3O4 (CaMn2O4-type) co-exist over a large pressure range and the transition
shows a considerable hysteresis. The transition is likely to be reconstructive and of
first order.
In general, the denser phase has greater bulk modulus. The bulk modulus of
h-Fe3O4 has been determined by shock-wave experiments at pressures 65– 130 GPa and
their corresponding molar volumes on extrapolation to ambient pressure have been
determined by different authors as:

Bulk modulus (GPa) Molar volume (cm3 mol21) Authors

340 38.27 Ahrens et al. (1969)

450 39.21 Anderson and Kanamori (1968)

The bulk modulus values (KTO) calculated from pressure – volume data using a
third-order Birch –Murnaghan EOS are obtained by Haavik et al. (2000) as:
Magnetite: 217(2) GPa
h-Fe2O3: 202(7) GPa
AB2X4 Structure 695

9.3.1.2. Néel temperature, TN

Magnetite has a ferrimagnetic ordering of magnetic moments of the iron atoms at
room temperature. Its magnetic ordering temperature, i.e., Néel temperature, TN, is
, 850 K (848.5 K; Grønvold and Sveen, 1974), when it becomes paramagnetic. Through a
super-exchange interaction, iron atoms interact through oxygen atoms lying in between.
This is responsible for the TN. The super-exchange interaction is a function of the lattice
parameter, a, and the oxygen positional parameter, u.
The TN is found to increase linearly with pressure, the slope dTN =dP being
20.5(10) K GPa21 while the pressure dependence of the oxygen positional parameter,
du=dP; is 5.8 £ 1024 GPa21 (Samara and Giardini, 1969). Evidently, there is a pressure
dependence of TN to the super-exchange interaction and the spin interaction is inversely
proportional to the inter-atomic distance.

TN: Bloch’s 10/3 law. Bloch (1966) determined the empirical relation between TN and
volume compressibility ðbÞ; which is known as the 10/3 law:
ðd ln TN =dPÞ ¼ ð10=3ÞbV ð9-8Þ
where P and bV denote the pressure and isothermal volume compressibility. Bloch’s
equation can also be rewritten as:
ðd ln TN =dPÞ ¼ 2bV ð2d ln W=d ln VÞ ¼ ð10=3ÞbV
where W is half the band-width and is proportional to the interaction between A and X and
between B and X (in tetrahedral A and octahedral B sites, respectively).
The volume compressibility is three times the linear compressibility of the unit-cell
edge, ba. Therefore, equation (9-8) becomes:

ðd ln TN =dPÞ ¼ 10ba

Since the linear compressibility, ba ; can be measured as {bðrAX Þ þ bðrBX Þ}=2;

where those in the parentheses () denote the compressibility of the distances rAX and rBX,
respectively, we can write the above equation as:

ðd ln TN =dPÞ ¼ 5{bðrAX Þ þ bðrBX Þ}

which means that the super-exchange interaction depends on the distances between A and
X and between B and X.
The Bloch’s equation has been modified by Nakagiri et al. (1986) to derive a
relationship between the Néel temperature and the oxygen positional parameter, u. The
modified Bloch’s equation gives a close agreement with the instrumental results.
The Néel temperature, TN, in super-exchange magnetism is related to the A – X – B
inter-atomic distance, d, in spinels (Weisz, 1951) as:
kTN ¼ C1 SA SB C12 C2 d
696 Chapter 9

where k is Boltzmann’s constant and SA and SB are the electronic spins of A and B atoms,
respectively. The distance A – X –B is equal to rAX þ rBX and the values of C1 and C2
determined empirically are 5.04 £ 1023 erg and 7 Å21, respectively.
The magnetic moments of A and B atoms are independent of pressure (Samara,
1969). Differentiating the above equation with respect to P and equating d as rAX þ
rBX{ ¼ (0.3765 þ 0.732u)a}, Samara and Giardini (1969) obtained:

ðd ln TN =dPÞ ¼ 2:42 £ 1022 GPa21

and for u:

ðdu=dPÞ ¼ 5:8 £ 1024 GPa21 :

9.3.2. Pressure dependence of u and a

The pressure dependence of the crystal structure of magnetite and the unit-cell
parameters have been studied by Nakagiri et al. (1986) up to 4.5 GPa (diamond
anvil/XRD). The isothermal bulk modulus (KT) and its pressure derivative KT0 determined
by fitting the pressure – volume data to the Murnaghan equation of state are 181(2) GPa
and 5.5(15), respectively. The values of the oxygen positional parameter, u, obtained at
different pressures, have been plotted in Fig. 9.2(b) and show that u does not vary
significantly with pressure because du/dP ( ¼ 2 1(15) £ 1025 GPa) is very small.
The linear compressibilities of inter-atomic distances of magnetite are shown in
column 4 in Table 9.8, where the inter-atomic distances were calculated using the values
a ¼ 8.3949(3) Å and u ¼ 0.2548(2) (Nakagiri et al., 1986).
The distances A – A, A – B and B –B, where A and B represent Fe atoms in the
tetrahedral and octahedral sites, respectively, are independent of u. The compressibilities
of these distances are the same as the linear compressibility, b, of the unit cell edge, which

TABLE 9.8
Linear compressibilities of inter-atomic distances of magnetite. A, B and X represent tetrahedral site Fe,
octahedral site and O, respectively (Nakagiri et al., 1986)

Atomic (1) Inter-atomic distance (2) Inter-atomic distance Compressibility Comments (5)
(ambient pressure) (3) 1023 GPa (4)
p
AA a 3 4 ; rAA 3.635(1) 1.84(2) Tet–tet cation
p separation
AB cation ap11 8 ; rAB 3.4803(1) 1.84(2) Tet–oct separation
BB a 2 4 ; rBB 2.9680(1) 1.84(2) Oct–oct cation
p separation
AX a 3(u 2 0.125) ; rAX 1.8873(29) 1.9(12) Tet bond
2 1/2
BX p 2 2u þ 0.375) ; rBX
a(3u 2.0592(16) 1.8(6) Oct bond
XX ap2(2u 2 0.25) ; rX1 3.0820(48) 1.9(12) Tet edge
XX a 2(0.75 2 2u) ; rX2 2.8541(47) 1.8(13) Shared oct edge
XX a(4u2 2 2u þ 0.375)1/2 ; rX3 2.6919(1) 1.84(3) Unshared oct edge
AB2X4 Structure 697

amounts to , 1.84(2) £ 1023 GPa21. The other inter-atomic distances, A – X, B – X and

X – X (where X represents the oxygen atom), depend on both a and u. The linear
compressibilities can be expressed as presented in the following:
     
1 drAX 1 1a 1 du
bðrAX Þ ¼ ¼2 2
rAX dp a dp u 2 0:125 dp
     
1 drBX 1 da 3u 2 1 du
bðrAX Þ ¼ 2 ¼2 2 2
rBX dp a dp 3u 2 2u þ 0:375 dp
     
1 drt 1 da 1 du
bðrXt Þ ¼ 2 ¼2 2 ð9-9Þ
rXt dp a dp u 2 0:125 dp
     
1 drXs 1 da 1 du
bðrXs Þ ¼ 2 ¼2 þ
rXs dp a dp 0:375 2 u dp
     
1 drXu 1 da 4u 2 1 du
bðrXu Þ ¼ 2 ¼2 2 2
rXu dp a dp 4u 2 2u þ 0:375 dp
where 2ð1=aÞðda=dPÞ is simply the compressibility of the unit-cell edge a, i.e., ba :
The bond angle A – X – B (i.e., a) is associated with the super-exchange interaction
and is measured as:
3u 2 1
cos a ¼ pffiffiffi ð9-10Þ
3ð3u 2 2u þ 0:375Þ1=2
2

The pressure dependence of the bond angle (a) of magnetite was calculated to be
6(88) £ 1025 deg/GPa, which is negligibly small.

9.3.3. Polyhedral bulk modulus, K

Using the values in Table 9.3, the volumes of octahedral and tetrahedra 3 polyhedra
have been found as:
Voct ¼ 11:61ð3Þ Å3
Vtet ¼ 3:45ð2Þ Å3
and the compressibility and the bulk moduli are determined to be
boct ¼ 5:4ð20Þ £ 1023 GPa; Koct ¼ 2:0ð8Þ £ 102 GPa and
btet ¼ 5:8ð35Þ £ 1023 GPa; Ktet ¼ 1:7ð10Þ £ 102 GPa

9.3.4. Ca-ferrite structure

In the calcium (Ca)-ferrite structure, the most dense structure for known crystalline
materials with AB2X4 stoichiometry (Reid et al., 1969), Mg2þ ions should be in eight-
coordinated sites and Al3þ ions occupy octahedral sites.
NaAlSiO4 also adopts this structure at 1,0008C and pressures above 18 GPa (Liu,
1977). However, Irifune and Ringwood (1969) demonstrated that Al3þ and Naþ ions
698 Chapter 9

are incorporated into majorite garnet at pressures corresponding to the transition region.
But majorite garnet is no longer stable under lower-mantle conditions, where the
aluminous phases that are stable include: 1-MgAl2O4 (Lu, 1978), aluminium-bearing
MgSiO3 and CaSiO3 perovskites (Irifune and Ringwood, 1987), an unidentified “Al-rich
phase” (Takahashi, 1987) and a hollandite-type structure (Ca0.5Mg0.5)Al2Si2O8 (Madon
et al., 1989).
Ca-ferrite-type MgAl2O4 may be another potential host for Al3þ and possibly Naþ
in the lower mantle.

9.3.4.1. CaMn2O4 and Mn3O4

In CaMn2O4, the local environment around Mn2þ is irregular due to JT deformation
of the MnO6 octahedra.
In h-Mn3O4 at 39 GPa, the Mn3þ – O22 bond lengths for the octahedral site and the
Mn –O22 bond lengths for the 8-fold coordinated site range from 177.2 to 223.6 pm and
2þ

from 210.0 to 249.7 pm, respectively (Paris et al., 1992).

The CaMn2O4-type structure is a deformed variant of the CaTi2O4 type and is
related to the CaFe2O4 type (discussed earlier in Section 9.3.1). These three, quite dense
structure types are compared by Haavik et al. (2000). From the deformed CaMn2O4 type
to the more symmetric CaTi2O4 type, the space-group symmetry is changed from Pbcm
to Pbmm.

9.3.5. Fe3O4, MgAl2O4 and g-Ni2SiO4 spinels

As stated earlier, all the three spinel-type compounds have bulk moduli of , 2 Mbar
but the individual cation polyhedral bulk moduli in different spinels differ greatly
(Table 9.9).
Divalent magnesium tetrahedra in spinels are twice as compressible as tetravalent
silicon tetrahedra in g-Ni2SiO4, whereas trivalent iron tetrahedra in magnetite have an
intermediate compressibility.
Divalent nickel octahedra in the silicate spinel are 50% more compressible than
trivalent aluminium octahedra in magnesium aluminate and the mixed Fe2þ – Fe3þ
octahedra of magnetite, with an average valence of 2.5, are again intermediate
in compressibility. In both octahedral and tetrahedral polyhedra, therefore, the

TABLE 9.9
Cation polyhedral bulk moduli ðKÞ and cation valences ðZÞ for g-Ni2SiO4, magnetite and spinel (Finger et al.,
1986)

Compound Octahedron Tetrahedron

Z K (Mbar) Z K (Mbar)

g-Ni2SiO4 þ2 1.7(1) þ4 .2.5

Fe3O4 þ2.5 1.9(2) þ3 1.9(2)
MgAl2O4 þ3 2.6(4) þ2 1.2(2)
AB2X4 Structure 699

TABLE 9.10
The K 0 0 values for spinel group crystals, two sets of K 0 0 values being calculated by power and exponential
functions for the inter-atomic repulsive potential

Experimental K 0 0 Calculated K 0 0 Data source

Power function Exponential function

MgO.2.6Al2O3 4.18 – – Schreiber (1976)

MgAl2O4 – 3.70 3.16 Anderson and Anderson (1970)
NiFe2O4 4.41 3.76 3.22 Liebermann (1969); Anderson
and Anderson (1970)
FeFe2O4 – 3.81 3.29 Anderson (1970)

inverse relationship is seen between the cation formal charge and the compressibility. The
high-pressure behaviour of spinel and magnetite, and that of g-Ni2SiO4, has been
investigated by Finger et al. (1979).
In the spinel structure, the tetrahedral bonds are stronger than the octahedral ones.
Finger et al. (1979), studying this structure up to 3.8 GPa, found that the tetrahedral Si –O
bond remained unchanged but octahedral Ni –O bond decreased. The K 0 0 values for this set
of spinels have been calculated by power and exponential functions for the inter-atomic
repulsive potential. These are presented in Table 9.10.

9.3.5.1. MgAl2O4 and MgO: elastic constants and sound velocities

Since both MgO and MgAl2O4 are cubic, there are only three unique elastic
constants C11, C12 and C44. The pressure dependence of these elastic constants is shown in
Table 9.11.
The pressure and volume dependencies of the sound velocities in MgO and
MgAl2O4 are listed in Table 9.12 (Chopelas, 1996). Table 9.13 compares the ultrasonic
results with those obtained from EOS study by different workers.

TABLE 9.11
Elastic constants of MgO and MgAl2O4

Constant M (GPa) (dM/dP)T (d2M/dP 2)Ta References

MgO
C11 296.8(2) 9.17(7) 20.118(60) Jackson and Niesler (1982)
C12 95.3(2) 1.61(12) 20.028(68)
C44 155.8(2) 1.11(1) 20.032(6)
MgAl2O4
C11 282.9 5.59 20.65 Yoneda (1990)b
C12 155.4 5.69 20.64
C44 154.8 1.44(10) 20.19(4)
a
The elastic constant M at pressure P is expressed as: MðPÞ ¼ Mo þ ðdM=dPÞTO P þ 1=2ðd2 M=dP2 ÞTO P2 where
pressure is in GPa.
b
Uncertainties not given.
700 Chapter 9

TABLE 9.12
Pressure and volume dependence of the sound velocities of MgO and MgAl2O4 (Chopelas, 1996a)

Pressure dependence in km/s Volume dependence in km/s

VS ðPÞ ¼ VS ð0Þ þ ðdVS =dPÞT VS ¼ VSO þ ½dVS =dðV=VO Þð1 2 V=VO Þ

MgO
VS ¼ 6.05(1) þ 0.0381(13)P 2 3.6(4) £ 1024P 2
VP ¼ 9.70(2) þ 0.0704(20)P 2 5.6(6) £ 1024P 2
MgAl2O3
VS ¼ 5.500(5) þ 0.001(1)P
VP ¼ 9.785(11) þ 0.047(5)P 2 0.0010(5)P 2
VS ¼ 6.05(1) þ 6.37(10)(1 2 V/VO)
VP ¼ 9.69(1) þ 12.5(2)(1 2 V/VO)
VS ¼ 5.49(1) þ 0.1(3)(1 2 V/VO)
VP ¼ 9.79(1) þ 8.0(4)(1 2 V/VO)

9.3.6. Electrical resistivity

At pressures between 55 and 63 GPa, Fe2O3 has been observed to transform from
the anti-ferromagnetic to the paramagnetic state (Mao et al., 1977). The electrical
resistivity is seen to start decreasing at pressures at , 50 GPa (Endo and Ito, 1982).
Up to a pressure of 48 GPa (with T between 258 and 300 K), the electrical
resistivities of Fe3O4 were measured by Morris and Williams (1997) for evaluating the

TABLE 9.13
Comparison of the pressure derivatives of the elastic moduli with ultrasonic data (Chopelas, 1996a)

Modulus (GPa) Pressure (GPa) Reference

MgO
KS ¼ 162.0(10) þ 4.08(9)P 2 (1/2)0.036(16)P 2 36.6 Chopelas (1996)b
KS ¼ 162.5(2) þ 4.13(9)P 2 (1/2)0.058(66)P 2 3.2 Jackson and Niesler (1982)a
KS ¼ 162.7(2) þ 4.24(7)P 2 (1/2)0.029P 2 7.8 Yoneda (1990)b
KS ¼ 162.5(2) þ 4.15(9)P 2 (1/2)0.022(4)P 2 27.0 Duffy and Ahrens (1995)a
G ¼ 130.9(5) þ 2.56(6)P 2 (1/2)0.030(10)P 2 36.6 Chopelas (1996)b
G ¼ 130.9 þ 2.53P 2 (1/2)0.066P 2 3.2 Jackson and Niesler (1982)a
G ¼ 131.1 þ 2.41P 2 (1/2)0.052P 2 7.8 Yoneda (1990)b
G ¼ 130.8(2) þ 2.5(1)P 2 (1/2)0.027P 2 27.0 Duffy and Ahrens (1995)a
MgAl2O4
KS ¼ 198.2(8) þ 5.05(9)P 2 (1/2)0.13(12)P 2 11.0 Chopelas (1996)
KS ¼ 197.4 þ 5.15P 2 (1/2)0.48P 2 1.0 Chang and Barsch (1973)
KS ¼ 197.9(2) þ 5.66(21)P 2 (1/2)0.65(10)P 2 6.2 Yoneda (1990)
G ¼ 108.6(5) þ 0.072(7)P 11.0 Chopelas (1996)
G ¼ 108.5 þ 0.51P 1.0 Chang and Barsch (1973)
G ¼ 108.4 þ 0.36(48)P 6.2 Yoneda (1990)

Note: MðPÞ ¼ MO þ ðdM=dPÞTO P þ 1=2ðd2 M=dP2 ÞTO P2 :

a
Values for ðd2 K=dP2 ÞTO are double those reported in this reference.
b
Elastic modulus and its first pressure derivative fixed at ultrasonic values are given by Jackson and Niesler
(1982).
AB2X4 Structure 701

compression-induced changes in electron exchange between divalent and trivalent

iron ions.
The resistivity of Fe3O4 decreases by more than an order of magnitude between
0 and , 20 GPa where it reaches a minimum, subsequently increasing by a factor of two
between , 20 and 48 GPa. This finding implies that the electronic exchange between Fe2þ
and Fe3þ ions is enhanced by the initial 7% of volumetric compression but is marginally
impeded at higher compressions.
The resistivity of Fe3O4 is 1 – 2 orders of magnitude less than that of Fe0.94O at
48 GPa, illustrating the profound effect of intervalence charge transfer (particularly at high
pressure) on the electrical properties of iron-rich oxides. The activation volume of
conduction is P-dependent for the low-P (magnetite) phase. Such P-dependent behaviour
implies that simulations of mantle resistivity that use a constant activation volume may
underestimate the resistivity of the lower mantle.

9.3.7. g-Fe2O3

At about 60 GPa, the X-ray diffraction d-values begin to reduce suddenly and, at
85 GPa, the diffraction lines corroborate the corundum structure (Yagi and Akimoto,
1982). The estimated bulk moduli and density of Fe2O3 obtained by different workers are
shown in Table 9.14.

9.4. Cr-spinels, MCr2O4 (M 5 Mg, Mn, Zn): decomposition

MCr2O4 series form a set of normal spinels such as FeCr2O4 (chromite), MgCr2O4
(picrochromite), MnCr2O4 and ZnCr2O4. Their high-pressure decomposition to phases
such as Cr2O3 (eskolaiite) and MO (rocksalt type) have been investigated by Catti et al.
(1999) using periodic unrestricted Hartree –Fock calculations.
The simulated behaviour of inter-atomic distances vs. pressure shows similar
compressibilities of M– O bonds in both octahedral and tetrahedral coordinations.

TABLE 9.14
Estimated bulk moduli and density for Fe2O3

K0 (GPa) Method

Low-pressure phase
Goto et al. (1982) 193 Shock wave
Liebermann and Schreiber (1968) 202.7 Ultrasonic
Wilburn et al. (1978) 199 High-pressure X-ray
K0 (GPa) r0 (g cm23) Method

High-pressure phase
Goto et al. (1982) 277 6.22 Shock wave

K0 is equal to 4 unless otherwise stated.

702 Chapter 9

Figure 9.3. Reaction enthalpy DHðPÞ vs. pressure for the decomposition of spinels into component oxides.
Results for MgAl2O4 are from Catti et al., 1999, q1999 Springer-Verlag.

The predicted decomposition pressures of these spinels are:

MgCr2O4: 19 GPa
MnCr2O4: 23 GPa
ZnCr2O4: 34 GPa
The high stability of ZnCr2O4 can be attributed to the higher CFSE of Zn2þ at
tetrahedral sites compared with other M2þ cations.
The decomposition of these chromates agrees excellently with the results obtained
from experimental (13 GPa) studies on MgAl2O4, which decomposed to Al2O3
(corundum) and MgO (periclase). The DHðPÞ curves for these spinels, obtained by the
mixed computational scheme and shown in Fig. 9.3, depict that the spinel structure
becomes increasingly more stable (with respect to component oxides) with the increase in
atomic number of the metals. The ZnCr2O4 curve is not only shifted downwards but also
shows a smaller slope than the other ones. This happens because of greater relative stability
of Zn in the tetrahedral site (in spinel) with respect to the octahedral site (in rock salt)
compared with other M2þ cations. Because of the high CFSE of Cr3þ in octahedral
coordination, such behaviour can be noted in the curves for MgCr2O4 and MgAl2O4.
A similar explanation holds for the pair Mn2þ/Mg2þ in MnCr2P4 and MgCr2O4 spinels.

9.4.1. Oxidation of Cr-spinel

Chromite ([4]Fe2þ[6]Cr2O4) with the cubic close-packed spinel structure (space

group Fd3m) has a unit cell in which eight of the total 64 tetrahedral (8a) and 16 of the 32
octahedral (16d) sites are occupied by cations.
Mössbauer investigations by a number of workers (Galvao de Silva et al., 1980;
Bancroft et al., 1983; Osbrone et al., 1984; Schmidbauer, 1987a,b; Zhe et al., 1988;
AB2X4 Structure 703

Dyar et al., 1989; Mitra et al., 1991a,b) record a wide range of deviation from such cationic
distribution.
From the study of synthetic chromite, Schmidbauer (1987a) showed that, in
chromite, Fe2þ could be converted to Fe3þ by oxidation which Pcould lead to the complete
conversion of Fe2þ to Fe3þ at the tetrahedral site. The Fe3þ / Fe ratio in chromite could
give an idea of the oxidation at the known temperature of formation.
Mitra et al. (1991a,b) showed that, in natural chromite, the octahedral site
occupancy of Fe2þ is explained as a product of an electronic oxidation process by which
electron from an oxidized A site [Fe2þ(A) ! Fe3þ(A) þ e2] moves to the B site and is
localized at the B site, converting the existing Fe3þ(B) to Fe2þ(B). Later, Pal et al. (1994)
determined the stages of oxidation of chromite as:
Primary stage: presence of Fe2þ partially at the A site and fully at the B site.
Intermediate stage: Fe3þ at the A site and Fe2þ located only at the B site.
Mature stage: Fe3þ at the A and B sites with no Fe2þ at either site.
An accurate quantitative determination of Fe2þ and Fe3þ and their site distribution
are hardly attainable from the use of wet-chemical, XRD P and microprobe analyses.
Mössbauer spectroscopy offers highly reproducible Fe3þ / Fe ratios and Fe2þ and Fe3þ
occupancies at different sites (cf., Mitra, 1992). However, single-crystal X-ray diffraction
is most reliable for site-occupancy determination in silicates (e.g., Princivalle et al., 1999)
and also in oxides.

9.4.2. OD in Cr-spinels

The thermodynamic properties of Cr-spinels become important for evaluation when

they make the furnace-smelting products or the high P –T component phase of igneous and
metamorphic rocks.
For understanding the thermodynamic properties of spinels, the characterization of
the order – disorder (OD) relation is central. Because of its high octahedral CFSE
(63 kJ mol21 cm21), trivalent chromium ion orders only in the octahedral site in the spinel
lattice while other cations, such as Mg2þ, Fe2þ and Al3þ, disorder between the octahedral
and tetrahedral sites. The ordering of Cr3þ at the octahedral site influences the partitioning
of the other cations between the octahedral and tetrahedral sites.
The high-temperature OD relation in (Mg2þ, Fe2þ)(Fe3þ, Al3þ, Cr3þ)2O4 solid
solutions have been evaluated on the basis of solid solutions of Fe3O4 –FeCr2O4, Fe3O4 –
MgCr2O4 and Fe3O4 – FeAl2O4 – FeCr2O4 from the data of electrical conductivity and
high-temperature thermopower studies (Nell and Wood, 1991). In such studies, it is
assumed that the spinels are n-type small polaron conductors with Fe2þ – Fe3þ electrons
hopping in octahedral B sites only. For further discussion on small and large polarons,
see Section 15.2.5.2.

9.4.2.1. Thermopower (Q) and conductivity

A systematic decrease in the absolute values of thermopower (Q) is noted with
Fe3O4 dilution in the solid solutions of Fe3O4 – FeCr2O4 and Fe3O4 – FeAl2O4 –FeCr2O4
704 Chapter 9

(Nell and Wood, 1991). This decrease results from the substitution of Fe3þ by Cr3þ and
Al3þ, causing a decrease in the Fe3þ – Fe2þ hopping at the B site.
In spinel solid solutions, a general decrease in conductivity occurs with magnetic
dilution, effecting a decrease in the number of conducting (Fe2þ þ Fe3þ) sites. The carrier
concentration in Fe3O4 solid solutions depends on temperature because of the thermal
disordering of Fe2þ and Fe3þ between octahedral and tetrahedral sites.
The temperature dependence of the carrier concentration is reflected by the
polynomial to which the thermopower (Q) is related

Q ¼ a þ bð104 =TÞ þ cð104 =TÞ2 ðmV=KÞ

where a, b and c are coefficients.

The conductivity (s) data fit to a polynomial

lnsT ¼ A þ Bð104 =TÞ

where the coefficient B is proportional to the apparent activation energy (Ea):

2Ea e0
B¼ ð9-11Þ
k £ 104
where k is Boltzmann’s constant, e0 the electronic charge and Ea the apparent hopping
energy in eV obtained from the conductivity data. When temperature and compositional
dependencies of carrier concentrations are neglected, the conductivity can be expressed as:

s ¼ exp½A 2 ðEa e0 =kTÞ=T ð9-12Þ

For further observations on conduction in minerals, readers are referred to Section 15.2.

9.4.3. Defects and electrical behaviour

The concentration of the conducting species, Fe2þ and Fe3þ, controls the electrical
behaviour. The total Fe content is determined by electron microprobe analysis. The Fe2þ
and Fe3þ contents and their ratios are obtained by Mössbauer spectroscopy. However, the
Fe3þ/Fe2þ ratios depend on the f O2 of formation.
Considering the ideal spinel stoichiometry as true stoichiometry, ðdÞ is measured
with reference to pure Fe3O4.
It is observed that, for Fe3O4, a well-defined plateau occurs in which both
thermopower and electrical conductivity are insensitive to variations in f O2. This plateau
extends over about five log units in f O2 and corresponds to ldl # 1 £ 1023 : This
observation corresponds to that expected for an n-type small polaron conduction, in which
the thermoelectric coefficient, Q, depends on Fe3þ/Fe2þ as (Wu and Mason, 1981):

Q ¼ 2k=e0 {ln½2ð½6 Fe3þ =½6 Fe2þ Þ þ A}V=K ð9-13Þ

where k is Boltzmann’s constant, e0 the electronic charges and A the vibrational entropy
associated with the ions surrounding a polaron on a given site. In a conducting system
AB2X4 Structure 705

where the conducting sites have nearly the same energy, as in the octahedral B sublattice,
the value of A is negligibly small (Austin and Mott, 1969; Emin, 1975).
The above relation clearly shows that the thermoelectric coefficient (Q) directly
gives the ratio (Fe3þ/Fe2þ) on octahedral sites. In the cases of Fe3O4 and FeCr2O4, where
only Fe2þ and Fe3þ are disordered between octahedral and tetrahedral sites, the value of Q
(in V/K) sufficiently characterizes the cation distribution.
In Fe3O4 – MgCr2O4 and Fe3O4 – FeAl2O4 – FeCr2O4 solid solutions, the sets of
three cations (Fe2þ, Fe3þ, Mg2þ and Fe2þ, Fe3þ, Al3þ, respectively) are disordered. In
such cases, electrical conductivity ðsÞ measurement data provide the additional
information required to derive the inter-site cation distributions.
Presuming that only octahedral hopping occurs and considering the relation of Wu
and Mason (1981), one obtains cation-site occupancies and activation energies as varying
monotonically with composition. Partitioning of Fe2þ and Fe3þ between octahedral and
tetrahedral sites is found to obey closely the spinel model of O’Neill and Navrotsky (1984).
In plots of activation energy of hopping vs. composition in the Fe3O4 – MgCr2O4
and Fe3O4 –Fe2TiO4 systems, the inflection points indicate transition from Fe2þ – Fe3þ
hopping between octahedral sites (activation energies of , 0.15 –0.25 eV) to a different
type of conduction mechanism (activation energies of , 0.4 – 0.5 eV). Because of the large
energy barrier involved for electron hopping between tetrahedral and mixed site
octahedral, the second conduction mechanism observed may be ascribed to electron
hopping between Fe2þ – Ti4þ and Fe2þ –Cr3þ. The data of Nell and Wood (1991) also
support the idea that Cr3þ participates in the electron hopping process.
In the join Fe3O4 – FeCr2O4, the activation energies for conduction show a
consistent value of about 0.5 eV in the region where XFeCr2O4 $ 0.75.
The plot shows three components. In the Fe3O4-rich region, the activation energy
increases the FeCr2O4 content. In the region between 0.6 and 0.75 XFe2Cr2O4, the activation
energy rapidly increases. Presumably, in the region 0 , XFe2Cr2O4 , 0.6, the conduction
occurs alone by octahedral hopping between Fe2þ and Fe3þ. In a more Cr-rich
composition, mixed electron transfer occurs along Fe2þ – Fe3þ – Cr3þ on the join
Fe3O4 – FeCr2O4. The octahedral Fe2þ – Fe3þ hopping continues down to 40% Fe3O4,
but not beyond, with higher concentrations of FeCr2O4. The conduction mechanism occurs
at such a composition where there is an inflection in the plots of activation energy vs.
composition. Such inflections in activation energy are seen in the binaries Fe3O4 –FeCr2O4
and Fe3O4 –MgCr2O4 and for the two joins in the Fe3O4 –FeAl2O4 – FeCr2O4 ternary.

9.4.3.1. Shocked chromite

Shock veins in Suizhou meteorite, chromite grain shows three zones: (i) a new
dense polymorph with a type structure of CaTi2O4 (CT), (ii) a type of CaFe2O4 (CF) and
(iii) the chromite itself.
Laser-heated DAC study showed that at 12.5 GPa CF phase appears in the spinel
form of chromite, and at 20 GPa the phase attains CT structure.
Oxide spinel actually transforms to post-spinel phases in the deep interior of the
Earth and occurs as hosts for some major and trace elements such as Cr, Al, Mg, Fe, Na, Si,
706 Chapter 9

and other transition and rare-earth elements. These thus play a major role in element
partitioning and geochemical evolution.

9.4.4. Chromite: post-spinel orthorhombic polymorph

For post-spinel transitions Ringwood group proposed orthorhombic CaFe2O4-type

(CF) and CaTi2O4-type (CT) structures. However, these phases have not been discovered
in terrestrial samples. In shock-metamorphosed meteorites many high-pressure minerals
have been identified such as ringwoodite, wadsleyite, silicate ilmenite, magnesiowustite,
(Na,K)AlSi3O8-hollandite, and post-spinel polymorphs.
Recent examination of the Suizhou meteorite L6 chondrite has revealed shock-
metamorphosed CT polymorph of chromite composition (Chen et al., 2003). Experiments
suggest that both CF and CT are quenchable polymorphs of chromite formed above 12.5
and 20 GPa, respectively.
The orthorhombic CT polymorph has density (5.639 cm23) which is 11.5% denser
than chromite. The transformation of chromite to CaTi3O4-structured polymorph is
estimated to occur at 20 – 23 GPa and 1,800– 2,0008C. The occurrence of ringwoodite,
NaAlSi3O8-hollandite and majorite-type pyrope garnet constrains the peak P, T to 20–
23 GPa and 1,800 –2,0008C (Xie et al., 2002).
This dense CaTi2O4-structured FeCr2O4 phase could be the host phase for Cr, Al,
Fe, Mg and Mn and other metallic elements in the deep Earth (Chen et al., 2003). The
quenched CF and CT phases could not be distinguished by microscopy, EPMA, SEM, or
even micro-Raman spectroscopy. Only X-ray from synchrotron source could come in use.
The orthorhombic lattice belongs to space gr. Pbnm, and the parameters (by XRD)
are
a ¼ 9:462ð6Þ
b ¼ 9:562ð9Þ
c ¼ 2:916ð1Þ
The orthorhombic CaTi2O4, CaFe2O4, and CaMn3O4-type structures are composed
of dodecahedral (AO8) and octahedral sites (BO6).

9.4.4.1. Chromite: Raman bands

Chromite shows characteristic Raman bands at 500, 595, 631 and 677 cm21. The
strong band at 677 cm21 can be assigned the A1g mode of the spinel structure (Gupta et al.,
1993).
Chromespinel (Mg , Fe) (Al,Cr)2O4 is an important accessory mineral in the
Earth’s mantle, e.g., In lherzolite coming from the upper mantle (Scarfe et al., 1979).
The natural CaTi2O4-structured FeCr2O4 phase in the Suizhou meteorite is found to
contain 6 wt% Al2O3 and 2.6 wt% of MgO, whereby the cations Al3þ and Cr3þ occupy
octahedral sites (BO6) and the Mg2þ and Fe2þ occupy the dodecahedral site (AO8).
Therefore CaTi2O4 structure can serve as an important host for Al3þ, Cr3þ and other
metals such as Mg2þ, Fe2þ, Ni2þ, Mn2þ, Zn2þ and Mn3þ in the deep Earth.
 707

(B) Sulfide Spinels

This page is intentionally left blank
AB2X4 Structure 709

9.4.5. ZnCr2S4 spinel

The compound ZnCr2S4, known as “kalininite”, is semi-conducting. By XRD

studies, the compressibility and amorphization of ZnCr2S4 spinel under high pressure have
been examined by Wittlinger et al. (1997). Deterioration of the single crystal was observed
around 10 GPa. The observed plastic deformation may have resulted from (i) a phase
transition, (ii) an irreversible structural change, (iii) chemical decomposition or (iv) a
hysteresis effect. After compression, the crystals show 30% reduction in unit-cell volume.
This page is intentionally left blank
 711

Chapter 10
ABX3, Perovskite– Ilmenite Structure

10.1. Introduction

Perovskites are ceramics (or rather, electro-ceramics), which constitute the Earth’s
most abundant mineral group and hold clues to the planet’s history. Ideal (defectless)
perovskites, like other ceramics, are electrical insulators in which all atomic sites are filled
and the ionic bonds are so strong that the electrons are tightly in place. The strong bonds
make the perovskite crystals resistant to scratch, deformation and melting but perovskites
are often distorted in structure and offer structural flaws to make them act as
superconductors, the properties of which attract considerable commercial interest.
Perovskites run the gamut from insulators (non-conductors) to semiconductors,
superionic conductors (in which whole ions, rather than electrons, flow through the
crystal), metal-like conductors and (now) high-temperature superconductors. The flaws in
the superconductor perovskites also allow them to manifest a wide range of electrical
properties.
Some perovskites manifest a ferroelectric property — the property seen in
materials which are polarizable and can reverse the polarity in an electrical field. The
atomic displacements in the ferroelectric mode of perovskite (ABO3) structure is shown in
Fig. 10.1. One such well-known ferroelectric perovskite is barium titanate (BaTiO3), also
the best-known electro-ceramic in commercial use. Under an applied electrical field, the
cations are energized and displaced and the crystal becomes more polarized. On release of
the applied field, the cations return slowly to their normal positions and release the stored
energy. BaTiO3 is thus used in capacitors, which store charge from pulses of current and
then release the stored charge between pulses, thereby producing a steady direct current. It
is employed as a voltage surge protector (which is often used in computers). When a surge
of electricity (say, from a lightning strike) reaches the BaTiO3 crystal, it absorbs the pulse
and slowly dissipates it. The eight-site model for perovskite ferroelectrics, particularly
BaTiO3 and KNbO3, is discussed later in Section 10.7.4.3.
The other interesting property shown by BaTiO3 perovskite is piezoelectricity. This
property arises because, when an external electric field displaces the Ti atoms from their
normal positions, the shape of the crystal changes. Conversely, mechanical pressure
deforming the crystal releases electrical energy. Hence, BaTiO3 perovskite also finds a use
in transducers, such as are employed in loudspeaker buzzers and microphones. Similar
properties are manifested by PZT perovskites (see Section 10.1.5).
712 Chapter 10

Figure 10.1. Atomic displacement in the “ferroelectric mode” of perovskite (ABO3) structure. Oxygen hatched
circles, B (octahedrally coordinated) cation (Rao and Gopalakrishnan, 1986).

Silicate perovskites can be classed into two structural types: (i) an orthorhombic
(Pbnm) structure, e.g., the Mg-rich perovskites and (ii) a cubic ðPm3mÞ high-symmetry
structure, e.g., pure CaSiO3 (Hemley and Cohen, 1992). Many geophysical observations
can be explained in terms of these two structural phases.
Silicate perovskite ((Mg,Fe)SiO3) is not only the predominant mineral phase of the
Earth’s lower mantle (,60 at% of the planet), its properties largely control the rheology
and the evolution and state of the planetary interior (e.g., Jackson, 1998). The
seismological, mineralogical and geochemical models of the Earth’s lower mantle can
now be unified on the basis of the measurements of perovskites under different pressures
and temperatures. Their structural properties and the equation of state (EOS) (viz. bulk
modulus and thermal expansivity) seem to attain a fair agreement. Also, the
seismologically determined density profiles, the 670 km discontinuity, D00 region and
the core – mantle boundary are better explained by employing the results of perovskites
under pressure.
The 670 km discontinuity is affirmed to coincide with the sharp transition boundary
between silicate spinel ((Mg,Fe)2SiO4) and perovskite þ magnesio-wüstite. But in the
perovskite-forming reaction, the P – T slope is not sufficiently negative to rule out any good
mixing of the upper and lower mantle (Ito et al., 1990). On this basis, the mineralogical
constitution of the peridotitic lower mantle is inferred to be 70% perovskite, 18%
magnesio-wüstite, 8% CaO-rich phase (in volumetric ratio) with a small amount of
stishovite and Al2O3-rich phase.
The zero-pressure model mineralogies can be compared with the adiabatically
decompressed seismic properties. Magnesio-wüstite is thermodynamically stable at zero
pressure but the silicate perovskite phase is not thermodynamically stable at pressures less
than about 22 GPa. However, a pure perovskite mineralogy is compatible with the
adiabatically decompressed density and temperature of the lower mantle.
Compositional constraints for the lower mantle that are based on an adiabatic
decompression of its seismic properties will systematically overestimate the pro-
portional abundance of the pyroxene component relative to olivine. This is because the
zero-pressure density and bulk modulus of MgSiO3 perovskite are greater than those of
ABX3, Perovskite – Ilmenite Structure 713

Figure 10.2. Goldschmidt diagram showing perovskite-forming compounds. The radii of 6-fold coordination
A and B cations are based on the ionic radii of Shannon and Prewitt (1976). The tolerance factors, t; are also
shown on this diagram, where these are calculated from the ionic radii of 8-fold-coordinated A and 6-fold
coordinated B cations. Solid circle, quenchable perovskites: open squares, unquenchable perovskites stable only
at high pressure (from Hattori et al., 1999, q 1999 Springer-Verlag).

MgO. The computed seismic parameters of a model for lower-mantle perovskite (Pv) and
magnesio-wüstite (Mw) mineral assemblages for XFe ¼ 0.10 at 30 GPa as a function of
temperature are shown in Fig. 10.3 (Wolf and Bukowinski, 1987). Superimposed on this
plot is the seismologically inferred density and seismic parameter of the Earth at a depth
corresponding to 25– 30 GPa (,800 km).

10.1.1. Magnetic ordering

Many cubic perovskite-structured ABX3 compounds, with A an alkali and B a 3d

metal, order anti-ferromagnetically at low temperatures with a collinear spin structure. In
spin ordering, the cubic-point symmetry is lowered and the cell becomes slightly distorted,
in close analogy to the ferroelectric phases of the cubic ABO3 perovskites. In many cases,
the distortion is so small that the resulting phase is potentially ferroelastic with a resulting
ferroelastic — anti-ferromagnetic coupling; both the properties originate simultaneously in
the same ordering process.

10.1.2. Perovskite and mantle convection

Wolf and Bukowinski (1987) concluded that, for the lower mantle to have a
pure perovskite mineralogy with XFe ¼ 0.10, a geotherm temperature near 2,400 K is
714 Chapter 10

required at a depth of 800 km, 25 –30 GPa consistent with temperature estimates for
layered-mantle convection (Richter and McKenzie, 1981). However, for a 3 : 2 or 2 : 1
assemblage of perovskite and magnesio-wüstite, approximately corresponding to a
pyrolitic composition, a geotherm temperature suggests a whole-mantle convection
(Brown and Shankland, 1981). But the adiabatically decompressed density and
temperature of the lower mantle is seen to be incompatible with a pyrolitic composition
and whole-mantle convection (Knittle et al., 1986).

10.1.3. Layered mantle

Indeed, by analyzing the density and elasticity of perovskite-dominated mineral

assemblages, one can conclude that the convection of the mantle is layered and that the
upper and lower mantle may not have intermixed significantly over geological time. The
lower-mantle properties can be satisfied by a different bulk composition (e.g., more
chondritic or Fe– Si enriched) than the upper mantle. Of course, any such layering has to be
imperfect or “leaky” throughout the evolution of the mantle.

10.1.4. Outlook

Silicate perovskite study under high pressure and high temperature is in its infancy
(Mao et al., 1991). Progress in this study would lead to better knowledge of the transport
properties and the additional constraints on mantle mineralogy.
The current knowledge of phase equilibria under lower-mantle conditions involve
(Mg,Fe)SiO3 and CaSiO3 perovskite, along with magnesio-wüstite and stishovite. But
evidence has been reported for new non-quenchable aluminous, calcic and magnesium
silicate phases, which are stable under lower-mantle conditions (Ito, 1989). Some other
phases have also been reported (Finger and Hazen, 1991) and, at deep-mantle conditions,
new classes of chemical reactions are likely (Knittle and Jeanloz, 1991). To fully describe
the elastic, rheological and transport properties of the materials occurring deep within the
Earth, some more research employing high P – T technology and advanced computational
techniques is needed.

10.1.5. Structure and types of perovskites

Perovskites of general formula ABX3 consist of cubes made up of three distinct

chemical elements (A, B and X) that are present in the ratio 1 : 1 : 3.
An A cation — the larger of the two kinds of metals — lies at the centre of each
cube, the B cations occupy all eight corners and the X anions lie at mid-points of the 12
cube edges. The formula is stated as ABX3 instead of AB8X12 because each B cation is
shared by eight neighbouring cubes and each X anion by four cubes.
In ABX3 perovskites, the following relation holds among the radii of ions:

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2ðrA þ rX Þ ¼ 2ðrB þ rX Þ
ABX3, Perovskite – Ilmenite Structure 715

where A is the cations at 8-fold coordination, B the cations at 6-fold coordination and
X the anion at 6-fold coordination.
In the perovskite structure, a variation of the ionic radius of B cation
(rB) is permitted only to an extent which is defined as the tolerance factor
(Goldschmidt, 1926), t:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
t ¼ ðrA þ rX Þ= 2ðrB þ rX Þ

Perovskite structure forms when t lies between 0.80 and 1.00. For t values smaller
than 0.8, the ilmenite (FeTiO3) structure is formed (Wells, 1984). Leinenweber et al.
(1994) opined that high-pressure perovskite phases with t < 0:84 are quenchable, while
those with t , 0:84 would back-transform to lithium niobate structure on release of
pressure. The Goldschmidt diagram (Fig. 10.2, after Linton et al., 1999, and Hattori et al.,
1999) for perovskite forming compounds show the relationships between A and B cations
and t (see caption).
In the ideal cubic structure for ABX3 perovskite, BX6 octahedra are located at the
corners of a cube surrounding the central A cation (Fig. 10.4a). The orthorhombic (Pbnm)
symmetry results from the cooperative rotation of octahedra about the indicated Q and F

Figure 10.3. Representative range of geotherm temperatures for the lower mantle, and their adiabatic
extrapolations to zero pressure, is indicated by the stippled region. Superimposed on these geotherms are the
computed stability fields for the orthorhombic, tetragonal, and cubic MgSiO3 perovskite polymorphs. Both the
adiabatic decompression of the lower mantle geotherm and the extrapolation of MgSiO3 perovskite’s thermal
expansivity to high temperatures cross the computed phase boundaries. The observed retrograde reversion
temperature of perovskite to enstatite is approximately equal to the computed lower critical temperature
(from Wolf and Bukowinski, 1987, q 1987 American Geophysical Union).
716 Chapter 10

Figure 10.4a. An idealised cubic structure of ABX3 perovskite. Surrounding the central A cation the BX6
octahedra are located at the corners of a cube. From the cooperative rotation of octahedra about the
indicated u and f axial directions an orthorhombic (Pbnm) symmetry results. the rotational sense about the
u axis is opposite for adjacent octahedra and is in the same sense for rotations along a particular f axes. In
orthorhombic MgSiO3 perovskite, the SiO6 polyhedra are slightly distorted from regular octahedra and the
Mg2þ cation is displaced from its central position (after Wolf and Bukowinski, 1987, q 1987 American
Geophysical Union).

axial directions. The rotational sense about the Q axis is opposite for adjacent octahedra
and is in the same sense for rotations along a particular F axis. Hypothetical rotations of
BX6 octahedra can lead to a transformation from perovskite to lithium niobate structure
(see also Fig. 10.21).
The orthorhombic ABX3 perovskites (Pbnm, Z ¼ 4), which include MgSiO3 and
NaMgF3 (neighborite), have structures that are distorted from ideal cubic structures by two
independent octahedral tilts, u and w (Fig. 10.4a).
The tilt w is about one of the diad axes, [110], of the BX6 octahedron; this is
equivalent to a tilt about the orthorhombic b-axis in the Pbnm phase. The tilt w is about the
[001] axis of the BX6 octahedron and corresponds to tilting about the c-axis. The u tilting
causes the bending of the (B– X –B) angle and thus a shortening of a and c, whereas the
w tilt causes a shortening of a and b.
On the basis of the assumption that the BX6 octahedra are regular, the unit-cell
volume of a centro-symmetrically distorted ABX3 perovskite with a space group Pbnm can
be readily written according to the formalism of O’Keeffe et al. (1979):
V ¼ abc ¼ 32½B – X3 cos2 F
where B –X is the bond length within the BX6 octahedron and f is the tilting about the
pseudo-3-fold axis of the octahedra (Fig. 10.4a). The quantitative relationship between the
angles u; f and F can be found in Zhao et al. (1993).
ABX3, Perovskite – Ilmenite Structure 717

The P- and T-induced dimensional change of perovskite structure can be

decoupled into two constituent parts: (1) the compression or expansion of the
octahedral bond length B –X and (2) the compression or expansion due to the BX6
octahedral tilting (e.g., Hazen and Finger, 1982). Thus, the volumetric expansion can be
expressed as:

3d½B – X 2d cos F
aV ¼ aV0 þ aV F ¼ þ ð10-1Þ
½B – XdT cos F dT

and the volumetric compression as:

23d½B – X 2d cos F
bV ¼ bV0 þ bV F ¼ 2 ð10-2Þ
½B – XdP cos F dP

Zhao and his co-workers (Zhao et al., 1993, 1994) demonstrated that this structural
model is applicable to thermal expansion of many perovskite structures.
A large number of elements may combine to form hundreds of perovskites. Ba,
K and REE (through 58Ce to 71Lu) are typical of the two dozen elements which can fill
the A position. Almost 50 elements — more than half of the stable entries of the
periodic table — can adopt the B sites. X is usually oxygen but may be occupied by
halogens (e.g., AgZnF3, CsCdBr3 and LiBaF3). Owing to ligand-field effects,
destabilization of transition metal ions occurs in 8- and 12-fold coordination relative
to the octahedral coordination site. In the perovskite structure, Fe2þ ion finds itself
unfavourable at the dodecahedral (8-fold) site because of its low crystal-field stabilizing
energy (e.g., Yagi et al., 1978). This explains the absence of transition-metal
perovskites.
In many perovskites, the central A cation is too small in relation to the B cations at
the corners of the cube. This disparity may cause the X anions along with the B cations to
move out of position. Such “off-centring” of positively charged cations can give
perovskite crystals electrical polarity, with one end charged positively and the other
negative. The possible cation – anion – cation interaction along the cubic edge is displayed
in Fig. 10.4b.
The A or B sites in perovskites may be occupied by more than one cation. For
example, in Ca2CaUO6, the octahedral B sites are occupied by calcium and uranium ions
(Fig. 10.5c). Again, differently charged ions of the same element may fill the octahedral
centres. One such “mixed valence” perovskite is BaBiO3 (or rather, Ba2Bi3þBi5þO6), in
which trivalent and pentavalent bismuth cations alternate in the octahedra throughout
the crystal.
By replacing Bi with an increasing amount of Pb, one obtains a BaBiO3 – BaPbO3
series. As the composition approaches BaPb0.8Bi0.2O3, the compound becomes a
semiconductor. Some of these intermediate compounds become superconducting when
cooled to absolute zero.
The group of perovskites commercially known as PZT encompasses the PbTiO3 –
PbZrO3 series. These exhibit a remarkably strong piezoelectric effect. This strong
718 Chapter 10

Figure 10.4b. The possible cation–anion–cation interaction along the cubic edge in perovskite structure. It also
shows the possible t2g and eg orbital directions in B cations, and ps and pp orbital directions in O22 ions in ABX3
perovskite.

piezoelectricity is employed in making loudspeakers, microphones, electrical relays,

pressure gauges, spark igniters and so on.

10.1.5.1. Defects in oxide perovskites

Oxides with perovskite structure of formula ABO3 have B ions at the centres of
regular octahedra of oxygen ions and A cations occupying holes between the octahedra.
Oxygen deficiency can be created when either A or B cations are substituted by other ions
of lower oxidation state, leading to the formation of non-stoichiometric compounds of
general formula ABO32y ð0 , y , 0:5Þ: A brownmillerite-type structure is frequently
obtained when y ¼ 0:5:
Non-stoichiometry in perovskite oxides from cation deficiency (at the A or B site)
and oxygen deficiency or excess leads to the defect ordering and superstructure exhibited
by these oxides. The large A cations at 12-coordinated sites can be missing, either partly or
wholly. This non-stoichiometry of the phases leads to the question of whether the A-site
atoms and vacancies are ordered or not. Random isolated defects or weakly interacting
defects can occur in metallic systems, unlike in their non-metallic counterparts. A-site
defective perovskite oxides are commonly known to be formed in cases where the B site is
occupied by such elements as Ti, Nb, Ta, etc.
The common examples for the oxygen-deficient calcium ferrites are: CaFeO2.5,
CaFeTiO5.5 (rather than CaFeTiO3 perovskite) and Ca2Fe1.33Ti0.67O5.33 (Fig. 10.5a,b
ABX3, Perovskite – Ilmenite Structure 719

Figure 10.5. In perovskites, lacking a full complement of oxygen atoms, layered structure may sometimes arise.
Entire sheets of octahedrons are replaced by smaller polyhedrons. In oxygen-poor compounds that result when
iron replaces some of the titanium in calcium titanate, such as (a) Ca2FeTiO5.5 and (b) Ca2Fe1.33Ti67O5.33, every
fourth and third layer, respectively, consists of iron-centred tetrahedrons. In (c) the octahedral B sites are occupied
by calcium and uranium ions (After Hazen, 1988).
720 Chapter 10

(see caption) from Hazen, 1988). In these, the oxygen deficit results in layers consisting of
Fe B cations that are surrounded by four oxygens rather than the usual six. In such cases,
the growth of the bulk crystals is far from uniform and they occur as flat, plate-like crystals,
reflecting a kind of layered arrangement.
In the Mn series of A0 Mn7O12, the Mþ, M2þ or M4þ cations are accommodated at
A0 , the charges being balanced by changes in the relative amounts of Mn3þ and Mn4þ in
the octahedral site (see for analogy Section 15.4.1 and Fig. 15.2). In the Cu series of
A0 Cu3X4O12, M2þ or M4þ cations occupy the A0 site and X accommodates Mn, Ge, Ti, etc.
(Chenavas et al., 1975).

10.1.5.2. Fe2O3 perovskite: TM and “magnetic hardening”

The high-pressure (,60 GPa) form of Fe2O3 attains an orthorhombic perovskite
structure (Pbnm). The orthorhombic unit cell contains four formula units of Fe2O3.
The unit parameters of the perovskite structure are related to the lattice constant, ac ; of
the cubic cell by the relations:
p
a < b < ac 2 and c < 2ac

Thus, the c=a and c=b ratios should be close to 2 ¼ 1.41 in orthorhombic distortion.
Williamson et al. (1986) investigated a-Fe2O3 in shock loading at peak pressures
from 8 to 27 GPa. Their studies employing Mössbauer and magnetic measurements
revealed that large fractions of the Fe sites do not exhibit TM transition. Small particles are
seen to have reduced hysteresis compared with bulk samples (Nininger and Schroeer,
1978). Their a-Fe2O3 powder had a “zero-field” magnetization Mo of 0.274 emu/g at
300 K. The zero-field magnetization above the Morin transition can be split into an
isotropic moment and an anisotropic “intrinsic” moment in the (111) plane, attributed as
the Dzyaloshink-y-Moriya spin-canting mechanism (Moriya, 1960).
Above TM, the magnetic response of hematite is determined by an in-plane
anisotropy, which is very sensitive to stress, and the large residual strain could produce
a “hardening”.
These observations by Williamson et al. (1986) can be a result of the reduced
crystallite size or large defect density associated with the residual strain. Large residual
strain leads to an apparent decrease in the intrinsic net magnetization above the Morin
transition, presumably due to strain-induced magnetic hardening. Under extreme
conditions, a small amount (<1%) of magnetite (Fe3O4) is formed by the shock.

10.1.6. MgSiO3 Perovskite

MgSiO3 perovskite (quenched from 28 to 38 GPa and ,2,000 K) shows an

orthorhombic structure of Z ¼ 4, having either a Pbnm (centrosymetric) or Pbn21 (non-
centro-symmetric) space group (Ito and Matsui, 1978). MgSiO3 perovskite at zero pressure
has a Pbnm orthorhombic structure (Ito and Matsui, 1978). In orthorhombic MgSiO3
perovskite, the SiO6 polyhedra are slightly distorted from regular octahedra and the Mg2þ
cation is displaced from its central position. There are 20 atoms in the orthorhombic unit
ABX3, Perovskite – Ilmenite Structure 721

cell: 10 parameters, collectively referred to as j, which are required to fully characterize

the structure, three lattice parameters and seven internal coordinates.
Displacive transformations in perovskites induce drastic changes in second-
order physical properties such as elastic moduli (Cij) and thermal expansivity (a)
(see Salje, 1989).

10.1.6.1. Atomistic simulation: MEG

The perovskite and ilmenite phases of MgSiO3 have been studied by an atomistic-
simultation approach (Matsui et al., 1987), empirical parametrization and energy
minimization. The elastic constants of perovskite that were predicted from this simulation
model agreed reasonably with the data obtained from volume-compression experiments.
Some, however, used parameter-free theoretical models based on quasi-harmonic
lattice dynamics and the modified electron-gas (MEG) theory (Wolf and Bukowinski,
1987). Following the approach (modified electron-gas) of Cohen and Gordon (1976), the
later workers chose to stabilize the O22 ion wave functions by using a Watson sphere with a
charge of þ 1 and a radius of 2.66 atomic units. In this were included the nearest-neighbour
cation, oxygen and oxygen – oxygen pair interactions up to the third shell. The thermal
contribution to the free energy was calculated using quasi-harmonic lattice dynamics.
The model predicts successive second-order phase transitions at elevated
temperatures to a tetragonal and then to a cubic structure. But calculations of the pressure
dependence of high-temperature transitions suggest that the orthorhombic phase would be
stable throughout most of the lower mantle. This model also predicts that the cubic
perovskite structure is stable to pressures of at least 70 GPa. The orthorhombic form of
MgSiO3 perovskite and the cubic form of CaSiO3 perovskite are seen as stable at zero
pressure. The latter, however, is predicted to be dynamically unstable beyond ,110 GPa.
The minor structural distortions (Hofmeister et al., 1999) are related to space
groups, the information on which is obtainable from vibrational spectroscopy. The
structural polymorphs of perovskites show distinctive IR signatures: tetrahedral phases
show five bands (eight modes) and the rhombic polymorphs show as many as 25 modes
(see Section 10.2.2.5).
In the MEG model, the equilibrium lattice constant for the cubic structure is
predominantly determined by a balance between the Madelung attractive and Si –O
repulsive forces.

10.1.6.2. Phonon spectrum

Orthorhombic perovskite. The perovskite phase is predicted to exist in an orthorhombic
structure throughout most of the lower mantle. For the high-temperature geotherms, the
perovskite phase at the top of the lower mantle exists in higher symmetry polymorphs,
although Fe, Ca and Al components may complicate the stability fields.
Figure 10.6 shows the computed quasi-harmonic frequency spectrum of
orthorhombic (Pbnm) MgSiO3 perovskite for the equilibrium structure at zero pressure
and 300 K. At ambient conditions, all mode frequencies are positive and the distorted
perovskite structure is dynamically stable in the quasi-harmonic approximation (Weng
et al., 1983).
722 Chapter 10

Figure 10.6. Quasiharmonic frequency spectrum of orthorhombic (Pbnm) MgSiO3 perovskite computed for
the equilibrium structure at zero pressure and 300 K. All mode frequencies are positive and the lattice is
dynamically stable. Observed infra-red vibrational frequencies are indicated at G (after Weng et al., 1983, q 1983
Carnegie Institution).

With pressure (at room temperature), all phonon frequencies increase and the
orthorhombic MgSiO3 lattice remains dynamically stable up to at least 150 GPa. With
temperature, however, phonons related to the “condensed” Rx25, Ry25 and Mz2 phonons
exhibit critical behaviour and the quasi-harmonic frequencies approach zero. This
behaviour is typical of anti-ferroelectric perovskites.

Cubic perovskite: R25 – T– M2 branch. The phonon spectrum for MgSiO3 in the cubic
perovskite structure at zero pressure was shown by Wolf and Bukowinski (1987). Mode
frequencies of cubic MgSiO3 along the R25 – T –M2 branch are imaginary at zero pressure.
This is in contrast to CaSiO3, where the branch is unstable only at high pressures. In both
these compounds, the associated atomic displacements for the phonons are related to the
cooperative rotation of SiO6 octahedra about the pseudo-cubic axes. The repeat distance is
shortest for the R25 mode, where the adjacent octahedra along the axial rotation direction
have an opposite sense of rotation. This is longest for the M2 mode when the rotations of
adjacent axial octahedra are in phase.
Mechanical instabilities in cubic perovskite structure lead to a variety of space-group
symmetries and unit-cell sizes. The resultant lower symmetry structure can be related to the
cubic structure through the condensation of particular unstable phonons and a “freezing-in”
of their associated octahedral rotations (Glazer, 1972). Free energy differences between
ABX3, Perovskite – Ilmenite Structure 723

Figure 10.7. Difference between self-consistent LAPW charge density and overlapping spherical ions was
(Mg2þ, Si4þ, O22) for cubic MgSiO3 along the (110) plane (P ¼ 155 GPa). The contour interval is
0.005 e 2/ bohr3. Negative contours are dashed (from Cohen et al., 1989, q1989 American Geophysical Union).

these various polymorphs are likely to be small, as is suggested by the fact that all modes
along the unstable R25 – T –M2 branch have nearly the same frequency.
Again, as pressure increases, the critical transition temperature (computed) of
MgSiO3 increases by ,28.6 K/GPa, as is known for ferroelastic phase transition. Near the
critical transition temperature, soft modes at the Brillouin zone centre in the orthorhombic
lattice approach the acoustic branches.
The total potential arises from the long-range Madelung energy and short-range
Si – O, O –O and Mg – O overlap energies. Because of an increase in its bond length with
distortion, the Si –O overlap repulsion decreases. However, this decrease in energy is
opposed by an increase in the Madelung energy and Mg – O repulsion.
The difference between self-consistent LAPW charge density and overlapping
spherical ions (Mg2þ, Si4þ and O22) for cubic MgSiO3 (at a pressure of 155 GPa) is shown
in Fig. 10.7.
On the potential surface, the presence of multiple minima indicates that the cubic
perovskite lattice is mechanically unstable relative to octahedral rotations. Although the
cubic perovskite lattice is dynamically unstable in the quasi-harmonic approximation, at
high temperature the cubic phase is thermodynamically stable relative to the distorted
phases through the octahedral rotation vibrations.

Rotation angles and distortions. When the SiO6 polyhedra are regular octahedral, the
structural parameters are reduced to five, e.g., (i) the Si – O distance, (ii) two parameters
describing displacement of the central Mg2þ or Ca2þ cation within the (001) plane and
(iii) two Eulerian angles, Q and F; needed to describe the coupled octahedral rotations
about the respective (110) and (001) pseudo-cubic axis (see Fig. 3.12).
724 Chapter 10

Figure 10.8. Contour plots of the total static lattice potential. U for MgSiO3 perovskite as a function of the
octahedral rotation angles, Q and F at constant volume. (a) V ¼ 1:05 V0 ; (b) V ¼ V0 ; and (c) V ¼ 885 V0 ;
where V0 is the zero-pressure static lattice volume for the cubic structure. With decreasing volume, the potential
minima deepen and occur at larger values of Q and F (Wolf and Bukowinski, 1987, q 1987 American
Geophysical Union).

Figure 10.8 (after Wolf and Bukowinsiki, 1987) shows the contours of the total
static lattice potential energy with respect to the octahedral rotation angles, Q and F; at
several fixed volumes for MgSiO3. MgSiO3 exhibits four structurally equivalent minima at
finite values of Q and F: With a decrease in volume (with respect to CaSiO3 perovskite),
the potential minima for MgSiO3 deepen and occur at larger values of Q and F:
The rotation angles can be obtained from the ratios of the cell parameters as:
p
a=b ¼ cos Q a=c ¼ cos F= 2 ð10-3Þ
These tilt angles, Q and F; can also be viewed as order parameters between the perovskite
structural polymorphs.
The rotation angle increases with pressure in the case of regular SiO6 octahedra
ðj ¼ 5Þ whereas, in distorted polyhedra, the values of Q and F; calculated from equation
(10-3), remain nearly invariant. The ratios c=a and b=a also increase with pressure in
cases when j ¼ 5 (regular octahedron). The tetragonal distortion of perovskite structure
occurs by (a) rotation of the oxygen octahedra about a 4-fold axis and (b) the trigonal
distortion generated by rotation of the oxygen octahedra about a 3-fold axis is presented
in Fig. 10.9a,b.
As the temperature increases, the amplitude of the rotation vibrations increases and
eventually the low potential barriers can be overcome. Thermal “well hopping” between
the potential minima can occur, resulting in a fluctuation of the structural properties over
time scales much longer than those that correspond to the vibrational frequencies. At
sufficiently high temperatures, the rotation vibrations are no longer impeded by the
potential barrier and the average structure of MgSiO3 perovskite becomes cubic. This is
how the silicate perovskite phase can undergo critical phase transformations at
high temperatures.
Wolf and Bukowinski (1987) listed the experimental and theoretical unit-cell
parameters, atomic positions and bond distances for Pbnm orthorhombic MgSiO3
perovskite at 300 K. The theoretical results derived from MEG theory at zero pressure and
ABX3, Perovskite – Ilmenite Structure 725

Figure 10.9. (a) Tetragonal distortion of the perovskite structure by rotation of the oxygen octahedra about a
4-fold axis (open circle oxygen, filled circle octahedrally coordinates cation hatched circles A cation). (b) Trigonal
distortion of the perovskite structure by rotation of the oxygen octahedra about a 3-fold axis. Two levels of
octahedra are shown. Cations at the centres of the octahedra are not shown (Rao and Gopalakrishnan, 1986).

150 GPa are also shown, both when the octahedra are regular ðj ¼ 5Þ and when they show
the full distortion ðj ¼ 10Þ:

Modelling of stability: basis sets, “pinning” twin boundaries. Using the periodic
quantum mechanical ab initio Hartree – Fock programme, CRYSTAL, the relative stability
of MgSiO3 – ilmenite, MgSiO3 –perovskite and (periclase þ stishovite) assemblages as a
function of pressure has been investigated by D’Arco et al. (1994). In the investigated
pressures up to 60 GPa, the mineralogical assemblage of periclase þ stishovite has a higher
enthalpy.
Relatively small basis sets successfully reproduced the experimental geometries.
However, larger basis sets (“triple zeta” quantity, plus polarization d functions) are needed
to yield significant thermochemical results.
Complexities such as multiple cations at sites, non-stoichiometry, octahedral
tilting, cation off-centring, point defects, etc., can occur together in any combination.
Defects on a large scale lead to twinning, which manifests dramatic effects on the
electrical properties. Electrical fields can cause twin boundaries to shift and can affect
absorption of electrical energy and reduce the polarization (of ferroelectric perovskites).
Artificially introduced impurities in the lattice may cause pinning of the twin
boundaries and thereby prevent them from moving. Thus, in BaTiO3 perovskite,
which has a strong tendency for twinning, manganese is routinely added to pin the
726 Chapter 10

boundaries. (For pinning of dislocation by impurity ions in a vibrating string model, see
Mitra and Bhattacharaya, 1982.)

10.1.7. Perovskite melting and bouyancy

In the early Earth, perovskite melting must have figured prominently as melting was
the most important mechanism for chemical differentiation. In principle, the solidus of
silicate perovskite provides bounds on the density of the liquid at high pressures. In the
interpretation of seismological observations and in the evaluation of geodynamic
buoyancy forces throughout the deep mantle, the thermal EOS of Mg –Fe silicate
perovskite is of special significance.
The melting curve of (Mg,Fe)SiO3 to 60 GPa (Kato and Kumazawa, 1985; Heinz
and Jeanloz, 1987) shows that, above ,30 GPa (Tm , 3; 000 K), the melting is mostly
independent of pressure (Fig. 10.10). However, there could be a weak increase in Tm with
pressure above 50 GPa (Knittle and Jeanloz, 1989). Ito and Katsura (1992) have observed
eutectic melting between magnesio-wüstite and perovskite and infer that the eutectic
composition shifts towards pyroxene composition with pressure and that an incongruent
melting occurs at deep-mantle pressures.
A decrease in the melting slope with pressure indicates that the volume change on
fusion approaches zero. The molecular dynamics calculations by Matsui and Price (1991) go
against any natural buoyancy of MgSiO3 melts. However, some others argue that bouyancy
does happen in the lower mantle and a shock-wave study on komatiite liquid by Miller et al.
(1991a,b) to 36 GPa suggests that liquidus perovskite is naturally bouyant at 70 GPa.

Figure 10.10. Melting curve for (Mg, Fe)SiO3. The solid line (HJ) is the experimental result of Heinz and Jeanloz
(1987) for (Mg0.9Fe0.1)SiO3 perovskite obtained to 60(^5) GPa, and the bold line at higher pressures (KJ) is the
result of Knittle and Jeanloz (1989a) determined experimentally to 96(^10) GPa. The error bars give a measure of
the estimated uncertainties in the measurements at the highest pressures (see original references for details). The
dashed curves are calculated results for MgSiO3 perovskite ising a variety of methods: O, Ohtani (1983); P, Poirier
(1989); SB, Stixrude and Bukowinski (1990); MP, Matsui and Price (1991). The lower pressure phase relations for
MgSiO3 composition are also shown (Kato and Kumazawa, 1985) (from Matsui and Price, 1991, q 1991 Nature
Publishing Group).
ABX3, Perovskite – Ilmenite Structure 727

10.2. MgO – (FeO) – SiO2 system: perovskites

The MgO – SiO2 system under P greater than 20 GPa (in DAC) manifests two
transformations:
MgSiO3 ðilmeniteÞ ! MgSiO3 ðperovskiteÞ ð10-4Þ
and
Mg2 SiO4 ðspinelÞ ! MgSiO3 ðperovskiteÞ þ MgO ðpericlaseÞ ð10-5Þ
These transformations may be chiefly responsible for the 660 km discontinuity in
the mantle. Thermodynamic data indicate a small free energy difference between the
silicate perovskite and its breakdown oxide: MgSiO3 ¼ MgO þ SiO2.
High-temperature and high-pressure phase-equilibrium studies in the system
MgO – FeO –SiO2 have shown that (Mg,Fe)SiO3 perovskite is the only stable Fe-bearing
silicate phase in the lower mantle (e.g., Liu and Bassett, 1986). Therefore, to understand
the electrical conductivity, heat transfer and such other properties of the lower mantle, a
knowledge of the crystal chemistry of iron in (Mg,Fe)SiO3 perovskite is essential. A
chemical composition model of Earth’s lower mantle, based on comparison of the
measured physical properties of the lower mantle minerals with the seismic data, has to be
consistent with the phase-equilibrium relations in the ternary system MgO – FeO – SiO2.
Based on the phase equilibrium, the mantle perovskites are assumed to have compositions
near Mg0.9Fe0.1SiO3 (e.g., Heinz and Jeanloz, 1987; Knittle and Jeanloz, 1987).
Kesson et al. (1995) showed that single-phase perovskite can be synthesized from a
starting material of composition almandine75 pyrope25. The synthesis of a single crystal of
MgSiO3 perovskite to as large a size as 300 mm in diameter by Ito and Widner (1986) has
helped to determine a host of parameters, such as: crystal structure at RT and at high
pressure, elastic moduli (Yeganeh-Haeri et al., 1989), thermal expansion (Ross and Hazen,
1989) and vibrational spectra (Hemley et al., 1989).
The structure of MgSiO3 perovskite projected on (001) is illustrated in Fig. 10.11.
Under pressure, the ratio of effective ionic radii, rA þ rO =ðrB þ rO Þ; would increase
because an oxygen anion is more compressible than cations. On that basis, Yagi et al.
(1978) suggested that orthorhombic MgSiO3 perovskite should tend toward cubic
symmetry with increasing pressure. O’Keeffe et al. (1979), on the other hand, argued that
the ratio ðrA þ rO Þ=ðrB þ rO Þ would decrease with pressure and would give rise to larger
tilting of SiO6 octahedra and increased distortion with pressure. They derived a
relationship between the angle of rotation of the SiO6 octahedron about the 3-fold axis
in terms of the relative compressibilities of the Mg – O and Si – O bonds.

10.2.1. Tolerence factor, t

Kudoh et al. (1987) observed that pressure moves the coordination of Mg towards
8-fold rather than 12-fold coordination. This predicts a general increase in the degree
of distortion with compression. Above 5 GPa, the SiO6 octahedra are observed to tilt
in addition to the compression of the Si – O and Mg –O bonds (Ross and Hazen, 1990).
728 Chapter 10

Figure 10.11. Room-pressure, room-temperature structure of MgSiO3 perovskite projected on (001); Mg sites are
ruled, Si sites are solid, O(1) sites are open and O(2) sites are doubled. Each atom is numbered with lower-case
Roman numerals (Ross and Hazen, 1990, q 1990 Springer-Verlag).

The tiltings and distortions in GdFeO3-type perovskites such as MgSiO3 can be described
in terms of the observed tolerance factor, tobs ; as (Sasaki et al., 1983):

kA – Ol
tobs ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð10-6Þ
2kB – Ol

where kA – Ol and kB–Ol are the mean inter-atomic distances with 12 and 6 coordination
for A (e.g., Mg) and B (e.g., Si) sites, respectively.
For an ideal cubic perovskite, tobs ¼ 1:0 and the degree of distortion from ideal
symmetry in MgSiO3 perovskite increases as a function of pressure (Hemley et al., 1987).
Above 5 GPa, the degree of distortion within the polyhedra also changes under
compression.
With increasing pressure, the observed tolerance factor (tobs) and the tilt in the [ab ]-
plane decrease, while the tilt in the [bc ]-plane and the bond-length distortion of the MgO12
site increase.
ABX3, Perovskite – Ilmenite Structure 729

10.2.2. Silicate perovskites

As stated earlier, silicate perovskite (plus magnesio-wüstite) is considered to be the

most abundant mineral in the lower mantle, accounting for more than half of the total
volume of the Earth. In particular, (Mg,Fe) SiO3 perovskite is regarded as the most
abundant mineral in the Earth as it is believed to constitute 80% or more of the Earth’s
lower mantle. This crystallizes in GdFeO3-type perovskite structure until its Fe/(Mg þ Fe)
content reaches ,20% but, at higher iron content, this breaks down to a mixture of
perovskite þ (Mg,Fe)O þ SiO2.
With increasing temperature, transitions to higher symmetry forms occur in
(Mg,Fe)SiO3 perovskite. If perovskite is cubic under lower-mantle P – T conditions, its
bulk modulus is expected to increase as a result of the loss of the degrees of freedom for
compression. But the bulk modulus could also increase by ,5% while going from
orthorhombic to cubic form (Cohen et al., 1989). As the transition is approached, the anti-
ferroelastic transitions, which may be soft-mode-driven, give rise to unusual shear
softening.
Results of extensive studies using XRD and 29Si MAS NMR spectroscopy
(Kirkpatrick et al., 1991) have shown that MgSiO3 perovskite has an orthorhombic
structure (space group Pbnm) with Si in very symmetrical octahedra and Mg in distorted
8- to 12-fold coordination polyhedra. Single-crystal XRD studies of orthorhombic
(Pbnm) MgSiO3 perovskite revealed that the SiO6 octahedra are nearly ideal size
and Mg generally occupies a 12-coordinated site (MgO12 polyhedron), albeit with
deviations from the central position. This causes the Mg – O distances and O –Mg –O
angles to range widely from 2.01 to 3.12 Å and from 50.08 to 70.98, respectively.
Effectively, the coordination of Mg varies from 8 to 12 and the point symmetry
becomes Cs :

10.2.2.1. Orthorhombic –tetragonal –cubic transitions

On the basis of high-temperature and -pressure experiments on MgSiO3, Wang et al.
(1991) concluded that Pbnm orthorhombic perovskite may not be the sole representative
phase in the Earth. Bukowinski and Wolf (1988) predicted that in the mantle, with
increasing temperature, the transitions orthorhombic ! tetragonal ! cubic may occur.
Recent results (e.g., Fiquet et al., 1998) indicate that the orthorhombic structure is stable to
at least 95 GPa at ,25,000 K. The EOS of three different polytypes of MgSiO3 perovskite
determined by LAPW in the LDA approximation (Stixrude and Cohen, 1993) is shown in
Fig. 10.12. The Pbnm structure is found to be most stable in the pressure regime of the
Earth’s mantle.
Ringwood (1966) first noted the possible importance of MgSiO3 and CaSiO3
perovskite in lower mantle petrology. Calculations have focussed on the elasticity and
EOS of these two perovskites. Lattice-dynamical calculations using empirical potentials
provide estimates of elastic properties as well as defect energies (Watt et al., 1976) and ion
migration paths (Miyamota, 1988). Models based on ab initio potentials (non-empirical,
i.e., assuming inter-atomic potentials independent of experimental data) have been
730 Chapter 10

Figure 10.12. The equation of state of three different polytypes of MgSiO3 perovskite as determined by LAPW
calculations in the LDA approximation (Stixrude and Cohen, 1993). The Pbnm structure is found to be most stable
throughout the pressure regime of the earth’s mantle. This is in agreement with experimental observations, which
so far have found no reproducible observations of other stable aristotypes. Circles and squares are experimental
data from Knittle and Jeanloz (1987) and Mao et al. (1991), respectively (Source: Stixrude et al., 1998, q 1998
Mineralogical Society of America).

successfully applied to silicate perovskites (Wolf and Bukowinski, 1987; Cohen, 1987;
Hemley et al., 1989).
As a function of temperature and pressure, the stability and thermodynamic
properties of CaSiO3 and MgSiO3 perovskites have been discussed by Wolf and
Bukowinski (1987). Theoretical properties are derived from quasi-harmonic lattice
dynamics employing parameter-free pair potentials derived from the MEG formulation.
Cubic CaSiO3 perovskite structure is dynamically stable at zero pressure and to pressures
of at least 70 GPa. At higher pressures, the static potential-energy surface of CaSiO3
exhibits minima that correspond to a distorted perovskite structure (see Section 10.4.1)
with rotated octahedra (like MgSiO3).

10.2.2.2. Ferroelectricity
The occurrence of ferroelectricity in perovskites leads one to suggest that silicate
perovskites may transform to ferroelectric phases under lower-mantle conditions
(Anderson, 1988). In BaTiO3, the Ti d-orbital hybridization is clearly responsible for
the ferroelectric behaviour (Cohen and Krakauer, 1990) but this explanation may not be
strictly valid for the silicate perovskites.
ABX3, Perovskite – Ilmenite Structure 731

10.2.2.3. MgSiO3 –FeSiO3 perovskites

The solubility of the FeSiO3 component into perovskite is limited to the
composition (Mg0.8Fe0.2)SiO3. With iron content higher than (Mg0.8Fe0.2)SiO3, an
assemblage of magnesio-wüstite (Mg2Fe0.76)O plus stishovite co-exists with perovskite
(Yagi et al., 1979). The solubility of the FeSiO3 component into ilmenite and perovskite is
found to be limited to nearly 10 mol%.
At pressures above ,25 GPa, the system MgSiO3 – FeSiO3 is divided into
three stability fields according to FeSiO3 content (XFe): (i) perovskite for 0 , XFe , 0.1,
(ii) perovskite (XFe , 0.1) þ (Mg05Fe 0.5)O þ stishovite for 0.1 , XFe , 0.5 and
(iii) (Mg,Fe)O þ stishovite for 0.5 , XFe , 1 (Ito and Yamda, 1982). This suggests
that some amount of oxide mixture of stishovite and relatively iron-rich Mg-wüstite
would be present with Mg –perovskite in the lower mantle.
For a bulk olivine composition with XFe ¼ 0.1 at a pressure of 23.5 GPa, there
occurs the sharp phase transition: spinel ¼ perovskite (XFe ¼ 0.02) þ magnesio-wüstite
(XFe ¼ 0.02). This corresponds to the sharp discontinuity at ,660 km. The dissociation of
spinel is completed within a very small depth interval; this would make the discontinuity
very sharp, especially in the upper portion. The dissociation of majorite (tetragonal garnet),
on the other hand, proceeds at a fairly large depth range, ,800 km. These features go in
agreement with seismic models in which the 660 km discontinuity is large and sharp but
there is still a gradient of velocities to a depth of 750 to 800 km.
Geophysicists have noted that the 660 km mantle seismic discontinuity coincides
with a perovskite phase-transition boundary and that (Mg,Fe)SiO3 perovskite is a
dominant lower-mantle mineral (e.g., Yagi et al., 1979; Liu, 1979). Orthorhombic
MgSiO3 perovskite has been investigated by many workers for its room-temperature
compressibility (e.g., Yagi et al., 1992), thermal conductivity (Knittle et al., 1986),
electrical conductivity (Li and Jeanloz, 1987) and vibrational spectroscopy (e.g.,
Hofmeister et al., 1987).

10.2.2.4. Shear moduli: ultrasonic interferometry

Theoretical and experimental studies of ambient elastic moduli reveal that MgSiO3
perovskite has an anomalously high shear modulus, G. A large temperature derivative of G
is required to satisfy seismic velocities in the lower mantle with reasonable petrological
models (e.g., Wang and Weidner, 1996).
For measurements of the shear elastic properties of MgSiO3 perovskite, Sinelnikov
et al. (1998) have adapted ultrasonic interferometry techniques to a DIA-type cubic anvil,
high-pressure apparatus (SAM-85) installed on the superconducting wiggler beam line
(X1781) at the National Synchrotron Light Source of the Brookhaven National Laboratory
(Chen et al., 1998). In this ultrasonic interferometry, a burst of high-frequency signal is
applied to the piezoelectric transducer. The resultant elastic wave propagating inside the
sample produces a series of echoes. The travel time of the elastic wave through the sample
is measured by observing constructive and destructive interferences among overlapped
echoes as a function of carrier frequency.
At each pressure and temperature, travel times are measured using energy-dispersive
XRD patterns for the sample and from the NaCl-confining media (pressure standards).
732 Chapter 10

TABLE 10.1
MgSiO3 perovskite density and shear elasticity

Source r (103/kg m3) G (GPa) ldG/dPlT 2 (dG/dT)P (1022 GPa/K)

Sinelnikov et al. (1998) ,4.06(2) 175(12) 1.8(4) 2.9(3)

Different workers 4.107a 177(4)b 1.6 –2.2c 2.0 –3.5d
a
MgSiO3 X-ray density from Joint Commission on Powder Diffraction Standard Card 341216.
b
MgSiO3 single-crystal Brillouin-scattering measurements (Yeeganeh-Haeri, 1994).
c
Elasticity systematics of dG/dP vs Ks/G for MgSiO3 perovskite (Wang et al., 1994).
d
Temperature-average dG/dT values of the lower mantle (Wang and Weidner, 1996).

Assuming hydrostatic compression, the travel times are converted to shear velocities and
shear moduli with the EOS coefficients from Wang et al. (1994). The density r; G and the
derivatives of G, determined from ultrasonic measurements of polycrystalline samples of
MgSiO3 perovskite, are shown in Table 10.1.
The acoustic measurements by Sinelnikov et al. (1998) produced the pressure and
temperature derivative of the shear modulus (G): ðdG=dPÞT ¼ 1:8 ^ 0:4 and
ðdG=dTÞP ¼ 22:9 ^ 0:3 £ 1022 GPa/K. The former value is found to be similar to
that of ScAlO3 perovskite (Kung and Rigden, 1997), which is mostly an analogue of
MgSiO3 perovskite. However, Jackson (1998) found the average ðdG=dTÞP value as
2 £ 1022 – 2.5 £ 1022 GPa/K. It may extend to the value of 3.5 £ 1022 GPa/K (Zhao and
Anderson, 1994).

10.2.2.5. Vibrational modes

In MgSiO3 perovskite (Pbnm), there are four formula units per unit cell, giving a
total of 60 vibrations at any one point in the Brillouin zone (see also Section “Vibrational
spectroscopy” in Section 4.4.3). At the Brillouin zone centre, these modes are distributed
among the following symmetry species (Vide eqs. 4 – 14):

7Ag ðRÞ þ 7Bg ðRÞ þ 5B2g ðRÞ þ 5B2g ðRÞ þ 8Au þ 8B1u ðIRÞ þ 10B2u ðIRÞ þ 10B3u ðIRÞ

Thus, there are three acoustic modes (B1u ; B2u and B3u symmetry), 24 Raman modes
and 25 infrared modes. Of the 24 Raman modes, only 11 modes have been identified in
MgSiO3 perovskite.
High-pressure Raman spectra of MgSiO3 perovskite reveal that there is a profound
change in the compression mechanism at about 40 GPa. Under pressure, the dominant
peaks disappear or new peaks appear to dominate and other modes change quite
dramatically, often decreasing to less than half their original values. This type of behaviour
is noted in all MgSiO3 phases.

Aluminous perovskite. It has been demonstrated that a fair amount of

aluminium can be dissolved in silicate perovskites (Liu, 1976) as a solid solution of
MgSiO3 –Al2O3.
ABX3, Perovskite – Ilmenite Structure 733

Figure 10.13. Selected Raman spectra of MgSiO3. Ten percentage Al2O3 –perovskite as a function of
pressure at 258C. The spectra were obtained using 514.5 nm excitation at 20 mW power (Liu et al., 1995, q 1995
Springer-Verlag).

According to a review article by Liu (1989), the substitution of Fe2þ for Mg2þ or
2Al for (Mg2þ þ Si4þ) in MgSiO3 – perovskite increases the three lattice parameters
3þ

(a, b and c). The pyrolite mantle model would suggest that the Earth’s lower mantle would
contain ,10% (by wt% or mol%) of Al2O3. Liu et al. (1995) synthesized MgSiO3 10%
Al2O3 – perovskite and investigated up to ,27 GPa (1088C/425 K) using Raman
spectroscopy (Fig. 10.13). Liu et al. (1995) observed that MgSiO3 10% Al2O3 – perovskite
is slightly more compressible than MgSiO3 – perovskite and that the volume thermal
expansion of the former is slightly greater than that of the latter.

Raman spectra. The laser Raman spectra at ambient and various pressure and
temperature conditions were recorded from 150 to 650 cm21 on a Microdial-28 Raman
microprobe using a spectral band pass of ,3 cm21. All spectra were obtained with the
514.5 nm line from a Spectra Physics model 2020 argon ion laser using 12 – 30 mW
734 Chapter 10

power at the sample. The Raman spectra were recorded using a 50 £ ULWD Olympus
microscope objective and 15 accumulations with 30 s integration time and those in the
high-pressure runs were recorded with a 32 £ UM Leitz microscope objective and 20
accumulations at 20 s integration time. The focused laser spot was ,2 –3 mm in
diameter. Wave numbers, as determined from plasma emission lines, were accurate to
^1 cm21. The details of high-pressure Raman experiments are available from Liu and
Mernagh (1993). For orthorhombically distorted perovskite structure, a total of 24
Raman modes are predicted but, for MgSiO3 – perovskite, only four Raman bands are
commonly observed. The most intense band occurs around 380 cm21 and this mode may
be similar to the Eg symmetric stretching vibration of isolated octahedron but may
also be strongly coupled to the F2g deformational vibrations. The low-frequency bands
at 281 and 252 cm21 are almost identical to those for MgSiO3 – perovskite and are
possibly due to Mg2þ translations coupled to octahedral deformations and rotational
motions. Like MgSiO 3 – perovskite, the Raman frequencies of MgSiO 3 –
perovskite increase non-linearly with increasing pressure (Fig. 10.13) and decrease
linearly with increasing temperature within the experimental uncertainties and the range
investigated.
A band broadening in this aluminous perovskite is observed, which can be
explained as a result of increasing disorder through substitution of 3Al3þ for
(Mg2þ þ Si4þ) or, as suggested by Chopelas and Boehler (1992), an increase in defect
sites in the lattice.
The Al –O vibrations produce weaker Raman intensities, as observed by McMillian
and Piriou (1982) in corundum and glasses (silica vibrations).

10.2.3. Elasticity: modelling

Orthorhombic (Mg,Fe)SiO3 perovskite is elastically anisotropic with the b-axis

having the smallest linear compressibility, bb ðba < bc Þ: It becomes elastically
anisotropic with pressure (Meade and Jeanloz, 1991). Indeed, the structural changes are
largely controlled by volumetric terms (e.g., Hemley et al., 1987).
Theoretical and computational methods have been used for an accurate prediction
of the density, EOS and structural properties of the crystal from a single unified
model. Calculations on first-principles method predict that the M- and R-point instabilities
would lead to the Pbnm structure for MgSiO3 (Hemley et al., 1985, 1987; Wolf and
Jeanloz, 1985; Wolf and Bukowinski, 1987). A MEG model has been employed
for examining the structural distortions at zero pressure and at high compression
(Hemley et al., 1987).
This model also showed good agreement with the bulk modulus determination
(Wolf and Jeanloz, 1985). The elastic constants of MgSiO3 – perovskite can be
theoretically derived from first principles at high pressure using the plane-wave pseudo-
potentials method. An ab initio MD technique allows the orthorhombic structure to be
efficiently optimized (Warren and Ackland, 1996; Karki et al., 1997b). Both plane-wave
pseudo-potential (Wentzcovitz et al., 1993) and linearized augment plane-wave
ABX3, Perovskite – Ilmenite Structure 735

calculations (Stixrude and Cohen, 1993) have shown that the orthorhombic phase is stable
throughout the pressure regime of the lower mantle.
Karki et al. (1997b) have reported the first-principles determination of the nine
stiffness coefficients of orthorhombic MgSiO3 –perovskite at lower mantle pressures up
to 140 GPa and discuss the geophysical implications of their results. Their results are
based on DFT (Kohn and Sams, 1965) within LDA and pseudo-potential theory (Payne
et al., 1992). Three orthorhombic strains and one shear (triclinic) strain are used to
calculate the nine elastic moduli C11, C22, C33, C44, C55, C66, C12, C13 and C23. The
magnitude of the strains is varied and the corresponding elastic constants are derived
from the resulting non-linear stress –strain relations. This method was also employed
for the determination of elastic constants of MgSiO– perovskite at zero pressure
(Wentzcovitch et al., 1995) and those of MgO and SiO2 at high pressure (Karki
et al., 1997a).
The relatively high values of the elastic moduli compared with other silicates reflect
the dense packing of the perovskite structure and the rigidity of the SiO6 octahedra, which
form a corner-sharing three-dimensional network. Covalent forces exert an important
influence on the elasticity of silicate perovskite.
Experimental Brillouin scattering data (Yeganch-Haeri, 1994) and pseudo-potential
results (Wentzcovitch et al., 1995) show that, at zero pressure, C22 . C33 . C11 :
However, Karki et al. (1997b) found that this relation ðC22 . C33 . C11 Þ is valid only
above 20 GPa, indicating that the a-axis is the most compressible at higher pressures. At
lower pressures, however, the c-axis is seen to be most compressible and the b-axis the
least (Mao et al., 1991; Karki et al., 1997) (Table 10.2). The elastic constants of MgSiO3 –
perovskite according to theory (Karki et al., 1997b) and experiment (Yeganeh-Haeri,
1994) are shown in Fig. 10.14a – c.

TABLE 10.2
Zero-pressure elastic moduli (GPa) of orthorhombic MgSiO3 –perovskite compared with calculations and
experiment

C11 C22 C33 C44 C55 C66 C12 C13 C23 K G

Calculations
Karki et al. 487 524 456 203 186 145 128 144 156 258 175
(1997b)
Wentzocovitch 496 560 504 151 198 171 132 136 156 267 179
et al. (1995)
Cohen (1987)a 548 551 441 241 253 139 54 153 175 256 196
Matsui et al. 460 506 378 162 159 112 139 184 177 260 140
(1987)
Experiment
Yeganeh-Haeri 482(4) 537(3) 485(5) 204(2) 186(2) 147(3) 144(6) 147(6) 146(7) 264(5) 177(4)
(1994)
a
Potential induced breathing (PIB) model.
736 Chapter 10

Figure 10.14. Elastic constants of MgSiO3 perovskite according to theory (Karki et al. 1997a) (lines) and
experiment (Yeganeh-Haeri 1994) (Source: Stixrude et al., 1998, q 1998 Mineralogical Society of America).
Symbols: C11 ; C12 ; C44 (W) C22 ; C13 ; C55 (K) C33 ; C23 ; C66 (A).

10.2.3.1. Wave velocities: anisotropy

The calculated single-crystal elastic wave velocities in different directions (Karki
et al., 1997a,c) show that the azimuthal anisotropy of shear waves (AS) is much stronger
than that of compressional (P) waves (AP) (Karki et al., 1997c). At 140 GPa, the P- and
S-wave velocities vary with the propagation direction by 12 and 21%, respectively.
The degree of elastic anisotropy of MgSiO3 – perovskite is found to be strongly
pressure dependent. The anisotropy first decreases with pressure and then increases,
showing corresponding changes in the propagation directions of the slowest compressional
(P) and shear (S) waves. The observed shear-wave splitting shows no detectable polarization
anisotropy throughout the bulk of the lower mantle (Meade et al., 1995). In the lower mantle,
the anisotropic single crystals of Mg – perovskites must be randomly oriented. However, this
isotropy is lost in the uppermost and lowermost parts of the lower mantle.
The maximum polarization anisotropy of the shear waves varies from 17% at zero
pressure to 14% at 140 GPa. The pressure dependence of the anisotropy is not monotonic:
the changes in pressure dependence (at ,20 and 40 GPa for AP and AS, respectively) are
ABX3, Perovskite – Ilmenite Structure 737

reflected in changes in the slowest propagation directions of the P- and S-waves beyond 20
and 40 GPa, respectively. At 0 GPa, the fastest and slowest longitudinal waves propagate
along [011] and [001], respectively, whereas the fastest and slowest shear waves both
propagate along [010] and are polarized in the (001) and (100) planes, respectively.
However, at 140 GPa, the longitudinal and shear waves are slowest along [100] and [110],
respectively, and the direction of maximum polarization S-wave anisotropy is [001] (Karki
et al., 1997c).
Comparison with the spherically averaged (radial) structure of the lower mantle
(Dziewonski and Anderson, 1981) with the theoretical P- and S-wave velocity curves of
Karki et al. (1997c) show parallelism. However, the theoretical P- and S-wave velocities
are, respectively, 6 and 8% higher than those observed seismologically. This difference can
be attained by the presence of Fe and high temperature (2,000 – 3,000 K) in the lower
mantle to cause a 1– 2% reduction of velocities. The presence of some other phases like
CaSiO3, magnesio-wüstite, etc., may also contribute to the rest of the difference.

10.2.4. Thermoelasticity and expansivity

Static compression studies are employed (when acoustic techniques are not
possible) to determine the isothermal bulk modulus, KT ; the pressure and temperature
derivatives of KT and thermal expansion, a: The values of a and lðdKT =dTÞP l for silicate
perovskites are found to be usually around 2 £ 1025 K21 and 2 £ 1022 GPa/K,
respectively (e.g., Fiquet et al., 1998). Values lower than these for a and lðdKT =dTÞP l
favour a chemically uniform mantle with a composition similar to pyrolite (Si/(Mg þ
Fe) ¼ 0.69) whereas analyses based on the higher values of a and lðdKT =dTÞP l favour a
hotter, heterogeneous lower mantle that is enriched in silica (Si/(Mg þ Fe) ,1). Higher
values for a and ðdKT =dTÞP have been reported by many (e.g., Stixrude et al., 1992).
For MgSiO3 perovskite, the values are (Jeanloz and Knittle, 1989):

a < 4 £ 1025 =KlðdK=dTÞP l < 5 £ 1022 GPa=K ð10-7Þ

Thermal expansivities are indeed very small (Chopelas and Boehler, 1992).
In modelling the thermoelastic properties, one needs to rely on the excitation
spectrum, harmonic and anharmonic vibrations. The contributing factors are defects.
Electronic and magnetic contributions, however, are known only approximately. A good
agreement between the finite- (static) and infinitesimal-compression (wave velocity)
measurement is important for the determination of K0 : However, the value of K0 is
determined independently from the Brillouin data.

10.2.4.1. Bulk modulus and EOS

The isothermal bulk modulus, KT ; is related to the adiabatic bulk modulus, KS ; by:

KS
KT ¼ ð10-8Þ
1 þ agT
738 Chapter 10

where a is the thermal expansion coefficient and g is the Grüneisen parameter. Since the
Brillouin spectroscopy experiment measures the elastic moduli directly, it can provide
more reliable results.
The EOS of MgSiO3 –perovskite has been experimentally measured at room
temperature under hydrostatic conditions to 8.5 GPa by Yagi et al. (1982). A fourth-order
Birch – Murnaghan EOS fitted to the theoretical points gives KOT ¼ 260 GPa and K 0OT ¼
4:01 assuming regular octahedra ðj ¼ 5Þ or KOT ¼ 258:2 GPa and K 0OT ¼ 4:05 for
distorted octahedra ðj ¼ 10Þ (Wolf and Bukowinski, 1987).
The EOS of low-pressure phases are strongly controlled by angle-bending forces
and compression of lower valent cation –oxygen polyhedra (because SiO4 tetrahedra are
relatively incompressible). In the cubic phases, the static lattice values of KOT are 274 GPa
for CaSiO3 and 821 GPa for MgSiO3 (Wolf and Jeanloz, 1985). If the octahedra are
allowed to shear, a reduction in the bulk modulus occurs (K0 ¼ 270:6 GPa for MgSiO3 –
perovskite).
The room-temperature compression of MgSiO3 –perovskite crystals show a linear
relationship with pressure and volume (Fig. 10.15; Ross and Hazen, 1990). The pressure –
volume data from the three crystals of Ross and Hazen (1990), when combined for a least-
square fit of a Birch –Murnaghan EOS, give the relation:

P ¼ 3=2KT ½ðV0 =Vp Þ7=3 2 ðV0 =Vp Þ5=3 ½1 2 3=4ð4 2 K 0T ½V0 =Vp Þ2=3 2 1 ð10-9Þ

where P; V0 ; Vp ; KT and K 0T are pressure, molar volume at room pressure, molar volume at
high pressure, bulk modulus and pressure derivative of the bulk modulus, respectively. By
fixing K 0T ¼ 4 and Vp =V0 ¼ 1:0 at room pressure, a value of 254 (13) GPa is obtained for
KT : Similar results are obtained by fitting a Murnaghan EOS to the data, but it should be

Figure 10.15. P – V relations of FexMg12xSiO3 perovskites at room temperature. Open squares ðx ¼ 0:1Þ; solids
circles ðx ¼ 0:2Þ; and triangles ðx ¼ 0:0Þ are measurements at 298 K. The solid curves are for 298 K isotherm
(From Ross and Hazen, 1990).
ABX3, Perovskite – Ilmenite Structure 739

remembered that, in the Birch –Murnaghan EOS (e.g., K 0T and zero-pressure volume),
there are many free variables.
MgSiO3 –perovskite undergoes a metastable-phase transition at 600 K and 7.1 GPa
(Wang et al., 1991a). From the measured P – V EOS of MgSiO3 and hydrostatic
compression, the zero-pressure bulk modulus ðKOT Þ was determined as 258(^20) GPa
(when dKOT =dP ; K 0OT ¼ 4 is assumed).
Hemley and Cohen (1996) listed the single-crystal elastic moduli (in GPa) for
MgSiO3 – perovskite at zero pressure using the theoretical PIB model and the
experimental Brillouin scattering spectroscopy. These results indicate that the value of
G under room T, P equals that of the lower mantle at 1,000 km depth (Dziewonski and
Anderson, 1981). If the lower mantle is perovskite-rich, the pressure and
temperature effects offset each other at that depth (40 GPa and 1,900 – 2,300 K). At
1,071 km depth, pressure G equals 255 GPa (at 300 K). These indicate that
dG=dT < 0:04 GPa/K is consistent with geophysical data when a perovskite-rich
lower mantle is assumed.

10.2.4.2. Lattice compressiblity and KT

The compressibilities of lattice parameters and the calculated isothermal bulk
modulus ðKT Þ by different workers on MgSiO3 perovskite are given in Table 10.3.
Ross and Hazen (1990) observed the compressibility of MgSiO3 perovskite to be
anisotropic, with b approximately 23% less compressible than a or c (which have similar
compressibilities). If an error of 0.1 GPa for pressure uncertainty is included, the axial
compressibilities for a, b and c are estimated as: 1.30(5) £ 1023, 1.04(4) £ 1023 and
1.24(4) £ 1023 GPa21, respectively. Thus, the compressibilities of a and c are not
significantly different.
The observed unit-cell compression gives a bulk modulus of 254 GPa using
a Birch – Murnaghan EOS with K 0 set equal to 4 and the V=V0 ratio at room
pressure equals 1 (see the earlier Section). Results from different workers confirm that
the b axis is the least compressible. Consequently, the axial ratios, a=b and c=b; increase
with pressure.

TABLE 10.3
Compressibilities and bulk modulus of MgSiO3

Compressibilities of Isothermal bulk Authors

lattice parameters modulus, KT (GPa)

a.b<c 258 Yagi et al. (1992)

a<b<c 266 Knittle and Jeanloz (1987)
c.a.b 247 Kudoh et al. (1997)
c.a.b 246 Yeganeh-Haeri et al. (1990)
c.a.b 272 Mao et al. (1989)
b,a<c 254 Ross and Hazen (1990)
740 Chapter 10

10.2.5. XRD results

Saxena et al. (1998) investigated the dissociation reaction, MgSiO3 (perovskite) !

MgO (periclase) þ SiO2 (HP phase) at 82(3) GPa by in situ heating to temperatures
between 300 and 1,780(50) K and XRD.
The orthorhombic perovskite changed to the pseudo-cubic phase between 1,280 and
1,485 K followed by a breakdown to the oxide mixture, which was observed to grow at
temperatures between 1,600 and 1,700 K. Orthorhombic perovskite was recovered upon
cooling at 1,140 K.
The XRD pattern of a synthesized perovskite was obtained by Saxena et al. (1998)
at 300 K and 82(3) GPa. The results of the experiment show that the orthorhombic
perovskite converted to a pseudo-cubic phase at 1,485 K, which is in accordance with the
data of Meade et al. (1995).
Meade et al. (1995) carried out successful XRD experiments for orthorhombic
silicate perovskite at 38(^2), 64(^3) and 70(^3) GPa. Silicate perovskite, (Mg,Fe)SiO3,
is seen to maintain its orthorhombic symmetry at 38 GPa and 1,850 K but, at pressures of
70 (^3) GPa and high temperature, (Mg,Fe)O mixed oxide and SiO2 stishovite are formed
at the expense of perovskite. The diffraction pattern consists of groups of lines that are split
in response to the distortion away from the ideal cubic perovskite structure (Glazer, 1975).
The triplet (020, 112 and 200 lines) is the strongest and most informative because it
constrains the unit-cell volume and the magnitude of the orthorhombic distortion. For a
cubic perovskite, these lines collapse to a single diffraction peak that is indexed to the 110
line of the high-symmetry phase.
At high temperatures, the diffraction lines shift to lower energies, indicating
expansion (more along the a and c directions than along b). There is a small decrease in
orthorhombic distortion (i.e., c=a ! 2 and a=b ! 1). At high temperatures, once the single
diffraction peak is formed, all the lines corresponding to the orthorhombic structure and the
tilting of SiO6 octahedra are seen to disappear. On cooling, however, these lines reappear
(Meade et al., 1995). The DV of the transition is seen to be negative.
However, these results are in contrast with the conclusions of LAPW calculations
(Stixrude and Cohen, 1993) and pseudo-potential calculations (Wentzcovitech et al.,
1993), which suggest that orthorhombic perovskite (Pbnm) is thermodynamically favoured
over cubic and tetragonal structures in the entire pressure range of the lower mantle.
However, it may be considered that the substitution and site occupancy of iron in
perovskite may alter the energetics of the distortion, resulting in the stabilization of the
high-symmetry phase.

10.2.6. (Mg,Fe)SiO3 – perovskite

The bulk modulus of (Mg,Fe)SiO3 – perovskite is now among the best determined
of all minerals. The bulk modulus at ambient conditions ðK0 Þ has been determined to
acceptable values by various methods, including Brillouin spectroscopy at zero pressure
and under both hydrostatic and non-hydrostatic compression to the 10 –100 GPa pressure
range (Mao et al., 1991; Weidner et al., 1993). Perovskites in the composition range of
ABX3, Perovskite – Ilmenite Structure 741

MgSiO3 to (Mg0.8 Fe0.2)SiO3 have been studied but no systematic effect of composition on
K0 values have been observed. It is even the case that neither the presence of trace elements
and additional mineral phases nor the possible occurrence of unusual partitioning of
crystal-site ordering in natural samples is likely to affect the observed results (O’Neill and
Jeanloz, 1993).
The bulk moduli of Mg12xFexSiO3 – perovskites with x ¼ 0; 0.1 and 0.2 have been
found independent of x (Mao et al., 1991). The EOS of (Mg0.9Fe0.1)SiO3 –perovskite was
determined at 100 GPa by XRD of laser-heated samples without a pressure medium
(Knittle and Jeanloz, 1987a). A third-order Birch– Murnaghan EOS fit to the data gave
zero-pressure parameters: KOT ¼ 266ð^6Þ GPa and K 0OT ¼ 3:9ð^4Þ: Mao et al. (1991)
obtained KOT ¼ 261ð^4Þ GPa using synchrotron radiation (to 30 GPa) for powder-
diffraction measurements. They also suggested that a relatively high value for K 0OT (,5.5)
must be assumed to reconcile the low bulk modulus obtained from Brillouin scattering
with the high-resolution diffraction data.
The pressure – volume relationship for (Mg,Fe)SiO3 perovskite at 300 K is now
determined (in parts in 103) throughout the pressure range of the lower mantle.

10.2.6.1. Iron in perovskite

Iron in perovskite plays an important role in determining the physical properties
(especially the electrical and thermal conduction) of the Earth’s mantle. When both Fe2þ
and Fe3þ are present, electrical conductivity increases several-fold because of electron
hopping. The inclusion of Fe may reduce the G value of perovskite in a manner similar to
the reduction of G in magnesio-wüstite (e.g., Wang and Weidner, 1996).
Using the EOS model and experimental data, Jeanloz and Knittle (1989)
determined the densities of perovskite and magnesio-wüstite at lower mantle conditions.
They concluded that, for the usually accepted geotherms, the upper mantle compositions
are incompatible with seismic data in the lower mantle, where an iron enrichment (,15%)
is required for an intrinsic density increase of 2.5% (and possibly even up to 5%) higher
than the upper mantle. Hemley et al. (1992) asserted the need for less iron at the top of the
lower mantle (with no thermal-boundary layer) when an amount of silica enrichment is
then required (Kuskov and Panferov, 1991). However, a change in chemical composition
leads to a thermal-boundary layer which may separate the two convecting systems into the
upper and the lower mantle.
Based on the measured bulk modulus and thermal expansion of (Mg,Fe)SiO3 –
perovskite, Jeanloz and Knittle (1989) proposed a lower mantle enriched in Fe. But
perovskite breaks down to oxides when the maximum solubility of Fe is reached (Hemley
et al., 2000). Jackson et al. (1987) synthesized (at 50 GPa/2,000 K) perovskites and
reported their extended X-ray absorption fine structure (EXAFS) measurements. Fe is in
6-fold coordination and, by inference, Si is in a 8– 12-coordination site (A site). However,
this was contradicted by a later synthesis (at 26 GPa/2,173 K) by Kudoh et al. (1990),
which indicated that Fe (with a small amount of Cr in the composition) substitutes for Mg
rather than for Si. Mg was observed to occupy the six-coordinated Si site (B site), thus
accounting for Si deficiency. The variation of the partition coefficient (K) of iron between
742 Chapter 10

Figure 10.16. Variation with pressure of the partition coefficient (K ) of iron between perovskite (filled square)
and magnesiowüstite. The error bars on the measurements are shown in the upper right-hand corner: (After Guyot
et al., 1989).

perovskite and magnesio-wustite is observed to show a dramatic change at 30 and 40 GPa

(Fig. 10.16, after Guyot et al., 1998).
The unit-cell parameters as a function of Fe content show that, with increase in Fe
content, the a-axis expands more than c (which expands more than b). This implies that the
distortion of the orthorhombic unit cell from the ideal cubic cell decreases with increasing
Fe content. Kudoh et al. (1990) also observed that the bond-length distortion, D; defined as
variance of the bond lengths around the coordination site (Sasaki et al., 1983; also from
equation (10-2)), increases with increasing iron (Table 10.4).

TABLE 10.4
Tilt angles (8), bond-length distortions and unit-cell distortions of (Mg12xFez)SiO3 perovskite (from Kudoh
et al., 1990)

Fe/Fe þ Mg 0.00a 0.04

Si–O(1)– Si 146.51(5) 145.6(7)

Si–O(2)– Si 147.04(5) 146.8(5)
O(2)–O(2) –O(2) 67.89(3) 67.5(5)
112.11(3) 112.5(5)
O(1)–O(2) –O(1) 67.89(3) 67.7(3)
110.77(3) 110.8(4)
Unit-cell distortion angle Fb 18.44 18.22
Bond-length distortion Dc
D A12 166.8 170.7
D A8 25.55 26.98
D A6 11.76 13.25
D B6 0.057 0.114
a
Horiuchi et al. (1987).
b
F ¼ cos21
P
ð2a2 =bcÞ (O’Keeffe et al., 1979).
c
D ¼ 1=n {ðri 2 rÞ=r}2 £ 103 (Sasaki et al., 1983).
ABX3, Perovskite – Ilmenite Structure 743

The unit-cell distortion, which describes the tilting of SiO6 octahedra (O’Keeffe
et al., 1979), decreases with increasing Fe content. In iron-bearing perovskites, the valence
band-site distribution of Fe is important for the characterization of defects and their effect
on transport properties such as electrical conductivity.
The Fe2þ cation occupies the A site where the mean M –O distance is 2.207 Å
(based on the eight shortest distances; Parise et al., 1990). Besides its content of Fe2þ and
Fe3þ, the (Mg,Fe)SiO3 – perovskite contains a third electronic configuration of iron, caused
by rapid electron transfer between Fe2þ and Fe3þ. The charge transfer is expected to take
place across the shared face between the A and B sites. The Fe2þ – Fe3þ distances range
between ,2.79 and 3.24 Å and, within this range, charge transfer can take place (Burns,
1993).
Addition of Al to (Mg,Fe)SiO3 – perovskite stabilizes iron in the structure and Al3þ
can enter either the A or B site with similar ease. The presence of Al3þ might stablize Fe3þ.
This could have a dramatic effect on the amount of Fe3þ present at a given oxygen
fugacity.
As stated earlier, EXAFS study on synthesized perovskite (at 50 GPa/2,000 K)
revealed that some Fe substitutes Si in the octahedral site and some Fe in
8 –12-coordinated site (Jackson et al., 1987). It has been conclusively proved by
Mössbauer study that Fe occurs in the distorted 8 – 12-coordinated site as well
(McCammon et al., 1992; Fei et al., 1993). Using optical absorption spectroscopy,
Burns (1993) also predicted the occupancy of Fe2þ at the 8 –12-coordinated site.

10.2.6.2. Fe21 in perovskites: A-site occupancy

The crystal chemistry of Fe2þ in perovskites intrigues the Earth scientists for its
implication in the structural, spectroscopic and thermodynamic behaviour of perovskites.
In the MgO – FeO – SiO2 system, iron ion, believed to be in the form Fe2þ, preferentially
partitions into (Mg,Fe)O relative to the perovskite phase (Ito et al., 1984; Fei et al., 1991;
Kesson and Fitzgerald, 1991).
In (Mg,Fe)SiO3 –perovskite, the EXAFS data (Jackson et al., 1987) indicated Fe2þ
to be in the B (octahedral) site, suggesting that the 8-fold A site is unfavourable to Fe2þ
because of a higher crystal-field stabilization energy (Williams et al., 1989) but later
crystallographic studies revealed Fe2þ to be located at the A site (Kudoh et al., 1990). Fe2þ
can thus occupy the large distorted site in the perovskite structure.
The structure and stability of the 1 : 1 and 1 : 3 compounds also show that the
perovskite A site is not always unfavourable for Fe2þ. (Note: The tetrahedral and square-
planar A-site geometries in the 1 : 1 and 1 : 3 compounds are quite different from the eight-
coordinated A site in the GdFeO3-type structure.) The high-pressure unquenchable
perovskite phase of FeTiO3 does adopt the GdFeO3 structure and the cell parameters are
consistent with the Fe2þ being on the A site (Leinenweber et al., 1991). During
decompression, this structure reverts to LiNbO3 structure. The same behaviour is seen in
non-transition-metal compounds such as MgGeO3 (Leinenweber et al., 1994). The balance
of energetics of Fe2þ –Mg partitioning between silicate perovskite and magnesio-wüstite
phases may limit the extent of the silicate perovskite stability field. The energy transition in
744 Chapter 10

Fe2þ in the perovskite structure (Section 10.2.6.3) and the temperature-induced valence
transition for Fe2þ are discussed in later sections.

Fe21 in a crystalline field: 8– 12-coordination sites. The effective charge values of

oxygen range approximately from 2 1.2 to 2 1.9 for many silicates (Sasaki et al., 1982). In
optical spectroscopy, the strong near-infrared band at 7,000 cm21 of (Mg,Fe)SiO3
perovskite is assigned to a spin-allowed transition of Fe2þ at the 8 – 12-coordinated site.
This is consistent with the spin-allowed bands of Fe2þ in eight-coordinated sites in
ferromagnesian silicates such as garnets (D2 site). The calculated spin-allowed transitions
for Fe2þ at different sites in perovskites are shown in Table 10.5.
Fe2þ ion with 3d6 configuration in an octahedral coordination is subjected to the
crystal-field potential:
X
6
V¼ Vi ð10-10Þ
i¼1

When only the spin-allowed transitions are considered, the system can
be treated as a one-electron system. Then the one-electron crystal-field potential can be
written as (Shen et al., 1994):
X X k ðbÞ ðbÞ
Vi ¼ r gka Zka ðui ; fi Þ
k¼2:4 a;b
rffiffiffiffiffiffi 
pffiffiffiffiffiffiffiffiffi c pffiffiffiffiffiffiffiffiffi s
p 5 2 pffiffiffiffiffiffiffiffiffi c
¼ 2g20 Z20 þ 3g22 c Z þ 3g22 s Z e r 2 g40 Z40 þ 2 5g42 c Z
22 22 42
5
 pffiffiffiffi
pffiffiffiffiffiffiffiffiffi s pffiffiffiffiffiffiffiffiffiffiffiffi c pffiffiffiffiffiffiffiffiffiffiffiffi s p 2 2
s c s
þ 2 5g42 Z42 þ 35g44 Z44 þ 35g44 Z44 Þ e r ð10-11Þ
12
ðbÞ
where Zka ðui ; wi Þ denotes real spherical harmonic functions, ri ; ui ; wi are the
ðbÞ
coordinates of the d electrons in Fe2þ, e denotes the electron charge and gka are
geometric parameters that can be determined from crystal-structure data.

TABLE 10.5
Calculated spin-allowed transitions for Fe2þ at differentiates in perovskites

CNa (effective chargeb) 8–12-coordinated site Octahedral site

8 (21.53) 10 (21.25) 12 (21.25) 6 (20.63)
c
Transition 7,023 7,034 7,050 7,050
4,432 4,783 4,951 6,761
4,198 3,026 3,504 100
404 735 595 57
a
CN denotes coordination number. Coordinates were chosen according to the bond lengths.
b
Effective charge of oxygen ion. Pure ionic state means effective charge of 22.0.
c
Energy is in wave number (cm21). The transition order was arranged according to the energy values.
ABX3, Perovskite – Ilmenite Structure 745

ðbÞ
When the geometric parameters, gka (Griffith, 1961), are known from crystal-
structure data, the spin-allowed transitions of Fe2þ at different sites in perovskites can be
calculated based on the nearest-point charge model.
The CFSE of Fe2þ in perovskite with coordination number 8 was calculated as
3,332 cm21, which is smaller than that obtained from magnesio-wüstite (4,320 cm21;
Goto et al., 1980). Indeed, Burns (1993) predicted that the CFSE of Fe2þ at the 8 –12-
coordinated site should have a lower value than that at the octahedral site. Thus, iron is less
stable in silicate perovskite and will tend to partition to mineral phases that have sites with
higher CFSE, such as the octahedral site in magnesio-wüstite. At deep-mantle conditions,
there may be high spin ! low spin transition of Fe2þ and also structural phase transitions
(Burns, 1993). Such phase transitions would strongly affect the iron partitioning among the
lower-mantle phases as well as between the lower mantle and the metallic core.

Fe31 in perovskites: Al31 content. In the perovskite structure, Fe2þ occupies the 8 –12-
fold coordination sites and Fe3þ is probably ordered on the octahedrally coordinated sites.
An electron-hopping mechanism possibly occurs between the face-sharing octahedra–
polyhedra, which offer the shortest M –M distances.
McCammon (1997), on the basis of her experimental studies (synthesis and
Mössbauer spectroscopy) with 3.3% Al2O3 in (Mg,Fe)SiO3 perovskite, concluded that the
proportion of Fe3þ in the lower-mantle perovskite phase is probably much higher than was
believed by the previous workers.
When Al is present in the perovskite structure, Fe3þ is stabilized (similar to the
observation in clinopyroxenes; see Mitra, 1992; Mitra et al., (1991b)). The amount of Fe3þ
in the lower-mantle perovskite phase is probably a function of both Al2O3 content and
oxygen fugacity. Fe3þ preferentially occupies the octahedral site (Keppler et al., 1994;
Fei et al., 1994).
McCammon (1997) argued, on the basis of her Mössbauer spectroscopic studies,
that Fe3þ occupies both sites because (i) the centre shift corresponding to Fe3þ in the
Al –perovskite phase (0.46 mm s21) is significantly higher than the value expected for
octahedral Fe3þ (McCammon, 1996) and (ii) a simple calculation shows that the
substitution Mg2þ þ Si4þ ! Al3þ þ Fe3þ is likely to stabilize Fe3þ in the structure. In the
lower mantle, relative to the (Mg,Fe)O phase, Fe3þ is likely to be concentrated in the
perovskite phase. High-pressure experiments on (Mg,Fe)O indicate that the solubility of
Fe3þ decreases dramatically at high pressure (McCammon et al., 1996). However, Fe3þ
may be stabilized in the spinel phase rather than in the (Mg,Fe)O phase. The former form is
probably the product of a high-pressure phase transition in the system Fe3O4 –MgFe2O4.
Since the oxidation state of iron in the perovskite phase affects the electrostatic
charge balance and equilibrium defect concentration, the presence of Fe3þ is likely to
affect significantly the physical and chemical properties of the lower mantle. These are the
phase relations, transport properties, mechanical behaviour, trace-element partitioning and
concentration of chemical species, e.g., hydroxide ions (McCammon, 1997). Fe3þ in
perovskite and conductivity in the mantle are discussed in Section 15.1.2.3.
746 Chapter 10

Crystal field and charge transfer: 5Eg – 5T2g triplets. Both XRD (Kudoh et al., 1990) and
Mössbauer data (McCammon et al., 1992) showed that Fe2þ enters the dodecahedral site in
the perovskite structure, whereas Fe3þ substitutes for Si4þ in the octahedral site
(McCammon et al., 1994).
The 7,000 cm21 band observed (Shen et al., 1994) in the optical and near-infrared
spectrum is ascribed to the crystal-field transition of Fe2þ. In Fig. 10.17 (from Keppler
et al., 1988), the broad band around 15,000 cm21 is likely to be due to Fe2þ – Fe3þ IVCT
(Burns, 1993).
In an undistorted dodecahedral environment, only one spin-allowed crystal-field
band of Fe2þ from the transition 5 Eg !5 T2g would be observed (because the magnitude of
the crystal-field splitting in a dodecahedral site is only half that in an octahedral site).
Under otherwise equal conditions (Burns, 1993), this band should occur at a significantly
lower frequency than the bands of [6]Fe2þ in silicate minerals. Also, because of
the monoclinic site symmetry (m) of the dodecahedral site in perovskite (Kudoh et al.,
1990), the 5T2g state should be split into three components, giving rise to three discrete
absorption bands.
Keppler et al. (1994) determined the individual components of the 5 Eg !5 T2g
transition to be at 6,283, 7,086 and 7,886 cm21 (all are separated by about 800 cm21).
Accordingly, if the frequencies of the three observed transitions are labelled as y 1
(¼ 6,283 cm21), y 2 (¼ 7,086 cm21) and y 3 ( ¼ 7,886 cm21), the crystal-field splitting
is given by (White and Moore, 1972):
1 1
D¼ ðy þ y 2 þ y 3 Þ 2 d
3 1 2
This yields D ¼ 6; 835 cm21. When d ¼ 0; D becomes 7,085 cm21. Acknowl-
edging that CFSE equals 3=5D þ 1=2d; one obtains CFSE ¼ 4,350 cm 21 for
d ¼ 500 cm21 and D ¼ 4; 251 cm21 for d ¼ 0: The broad band around 15,000 cm21 has
a half-width characteristic of Fe2þ $ Fe3þ IVCT transitions (Mattson and Rossman, 1987)
and occurs in the frequency range predicted by Burns (1993). Since, in perovskite
structure, the dodecahedra and octahedra share faces, charge transfer can take place at
much lower energies, below 10,000 cm21 (Burns, 1993). The near IR bands for Fe2þ in
silicate perovskite in comparison with garnet and periclase are tabulated in Table 10.6.
The strong absorption over the entire IR and visible range in the optical spectrum
(Keppler et al., 1994) of (Mg,Fe)SiO3 – perovskite indicates that it would show poor
thermal conductivity at high temperatures where radiative transport dominates but,
because of thermal conduction through electron hopping (polaron conductivity), it will
manifest high electrical conductivity. This possibly explains the electrical semi-
conductivity in the lower mantle. Since perovskite with Fe3þ is in equilibrium with
metallic iron (McCammon et al., 1992), CT bands contribute to the optical absorption
throughout the mantle.

10.2.6.3. Temperature-dependent electron delocalization

The 57Fe Mössbauer spectra of Fe-bearing silicate perovskite show three resolvable
peaks. The two doublets with the largest quadrupole splittings are assigned to Fe2þ,
ABX3, Perovskite – Ilmenite Structure 747

Figure 10.17. Spectrum of polycrystalline Mg0.94Fe0.06SiO3 perovskite alter background correction and
deconvolution into Gaussian components. Measured data are shown as dots (Keppler et al., 1994, q 1994
Mineralogical Society of America).
748 Chapter 10

TABLE 10.6
Observed near-infrared bands for Fe2þ in silicate perovskite, periclase and garnet structures

Mineral Site Co-ordination number Bond length (Å) Band (cm21)

Silicate A 6 1.78–1.80a
Perovskite B 8 –12 2.01–3.11a 7,000b
Periclase 6 6 2.11c 11,600d
10,000d
Garnet D2 8 2.20–2.34c 7,800e
6,100e
4,500e
a b c d e
Ross and Hazen (1990); Burns (1993); Smyth and Bish (1988); Goto et al. (1980); White and
Moore (1972).

whereas the inner-fitted doublet is assigned to Fe3þ. But the isomer shift (IS) of the Fe3þ
doublet was seen to increase with increasing temperature (Fei et al., 1994) rather than
decrease, as expected for the second-order Doppler shift variation with temperature.
Again, of the two Fe2þ doublets, the area of the doublet with larger quadrupole splitting
(QS) increases with decreasing temperature and the area of the Fe2þ doublet with the
smaller splitting decreases with decreasing temperature. This intensity inversion may
imply a change in the positional order – disorder of Fe2þ in the structure as a function
of temperature.
In a four quadrupole split doublet fit (Fig. 18.1 (Fei et al., 1994)), the observed
changes in relative intensities and line-broadening with decreasing temperature are
explained as being caused by different NNN populations, which in effect cause a
distribution of hyperfine parameters. Indeed, fitting the broad Fe2þ absorption band with
multiple fixed half-width doublets showed the area ratios of Fe3þ/Fetot, Fenþ/Fetot and
Fe2þ/Fetot obtained from the multiple-doublet fits remained the same (as obtained from
four-doublet fits). Magnetic relaxation as observed at low temperature (,80 K) indicated
that the onset temperature of magnetic relaxation increases with increasing Fe content in
perovskite.
In the fits with four quadrupole or split doublets, the Fe3þ absorption in perovskite
samples ((Mg0.95Fe0.05)SiO3 and (Mg0.9 Fe0.1)SiO3) is characterized by almost constant
values of QS (0.91 –1.23 mm/s) and IS (0.34 – 0.43 mm/s) as a function of temperature.
These indicate that Fe3þ is ordered in a single crystallographic site.
The doublet shows IS and QS values intermediate between those for Fe2þ and Fe3þ
with increasing temperature. The relative intensity of this doublet increases systematically,
indicating the presence of a mixed-valence state arising from a thermally activated
electron-delocalization process. An increase in intensity of this doublet corresponds to a
decrease in the cumulative intensity of Fe2þ and Fe3þ absorption bands.

10.2.6.4. Defect equilibria and M31: physical properties

The defects and trivalent ionic substitutions in perovskites are seen to control
several physical properties, as enumerated below.
ABX3, Perovskite – Ilmenite Structure 749

(a) Only 5% Al3þ substitution (along with oxygen vacancies) softens the elastic
properties, thermal expansivity and Anderson– Grüneisen parameter of MgSiO3 perovskite
(Zhang and Weidner, 1998). When 2.89% Al2O3 is dissolved in a perovskite with a small
iron content, the electrical conductivity increases by a factor of 3.5, as does the ferric iron
concentration (Xu et al., 1998). This can be accounted for by considering the presence of
the ionic substitution of ferric iron and aluminium and excess magnesium and oxygen
vacancies. (Note: Al2O3 is seen to dissolve in Mg – perovskite up to 25 mol%; Weng et al.,
1982.)
Kato et al. (1988) suggested that under lower mantle pressure (,25 GPa) Al
possibly replaces Si6þ at the octahedral site in MgSiO3 – perovskite with the simultaneous
accompaniment of one oxygen defect for every two Al atoms substituted.
(b) In synthetic perovskite, defect equilibria can be dominated by the substitutions:

2B4þ 3þ
B ¼ 2MB þ VO ð10-12Þ
VO þ O22
O þ H2 O ¼ 2OH2
O ð10-13Þ
where the subscript B denotes sites occupied by tetravalent B atoms in the original
structure and subscript O denotes oxygen sites. The first equation creates oxygen vacancies
and the second fills them. Hydration according to reaction (10-13) is highly exothermic by
100 –170 kJ/mol of water.

10.2.7. Glassy phase

MgSiO3 –perovskite transforms fairly easily to glassy phases at low pressures

ðP ! 0Þ on being subjected to heating, laser illumination and grinding or polishing (Wolf
et al., 1990). Preliminary Raman spectra as a function of temperature to the point of
vitrification (at zero pressure) has been reported by Durben and Wolf (1991). The
frequencies ðnÞ decrease with temperature but no anomalously soft-mode behaviour
(defined as n ! 0) has yet been observed.

10.3. Transformations

10.3.1. Activation energy

Olivine, pyroxene and garnet transform to perovskite-structured mineral at

P . 22 GPa and T . 1,700 K (Liu, 1979). The activation enthalpy and (internal) energy
are identical at P ¼ 0: Knittle and Jeanloz (1989) did not distinguish between the
activation enthalpy and Gibbs free energy. The effect of P is to change the relative Gibbs
free energies and hence to change the stability between enstatite and perovskite.
Knittle and Jeanloz (1989) have measured the activation energy for the back
transformation of (Mg0.9Fe0.1)SiO3 in the perovskite structure to enstatite structure at zero
P. The activation energy is 70(^20) kJ/mol, which is small in comparison with the values
for creep, diffusion and phase transitions in many other silicates. They suggested that the
transition may be activated by a movement of central Mg ion for the corresponding
750 Chapter 10

vacancy through the four oxygen ions of the dodecahedral face in the perovskite structure.
From a simple harmonic model of the amplitude of vibration of Mg –O bonds at elevated
T; this mobility of the Mg ions is expected to occur with an activation energy of 60 kJ/mol
and an activation volume near 1 cm21/mol.
All these imply that any silicate perovskite placed in the Earth’s surface in a
xenolith would revert fairly easily to enstatite in a geologically brief time (3 – 100 yrs).
Therefore, the chance of finding this high-pressure phase as a natural sample on the Earth’s
surface is indeed very remote.

10.3.2. Perovskite breakdown: volume change

In the breakdown relation of MgSiO3 ! MgO þ SiO2, the DV value is positive at

all pressures. DV increases from ,1 to 4% of the perovskite volume at 25 –120 GPa,
respectively. However, high temperature or the presence of some impurities may cause an
inversion of the sign (i.e., 2ve). Experimental data are, however, inadequate to prove this.
The recent EOS data of MgSiO3 show (Fiquet et al., 1998) the following:
V ¼ 162:6 Å3, K0 ¼ 256 GPa, K 0 ¼ 4; and those of MgO are (Fiquet et al., 1996):
V ¼ 74.70 Å3, K0 ¼ 167 GPa, K 0 ¼ 4:5:
It can be derived that the density of silicate perovskite is higher than the mixture
of its oxides down to CMB depth. Therefore, when excess of magnesio-wüstite is
present in the lower mantle, the presence of free silica is less probable at the CMB
(Andrault et al., 1998).

10.3.3. Vibrational models: intrinsic anharmonic effects

In the thermodynamic modelling of high-pressure phase relations in the MgO – SiO2

system (Fei et al., 1990), vibrational models are commonly used to extrapolate the low-
temperature thermodynamical data obtained from high-pressure phases to the temperature
conditions pertinent to the Earth’s mantle. These models always assume the quasi-
harmonic approximation — hence the Dulong and Petit limit for the heat capacity at high
temperature. Combined calorimetric and vibrational studies of olivine have shown that the
approximation fails at high temperature because of the occurrence of intrinsic anharmonic
effects (Fillet et al., 1991; Fiquet et al., 1992). These effects can be included in vibrational
modelling provided that high-temperature, high-pressure Raman spectroscopic data are
available. An evaluation of intrinsic anharmonic effects in high-pressure phases is thus
needed to refine the vibrational modelling of these high-temperature thermochemical
properties.

10.3.4. Raman study

Perovskite studies by vibrational Raman scattering become very relevant. Cubic

perovskite shows no Raman-active mode and the activity of the first-order Raman spectrum
is strongly dependent on the distortions of the crystal from its high-symmetry cubic form.
This is because all Raman-active vibrations in the distorted (e.g., orthorhombic)
ABX3, Perovskite – Ilmenite Structure 751

perovskite structures are derived from modes that occur at the edge of the Brillouin zone in
the cubic form. Certain of these modes are associated with displacive transitions
between low- and high-symmetry perovskite structures. These may be the soft modes
whose frequencies are seen to decrease to vanishing point with pressure or temperature at
the transition.
There are 20 atoms in the primitive unit cell in orthorhombic (Pbnm) perovskite.
There are, therefore, three translational modes and 60 zone centre modes (see Section
10.2.2.5). The optic modes decompose as 7Ag þ 7B1g þ 5B2g þ 5B3g þ 8Au þ 7B1u þ
9B2u þ 9B3u : There are 24 Raman-active modes (Ag ; B1g ; B2g and B3g ) and 25 infrared-
active modes (B1u ; B2u and B3u ).
In MgSiO3 –perovskite, a decrease in intensity and a small change in pressure shifts
of the infrared and Raman modes at 37 GPa offer evidence for phase transition (Chopelas
and Boehler, 1992) but XRD study could hardly establish this phase transition. A strong
coupling of the modes of two polyhedral vibrations occurs in the perovskite structure,
associated with the face-sharing SiO6 octahedra and MgO12 distorted dodecahedra.
Possibly, this corresponds to the strongest Raman bands and to Ag symmetry.

10.3.4.1. Soft-mode transition

High-temperature phase transitions of distorted perovskite would involve soft-
mode behaviour (with some mode frequencies vanishing) and hence Raman spectroscopy
is usefully employed for studying this vibrational dynamics. A soft-mode transition from
an orthorhombic phase to a tetragonal phase would involve a probable softening of c11 – c12
and the transition from the orthorhombic phase to a monoclinic phase would involve
softening of a shear mode (e.g., c66 ). Also, against pressure and/or temperature, the average
shear modulus may change non-monotonically (Hemley and Cohen, 1992).
In Pbnm perovskite, it is important to know which modes correspond to the R25 and
M2 modes in cubic perovskite and whether these modes would go soft at phase transitions.
It can be seen that only Ag symmetry modes are compatible with tilt distortions that can
lead to continuous phase transitions at high temperatures. There are two modes compatible
with M2 symmetry and two Ag modes compatible with R25 symmetry.
In general, a soft-mode behaviour is not required for a second-order
continuous phase change; for example, transitions in La22xBaxCaO4 (Pickett et al.,
1991). Order –disorder character may characterize a phase transition. Above the phase
transition temperature, Tc ; there would be uncorrelated local tilts. At phase
transition, all tilts are in phase. Such transitions can cause soft-mode transitions and
elastic anomalies.
Using the PIB model, Cohen (1987) showed that the energy difference (expressed
as temperature) between cubic and orthorhombic MgSiO3 rose rapidly by 60 K/GPa.
Matsui and Price (1991), however, found a continuous transition from orthorhombic to
cubic, which is not allowed in a soft-mode transition and indicates order –disorder
character. A transition from orthorhombic to tetragonal should correspond to a softening of
the shear modulus c11 – c12 at the transition. Samples annealed at high temperatures have
the possibility of forming coarser twins.
752 Chapter 10

Because the ilmenite phase of MgSiO3 is metastable at ambient pressure, the

thermodynamic properties such as CP ; S and a are limited below 700 K (Ashida et al.,
1988). Thus, vibrational models have been used to calculate the heat capacities and
entropies in the temperature range relevant to the mantle (1,000 – 2,500 K).
However, vibrational modelling is carried out under harmonic approximation and
the calculated heat capacities at constant volume ðCV Þ tend towards the classical Dulong –
Petit limit ð3nRÞ at high temperature. High-temperature calorimetric data and HPHT
Raman spectroscopic measurements have shown that this approximation fails at high
temperature in many oxides, silicates and their analouge germanates (Fiquet et al., 1992)
due to intrinsic anharmonic effects.

10.3.5. Ilmenite structure (R3̄)

Under pressure, silicates and germanates of Fe, Mg, Co, Zn and also titanites of
these elements occur in ilmenite as well as in perovskite structure. These structures can
also contain large cations such as Ba, Sr, Ca and Cd but all other transition elements are
destabilized in the 8- and 12-fold coordination sites relative to octahedral sites because of
the ligand field effects (viz. less CFSE, etc.). This explains why we do not expect
transition-metal perovskites.
Pressure disordering in the ilmenite structure may cause its breakdown to a slightly
more dense corundum structure. The perovskite and corundum phases are seen to have
higher entropies than the ilmenites, leading to negative dP=dT slopes for the ilmenite –
perovskite and ilmenite –corundum transitions.

10.3.5.1. MgSiO3 ilmenite

MgSiO3 –ilmenite (the Mg end-number of (Mg,Fe) SiO3 –akimotoite) is a high-
pressure polymorph of enstatite that is characterized by a relatively narrow stability field
(e.g., Ito and Navratosky, 1985; Sawamoto, 1977) in the range of 20 –25 GPa. It is bounded
at lower pressures by the b-Mg2SiO4 – stishovite assemblage stability field, at high pressures
and temperatures by the perovskite-type MgSiO3 stability field and marginally at high
temperatures (.2,200 K) by the majorite stability field (Ito and Navrotsky, 1985;
Sawamoto, 1977). Its stability at low temperatures makes it a candidate for the mantle-
forming phase in subducting slabs in the 600 – 700 km depth range. A detailed knowledge of
its stability is important for understanding the dynamics of subducted cold slabs and for
refining thermodynamic databases for phase-diagram calculations.

In subducting slab. Because MgSiO3 ilmenite is stable at relatively low temperature, in

the subducting slabs at the 600 –700 km depth range of the mantle, it is potentially an
important constituent of the mantle. In subducting lithosphere (at low temperature),
clinoenstatite may persist metastably in the b-MgSiO4 þ stishovite stability field before
transforming directly to ilmenite (Hogrefe et al., 1994). This metastable clinoenstatite-to-
ilmenite transformation causes large volume change, which affects the stresses and
ABX3, Perovskite – Ilmenite Structure 753

buoyancy forces in the subducting slab and may help the mechanisms for deep-focus
earthquakes.
Again, the ilmenite-stability field at low temperature extends several tens of
kilometres deep within the lower mantle. The presence of MgSiO3 ilmenite at this depth
would cause a negative density anomaly within the surrounding mantle where perovskite is
stable. This inhibits slab penetration into the lower mantle.
High-pressure powder XRD data for MgSiO3 ilmenite have been collected by
Reynard et al. (1996) up to 28 GPa, using a diamond-anvil cell in an energy-dispersive
configuration at the storage ring DCI at the LWRE (Orsay, France).

 with two formula units in the primitive cell, yields 30

Vibrational modes. Ilmenite, R3;
normal modes of vibration at any one point in the Brillouin zone distributed at the zone
centre as follows:
5Ag ðRÞ þ 5Eg ðRÞ þ 5Au ðIRÞ þ 5Eu ðIRÞ
From symmetry analysis, the Raman spectrum of MgSiO3 ilmenite should give 10
bands: five Ag and five Eg bands (McMillan and Ross, 1987). Reynard and Rubie (1996),
however, could observe nine low-intensity bands at 294, 351, 402, 413, 484, 499, 620, 680
and 789 cm21. The intensity of the bands (mostly low) decreased dramatically with
increasing pressure, making it difficult to record the spectra above 7 GPa. No hysteresis is
observed on decompression.
At high temperature (up to 880 K), frequency shifts and line width increase of the
nine Raman bands were determined by Reynard and Rubie (1996). Frequencies and line
widths were obtained by deconvolution of the high-temperature spectra using Voigt-type
profiles for the Raman bands.
Above 880 K, a partial transformation of MgSiO3 ilmenite to a glass is observed,
marked by the appearance of broad bands in the 850 –1,050-cm21 range. When heated
above 973 K (for 1 h), ilmenite shows a complete decomposition to a mixture of glass and
various crystalline phases, such as pyroxene and an assemblage called X, a disordered hcp
phase (Ito and Navrotsky, 1985). At temperature near to 1,000 K, pyroxene bands appear
and grow in intensity with time. In general, a mixture of glass, ilmenite and enstatite is
observed. A transformation to glass phase is seen before recrystallization to enstatite; a
similar “amorphization” of a high-pressure phase at ambient pressure and high temperature
is known. Reynard et al. (1996) have measured the lattice parameters as a function of
pressure up to 28 GPa by using XRD.
For MgSiO3 – ilmenite, the elasticity calculations have shown that C33 is
much smaller than C11 at zero and higher pressures, indicating that the c-axis
remains substantially more compressible than the a-axis under compression
(da Silva et al., 1999).

Intrinsic anharmonic parameter. At ambient conditions, the absolute values of the

anharmonic parameters of geikielite are small compared with those of many
multiple oxides. This suggests that this ilmenite has a nearly quasi-harmonic behaviour,
and also that other ilmenites may behave quasi-harmonically as well.
754 Chapter 10

In their study of geikielite (ilmenite-type MgTiO3), Reynard and Guyot (1994)

suggested that these effects may be negligible for the ilmenite structure whereas
they are important in the garnet structure (Gillet et al., 1992). This would lead to
significant intrinsic anharmonic corrections to the entropy and enthalpy changes
associated with the ilmenite –majorite transformation.
The effects of intrinsic anharmonicity have been introduced into vibrational
calculations of the heat capacity of geikielite. An estimation of the intrinsic anharmonic
effects for the ilmenite ! majorite transition has been attempted by Reynard and Guyot
(1994). They assumed that the intrinsic anharmonic behaviour of MgSiO3 – ilmenite is
similar to geikielite. While the intrinsic anharmonic behaviour of majorite is similar to that
of pyrope, grossular and andradite (Gillet et al., 1992). The calculations showed an average
intrinsic anharmonic parameter to be about 2 1 £ 1025 K21 for ilmenite and
2 £ 1025 K21 for majorite.

Compressibility. Determination of the structural behaviour of high-pressure forms

are based on molecular dynamics with empirical potentials (e.g., Matsuiand Price, 1992)
and on the Hartree –Fock method (D’Arco et al., 1994). Karki et al. (2000) cal-
culated elastic-wave velocities of ilmenite at high pressure by first principles (da Silva
et al., 1999).
The compressibility of MgSiO3 ilmenite is anisotropic, being twice as
incompressible along the c-axis as along the a-axis. In ilmenite structure, the
compression along the c-axis is achieved through a flattening of the M2þO6
octahedron, which is more compressible than the M2þ –M4þ distances across face-sharing
octahedra and the M4þ – M4þ distances across a vacant site. On the other hand,
compression of the a-axis causes a reduction of the M2þ –M2þ and M4þ – M4þ distances
across shared edges within the monocationic layer. However, because of large
repulsion, the M4þ – M4þ distances are less compressible (Ross et al., 1993). This
is a classic behaviour of the ilmenite structure in contrast to the corundum-type
structure under pressure, where a and c show similar compressibilities (Finger and
Hazen, 1980).
Because of the importance of MgSiO3 –ilmenite in mantle studies a precise
knowledge of the EOS is very important. Employing Brillouin scattering (Weidner and Ito,
1985) and high-pressure XRD (Reynard et al., 1996), the EOS of it has been evaluated.
This is discussed in the following.
Reynold et al. (1996) used P – V data for the EOS determination of MgSiO3 –
ilmenite. Pressures were better estimated from the EOS of ice VII, rather than from ruby
fluorescence. The measurements were consistent with the Brillouin-scattering data of
Weidner and Ito (1985). However, the derived K 00 value was found to be significantly
higher than the value of 4 (second-order truncation of the Birch – Murnaghan EOS),
commonly used to extrapolate ambient-pressure determinations of K0 : Indeed, K 00 values
higher than 4 have been determined for close-packed structures such as Mg-rich silicate
spinels (Hazen, 1993; Rigden et al., 1993).
ABX3, Perovskite – Ilmenite Structure 755

10.4. CaO –SiO2 system

The P – T phase diagram for CaSiO3 perovskite is shown in Fig. 10.18, which
presents the phase stability of Ca– perovskite in complex systems.

10.4.1. CaSiO3 perovskite

In the lower mantle, next only to ferromagnesian silicate perovskite and magnesio-
wüstite, CaSiO3 perovskite occurs as a major phase (,6 –12%) (see Section 10.2.2.1). The
observed increase in Ks in the lower mantle could perhaps be due to an appreciable
enrichment of calcium, since CaSiO3 has a higher (by ,3%) bulk modulus than
(Mg,Fe)SiO3 (Mao et al., 1991). However, the thermal expansivity of CaSiO3 perovskite
has not yet been measured. Ito and Stixrude (1991) suggested that CaSiO3 perovskite
transition can also result in the seismic structure observed near 520 km (i.e., within the
transition zone) (Shearer, 1990).
CaSiO3 crystallizes in a perovskite structure at pressures in excess of 15 GPa
and temperatures near 1,500 K (Liu and Ringwood, 1975). In the cubic structure, fully
linked SiO6 octahedra surround the central Ca2þ ion, which is dodecahedrally coordinated
with oxygen. A distortion from the ideal cubic structure occurs through the cooperative
rotation of the octahedra. At higher pressures, cubic structures are seen by in situ
diffraction study.
As a function of temperature and pressure, the stability and thermodynamic
properties of CaSiO3 – and MgSiO3 – perovskite are discussed by Wulf and Bukowinski
(1987). Theoretical properties are derived from quasi-harmonic lattice dynamics
employing parameter-free pair potentials derived from the MEG formulation.

Figure 10.18. The P – T phase diagram for CaSiO3 –pvs. Filled-square: break down of cpx, half-filled square:
CaSiO3 –pvs þ Mg –garnet. Circles: exsolution of CaSiO3 –pvs from Mg–garnet near 520 km depth along two
postulated geotherms (Source: Canil, 1994).
756 Chapter 10

Cubic CaSiO3 – perovskite structure is dynamically stable from zero pressure to pressures
of at least 70 GPa. At higher pressures, the static potential energy surface of CaSiO3
exhibits minima that correspond to a distorted perovskite structure.
Ca –perovskite forms at significantly lower pressure than does Mg – perovskite
(11 – 13 GPa) (Mao et al., 1989; Tamai and Yagi, 1989; Irifune et al., 1989).
However, CaSiO3 is stable and retains its cubic structure to the pressure prevalent at the
core – mantle boundary (Mao et al., 1989a). The wide stability field of the cubic form is in
general agreement with theoretical predictions (Hemley et al., 1987; Wolf and
Bukowinski, 1987). Thus, CaSiO3 perovskite may constitute a significant (but invisible)
component in the lower-mantle mineralogy (Tarrida and Richet, 1989). A diagram
summarizing CaSiO3 – perovskite stability in some simple and complex systems is
presented in Fig. 10.18.
CaMgSi2O6 (diopside) breaks down at .24 GPa to orthorhombic Mg-rich and
cubic Ca-rich perovskite. But a complete solid solution between CaMgSi2O6 and
CaSiO3 in the cubic perovskite structure is also possible (Liu, 1987). Indeed, in garnet
peridotite, a cubic perovskite with a nearly ideal diopside composition has been
reported by Ito and Takahashi (1987). CaFeSi2O6 (hedenbergite) at 19 –26 GPa forms Ca–
perovskite (Kim et al., 1991). Rare-earth elements seem to partition in a greater
measure in CaSiO3 perovskite than in (Mg,Fe)SiO3 perovskite (Kato et al., 1988).
Therefore, cubic CaSiO3 perovskite could be the largest reservoir of rare-earth elements in
the mantle.
Heavier elements such as alkaline-earth, Na, Ba, Sr, etc., are stable (dynamically)
only at high-pressure perovskites, which on decompression become dynamically unstable
because of the presence of large cations in the A site. Such phases consequently become
unquenchable from high P – T condition. The non-quenchable cubic perovskite phase of
omphacite composition has been synthesized by Liu (1987).

10.4.1.1. LAPW calculations: phonon spectrum and transition temperature

In the pressure range available at depths between 500 and 2,900 km, the CaSiO3
perovskite is seen by XRD studies to be cubic with little deviation from Pm3m symmetry.
Theoretical studies based on ionic models, periodic Hartee– Fock calculations and pseudo-
potential calculations have supported this view.
Stixrude et al. (1996) carried out investigation from first principles of the full
phonon spectrum of cubic CaSiO3 – perovskite from low pressures to those available in the
lower mantle. The calculations were based on LAPW method, regarded as most accurate
for solving band structure and the total energy problem. The calculations showed the cubic
phase to be dynamically unstable and the ground state to be lower than cubic symmetry. It
may be that the cubic phase is thermodynamically stable at high temperatures because of
its greater entropy (e.g., Salje, 1990).
The calculated dynamical matrix of the cubic Pm3m phase of CaSiO3 – perovskite
is represented at four points in the Brillouin zone (G, X, M, R). Computational variables
(K-point mesh: 4 £ 4 £ 4 for computations of the dynamical matrix elements, 8 £ 8 £ 8
for 1 (dielectric constant) and Z p (Born effective charges) and the number of basis function
ABX3, Perovskite – Ilmenite Structure 757

per atom <150 were chosen such that phonon frequencies converged to better than one
wave number. The full phonon-dispersion curve was determined using an interpolation
scheme that separates short-range forces from long-range Coulombic interactions (Yu and
Krakauer, 1994).
The full phonon-dispersion curves reveal dynamical instabilities in the cubic
structure along the zone boundaries. Instabilities occur at the M and R points and along the
zone edges from M to R:
Using LAPW total energy calculations, Stixrude et al. (1996) found the total energy
as a function of displacement along the most unstable mode eigenvector ðR25 Þ and the
associated minimum-energy displacement (frozen phonon approach; Cohen, 1992). For
this finite displacement, the symmetry becomes tetragonal I4=mcm with 10 atoms in the
unit cell. The results at mid-mantle pressure (P ¼ 80 GPa) show that the minimum-energy
displacement corresponds to an octahedral rotation angle of 78 and is 360 K lower in
energy per octahedron than the cubic phase.
The transition temperature can be estimated from the distorted I4=mcm structure to
the cubic Pm3m structure. The transition temperature can be estimated as

4 JlAl
Tc ¼ < 2; 200 K
3qð3Þ KB B

where KB is the Boltzmann constant and qð3Þ ¼ 0:5054: The result corresponds to
P ¼ 80 GPa (< 1,850 km) and a temperature of ,2,500 –3,000 K.
The elastic anomaly that causes the subset of the reflective features near 700, 900
and 1,200 km depth in the lower mantle may be associated with phase transitions in
CaSiO3 – perovskite. (Note. At this pressure range, SiO2 undergoes phase transition from
the stishovite to CaCl2 structure.)

10.4.1.2. Density and acoustic velocity

The density of CaSiO3 perovskite is higher than that of MgSiO3 perovskite,
whereas its bulk sound velocity is lower. An addition of 10% iron to MgSiO3 perovskite
would make its density and bulk sound-velocity profiles nearly identical with those of
CaSiO3 perovskite (Karki and Crain, 1998a). The shear-wave velocity of CaSiO3 –
perovskite in the lower half of the lower mantle is fairly high.

10.4.2. Pseudo-wollastonite

At temperature .1,1258C, the pseudo-wollastonite form of CaSiO3 is seen to be

stable and it commonly occurs in slags, cement and ceramic materials. At high P, T, a
transformation to wahlstromite-II (,1,1008C and 3 GPa) occurs, yielding to perovskite
form at pressure .10 GPa (see discussion under Section 10.4.1.1).
Pseudo-wollastonite structure shows several polytypes:
(i) a two-layer polytype;
(ii) a three-layer polytype, synthesized at 6.5 GPa, 1,3008C (Trojer, 1969);
758 Chapter 10

(iii) a four-layer plus a six-layer disordered structure; monoclinic, C2=c (Yang and
Prewitt, 1999);
(iv) a six-layer polytype.
Yang and Prewitt (1999) have investigated the two-layer pseudo-wollastonite
polytype under pressures up to 9.94 GPa. It is monoclinic with space group C2=c; showing
the following cell parameters:
a ¼ 11:8322ð6ÞÅ,
b ¼ 6:8624ð8ÞÅ,
c ¼ 10:5297ð5ÞÅ,
b ¼ 111:245ð8ÞÅ, and
V ¼ 796:9ð1ÞÅ3.
Under increasing pressure, the compressibilities are noted as below:
ba : bb : bc ¼ 2:25 : 2:36 : 1:00
KT ¼ 86ð1Þ GPa,
K 0 ¼ 3:8ð4Þ:

10.4.3. CaSiO3 – CaTiO3 join

Ringwood and Major (1971) reported that the CaSiO3 – CaTiO3 – perovskite
system were quenchable at ambient pressure for compositions up to 83 mol% CaSiO3.
Kubo et al. (1997) studied this system at pressures of 5.3 –14.7 GPa and temperatures
of 1,200 – 1,6008C to investigate the solid solution. They observed that, as the
pressure increases, the stability field of perovskite solid solution extends from CaTiO3
to CaSiO3 and the perovskite becomes stable for the entire composition range above
,12.3 GPa.
The stability field of Ca(Ti12xSix)2O5 ð0:78 , x # 1Þ titanite solid solution þ
Ca2SiO4 larnite exists in the CaSiO3-rich composition range at 9.3– 12.3 GPa and 1,2008C.
Perovskite solid solutions containing a CaSiO3 component of 0 – 66 mol% could be
quenched to 1 atm.
The composition –molar volume relationship of perovskite solid shows that the
molar volume of perovskite solid solution linearly reduces from the value of CaTiO3 to that
of CaSiO3.

10.5. MgO(CaO)– SiO2(GeO2) – Al2O3 system

The possible perovskite stability field in part of the MgO –SiO2 – Al2O3 system is
shown in Fig. 10.19 (Navrotsky, 1999). The MgSiO3 – Al2O3 and MgSiO3 – MgAlO2.5 joins
are the likely limits to the substitution of aluminium in perovskite. The shaded area
schematically shows the region where a single-phase perovskite could form. The filled
ellipse represents likely mantle compositions and the dot represents the ideal composition
of garnet pyrope. Excess MgO favours the substitution along the MgSiO3 –MgAlO2.5 join.
Which of the assemblages should be denser depends critically on the thermal expansivity,
ABX3, Perovskite – Ilmenite Structure 759

Figure 10.19. Part of the MgO–SiO2 –Al2O3 system showing the possible perovskite stability field. The ruled
triangular area is where a single-phase perovskite could form. The elliptical area is for the mantle composition.
The small circles represents ideal composition of garnet pyrope (from Navrotsky, 1999).

compressibility and molar volume of the aluminous perovskite formed by each substitution
mechanism.

10.5.1. Ca – Al perovskite

In the Ca –Al –Si system, a continuous solid solution may be present in cubic
perovskite structure with a0 varying between 3.56 and 3.71 Å for the compositional range
CaSiO3 to Ca2AlSiO5.5 (or 2Ca2SiO4, Al2O3). Identically, brownmillerite (Ca2FeAlO5) is
continuous across a cation solid-solution range covering Ca2Fe2O5.
Within the parent structure of cubic perovskite, superstructures may develop in the
arrangement of Si, Al, Ca atoms and O vacancies. Moreover, an entry of Al2O3 into solid
solution in CaSiO3 perovskite is associated with an oxygen-deficient structure. The
intensities of super-reflection are expected to be dominated by electron-density
modulations associated with the heavy atoms such as Si, Al and Ca. Oxygen deficiency
in cubic and cubic-based perovskites up to 1/6 of the oxygen position is known for many
compounds.
The 5-times superstructure, if it is a compositional modulation, presumably
involves not only oxygen vacancies but also Ca, Al and Si atoms distributed among the
760 Chapter 10

groups of related A and B sites within each 5-times super-cell. A rhombohedral distortion
of the cubic perovskite cell is consistent with the existence of reflection arising from the
5-times modulation.
Gerald and Ringwood (1991) synthesized a perovskite with composition
CaAl2SiO5.5 using an MA-8 apparatus operating at 1,7008C and 16 GPa. The XRD
study revealed a cubic perovskite cell (a0 ¼ 3:706 ^ 0:003 Å). Using this parameter and a
transformation matrix, a rhombohedral cell can be derived which is based on the parent
cubic perovskite as:
  0 12 3
u 2 1 2 u
 
  B C6 7
v ¼B C6 7 ð10-14Þ
  @ 2 2 1 A4 v 5
 
w 1 2 2 w rhomb
cubic

This also satisfies the requirements for electron diffraction observations for a 5-times cell
along the 111cubic cube diagonal. The resultant rhombohedral cell has a ¼ 11:118 Å and
a ¼ 27:2668:
Liu (1978) compressed gehlenite (Ca2Al2SiO7) to 15 –20 GPa, laser heated to
1,0008C in a diamond-anvil cell and obtained a structure with cubic cell
(a0 ¼ 14:88 ^ 0:02 Å). He proposed that the high-pressure Ca2Al2SiO7 phase possessed
elements of a sodium titanate-type structure in which packing density is increased via
blocks of edge-shared octahedra with tetragonal or orthorhombic distortions.

10.5.2. Ca – Ge perovskites

10.5.2.1. IR modes: Ca translation

The infrared profile for CaGeO3 strongly resembles that of MgSiO3, allowing IR
modes to be easily correlated. Under pressure, the modes shift linearly to higher
frequencies at rates of 1.28 –2.09 cm21/GPa (Lu and Hofmeister, 1994). The Grüneisen
parameters decrease monotonically with increasing mode frequency (excepting the modes
at ,265 and 386 cm21). The mode Grüneisen parameter is defined as:
   
V dni K dni
gi0 ¼ 2 ¼ T ð10-15Þ
ni0 dni ni0 dP
where V is volume, KT bulk modulus and ni a mode frequency (Grüneisen, 1912). The
average value of ni0 is observed to be 1.13 (Lu and Hofmeister, 1994).
Vibrational modes in CaGeO3 at 1 atm are restricted to a very narrow region (155 –
786 cm21) than in MgSiO3 – perovskite (180 – 950 cm21). Among the spectra of perovskite
polymorphs, systematic relationships exist. The strong F1u modes of the cubic polymorph
are retained in all polymorphs and each transition brings in weaker and narrower peaks
(i.e., fine structure). The intense modes originating in the cubic phase dominate the profile
of an IR spectrum for the orthorhombic phase of lower symmetry.
The lowest frequency Raman modes correspond to Ca translations. The Raman
spectrum has a lower limit of 151 cm21 (Durben et al., 1991), which correlates with the
ABX3, Perovskite – Ilmenite Structure 761

lower frequency IR band occurring between ,155 and 173 cm21. The IR spectrum of the
tetragonal phase should possess the five most intense bands of the orthorhombic phase
without the superimposed detailed structures. The cubic phase will have only IR modes at
,155 –173, 370 –559 and 667– 786 cm21.
The relative positions of the Ca translation and mode C are traversed from those of
the Mg translation and mode C in MgSiO3 –perovskite (Lu et al., 1994). The much lower
frequency of 155 cm21 for the Ca translation, as compared with 282 cm21 for the Mg
translation in MgSiO3 perovskite, is obviously due to the difference in mass. The
frequencies for mode C are nearly identical (226 vs 224 cm21) and this is consistent with
the oxygen lattice dominating the deformations.
The linear dependence of mode Grüneissen parameters and frequency can be
observed and the frequency can be described in terms of a force constant K and reduced
mass m:
sffiffiffiffiffiffiffi
Ki
ni ¼ ð10-16Þ
mi

Substituting equation (10-16) in equation (10-15) gives

   
KTO dKi KTO dKi
gi0 ¼ ¼ ð10-17Þ
2Ki0 dP O 2mi n2i0 dP O
This linear trend between mode Grüneisen parameter and frequency, as is found in
CaGeO3 (Lu and Hofmeister, 1994), should also hold for MgSiO3 – perovskite.
A structural transition at ,12 GPa from orthorhombic Pbnm to a tetragonal phase
has been suggested based on EXAFS data (Andrault and Poirier, 1991). No significant
change in IR spectrum was seen up to 24 GPa. However, a phase transition between Pbnm
and Cmcm (Aleksandrov, 1976) cannot be ruled out because these polymorphs have the
same symmetry class ðD2h Þ and should have similar IR spectra, although the distribution of
Raman modes should differ.

10.6. Alkaline-earth perovskites

10.6.1. Li(Nb,Ta)O3 ferroelectrics

LiNbO3 and LiTaO3 ferroelectric materials, discovered in 1949, have been

applied in optical, electro-optical and piezoelectric devices. These show only one
structural phase transition. The transition temperatures, Tc ; for LiNbO3 and LiTaO3 are
1,210 and 6708C, respectively, which are among the highest known for ferroelectric
oxides. The transition temperatures are, however, dependent on sample stoichiometry
(Mehta et al., 1993).
The structure of ABO3 – perovskite was discussed in Section 10.1.5. The structure
of sesquioxide, such as lithium niobate (LiNbO3) structure, consists of cations occupying
octahedral sites. In the former, the cations are ordered in alternate layers whereas, in the
762 Chapter 10

Figure 10.20. Hypothetical rotations of BO6 octahedra leading from perovskite hettotype (bottom) to fully
collapsed lithium niobate (top), circles represent positions of A cations (Leinenweber et al., 1995, q 1995
Springer-Verlag).

latter cation, ordering occurs within each layer (Megaw, 1973). A rotation of BO6
octahedra may lead a transition from hettotype perovskite to fully collapsed LiNbO4
structure (Fig. 10.20).
LiNbO3 structure consists of chains of oxygen octahedra with Li ions shared
between two octahedra; at low temperatures, they occupy one set of octahedra. At high
temperatures, Li atoms hop between two octahedra and this motion of Li drives the
ferroelectrcity. Inbar and Cohen (1995) performed frozen phonon computations and found
that the double wells for Li displacements are quite shallow and that the double wells for
oxygen displacement are much deeper. They observed that the ferroelectric instabilities in
LiNbO3 and LiTaO3 are quite similar to the instabilities in the oxide perovskites and that a
primary driving force is the d0 configuration of the Nb5þ and Ta5þ ions. Hybridization with
the surrounding oxygens allows the d0 cation to go off-centre. This sets up local fields that
drive or order the Li off-centring.
Li(Ta,Nb)O3 systems do not show a large sensitivity to pressure (Jayaraman
and Ballman, 1986), in contrast to the great sensitivity of the potential energy
ABX3, Perovskite – Ilmenite Structure 763

surface to pressure changes displayed in perovskites. In BaTiO3 and PbTiO3

systems with increasing pressure, the frequencies decrease (Curie – Weiss-type
behaviour) and a soft mode is observed whereas in Li(Ta,Nb)O3 the frequencies
increase with pressure.

10.6.1.1. Ferroelectric and para-electric structures

A rhombohedral ferroelectric structure has a 10-atom unit cell with space group
 in which oxygen octahedra share faces along the polar trigonal axis. The transition-
R3c;
metal ion is displaced from the centre of the oxygen octahedra along the trigonal axis. The
next oxygen octahedron along this axis is empty. In LiTaO3 ferroelectrics, the adjacent
octahedron has a Li ion, ferroelectrically displaced from the oxygen position in the
spontaneous polarization direction.
Ferroelectric systems are thought of as exhibiting both displacive (far from the
transition-temperature region) and order – disorder (near Tc ) characters. However, the
neutron scattering data serve as the cornerstone for all theories modelling ferroelectric
transition as an order –disorder mechanism. In the case of LiTaO3, the Li atom hops
between the centro-symmetric and the interstitial sites (Frenkel defect; Birnie, 1991) or
between two adjacent oxygen octahedra (Bakker, 1993). Bakker et al. (1993) predicted and
observed a 32-cm21 excitation, which can be ascribed to Li motions between the central
and lowest wells.
Ferroelectric phase transitions occur with increasing temperature from one
ferroelectric state to another, or to the high-temperature para-electric state. Ferroelectricity
is very sensitive to volume. Dielectric and piezoelectric properties are strongly dependent
on temperature, especially around the phase transitions.
In the para-electric structure, the space group of the average structure is R3c; 
wherein the transition-metal ions occupy the centres of the octahedra and the Li ions divide
the occupancy time between the two adjacent octahedra. In the para-electric phase, the
order –disorder character of oxygens causes oxygen octahedra to change locally. These
changes in oxygen positions, caused by the electrostatic interaction between Li and
oxygens, drag Li atoms from their centro-symmetric sites.
Inbar and Cohen (1995) calculated the potential energy surfaces along the
experimental soft-mode coordinate for LiTaO3 (aH ¼ 5:15428 Å and cH ¼ 13:7851 Å)
and LiNbO3 (aH ¼ 5:14829 Å and cH ¼ 13:8631 Å). In Fig. 10.21, they present the
potential energy surfaces of LiTaO3 and LiNbO3 with respect to Li displacement only
(upper curves) and Li þ O displacements (lower curves). The Li displacements along the
soft-mode coordinate result in a single anharmonic well. Also, the energy changes
involved in these displacements are very small compared with the energy changes that
result from the oxygen displacements. The deep double wells are the result of the oxygen
displacements along the experimental ferroelectric coordinate against the transition-metal
atom. This indicates that the structural-phase transitions are determined energetically by
the oxygen distortions, rather than by the displacements of the Li atoms. This finding by
Inbar and Cohen (1995) stands in contrast to the theories that assume that the Li ion drives
the phase transformation. But, since these wells are formed due to oxygen displacements, it
seems that the mechanism for this phase transformation is the order – disorder of oxygen
764 Chapter 10

Figure 10.21. Potential energy surfaces of LiTaO3 and LiNbO3. The upper curves represent the displacement of
the Li atoms along the soft mode coordinate and the bottom curves represent the displacement of oxygens and Li
atoms along the same coordinate. The curves represent a fourth order polynomial fitting to the data. They were not
constrained to go through the zero of energy. The displacements, Q, are in the units of (amu)1/2 £ Bohr
(Source: Inbar and Cohen, 1995).

displacements, not Li atoms. At high temperatures, however, the order –disorder character
of the oxygen atoms causes local changes in oxygen octahedra, which in turn displace the
Li atoms from their centro-symmetric site. The large Li amplitude probably results in a
different type of transition.
All Coulombic lattices are unstable with respect to ferroelectric displacements. The
short-range repulsive forces tend to stabilize crystals with respect to off-centre
displacements.
In ferroelectric perovskites, there is a very strong hybridization (covalency)
between the transition-metal d states and the oxygen 2p states which drives the
ferroelectric phase transitions. This hybridization also causes symmetry-lowering
distortions to occur. Perovskites have a 3d soft-mode axis and exhibit a first-order
phase transformation, but Li(Nb,Ta)O3 uniaxial ferroelectrics show second-order phase
transitions. When M(Ti, Ta, etc.) –O distance increases, short-range repulsion decreases
and hybridization decreases, but incipient ferroelectric behaviour may persist as in KTaO3,
which retains its high-symmetry phase.

10.6.2. RE orthoferrites

Rare-earth orthoferrites in general adopt the GdFeO3 structure with an

orthorhombic distortion and space group Pbnm, involving tilting of the octahedra about
each of their three 4-fold axes. The A site is distorted significantly from cubic symmetry,
ABX3, Perovskite – Ilmenite Structure 765

with eight cation –anion distances significantly shorter than the other four. Iron as Fe3þ
occupies the octahedral site.
As the size of A decreases from LaFeO3 to LuFeO3, the value of the tolerance factor
ðtÞ decreases from t ¼ 0:88 (for LaFeO3) to t ¼ 0:82 (for LuFeO3) and the octahedra
are increasingly tilted with the decrease in the size of the RE cation from La to Lu.
An increased octahedral tilting angle reduces the Fe3þ –O –Fe3þ angle associated with
super-exchange (Inbar and Cohen, 1995). The distortion of the crystal structure from cubic
symmetry produces a slight canting of otherwise anti-ferromagnetically arranged spins,
resulting in weak ferromagnetism.

10.6.2.1. (Sr/Ca) FeO3 perovskite

High-pressure experiments on alkaline-earth perovskites show decrease in Tc with
pressure and tri-critical points are reached where the transition changes from first order to
higher order.
SrFeO3 ðt ¼ 1:01Þ adopts a cubic perovskite structure (MacChesney et al., 1965)
while CaFeO3 ðt ¼ 0:98Þ shows a slight tetragonal distortion from the cubic form
(Takeda et al., 1978). SrFeOx adopts a cubic perovskite structure when 2:9 # x # 3;
a tetragonal structure when x ¼ 2:86 and an orthorhombic perovskite structure when
2:68 # x # 2:73 (Takeda et al., 1986). SrFeO2.5 adopts a brownmillerite structure,
a derivative of perovskite structure, where ordered vacancies create tetrahedral
cation sites.
The divalent alkaline-earth cation stabilizes the unusual oxidation state Fe4þ in
the structure. In SrFeO3 and CaFeO3 structures, Fe4þ in octahedral coordination has a
high-spin d4 configuration and the t2g electrons are localized at the iron atom, while the eg
electron is delocalized into a partially filled band.
At low temperature, CaFeO3 shows a second-order transition to a localized
behaviour as:
2Fe4þ , Feð4þdÞþ þ Feð42dÞþ
where d approaches 1 at low temperature (Takano et al., 1991).
At high pressure, CaFeO3 undergoes a change in both crystal and electronic
structure (Takano et al., 1991). Both XRD and Mössbauer spectroscopy show a change
at ,30 GPa and magnetic splitting occurs, indicating a high-spin to low-spin transi-
tion in Fe4þ. Since the symmetry reduction to orthorhombic structure implies a
lower tolerance factor, the size of the A cation site also possibly changes through the
transition.
The slightly lower Néel temperature in CaFeO3 compared with SrFeO3 can be due
to greater covalency as well as the Fe – O –Fe angle (less than 1808), which reduces the
strength of anti-ferromagnetic exchange.

10.7. Titanate perovskites and ilmenites

The phase relationships in FeTiO3 are of interest in the fields of Earth and material
sciences. Such ABO3 compounds crystallize either in perovskite structure or in one of the
766 Chapter 10

Figure 10.22. Two representations of the CaTiO3 –perovskite structure. In (a) the Ti is at the cell vertices and the
Ca in the center site. In (b) the Ti is at the center.

sesquioxide structures. Perovskite structure, such as CaTiO3, can have two representations
(Fig. 10.22, see caption). The phase transitions involving displacement of cations or
rotations of octahedra are presented in Fig. 10.20.
Syono (1981) quenched FeTiO3 from 20 GPa (and 1,273 K) and obtained almost
ilmenite structure but systematic absences in the X-ray pattern indicated a c-glide plane.
The density of the quenched phase was 0.93% higher than that of ilmenite. This phase
showed a lithium niobate structure ðR3cÞ:  Lienenweber et al. (1991) found that the
lithium niobate phase of FeTiO3 transforms reversibly at room temperature and
16 ^ 1 GPa to a perovskite-like phase. Later, Mehta et al. (1994) found that the form of
FeTiO3 at high temperature and pressure is indeed perovskite.
Ilmenite (FeTiO3) occurs as a common accessory mineral in many igneous and
metamorphic rocks where ilmenite –geikielite solid-solution minerals, in conjunction with
such minerals as hematite and olivine, can serve as an indicators of T; P; fO2 of
crystallization (e.g., Frost et al., 1988).

10.7.1. CaTiO3 – FeTiO3 join

Both CaTiO3 and FeTiO3 end-members form perovskite phases under high P, T
conditions. An exploratory high-pressure study of the CaTiO3 – FeTiO3 join by
Leinenweber et al. (1995) uncovered two intermediate perovskites with the compositions
CaFe3Ti4O12 and CaFeTi2O6, termed respectively as 1 : 3 and 1 : 1 phases (Ca : Fe
stoichiometry). These perovskites have ordering of Ca2þ and Fe2þ on the A sites. Both
these perovskites are unusual in that the A sites containing Fe2þ are either square planar or
tetrahedral arising from particular tilt geometries of the octahedral frameworks
(Fig. 10.24). The presence of a transition metal (like Fe2þ, Cu2þ or Mn3þ) in these
four-coordinate sites possibly stabilizes the structure of the 1 : 1 and 1 : 3 compounds.
ABX3, Perovskite – Ilmenite Structure 767

The crystallographic parameters of these phases are compared below with those of
FeTiO3 ilmenite.

Ilmenite 1 : 3 Titanatea 1 : 1 Titanateb

SG. R3c Im3 P42 =nmc

a ¼ 5:1233ð1Þ 7.4672 Å 7.5157(2)
c ¼ 13:7602ð2Þ 7.5548(2)

a
Isostructural with NaMn7O12. Fe2þ is in a square planar A site. Fe–O ¼ 2.042(3) Å, second neighbours in a
rectangle at Fe –O ¼ 2.780(6) Å, Ca in distorted icosahedron with Ca–O ¼ 2.635(5) Å,
b
A-site Fe is square planar; Fe–O ¼ 2.097(2) Å and tetrahedral Fe– O ¼ 2.084(2) Å, second neighbour oxygens
near 2.8 Å.

The 1 : 3 phase, CaFe3Ti4O12 is the only Im3 perovskite which takes iron in the
square planar A site (common cations are Mn3þ and Cu2þ), which has 12 oxygen
neighbours with four nearest neighbours in a slightly distorted square at 2.038(3) Å, four
second neighbours in a rectangle at 2.770(7) Å and four distinct third neighbours in a
slightly distorted square at 3.244(2) Å (Fig. 10.23). The Fe2þ ion is coplanar with all three
sets of oxygens. The O –Fe – O angles in the distorted first-neighbour square are 878 and
938. This structure can be derived from perovskite structure by rotation of each octahedron
in the connected TiO6 octahedral framework, which leads to the regular Ca site and the
square planar Fe site. In the notation of Glazer (1972), the tilt system is aþ aþ aþ :
The ordered perovskites appear to be stable at 12 – 15 GPa but CaFe3Ti4O12 is
found to be stable at as low a pressure as 5 GPa. Therefore, these perovskites are certain to
bear important implications for upper-mantle mineralogy, particularly for kimberlites. The
reported anomalous ilmenite compositions in kimberlites (Haggerty, 1973) need to be

Figure 10.23. Fe2þ environment in CaFe3Ti4O12 perovskite. Four nearest Fe–O distances (thick lines, 2.038(3) Å
and second nearest distances (thin lines, 2.770(7) Å) are indicated (Leinenweber et al., 1995, q 1995
Springer-Verlag).
768 Chapter 10

re-examined to evaluate if these represent the high-pressure ordered perovskites or their

decompression products.
These titanate perovskites are the known Fe-bearing perovskites which are
quenchable. Indeed, the 1 : 3 phase is one of the Im3 perovskites to take iron in the square
planar A // site, which is known to be favoured by Mn3þ or Cu2þ ions (because of the
Jahn – Teller effect).
In the manganese series, A0 Mn7O12, the Mþ, M2þ or M3þ cations are
accommodated at A0 ; with charge balance from Mn3þ and Mn4þ proportions in octahedral
sites. As a result of ordering in octahedral sites, many members of this series (A0 ¼ Ca, Cd,
Sr, La, Ns) belong to trigonal or monoclinic subgroups of Im3 (Bochu et al., 1974).
CaFeTi2O6 shows a perovskite-type structure in which Ca and Fe are ordered into columns
along the c-axis. Ca is in a somewhat irregular site on the 42 screw axis, with Ca –O
distances ranging from 2.3 to 2.7 Å while iron is on two distinct sites with m2 symmetry
that alternate along the 4-axis. Fe is coplanar with the four oxygens and the square is
undistorted. In Glazer’s (1972) notation, the tilt system is aþ aþ c2 :

10.7.2. MgTiO3 –FeTiO3 join


FeTiO3 (ilmenite) and MgTiO3 (geikielite) are stable phases in the ilmenite ðR3Þ
structure at ambient conditions and show complete solid solution (see Fig. 10.24)

Figure 10.24. Calculated phase diagram for the FeTiO3 –MgTiO3 join at 8008C. The two phase loops for the
ilmenite to perovskite transition were calculated using WG(il) ¼ 4,300 J/mol and WG(pv) ¼ 4,300 J/mol (dotted
curve), WG(il) ¼ 2,278 J/mol and WG(pv) ¼ 2,163 J/mol (dashed curve). The solid curve represents ideal mixing
(Source: Linton, 1999, q 1999, Mineralogical Society of America).
ABX3, Perovskite – Ilmenite Structure 769

(e.g., Wood et al., 1991). The high-pressure phase relations between ilmenite and geikielite
have aroused considerable interest because of the observations on such minerals obtained
from ultra-high-pressure rocks, such as Alpe Arami mantle peridotite, wherein ilmenite
with unique structures is seen in olivine (Dobrzhinetskaya et al., 1996).
In the case of FeTiO3, it was observed (Leinenweber et al., 1991) that, at 16 GPa
(RT), in the lithium niobate structure it transformed to an orthohombic perovskite
polymorph. During quenching, this FeTiO3 perovskite reverted back to the lithium niobate
structure (Leinenweber et al., 1991).
MgTiO3 –FeTiO3 ilmenite solid solution can be synthesized from mixtures of pre-
dried Fe2O3, TiO2 and MgTiO3. In FeTiO3, a phase transition is observed at high pressures
from the ilmenite structure (FeTiO3-I) to a lithium niobate structure (FeTiO3-II). At
16 GPa, this FeTiO3-II transforms from lithium niobate structure to an orthorhombic
perovskite polymorph (FeTiO3-III) (Leinenweber et al., 1991). On quenching, this
perovskite (FeTiO3-III) reverts to the lithium niobate structure.
Thermodynamic calculations based on measured enthalpy of II ! I transition show
that, at high pressure, the perovskite (FeTiO3-III) is stable while its quenched phase with
lithium niobate structure ðR3cÞ is metastable (Mehta et al., 1994).
The perovskite (III) phase is a unique example wherein complete solid solution
between Mg and Fe occurs in perovskite structure.
From the thermodynamic data determined through calorimetry (Table 10.7) and
some estimated values of physical parameters (Table 10.8), the phase relations in the
FeTiO3 –MgTiO3 system can be calculated.
The phase diagram of the FeTiO3 – MgTiO3 join at pressure can be calculated by
simultaneously solving the following equations:
ðP
0
DHMgTiO 3
ðil ! pvÞ 2 TDS0MgTiO3 ðil ! pvÞ þ 0
DVMgTiO 3
ðil ! pvÞdP
l atm
! ð10-18Þ
apv
MgTiO3
þ RT ln ¼0
ailMgTiO3

TABLE 10.7
Thermochemical data used in FeTiO3 – MgTiO3 phase-equilibrium calculations (from Linton et al., 1999)

Transition DH 0298 (kJ/mol) S 0 (J/K mol)

FeTiO3(I) ! FeTiO3(II) 14.5 ^ 2.2a 4.9 ^ 3.1

MgTiO3(I) ! MgTiO3(II) 28.8 ^ 1.0b 4.0 ^ 2.6
FeTiO3 (I) ! FeTiO3(III) 27.1 ^ 4.0c 8.9 ^ 3.1c
MgTiO3(I) ! MgTiO3(III) 33.1 ^ 5.1d 5.5 ^ 2.6d
FeTiO3(II) ! FeTiO3(III) 12.3 ^ 5.9a,c 4.0 ^ 4
MgTiO3 (II) ! MgTiO3(III) 4.4 ^ 6.1b,d 1.5 ^ 2
a
Calorimetry (Mehta et al., 1994).
b
Calorimetry (Linton et al., 1999).
c
High-pressure –temperature phase equilibrium (Syono et al., 1980).
d
High-pressure–temperature phase equilibrium (Linton et al., 1999).
770 Chapter 10

TABLE 10.8
Physical properties used in phase-equilibrium calculations (from Linton et al., 1999)

Compound Structure V8298 (cm3/mol) K (GPa)

a
MgTiO3 Ilmenite 30.87 169b
FeTiO3 Ilmenite 31.70c 177(3)c
MgTiO3 Lithium niobate 30.71d 172e
FeTiO3 Lithium niobate 31.40f 182(7)g
MgTiO3 Perovskite 29.31e 225e
FeTiO3 Perovskite 30.15e 236h

Note: dK=dP assumed to be 4. Thermal expansion was assumed equal to that of FeTiO3 ilmenite, 27.9 £ 1026 K21
Wechsler and Prewitt (1984).
a
Wechsler and Von Dreele (1989); b Liebermann (1976); c Wechsler and Prewitt (1984); d Linton et al.
(1997); e Estimated, Linton et al. (1999); f Leinenweber et al. (1995); g Mehta et al. (1994); h Leinenweber
et al. (1991).

ðP
0
DHFeTiO 3
ðil ! pvÞ 2 TDS0FeTiO3 ðil ! pvÞ þ 0
DVFeTiO 3
ðil ! pvÞdP
l atm
!
a0FeTiO3
þ RT ln ¼0 ð10-19Þ
a0FeTiO3

where S is entropy and a is activity. Linton et al. (1999) modelled the activity by
assuming the ideal case g ¼ 1; where the activity, aj ; simply equals the mole fraction, Xj :
Another approach employs non-ideal mixing models. The MgTiO3 – FeTiO3 join for the
ilmenite structure has been described by a symmetrical regular mixing model, RT ln gj ¼
WG ¼ ð1 2 X aÞ2 ; where WG ¼ 4,300 J/mol, based on partitioning of Mg and Fe between
co-existing ilmenites and olivines (Anderson et al., 1991). Using the method of Davies
and Navrtosky (1983), which correlates the interaction parameter with the volume
mismatch of the two end-members, the symmetric WG values have been estimated as
2,278, 2,163 and 1,824 J/mol, respectively, for Mg –Fe mixing in the ilmenite, perovskite
and lithium niobate structures. With these non-ideal mixing models, equations (10-18)
and (10-19) become:
ðP
0 0 0
DHMgTiO 3
ðil ! pvÞ 2 TDS MgTiO3 ðil ! pvÞ þ DVMgTiO 3
ðil ! pvÞdP
l atm
!
0
XMgTiO il
þ RT ln il
3
2 WG ðXFeTiO 3
Þ2 þ WGPV ðXFeTiO
PV
3
Þ2 ¼ 0 ð10-20Þ
XMgTiO 3

and
ðP
0
DHFeTiO 3
ðil ! pvÞ 2 TDS0FeTiO3 ðil ! pvÞ þ 0
DVFeTiO 3
ðil ! pvÞdp
l atm
pv
!
XFeTiO il pv1
þ RT ln il
3
2 WG ðXMgTiO 3
Þ2 þ WGpv ðXMgTiO Þ2 ¼ 0 ð10-21Þ
XFeTiO 3
3
ABX3, Perovskite – Ilmenite Structure 771

Figure 10.25. Typical spectra of geikielite at high pressure. No phase change can be evidenced. Line broadening
is probably due to increase of deviatoric stress with increasing pressure (Reynard and Guyot, 1994, q 1994
Springer-Verlag).

These two equations are solved for different values of WG (il) and WG (pv)
using Newton’s method, with starting values for composition from the exact solutions
of equation (10-15) assuming ideal mixing, to obtain the phase diagram at 8008C
(Fig. 10.25, see caption).
In the (Mg, Fe, Mn)TiO3 – perovskites, the DH 0 of transition between II and I
increases linearly with Mg content (Linton et al., 1999). Similarly, DH 0 (I ! II) should be
a linear function of “A” cation radius (in 6-fold coordination). CdTiO3 is the only one of
the four titanates (CdTiO3, FeTiO3, MgTiO3 and MnTiO3) which, in perovskite form, can
be synthesized in ambient condition and it alone does not show a lithium niobate quench
phase. The upper limit of A cation radius for compounds with a lithium niobate quench
phase is ,0.86 ^ 0.15 Å. The ionic radii for Ni2þ ¼ 0.70 Å, and Co2þ ¼ 0.735 Å would
allow their high-pressure perovskite polymorph to quench to lithium niobate structures on
release of pressure.

10.7.2.1. MgTiO3, geikielite

Geikielite has the ilmenite structure, R3 (space group C3 ), which is based on a
distorted hcp oxygen sublattice and consists of alternate layers of edge-sharing TiO6 and
MgO6 octahedra perpendicular to the c-axis. MgTiO3 is one of the several ABO3
compounds, including FeTiO3, MnTiO3, MnSnO3 and MgGeO3, that form lithium niobate
structure at high pressure (e.g., Leinenweber et al., 1991).
Geikeilite is one end-member of the natural (Fe, Mn, Mg, Zn)TiO3 ilmenite solid
solution series which occurs in magmatic, metamorphic and ultrabasic rocks. Geikielite
(MgTiO3) content in a typical deep-source kimberlitic ilmenite ranges between 25 and
60% (Haggerty, 1991). Dobrzhinetskaya et al. (1996), in studies of Alpe Arami peridotite,
772 Chapter 10

found topotaxially oriented (Fe, Mg)TiO3 rod-shaped precipitates with orthohombic

crystal structures in olivine, which they proposed as being metastable decomposition
structures from (Fe, Mg)TiO3 perovskite.
Geikielite was synthesized from stoichiometric quantities of MgO and TiO2. By
adding dry Fe2O3 and TiO2 with MgTiO3, samples of MgTiO3 –FeTiO3 series were
prepared by Linton et al. (1997). From experiments at 21 GPa and 1,2008C, a new high-
pressure MgTiO3 phase with lithium niobate structure has been observed by Linton et al.
(1997).
At high pressure (RT), FeTiO3, MnSnO3 (Leinenweber et al., 1991) and MnTiO3
(Ross et al., 1989) have been observed to transform from LiNbO3 structure to perovskite
structure, which transforms back to LiNbO3 form upon decompression. The small
difference in the radii of A and B cations in these perovskite polymorphs may cause the
instability of the perovskite polymorphs. This instability is not seen in CaTiO3 ilmenite in
which the difference in radii is not so small.
At high pressure, MgTiO3 – FeTiO3 lithium niobate polymorphs form inter-
mediate compounds, suggesting that MgTiO 3 and FeTiO 3 have isostructural
high-pressure forms. However, thermodynamic calculations indicate that the stable
FeTiO3 phase at high pressure has the perovskite structure while the lithium niobate
polymorph is a metastable quench phase. This suggests that these Mg –Fe –titanates with
lithium niobate structures may also be metastable quench phases from perovskite.
The heat capacities and entropies of MgTiO3 ilmenite are presented in
Table 10.9.

Vibrational modes. The ilmenite structure is dense (2/3 of the octahedral sites of the
pseudo-hcp oxygen sublattice are occupied) and MgO6 and TiO6 octahedra share faces
between the successive layers of the structure. Strong mixing between Ti, Mg and O
motions is thus expected.

TABLE 10.9
Heat capacities and entropies of MgTiO3 – ilmenite (from Reynard and Guyot, 1994)

T CVh CVa CPh CPa Sa CPmes Smes

(K) (J/mol/K) (J/mol/K) (J/mol/K) (J/mol/K) (J/mol/K) (J/mol/K) (J/mol/K)

100 24.1 24.2 24.4 24.5 10.7 25.29 10.99

200 66.6 66.9 67.3 67.6 41.4 68.00 42.50
298.15 90.5 91.0 91.9 92.4 73.5 91.76 74.56
500 110.2 111.1 112.6 113.5 127.3 114.8 129.0
700 116.9 118.2 120.2 121.5 167.0 121.8 168.9
900 119.9 121.6 124.1 125.8 198.2 126.5 200.1
1,200 122.0 124.2 127.7 129.9 235.6 132.6 237.3
1,500 122.9 125.8 129.0 132.8 264.3 138.5 267.5
1,800 123.5 126.9 132 135.4 288.8 144.2 293.3

The subscripts h and a refer to properties calculated in the harmonic approximation and with anharmonic
corrections, respectively. All calculated values use DOS. The correction from CV to CP is calculated with
a ¼ 3.0 £ 1025 K21, KT ¼ 169 GPa and V ¼ 3:00855 £ 1026 J/Pa, all assumed constant with temperature.
ABX3, Perovskite – Ilmenite Structure 773

The primitive unit cell has two formula units giving 30 normal modes of vibration.
Factor-group analysis (Farmer and Lazarev, 1974) gives the symmetry species for the 30
normal modes at K ¼ 0 as:

Gvib ¼ 5Ag ðRÞ þ 5Eg ðRÞ þ 5Au ð1A þ 4IRÞ þ 5Eu ð1A þ 4IRÞ ð10-22Þ

where A refers to acoustic modes, R Raman mode and IR infrared mode.

HPHT Raman investigation of MgTiO3 was undertaken by Reynard and Guyot
(1994) in order to determine the extent of intrinsic anharmonic effects in ilmenite structure.
These effects were taken into account in the vibrational modelling of heat capacities of
MgTiO3 at high temperature and their bearing on various properties (e.g., Cp ; S) of
MgTiO3 ilmenites susceptible of undergoing an order –disorder phase transition (ilmenite
to corundum) at high temperature.
The typical spectra of geikielite at high pressure are shown in Fig. 10.25 (Reynard
and Guyot, 1994). Reynard and Guyot (1994) found positions of all the 10 Raman active
modes from a powder spectra of geikielite at 225, 281, 306, 328, 352, 398, 487, 501, 641
and 716 cm21.
The isothermal mode Grüneisen parameters (Grüneisen, 1912) are calculated from
the wave number shifts with pressure:

giT ¼ ðd ln ni =d ln VÞT ¼ KT ðd ln ni =dPÞT ð10-23Þ

with KT ¼ 169 GPa (Liebermann, 1976). The values of giT are scattered between 1.35 and
0.7, yielding a mean Grüneisen parameter
X
giT Cvi
kgl ¼ iX ð10-24Þ
Cvi
i

of 1.3 at ambient temperature and of 1.1 in the high-temperature limit.

Having determined the wave number shifts with temperature, it is possible to
calculate the isobaric microscopic parameter:
gP ¼ ðd ln ni =d ln VÞP ¼ 21=aðd ln ni =dTÞP ð10-25Þ
25 21
where the thermal expansion coefficient, a; is estimated as 3 £ 10 K , i.e., equal to that
of FeTiO3 (Wechsler and Prewitt, 1984).

Order –disorder transition. Reynard and Guyot (1994) observed no evidence of order –
disorder transition in MgTiO3 – and MgSiO3 – ilmenite quenched from high-pressure
experiments. However, in FeTiO3, an order – disorder transition (random distribution of Ti
and Fe on their octahedral sites) is expected to be around 1,670 K, based on
thermodynamic modelling of the Fe2O3 – FeTiO3 (hematite –ilmenite) solid solution
(Ghiorso, 1990).
The different behaviour of ilmenite and geikielite may arise from the different
electronic structures of Fe and Mg. No charge transfer occurs between Mg and Ti in
774 Chapter 10

geikielite or in silicate ilmenites that favour an order –disorder transition at high

temperatures. No extensive cation disordering can occur in MgSiO3 – ilmenite within its
stability field. However, the presence of a trivalent cation like Al may favour an order –
disorder transition, as happens in majorite garnets (McMillan et al., 1989), and that may
significantly affect the stability field of silicate ilmenite in the MgO – Al2O3 –SiO2 and
natural systems (Reynard and Guyot, 1994).

10.7.2.2. High-temperature phase transition (without order –disorder)

No phase change (or rather, an order –disorder transition) is seen in high-
temperature spectra of giekeilite. Such a change would be accompanied by a symmetry
change from R3 (ilmenite structure) to R3c
 (corundum or hematite structure), which would
reduce the number of Raman active modes to seven. Three modes of Ag symmetry in the
ilmenite become inactive (A2g symmetry) in the disordered corundum structure and two
(A1g symmetry) remain Raman active (Reynard and Guyot, 1994).
In an order –disorder transition occurring below the melting point, some static
disorder should be quenched with a resulting increase in Raman linewidths. But no
such feature, nor the disappearance of three modes of Ag symmetry, could be tracked
by Reynard and Guyot (1994) — possibly because of the anharmonic broadening and
thermal emission from their sample. The intrinsic anharmonic mode parameters are
calculated as:

ai ¼ ðd ln ni =dTÞn ¼ aðgiT 2 giP Þ ð10-26Þ

The average behaviour of MgTiO3 at low temperature is close to that of a

quasi-harmonic solid (Note: The anharmonic parameters have, in general, higher
absolute values for low wave-number modes (typically lattice modes) than for high
wave-number modes (typically internal modes of “rigid” molecules such as SiO4 or
CO3 groups in silicates and carbonates). Also, the quasi-harmonic approximation is
valid as long as only the lowest energy levels of the oscillator are occupied.)
Reynard and Guyot (1994) observed a significant curvature of the wave-number
shifts and line-width increase at temperatures in excess of 1,200 K, irrespective of the
functional used in the fitting procedure. This deviation may be due to an increase in the
intrinsic anharmonic effects with temperature, as seen in several compounds that do not
undergo phase transitions (e.g., Gervais et al., 1973).

10.7.2.3. FeTiO3 structure

FeTiO3 ilmenite is one of the five high-pressure titanate, stannate and germanate
compounds which show reversible transitions from lithium niobate structure to
orthorhombic perovskite at high pressure. In its structure, both Fe and Ti are octahedrally
coordinated. Each octahedron shares a face with a vacant octahedron, leading to large
displacements of the cations away from the centres of the octahedra due to second-
neighbour repulsions. Thus, the arrangement of Fe and Ti remains eccentric, causing
ferroelectric behaviour (as in LiNbO3). In one of the two O6 octahedra, the cation is off-
centre, reducing the symmetry from R3 to R3c.

ABX3, Perovskite – Ilmenite Structure 775

TABLE 10.10
Structural parameters for FeTiO3

Atom Site x y z

Fe 6a 0.0 0.0 0.2871(1)

Ti 6a 0.0 0.0 0.0000
0 18b 0.0449(5) 0.3446(4) 0.0641(3)

The selected bond lengths are: Fe – O: 2.184(3) £ 3, 2.061(4) £ 3; Ti – O: 1.882(3) £ 3, 2.112(3) £ 3.

a ¼ 5.12334(5) Å, c ¼ 13.76020(17) Å, Z ¼ 6; R ¼ 0:034:

The TiO6 octahedral framework of FeTiO3 follows Glazer’s tilt system a2 a2 a2

(SG R3c). In the fully collapsed, high-temperature lithium niobate structure,
which occurs for a rotation of 308 away from the cubic perovskite, the A cation is
in the centre of a triangle of oxygens which is the shared face between two O6
octahedra. A c-axis projection of a layer in this structure is shown in Fig. 10.20 for
three rotation angles. In the FeTiO3, as in low-temperature lithium niobate, the cation is
off-centre.
The structural parameters for FeTiO3 are presented in Table 10.10.
At high pressure, the FeTiO3 tilt system changes reversibly to aþ b2 b2 (SG Pmnb);
this is in accord with in situ XRD results (Leinenweber et al., 1991).
The ordered perovskite appears to form at pressures much lower than that for the
presumed ilmenite – perovskite transition in FeTiO3 (,13 GPa at 1,2008C). Both the 1 : 1
and 1 : 3 phases form readily at pressures of 12.5 – 15 GPa at temperatures between 1,150
and 1,4008C.

(Fe, Mg) – Ilmenite transformation. The ilmenite structure has been suggested as
the successor to pyroxene and garnet in the Earth’s transition zone. Pyroxene –
ilmenite transformation is indicated by the hexagonal form of MgSiO3 reported by
Kawai et al. (1974). The garnet –ilmenite transformation was supported from the studies of
Mg3Al2(Ge,Si)3O12 garnets (often stated as Mg – Al– germanate garnets).
Natural ilmenite from kimberlite of Kentucky (USA) was studied by Liu (1975). It
was a solid solution of three major end-members, FeTiO3 –MgTiO3 – Fe2O3. The sample in
fine polycrystalline form was compressed between two flat anvil faces and heated by a
continuous YAG laser and the pressure was calibrated according to NaCl pressure scale at
RT (accuracy ^ 10%). The temperature of the laser heating, estimated by using an optical
pyrometer, ranged between 1,400 and 1,8008C. The observations at 14 and 25 GPa are
discussed below.

14 GPa. In the loading pressure range 12– 25 GPa, ilmenite displays two steps of phase
transformations. The XRD data for the sample quenched from loading pressure of
,14 GPa and a temperature of ,1,400 –1,8008C are shown in Table 10.11 (Liu, 1975).
776 Chapter 10

TABLE 10.11
Room-temperature and 1bar pressure XRD data for ilmenite quenched from a loading pressure of about 140 kbar
and temperature of 1,400–1,8008C (Co-Ka) (Liu, 1975)

Observed Ilmenitea Perovskiteb

I/I100c d (Å) I/I100 d (Å) hkl d (Å) hkl

60 3.716 60 3.718 102 3.715 002

30 3.515 3.530 110
25b 3.303 3.287 (112)
30 2.877 2.877 020
100 2.738 100 2.736 104 2.738 (140)
20 2.712 2.710 (231)
90 2.547 80 2.539 110 2.558 112
5 2.476 2.476 003
90 2.230 70 2.228 113 2.236 200
5 2.155 2.155 (322)
70 1.859 70 1.860 204 1.857 004
5 1.717 80 1.715 116 1.715 131
50 1.701 1.704 032
10 1.662 1.659 203
5 1.619 40 1.622 108 1.621 (530)
1.611 (502)
5 1.594 1.594 213
1.592 132
30 1.503 50 1.500 214
30 1.469 50 1.463 300 1.466 (105)
5 1.306 1.303 321
5 1.266 15 1.269 220
10 1.149 50 1.151 314 1.151 (355)
1.143 (505)
10 1.112 40 1.113 226 1.118 400
a
Measured from standard power method (Fe–Ka).
b
Calculated from a0 ¼ 4:471 Å, b0 ¼ 5:753 Å and c0 ¼ 7:429 Å; indexes in parentheses correspond to reflection
for cell with a and b axes doubled.
c
Estimated visually; the letter b denotes broad line.

The observed d-spacing can be indexed as a two-phase mixture of ilmenite and an

orthorhombic perovskite-like structure. The three reflections, 3.303, 2.712 and 2.155 Å,
are not observed in ilmenite and belong to perovskite structure.
The zero-pressure volume (28.77 ^ 0.08 cm3/mol) for the high-pressure perovskite
phase is 8% smaller than that of the starting material, ilmenite (FeTiO3). Similarly, the
perovskite phase of MgSiO3 found earlier by Liu (1974) indicated a volume change of
7.4% for the ilmenite –perovskite transition in MgSiO3.
The high-pressure lithium niobate phase of ilmenite was synthesized by Mehta et al.
(1994) at 15 GPa and 1,473 K, using uniaxial split-sphere high-pressure apparatus (USSA,
2000). Thermodynamic calculations using the enthalpy of the ilmenite to lithium niobate
transition indicate that the ilmenite to lithium niobate phase boundary is metastable with
ABX3, Perovskite – Ilmenite Structure 777

TABLE 10.12
Room-temperature and 1-bar pressure XRD (Co-Ka) data for ilmenite quenched from a loading pressure of about
250 kbar a temperature of 1,400–1,8008C (C, cubic; R, rhombohedral) (Liu, 1975)

Observed Ilmenitea Mixed oxidesb

I/I100a d (Å) I/I100 d (Å) hkl d (Å) hkl

Co-Ka
25 2.733 100 2.736 104
30 2.604 2.572 C (111)
30 2.464 2.460 R (111)
80 2.223 70 2.228 113 2.228 C (200)
100 2.129 2.130 R (200)
50 1.508 50 1.500 214 1.506 R (220)
Mo-Ka
20 2.73 100 2.736 104
30 2.58 2.572 C (111)
30 2.46 2.460 R (111)
100 2.221 70 2.228 113 2.228 C (200)
70 2.129 2.130 R (200)
50 1.568 1.575 C (220)
40 1.505 50 1.500 214 1.506 R (220)
80 1.284 1.286 C (222)
1.284 R (311)
5 1.235 1.230 R (222)
20 1.116 40 1.1133 226 1.1138 C (400)
20 0.9954 0.9962 C (420)
5 0.9549 0.9526 R (420)
a
Estimated visually.
b
Calculated from a0 ¼ 4:260 Å for the rock-salt structure and a0 ¼ 4:455 Å for the cubic phase.

respect to the ilmenite to perovskite phase boundary and the stable phase at high pressure is
probably the perovskite phase.

25 GPa. XRD of the ilmenite quenched from a loading pressure of about 25 GPa (and
T ¼ 1,400 – 1,8008C) are shown in Table 10.12 (Liu, 1975). Co-Ka radiation gives better
information for low-angle reflections and the Mo-Ka radiation provides the high-angle
reflections.
The products are mainly a mixture of (Fe, Mg)O of rock-salt structure and another
of cubic form, together with some ilmenite. The cubic form (TiO2) can be indexed in the
same space group as that of the fluorite (CaF2) structure, Fm3m:
Therefore, at high pressure, the transformation is:

ðFe; MgÞTiO3 ! ðFe; MgÞTiO3 ! ðFe; MgÞO þ TiO2

ilmenite perovskite mixed oxides

The values of the lattice parameters found for the rock salt and the cubic phase
indicate a decrease of 13.3% in the zero-pressure volume from the perovskite to the oxide
778 Chapter 10

mixture. This corresponds to a 20.2% volume decrease from ilmenite to the mixture. These
are in excellent agreement with the apparent change in volume in the shock-Hugoniot data.

Comments. Natural ilmenite with composition (Fe, Mg) TiO3 from a kimberlite pipe was
found to transform to the perovskite structure at a loading pressure of ,14 GPa and
T between 1,400 and 1,8008C. The perovskite phase of (Mg, Fe)TiO3 was further observed
to disproportionate into a mixture under a loading pressure of about 25 GPa and a
temperature between 1,400 and 1,8008C.
The hypothesis that the core consists of a metallic form of mantle material, originally
proposed by Bernal (1936) and revived by Ramsey (1948), has been discarded in recent
years in favour of an iron core with light constituents. The hypothesis of a metallized silicate
or oxide core has been revived on the basis of the discoveries of conductive Fe2O3, FeO,
TiO2 and SiO2 under static high-pressure conditions and also in shock studies. The phase
transformation of ilmenite to perovskite structure then to the cubic high-pressure phase of
TiO2 plus (Fe,Mg)O provides a basis for appraisal of these conclusions.

MnTiO3 – ilmenite. MnTiO3 has the ilmenite structure (a ¼ 5:137 Å, c ¼ 14:283 Å)
belonging to space group R3 and has six formula units in the unit cell. The structure is
anti-ferromagnetic with anti-parallel spins in each layer, at z ¼ 0 and 1/2, directed
along the c-axis. Below TN ; the two-dimensional order decreases as the three-dimensional
anti-ferromagnetic long-range order increases.

10.7.2.4. Shocked FeTiO3 – ilmenite: Mössbauer study

Ilmenite (FeTiO3) structure is a sensitive indicator of shock events in nature.
Lunar ilmenites show complex lamellar twinning (e.g., Haggerty et al., 1970), which is
ascribed to shock effects by meteorite impacts. In ilmenite, the deformation becomes
more complex with increasing strain rates and the most complex deformation can be
introduced by shock loading. The line widths of diffraction peaks of shocked ilmenite
for shock pressures up to 80 GPa are shown in Fig. 10.26. Sclar et al. (1973) claimed by
neutron diffraction study that a partial cationic disordering is introduced in artificially
shocked ilmenites.
Shock-compression experiments of single-crystal ilmenite to pressures of 93 GPa
(King and Ahrens, 1976) revealed an anomalous large compression above ,30 GPa and a
remarkable compression anisotropy in the crystal between 30 and 60 GPa. This is also
reflected in the change in line widths of XRD lines (Fig. 10.26). They interpreted the
anomalous compression as the onset of phase transition, possibly the same disproportiona-
tion reactions reported in the static high-pressure experiments (Liu, 1975).
In another experiment, powdered specimens of ilmenites were subjected to pressures
of 33, 55 and 80 GPa and studied by Mössbauer spectroscopy (Syono et al., 1981). The
patterns for both the shocked and unshocked samples are shown in Fig. 10.27. Remarkable
line-profile broadening in shocked ilmenites is clearly observed. While the IS value does not
change remarkably, the QS and line width both increase with the shock pressure.
The contribution from the highly disordered regions induced by shock loading is
estimated by subtracting the normalized spectrum of the unshocked specimen from the
ABX3, Perovskite – Ilmenite Structure 779

Figure 10.26. Half line-width of diffraction lines of shocked ilmenite vs peak loading pressure. Fe-Ka radiation is
used.

shocked one, the difference being shown as a hatched area in Fig. 10.27. It is interesting to
compare the MS of the shocked ilmenite with that of the disordered ilmenite (FeTiO3 II)
synthesized at static high pressures by E. Ito. In that material, divalent (Fe2þ) and
quadrivalent (Ti4þ) ions are randomly distributed among octahedral sites in the
hexagonally close-packed oxygen framework with overall crystal symmetry R3c  (the
same as corundum structure).
The results of Mössbauer experiments of shock-loaded ilmenite are consistent with
shock residual effect observed by X-ray analysis. The increase in QS observed in shocked

Figure 10.27. Mössbauer spectrum of the unshocked (top) and shocked (bottom) ilmenites measured at room
temperature. In the bottom figure the hatched region is obtained by subtracting normalized spectrum of on
shocked ilmenite (thin line) from the observed highly strained regions is the shocked specimen with larger
quadrupole splitting. The position of the Mössbauer spectrum of the disordered ilmenite. FeTiO3 II measured at
room temperature is also shown. The velocity is referred with respect to pure iron metal (Source: Co57 in Rh.)
(Syono et al., 1981, q 1981 Springer-Verlag).
780 Chapter 10

crystal is due to atomic disorder induced by shock loading (Syono et al., 1981) but
the Mössbauer experiment is not sensitive to the long-range coherence, as is revealed by
X-ray data.
An increase in QS, in addition to line broadening, observed in shocked crystals may
reflect the disordering; QS is expected to show increased values owing to an increase in
EFG at disordered regions. Short-range configuration in the sheared region produced by a
heterogeneous yielding process during shock loading is probably similar to those in
disordered ilmenite.

10.7.3. Xenoliths in Kimberlites

Some garnet xenoliths in kimberlites originate from pressures as high as

12 –13 GPa (,400 km; Haggerty and Sautter, 1990). Kimberlites commonly contain
xenoliths of FeTiO3. The pressure –temperature regime in which kimberlite originates
cause the FeTiO3 phase to transform from ilmenite to perovskite structure.
The pressure of the ilmenite to perovskite transition in the mantle can be determined
from a superimposition of the mantle geotherm on the FeTiO3 phase diagram. The Clark
and Ringwood (1964) shield geotherm and the Lesotho (Boyd and Nixon, 1975) geotherm
are superimposed on the ilmenite –perovskite phase boundary in Fig. 10.28. The transition
occurs along the geotherm in the pressure range 7 –10 GPa and the temperature range
1,600 –1,800 K (Mehta et al., 1994). Experimental geopiezometers and growth-rate data

Figure 10.28. Comparison of selected geotherms for kimberlites with FeTiO3 phase boundary (after Mehta et al.,
1994, q 1994 Springer-Verlag).
ABX3, Perovskite – Ilmenite Structure 781

Figure 10.29. Eight-site model for perovskite ferroelectrics, particularly BaTiO3 and KNbO3. The cube
represents the A-cations Ba or K. The oxygen atoms (not shown) lie approximately at the of each face. The
dark spheres in the inner cube indicate the preferred “site” for the B-cations Nb. The cubic phase of PbTiO3
may also be disordered as shown here, but the tetragonal phase PbTiO3 most likely is ordered with Ti
displacements along a cubic axis as given by the average structure. The tetragonal phase of PbTiO3 is ordered
due to the large lattice strain ðc=a ¼ 1:0Þ stabilize the [001] displacement of the Ti over the [111]
displacement. Both the oxygens and cations move by similar amounts (After Dougherty et al., 1992 and
Cohen, 2000).

for olivine recrystallization (Mercier, 1979) suggest that kimberlites rise very rapidly to the
surface from mantle sources.

10.7.4. BaTiO3

The structural configuration of ABO3 perovskite family (A includes Na, K, Ca, Sr,
Ba and B includes the transition metals) can best be illustrated with BaTiO3. In its highest
symmetry, it is cubic (space group Pm3m) with a ¼ 4:012 Å at 435 K. Below 408 K, the
structure transforms to tetragonal (space group P4mm) with a ¼ 3:9920 Å and
c ¼ 4:0361 Å at 293 K. The elastic and the electric behaviour of BaTiO3 arising from
its structural properties are discussed below.
In perovskite-like BaTiO3 and PbTiO3, the O2p states strongly hybridize with the
d0 cation, in this case Ti4þ, reducing the short-range repulsions and thus allowing off-
centre displacements (see Section 10.6.1.1).
782 Chapter 10

Cohen (1992) determined that the shortest Ti –O distances in the tetragonal

structures of BaTiO3 and PbTiO3 are 1.86 and 1.78 Å, respectively, significantly smaller
than the 2.00 Å expected from the Shannon and Prewitt (1969) ionic radii.
On BaTiO3, there are two types of surfaces with either TiO2 or BaO terminations.
On the TiO2 surface, the dangling Ti bond relaxes back onto the surface but
atomic relaxations are quite small. A surface state was found in the band structure. The
surface energy of BaTiO3 is high, explaining the poor cleavage in BaTiO3 and the
rough surface that forms on the fracture. Defects, domain boundaries and surfaces
play important roles in ferroelectrics. Oxygen vacancies in PbTiO3 may be important
in fatigue.
In the ideal order –disorder scenario, the atoms are in off-centre sites at all
temperatures and there should be a large configurational entropy change at the ordering-
phase transition.
When the strain effects are absent or minor, the lowest energy off-centre
displacements are along the (III) directions. Thus, the ground state is rhombohedral. As
temperature is raised, the off-centre displacements disorder over two directions, giving an
average orthorhombic symmetry, then there is disordering over four directions giving
tetragonal symmetry and, finally, a cubic phase at high temperatures. The same features are
observed in BaTiO3 and KNbO3 (see Figs. 10.29 and 10.32).

Figure 10.30. Records of ferroelectric hysteresis loops of BaTiO3 single crystal at room temperature at
varying pressures. The polarization (g-axis) is seen to increase first slightly with pressure then to decrease,
passing through a maximum around 2.5– 3.0 GPa (after Schloessin and Timco, 1977, q 1977 Elsevier
Science Ltd.).
ABX3, Perovskite – Ilmenite Structure 783

Figure 10.31. Difference of self-consistent LAPW charge density for tetragonal BaTiO3 relative to overlapping
ions having (a) nominal charges, Ba2þ, Ti4þ, and O22 and (b) Ba2þ, Ti2.98þ and O1.632. This is a (110) slice is
shown with Ti in the center, Ba at the corners and O at the top and bottom centers. The contour interval is
0.005 e2/bohr3 and the dashed contours are negative. Note that the differences are centered around the Ti atom or
in the Ti –O bond, and are not localized around the oxygen, as are assumed in the shell model (From Cohen and
Krakauer, 1990).

Figure 10.32. Displacive phase transitions in perovskites, involving displacement of cation or rotation of
octahedra. BaTiO3 shows eight sites for Ti4þ cations in the cubic phase (from Dove, 1997).
784 Chapter 10

10.7.4.1. Ferroelectricity and ferroelasticity

The eight-site model for perovskite ferroelectrics is shown in Fig. 10.29
(see caption; from Cohen, 2000). The Ba and Ti atoms are each displaced from their
mean oxygen planes by 0.074 and 0.128 Å, respectively. A spontaneous polarization of
about 26 £ 1022 cm22 originates in the dipoles associated with these displacements along
the polar axis. The c-axis may be transformed into an a-axis at room temperature by the
application of uniaxial pressure. This interchange depicts that the phase is ferroelastic.
This displacement, parallel to the polar axis of the Ba and Ti atoms from the mean oxygen
planes, is nearly identical with the tetragonal displacement. For BaTiO3, the transition
pressure is obtained from the linear relationship:
Ptr ðMPaÞ ¼ 0:17 2 21:25½ðTtr ÞP 2 ðTtr ÞLV 
where LV is the liquid –vapour pressure of pure H2O. The data of Decker and Zhao (1989)
indicate that the equation given above is probably valid to the tri-critical pressure for
BaTiO3 of about 3.5 GPa. In the stability field of BaTiO3 in its tetragonal modification, the
FE transition temperature decreases with increasing pressure. In an isolated crystal at room
temperature, the FE hysteresis loop is no longer observable beyond about 2.5 –3.0 GPa
(Fig. 10.31).
In both tetragonal and orthorhombic phases, the interchange of axes results in
reorientation of the polar axis by 908. Full coupling is expected between ferroelasticity and
ferroelectricity in orthorhombic BaTiO3. The cubic to tetragonal phase transformation of
BaTiO3 is seen to be associated with an instability in the excitation spectrum, which
displays the dynamical nature of this transition.
Ferroelectricity is a bulk property, although surface effects may be crucial and size
effects may be huge. Photo-emission studies on BaTiO3 (sputtered and annealed) reveal
significant surface defects (Cord and Courths, 1985). Molecular Xa calculations give some
insights into possible surface effects and the effects of Madelung fields and charge
relaxation at surfaces (Tsukada et al., 1980). First-principles electronic-structure
approaches have been very successful for bulk ferroelectrics (Cohen, 1993; Zhong et al.,
1994; Yu and Krakaner, 1995; Singh, 1995) but the application of first-principles band-
structure methods to surface properties is discouraging because of the huge computational
problem involved.
Schloessin and Timco (1977) employed two independent techniques for studying
the ferroelectric properties (of titanate perovskites) under pressure:
(1) Determinations of the Curie temperature, dielectric constant (permittivity) values and
dissipation factors by means of a capacitance (General Radio 1620 AP) measuring
assembly.
(2) Hysteresis loop observations, using a Sawyer-Tower circuit, which allowed
determination of the Curie temperature and gave semi-quantitative information
about polarization, coercivity and possible perturbations arising from lattice defects.
In the experiments, well-developed ferroelectric hysteresis loops were seen to occur
as persistent up to pressures of ,5 GPa (i.e., beyond the Moon’s centre pressure).
In an isolated crystal under purely hydrostatic pressure, the ferroelectric transition
is displaced below room temperature as the pressure increases. However, the properties of
ABX3, Perovskite – Ilmenite Structure 785

a composite system are far more complex. The measurements are made in pyrophyllite as
the pressure-transmitting medium. For the tetragonal modifications, the peak positions are
shifted towards lower temperatures and shifts are accompanied by varying degrees of
peak broadening depending on the composition and internal stresses (Timco and
Schloessin, 1976b). The experiments (on BaTiO3) were performed in the temperature
range of the tetragonal modification for which ðdTc =dPÞ is negative. The temperature
for the tetragonal – orthorhombic transition in BaTiO3 is seen generally to increase
with pressure.
In the case of BaTiO3 and PbTiO3, the hybridization of Ti 3d states and O2p states
is essential for ferroelectricity. However, the difference in ferroelectric-phase behaviour of
two materials evidently lies in their structures. In PbTiO3, the states of Pb and O hybridize,
leading to a large strain that stabilizes the tetragonal phase whereas, in BaTiO3, the
interaction between Ba and O is completely ionic, favouring a rhombohedral structure
(Cohen, 1992).
Both are para-electric (non-polar) at high temperatures and have the cubic
perovskite structure (which is simple cubic; space group Pm3m) with “A cations” (e.g., Pb
and Ba) in the large 8-fold coordination site at (0,0,0), “B cations” (e.g., Ti) in the
octahedrally coordinated site at (0.5, 0.5, 0.5) and oxygens at the equipoint (0.5, 0.5, 0).

BaTiO3 vs PbTiO3. BaTiO3 has three ferroelectric-phase transitions at different

temperatures:
393 K: cubic ! tetragonal
278 K: tetragonal ! orthorhombic
183 K: orthorhombic ! rhombohedral.
On the other hand, PbTiO3 has only one temperature for ferroelectric-phase
transition:
766 K: cubic ! tetragonal.
As stated above, BaTiO3 shows a series of phase transitions, cubic ! tetragonal !
orthorhombic and to rhombohedral with decreasing temperature (see also Fig. 10.29),
whereas PbTiO3 shows a single phase transition to tetragonal. When tetragonal strain is
included, PbTiO3 shows a much deeper well whereas, in BaTiO3, the rhombohedral phase
has the lowest energy. Thus, the tetragonal strain is responsible for the tetragonal ground
state in PbTiO3.
Analysis of charge densities and DOS show that the ferroelectric instability is due to
hybridization between the O2p states and the Ti 3d states. The hybridization between the
O2p and Ti 3d can lead to a multiple-well potential surface with minima along the eight cube
diagonals with order – disorder transitions between the four phases of perovskite BaTiO3.
In BaTiO3, the Ba is quite ionic and spherical whereas the Pb in PbTiO3 is not very
spherical in the ferroelectric phase. The polarization of the Pb helps stabilize the large strain
and the tetragonal ground state in PbTiO3. PbTiO3 has a very large strain (6%) whereas
tetragonal BaTiO3 has only a 1% strain. The polarizability of Pb plays a special role in
ferroelectrics.
786 Chapter 10

10.7.4.2. Linearized augmented plane wave (LAPW) calculations: surface effects

For predicting ferroelectric behaviour, more and more computational methods and
experimental techniques are being employed in the hope of developing the ability to
predict ferroelectric behaviour. In the LAPW method, an all-electron method is employed
with no shape approximations for the charge density or potential and no pseudo-potential
approximation. The difference in the self-consistent LAPW charge density and
overlapping spherical ions for tetragonal BaTiO3 is presented in Fig. 10.31. The frozen-
phonon calculations indicate gigantic sensitivity of the potential surface for ferroelectric
distortions to volume and shear strain. This indicates the possibility of tuning the phase
transition by applying external stresses to crystals.
For BaTiO3 and PbTiO3, the LAPW frozen-phonon calculations have shown that
hybridization between the d0 cation and oxygen is the dominant mechanism for phase
transformation. This hybridization softens the short-range interaction and allows the
instability, driven by the long-range Coulombic interactions.
To evaluate the effects of surface on ferroelectric BaTiO3, Cohen (1996) performed
LAPW calculations for periodic (001) and (111) slabs of it. The (111) slab is found to be
much less stable than the (001) slab. The more stable (001) slab was studied in the ideal
configuration, with a tetragonal ferroelectric distortion and with some surface relaxation.
The (001) slab has 4-mm tetragonal symmetry with eight space-group operations and it has
two types of charge-balanced surfaces. One surface contains Ti and O in a ratio of 1 : 2 and
the other consists of equal amounts of Ba and O.
In bulk BaTiO3, the tetragonal phase consists of displacements towards the cube
diagonals and the tetragonal structure is a dynamical average with hopping among
four [111] directions. Ferroelectric distortion moves Ba out of the Ba – O plane
by 0.06 Å and displaces the surface Ti by about 0.056 Å. Using periodic slab LAPW
computations for BaTiO3 ferroelectric, Cohen (1996) found that the surface Ti bond
relaxes back into the surface and self-heals, making the Ti – O surface an excellent
substrate for epitaxial growth. The electronic surface effects do not extend deeply
into the solid. The muffin-tin radii were taken as 2.3 Bohr for Ba, 1.75 Bohr for
Ti and 1.75 Bohr for O, and the convergence parameter, RKmax ; was set to 7.0. A
4 £ 4 £ 2 special K-point mesh was used which gives 3K points for the (001) slab. The
matrix order (number of basis functions) for the slab was about 2,900. These
calculations were computationally intensive and extremely time consuming
(Cohen, 1996).

10.7.4.3. Multiple-site model for perovskite ferroelectrics

In perovskites, there is often more than one phase transition with decreasing
temperature. In BaTiO3 and KNbO3, the observed sequence is cubic ! tetragonal !
orthorhombic ! rhombohedral with the macroscopic polarization pointing along the
[001], [110] and [111] directions (see Fig. 10.29).
For displacements of the octahedral B-cation (Ti or Nb), a local minima may be
assumed to lie along the diagonal [111] directions. In the eight-site model, the cubic phase
is a disordered one because all the eight sites are randomly occupied (Fig. 10.29). The
tetragonal phase has four sites preferentially occupied, while the orthorhombic phase has
ABX3, Perovskite – Ilmenite Structure 787

Figure 10.33. Schematic diagram of the evolution of the optic mode frequencies in BaTiO3 and KNbO3. In
PbTiO3 only the cubic to tetragonal transition occurs (After Fontana et al., 1984).

disorder between the two sites. The ground-state orthorhombic structure is the only ideal
ordered phase. However, this model ignores the long-range coupling and the fact that
oxygen moves, as do the B-cations.
The occasional hopping of atoms between different local minima causes a low
frequency — too low for a study of the soft-mode dynamics using normal Raman
techniques. This low frequency is typically at MHz to GHz range, whereas typical phonon
frequencies are in THz range.
Orthorhombic KNbO3 shows low-frequency relaxation features with the A21
symmetry response and displacements along the polar axis, c. Its soft-mode symmetry ðB2 Þ
with displacements perpendicular to c shows only a soft-mode response. In Fig. 10.33
(Fontana et al., 1984), the evolution of optic-mode frequencies in BaTiO3 and KNbO3 are
shown (see the caption).

10.7.4.4. Ferroelectric instability: “rattling-ion” model

It was proposed that the B-cation in perovskite structure experiences an octahedral
site too large for it and it rattles to be displaced off-centre. This is favoured by Coulomb
forces, when the attractive and repulsive forces balance each other. In the ideal position of
B-cation, the Madelung energy is at a maximum.
788 Chapter 10

The overlaps of the spherically symmetric closed shells of Ba2þ or Pb2þ, Ti4þ and
22
O ions in a non-empirical potential induced breathing (PIB) model make the cubic
structure stable (but the soft mode is of high frequency; Gong and Cohen, 1992).
In electronic structure calculations using the LAPW method, the nuclei are placed
on the cubic perovskite positions and self-consistent interactions with polar electronic
charge density are considered. The charge density always relaxes to a symmetric cubic
structure. The results also correspond to those obtained from using cubic symmetry.
However, use of the ionic radii or well-constrained potential models can hardly help
in correctly predicting ferroelectricity. The large polarizability of O22 ion stabilizes the
ferroelectric phase and causes ferroelectric instability. The ferroelectric instability in
perovskite titanates originates from the covalent hybridization between O2p and Ti 3d.
This holds for other perovskite ferroelectrics with d0 B-cations. This hybridization differs
from a vibronic mechanism in which a pseudo-Jahn– Teller effect is proposed to be
responsible for ferroelectricity (Bersuker and Polinger, 1989). But no evidence is found to
support the contention that only the changes in hybridization, as in the pseudo-Jahn –Teller
effect, drive the ferroelectric instability (Cohen, 1993).
In ferroelectrics, dopants are important in tuning the transition temperatures and
dielectric properties and also in enhancing the poling (or depoling) of crystals. Some
chemically disordered (intra-crystalline) or varieties of solid solutions (“relaxors”) serve as
useful ferroelectrics and some defects may switch fatigue in ferroelectrics. The order –
disorder problem in ferroelectrics may be studied using the PIB model (Burton and
Cohen, 1993).

10.7.5. PbTiO3

In a synthesized PbTiO3 sample, the tetragonal – cubic phase transition temperature

was determined at 1 atm and along six isochores of H2O through heating and cooling
experiments (Chou and Haselton, 1994). Least square regression of the data gave:
Ptr;h ðMPaÞ ¼ 7;021:7 2 14:235Ttr;h ð8CÞ

Ptr:c ðMPaÞ ¼ 6;831:3 2 14:001Ttr;c ð8CÞ

where the subscripts (tr,h) and (tr,c) represent the pressure or temperature at which the
movement of the phase fronts starts (or ends) on heating (h) or cooling (c) and stands as the
transition pressure (Ptr) or transition temperature (Ttr).
Long-range strain can affect the stability and the phase diagram. Thus, tetragonal
strain stabilizes the tetragonal phase in PbTiO3. The polarizability of the Pb 6S states leads
to different ferroelectric behaviour and greater strain in PbTiO3 (in contrast, Ba2þ is a
perfect spherical ion). The Pb 6S state is hybridized with the O2p states and distorts in the
ferroelectric structure (Cohen and Krakaner, 1992). For PbTiO3, ionic PIB potential gives
no ferroelectric instability. The modified PIB þ potential and the self-consistent LAPW
results do show a ferroelectric double well (Cohen, 1993).
However, not all d0 perovskites show the same phase ferroelectric transitions. Real
ferroelectrics form domains or have charge relaxations and impurities at the surfaces that
ABX3, Perovskite – Ilmenite Structure 789

cancel the depolarization field and stabilize the ferroelectric. An introduction of surface
defects cancels the depolarizing field.
Large size effects and macroscopic polarization are seen for ferroelectrics and, in
small crystals, ferroelectricity is lost (Uchino et al., 1990).

10.7.6. Other titanates

Member mineral phases of the MgTi2O5 (karrooite) –FeTi2O5 (ferropseudobroo-

kite) – Fe2TiO5 (pseudobrookite) system occur in many volcanic and metamorphic rocks.
These also constitute important refractory ceramic materials and magnetic compounds.
A knowledge of the cation order – disorder in (Mg, Fe)Ti2O5 compounds is of
significance because of its strong effects on the crystal structure and high-temperature
stabilization (e.g., Brown and Navrotsky, 1989).

10.7.6.1. MgTi2O5 karrooite: order – disorder

The crystal structure of karrooite consists of two crystallographically distinct,
highly distorted octahedral sites, M1 and M2, with M1 larger and more distorted than M2.
The former is predominantly occupied by Mg, and M2 by Ti. A wide range of non-
convergent Mg – Ti order – disorder between M1 and M2 has been reported (e.g., Brigatti
et al., 1993). This order – disorder reaction is strongly temperature dependent and the
quenched higher temperature samples are more disordered. The fully ordered and
disordered structures are:
ordered: (Mg)M1(Ti2)M2O5 and
disordered: (Mg0.333Ti0.667)M1(Ti1.333Mg0.667)M2O5.
Increasing cation disorder is seen to increase the compressibility of the M2
octahedron and decrease that of the M1 octahedron. All mean kM –Ol bond lengths
decrease linearly with increasing pressure, with an increase in the linear compressibility of
the mean kM2 – Ol distance and a decrease in that of the mean kM1 – Ol distance in the
more disordered structure.
The more Ti in a site, the less compressible is the mean kM – Ol bond. Yang and
Hazen (1999) found a linear function of the Ti contents ðXÞ in the octahedral sites as
K (GPa) ¼ 168(4) þ 88(6) X.
The relationship between bm (compressibility) and X has been expressed by the
equation (Yang and Hazen, 1999):

bm ðGPa21 Þ ¼ 0:00248ð8Þ 2 0:00108ð4Þ X

From this, the extrapolated bm values for an octahedral site, occupied fully by Mg
and Ti, are 0.00140 and 0.00248 GPa21, respectively. Increasing cation disorder with
pressure increases the compressibilities of the mean bond lengths, kM –Ols, in a sample as:

kM2–Ol bond lengths kM1–Ol bond lengths

21
Ordered: 0.000148(2) GPa 0.00243(5) GPa21
Disordered: 0.00163(7) GPa21 0.00193(12) GPa21
790 Chapter 10

Both compressibilities of the mean kM –Ol bond lengths and the octahedral
volumes decrease linearly with increasing Ti in octahedral sites. Under pressure,
the more distorted site shows a larger decrease in distortion (in terms of
octahedral angle variance (OAV)). The compression of the karrooite structure is seen
to be controlled primarily by the bond-length shortening rather than by bond-angle
bending.
The strong compressional anisotropy of the structure is a consequence of
the differential compressibilities of the weaker Mg2þ – O and stronger Ti4þ – O bonds and
the complex edge-sharing linkage involving the M1 and M2 octahedra (Yang and
Hazen, 1999).

10.7.7. Ti/Nb perovskites

Silicate perovskite and ilmenite phases constitute an important volume of the

Earth’s mantle. At mantle pressures, the complete solid solution of Mg and Fe, so common
in crustal minerals, often breaks down, as in the limited substitution of Fe in (Mg, Fe)SiO3
ilmenite and perovskite.
The composition of naturally occurring perovskites (with Ti/Nb) can be expressed
in terms of seven end-member compounds (Mitchell, 1996), five with no vacancies and
two with cation deficiencies:
No vacancies:
Ca2þTi4þO3 (perovskite)
NaþNb5þO3 (lueshite)
Sr2þTi4þO3 (tansonite)
NaþCe3þTi4þ 2 O6 (loparite)
Ca2þ 5þ 3þ
2 Nb Fe O6 (latrappite)
Cation deficient:
Ca3þ 4þ 2þ
2 Ti2 O7 and Ca2 Nb2 O7
5þ

The commonly reported natural perovskite has a general formula like CaTiO3.
The exact chemistry of two such samples reported by Burns (1989) and Muir et al.
(1984) are stated below to show the nature of site occupancies of minor elements in
its structure:
(a) (Ca0.954REE0.046Sr0.007) (Ti0.911Fe0.052 Nb0.025Al0.018)O3· · · (Burns, 1989)
(b) (Ca1.07Na0.006Mn0.004K0.001)(Ti0.818Fe0.127Nb0.034Si0.021Al0.018Mg0.001)O3· · · (Muir
et al.,1984)
In (b), the presence in significant amounts ($ 0.1%) was noted for Y, Zr, La, Ce, U,
Th, Pb and F.
Perovskites of higher symmetry are expected to be of high-temperature genesis
(e.g., Liu and Leibermann, 1993). The mineral latrappite is a complex solid solution
between several end-members, with the perovskite, Ca2Nb2O6, as the dominant end-
member component (Mitchell, 1996). It has an orthorhombic cell with a general formula
close to ABO3.
ABX3, Perovskite – Ilmenite Structure 791

10.8. Mn –oxide perovskites

The manganese oxide system, called manganites, constitutes a richly varied and
fascinating system. These offer scope for studies on the relation between electron
correlations, local structure, long-range order and physical properties. In manganites, the
interaction between the electrons and lattice vibrations (phonons) is unusually strong.
In these, a high density of electrons is strongly coupled to phonons. These, in consequence,
elucidate the interplay between local structural deformations and global properties (Teslic
et al., 1997).
Manganese oxides of general formula A12xBxMnO3 (where A and B are trivalent
and divalent cations, respectively) have attracted much attention by virtue of their unusual
magnetic properties. As stated earlier, on cooling, these compounds show a large decrease
in resistivity associated with a para-magnetic to ferromagnetic transition. This transition
causes a large negative magneto-resistance near the Curie temperature, Tc. In some of these
materials, magnetic fields can drive insulator to metal transitions, where both the
conductivity and magnetization change dramatically — an effect termed “colossal
magnetoresistance” — and alerted the magnetic recording industry to its potential. In the
manganese-oxide perovskites, showing intrinsic colossal magneto-resistance (CMR) in the
neighborhood of their ferromagnetic transition temperature, Tc, there is a dramatic
decrease in resistivity with the applied magnetic field. Recent theoretical and experimental
studies indicate that small polaron effects (including the Jahn – Teller distortion), as well as
the double-exchange model, are required for understanding the transition and the transport
mechanisms.
The Mn –perovskite is mixed valent with Mn3þ (3d4) and Mn4þ (3d3). For
3 1 5
the octahedral site symmetry of the MnO6 complex, the configuration becomes t2g eg ð EÞ
3þ 3 4 4þ
for Mn and t2g ð A2 Þ for Mn . In the double exchange (DE) mechanism, the eg electrons
are considered as mobile charge carriers interacting with the localized Mn4þ(S ¼ 3/2)
spins.
The carrier hopping avoids the strong on-site Hund rule exchange energy Jex
when the Mn spins are aligned ferromagnetically. Jex (,2.5 eV) is much larger than
the eg band-width (,1 eV). Thus, the electrons are expected to be highly spin polarized
and a possibility exists for 100% spin polarization, i.e., almost a half-metallic state
as reported from La0.7Sr0.3 MnO3 by Park et al. (1998). For further discussion, see
Section 15.2.
The hydrostatic pressure dependence of the ferromagnetic Tc has been measured
by Neumeier et al. (1995). In cases when the Jahn –Teller effects become important,
the effects of volume-preserving uniaxial strain become much larger. A quantitative
understanding of the strain effects has been made possible by Millis (1998).
The manganite perovskite has begun to attract current interest because of its giant
magneto-resistance effect (e.g., Kuster et al., 1989). These exhibit a striking magnetic field
and temperature dependence on the resistivity. The motivation for studies of manganite
perovskite lies in the fact that there are numerous potential sensor applications for high-
temperature, high-MR materials. There is also a drop in resistivity upon application of a
magnetic field.
792 Chapter 10

In general, the A12xBxMnO3 materials appropriately doped with 2þ valence B

species undergo a semiconductor-like to metallic phase transition upon cooling. The
transition is associated with a sharp peak in the electrical resistivity. Magnetization studies
show that this peak-resistivity temperature (TPR) is very close to the Curie temperature
(Tc). A peak in the negative magneto-resistance also occurs very close to Tc :

10.8.1. “Ruddlestone-Popper” series

The perovskite family constituting the “Ruddlestone-Popper’series has the

chemistry REnþ1MnnO3nþ1. In this series, the important electrons are the Mn d electrons,
which are 4 2 x in number, and the magnetic moment of Mn ion arises from its parallel
spin. Of the colossal magneto-resistance manganites, the most studied family has the
formula RE12xAx MnO3 where RE is a rare earth such as La or Nd and A is a divalent
alkali such as Sr or Ca. This has been further discussed in Section 15.4. In 1989, a very high
peak MR was first reported from single crystals of Nd0.5Pb0.5MnO32d (Kuster et al., 1989).
 793

Chapter 11
Silicate Melts and Rocks

11.1. Introduction

Determination of the chemical structure of silicate melts and aqueous fluids are
important because these pertain to processes involving heat and mass transfer in the
terrestrial planets. During cooling of the silicate planetary bodies, materials constantly
recycled throughout the crust and mantle result in chemically diverse rock compositions.
Since melts served as the main agents in recycling, it is necessary to understand the
chemical structure of both silicate melts involving alkali and alkaline earths, aluminium
and also hydrothermal fluids. The fields of major rock types in the system SiO2 –Al2O3 –
MnOn22 (where M ¼ K, Na, Ca, Mg and Fe2þ) are shown in Fig. 11.1.
Volatiles such as H2O and CO2 significantly influence melting behaviour. Melts
generated below a critical depth would remain neutrally buoyant and thus not rise to the
surface. With pressure, the melting curve rises continuously and the melting line may (in
general) terminate in a critical point. The melting curve may rise to a limiting or
asymptotic value. Possibly, at sufficiently high pressure, the distinction between a solid
and a liquid would disappear and the material would persist in an amorphous state.
Interactions between silicate melt and water influence the fundamental physical
and chemical processes within the Earth. At high pressure, water is released from
silicate melts, altering the surrounding volume of fluid, which may govern the explosive
nature of many volcanic processes. Evidently, a relationship exists between the mixing
behaviour of solutions in silicate melts (with or without silicate-bearing aqueous solutions)
and the transport properties or rheology of fluids and melts. These fluids may also act as
solvents for silicates and serve as the principal transporting agent within the Earth.
This transport may cause enrichment of elements to cause deposition of important
ore bodies.
Magmatic liquids are regarded as the main agents for heat and transfer within the
Earth and other terrestrial planetary bodies. In magmatic melts positive correlation exists
between melt viscosity or polymerization and silica content. Positive correlation also
exists between electrical conductivity and metal/silicon ratio in the melt. Conductivity is
related to diffusivity.
Nernst – Einstein equation relates conductivity of a component, si ; to diffusivity, Di ;

F 2 zi D i
si ¼
kT
794 Chapter 11

Figure 11.1. Approximate fields of major rock types expressed in terms of the pseudo-ternary system MnOn/2 –
Al2O3 –SiO2, where M ¼ K, Na, Ca, Mg, and Fe2þ (after Mysen, 1990).

where F is Faraday’s constant, zi the electric charge of a particle i, k the Boltzmann’s

constant and T the absolute temperature.
Diffusivity ðDi Þ is related to viscosity ðhÞ using the Stokes– Einstein equation
KT
h¼
6pri Di
where ri is the radius of the moving particle.
For silicate melts the equation stated below is found more reliable than the Stokes–
Einstein equation
KT
h¼
ai D i
where ai is the jump distance.
Bond strength activation energy. For melts in system Na2O – SiO2 activation energy
decreases rapidly from a value of pure SiO2 that is similar to the Si – O bond energy
(, 600 kJ/mol) to a nearly constant value for Na2O contents above , 15 mol% (Mysen,
2003).
The structural control of viscous flow may be breakage of bridging Si –O – Si bonds.
However, viscous flow does not depend solely on the energetic of bond breakage.
Arrhenius expression for viscosity and activation energy is
lnh ¼ lnh0 þ ðEh =Rt Þ
where h is the viscosity, Eh the activation energy of viscous flow, T the temperature in
Kelvin and R the gas constant. Viscosity of most silicate systems including melt and glass
do not follow Arrhenius expression. Rather viscosity displays a distinct curvature in ln h vs
1/T space.
For describing viscosity of melt and glass the most successful model involves
the theory of configurational entropy, first described by Adam and Gibbs (1965).
Silicate Melts and Rocks 795

For describing viscosity Richet (1984) used the relation

lnh ¼ Ae þ Be =TSconf

where Sconf is the configurational entropy, Ae a constant and Be the molar free hindrance
energy, which is the energy required for changing from one configuration to another in
melt. This energy is positively correlated with the silica content.

11.1.1. Magmatic melt under pressure

Under pressure, the polymerized silicate melts behave anomalously: their

viscosities decrease with increasing pressure.
Viscosities can be estimated from the diffusivities of network-forming ions using
the Eyring relation. Poe et al. (1995) have determined oxygen self-diffusion coefficients up
to 15 GPa for three melts with varying degrees of polymerization. They found that, for
sodium tetrasilicate (Na2Si4O9), which nominally has an average of 0.5 non-bridging
oxygens per tetrahedral cation (NBO/T), oxygen diffusivities continue to increase with
pressure up to 15 GPa. Results from molecular dynamics simulations of Na2Si4O9 liquid
have indicated that diffusivities pass through a maximum near 20 GPa but, for the slightly
more polymerized Na3AlSi7O17 (NBO/T ¼ 0.25), the experimentally determined
diffusivities pass through a maximum near 5 GPa. These results would thus suggest that
viscosities pass through a minimum with increasing pressure, and that the minimum is
polymerization dependent.
Solid-state NMR studies have shown that glasses quenched from high pressures
contain increasing amounts of high-coordinate Al (in alumino-silicates) and Si (in binary
silicates) with increasing pressure. This is interpreted as a densification mechanism which
initially enhances oxygen diffusivities. NBOs are likely to participate in the formation of
high-coordinate Al and Si species. However, at higher pressures, after all NBOs have been
consumed by this reaction, bridging oxygens are required to continue the increase in the
coordination of Al and Si (thus forming three-coordinate oxygen). This marks the
diffusivity maximum (and viscosity minimum) and accounts for its polymerization
dependence.
Such a diffusivity maximum is observed for the fully polymerized albite melt. This
possibly indicates that oxygen self-diffusion is sensitive to trace amounts of water.
However, there is little variation in water content as a function of pressure. The estimated
concentrations (, 0.1 wt% H2O) would reflect only a minor extent of depolymerization by
incorporation of H2O. Finally, increasing oxygen diffusivity with pressure up to 5 GPa is
consistent with decreasing viscosity up to 3 GPa. Experimental studies on rock melts
provided varying melt products. The salient observation on peridotite study may be
outlined as below.
In the low-pressure range (, 5 GPa), the chemistry of the melt is dependent on the
exerted pressure and the relative proportions of the volatiles (H2O and CO2) present. With
increasing pressure, the products of peridotite melt range from basalt through picrite and
796 Chapter 11

komatiite to lherzolite, accompanying a change in the chemical content of MgO and

SiO2 as:

Oxidewt% Picrite ! Komatiite ! Lherzolite

MgO . 10 . 20 . 35
SiO2 45 – 50 35 – 45 40

Thermodynamic modelling of mantle melting shows that garnet –spinel and

spinel – plagioclase transitions are regions of low freezing but not of enhanced melting. The
implications of this for problems in MORB petrogenesis have been discussed by Ashimow
(1995).
Kimberlites could be derived from relatively low-temperature melting of an H2O-
rich mantle at depths of 150 –300 km. Thus, the diamond-producing deposits could be
associated with hydrous melting processes. Abundant evidence from xenoliths indicate
that some regions of the sub-continental upper mantle may be significantly hydrated.

11.2. Alumino-silicate melts

A knowledge of the effect of pressure on alumino-silicate glass and liquid structure

is critically important to understand the magma flow up from depth and the melting of the
igneous rocks. A pressure-induced coordination change is likely to be a very important
mechanism in influencing the properties of natural magmas in the Earth’s upper mantle
(, 300 km depth).
The anionic structure of most magmatic liquids will consist of units that have
average values of NBO/T equal to 2, 1 and 0. Alumino-silicate melt shows a change of
Al3þ coordination from fourfold at room pressure to sixfold coordination at upper mantle
pressures.

11.2.1. CaO –Al2O3 – SiO2 melts: compressibilities

The melts in the CaO – Al2O3 – SiO2 system were studied by Webb and Courtial
(1996) for their compressibilities over the range 1,350 –1,6008C. For these melts, the bulk
modulus increases with the addition of CaO and decreases with the addition of Al2O3 and
SiO2. The compressibilities of CaO – SiO2 and CaO – Al2O3 – SiO2 melts are lower than
those of Na2O – Al2O3 –SiO2 and binary alkali –silicate melts in general.
The mechanism of compression of CaO –Al2O3 – SiO2 melts is different from that
of Na2O – Al2O3 –SiO2 melts. The compressibility of the former is a complex function of
the structural geometry variations arising from the presence of CaO, the CaO –O – Si bond
interactions and the packing of cations around the anions in a melt, having clusters of
preferred Al –Si –Al bonding. The compressibility of the latter melts appears to be mainly
a function of the Al– O interactions.
Silicate Melts and Rocks 797

11.2.2. Na2O –Al2O3 – SiO2 melts: Ab50NTS50

Yargar et al. (1995) investigated a model composition between albite (Na AlSi3O8)
and sodium tetrasilicate (NTS, Na2Si4O9). a melt of 50 : 50 of these produced a glass of
composition Na3AlSi7O17 (Ab50NTS50). The 27Al MAS NMR spectra (Fig. 11.2) for the
glasses prepared at high pressure showed the resonance lines ascribed to as AlIV, AlV and
AlVI species (Sato et al., 1991). The relative abundance of AlVI markedly increases with
pressure while the amount of AlV goes through a maximum (, 28%) around 8 GPa
(Table 11.1). In Table 11.1, the abundance of AlV and AlVI in quenched glasses was
estimated from the asymmetric peak areas of the 2 1/2 ! 1/2 (central transition) 27Al
NMR resonance. The diso was estimated near the base of the asymmetric peaks.
A convolution minimization fit to all three resonances in the spectra, allowing the
amplitude and width to vary, gave a reasonable representation of the original spectra.
Relative errors for the quantification of Al species were ^ 6%.
Figure 11.2 shows that, at high pressure, the line of the 23Na resonance (in
Ab50NTS50) becomes narrower. This is caused by a decrease in quadrupole coupling

Figure 11.2. 23Na and 27Al MAS NMR spectra for glasses under pressure showing the resonance liner for AlIV,
AlV, and AlVI spices (Sato et al., 1991).

TABLE 11.1
The abundance of AlIV, AlV and AlVI at varying pressures in Ab50NTS50

Species dISO (ppm) Integral (%)

6 GPa 8 GPa 10 GPa 12 GPa
IV
AlO4 (Al ) 77 80 49 42 35
AlO5 (AlV) 40 12 28 20 17
AlO6 (AlVI) 8 8 23 38 48
798 Chapter 11

parameters or by a reduction in the range of chemical environment, i.e., lesser distortion

(more symmetrical average coordinate of Na) at high pressure.

11.3. Viscosity: controlling factors

Viscosity and density are related to the number of NBOs and also to the difference
between total positive and total negative electrical charge. The property of interest in most
high-pressure experimental investigations of melt rheology is the Newtonian viscosity
(see Webb and Dingwell, 1995).

11.3.1. Diffusivity: Stokes – Einstein equation

The Stokes– Einstein equation was formulated to explain the inverse proportion-
ality between diffusivities of specific components and viscosity of the liquid:
hD ¼ kB T=l ð11-1Þ
where h is the viscosity, D the diffusivity, kB the Boltzmann’s constant and l an effective
length scale or jump distance for diffusion, which is also represented as 2pa; where a is the
diameter of a Brownian particle (in liquid) (see also Sections 15.14.1 and 15.14.3). The
determination of diffusivities can be accomplished by the detection of very short
concentration profiles within a binary diffusion couple.
Silicate melts posses a viscoelastic behaviour, which can be usefully approximated
by a Maxwell body with a distribution of relaxation times. The relaxation mode controlling
viscous flow in silicate melts has been linked quantitatively in temperature – time –space to
the exchange of Si and O atoms in the melt, as determined by motional averaging in
spectroscopic experiments.
Many homogeneous equilibria involve reactions in the silicate melt phase whose
kinetics are closely reflected in the relaxational time scale of resultant properties such as
enthalpy, volume and shear stress. Hence, a spectroscopic investigation of relaxation in
melts can be used to derive the viscosity data.
Time-domain longitudinal dilatometric experiments reveal an onset of non-
Newtonian rheology at approx. 2.5 log units of strain below the relaxation strain rate
derived from the Maxwell equation:
t ¼ h=G ¼ 1=1t ð11-2Þ
where h is the shear viscosity, G the shear modulus, t the shear relaxation time and 1t
is the relaxation strain rate. (Note: Experimental strain rate , 1,000 strain rate (relaxation).)

11.3.2. Temperature dependence: Arrhenian approximation

Over the restricted temperature ranges defined by the phase equilibria of melting
within the Earth, the Arrhenian approximation
logh ¼ a þ b=T ð11-3Þ
Silicate Melts and Rocks 799

holds in a limited way but the temperature dependence of liquid viscosities over the range
of viscosities encountered from liquidus temperatures (as low as 10 Pa s) to the glass
transition (,1012 Pa s at typical dilatometric experimental timescales) reveals enormous
departures from the Arrhenian approximation (see Fig. 3 of Dingwell, 1998).
In the case of the diffusion of silicon, the Arrheninan relationship such as the
Tamann– Vogel – Fulcher equation:

logh ¼ a þ b=ðT 2 cÞ ð11-4Þ

stands valid over a wide range of viscosities (Dingwell, 1990). Relaxational spectroscopic
data also reveal oxygen-diffusivity change with viscous flow at high viscosities.

11.3.3. Alkali oxides

The viscosity can be seen to decrease strongly and non-linearly with addition of
alkali oxide to silica melt. The non-Arrhenian temperature dependence is typical of very
basic melts of geological relevance. In the high-viscosity range, the apparent activation
energy increases with addition of alkali oxide to silica.

11.3.4. Water effect

Traces of water, typically in the range of tens of hundreds of ppm, can result in a
reduction in melt viscosity of several orders of magnitude. However, the effect of water is
more pronounced in a high-viscosity range rather than in a low-viscosity range (e.g., Hess
and Dingwell, 1996).

11.3.5. Pressure effects on viscosity

The effect of pressure on the viscosity of silicate melts has been determined by
using the Stokesian falling sphere law:

V ¼ 2r 2 Drg=2h ð11-5Þ

where V is the settling viscosity, r the sphere radius, Dr the density contrast between
sphere and liquid and g the acceleration. The influence of pressure on melt structure and
properties has been discussed by Wolf and McMillan (1995). In the CaO – Al2O3 – SiO2
system, there is a transition to a positive pressure dependence of viscosity as the SiO2
content is reduced below 50 mol%. The degree of polymerization of the melt determined
from the NBO/T (Mysen et al., 1982) content shows a trend of less viscosity decrease with
pressure as SiO2 decreases. Melt compositions approaching ultra-basic chemistry show
little influence of pressure on the viscosity. The general nature of the transition from
negative to positive pressure dependence of viscosity in more depolymerized melts was
investigated by Brearley et al. (1986).
In magmatic processes under pressure, the viscosity of many silicate melts shows a
decrease. Again, pressure may enhance the fragility of silicate melt. In alumino-silicate
800 Chapter 11

melts, the type of charge-compensating cation affects the pressure dependence of the
viscosity as well as the compressibility of the melt (Kushiro, 1981).

11.3.6. Silicate polymerization

A melt of SiO2 shows a high viscosity because of the presence of a fully

polymerized tetrahedra. Viscous flow in pure SiO2 requires breaking strong Si – O bonds
(452 kJ/mol), which form a high-viscosity network. When a network modifier like Na2O is
added to it, the Si – O –Si linkages break to form Si – O2  Naþ. This constitutes what are
called NBOs and causes a lowering of the viscosity. Thus, adding alkali creates NBOs,
which bar the formation of high-coordinate network cations. This in turn reduces the
viscosity but, if Al2O3 is added to the melt (say, an alkali silicate) to such an extent that
Na2O:Al2O3 $ 1, the NBOs are removed, the activation energy is increased and the
tetrahedral network is reconstructed with increased viscosity.
Under pressure, the formation of high-coordinated species is favoured with
consequent lowering of the activation energy for oxygen exchange. High-pressure
lowering of the viscosity can be related to the increase in the number of isoergic
configuration states in the melt, which results in an increase in the number of channels
for stress relaxation (Dickinson et al., 1990). By incorporating these conditions in the
Adam –Gibbs expression for cooperative relaxation in polymer melts, one obtains the
expression for shear viscosity:
h ¼ h0 expð2C=TSc Þ ð11-6Þ
where C relates the activation energy barrier for relaxation and Sc is the configurational
entropy of the melt.
NMR studies have revealed the presence of substantial amounts of high-
coordinated silicon species (SiV and SiVI) in partially depolymerized alkaline silicate
glasses quenched from high-pressure melts (Xue et al., 1991).
In most natural melts, the ratio of metal oxide to Al2O3 is higher than 1 and Al
enters the network in fourfold coordination. Any increase in Al coordination with pressure
is likely to lead to a decrease in the high-pressure viscosity of these alumino-silicate melts.
Alumino-silicates, covering the chemistry of andesitic to basaltic magmas, form
tetrahedral networks with extremely high viscosities. Under pressure, the viscosity
decreases and for this reason, at depth, the mobility of the melt can be higher by
several orders of magnitude. This viscosity decrease is related to the pressure-induced
increase in the Al coordination and the resultant weakening of the Al – O bond strength by a
change in the bond angles in the alumino-silicate tetrahedral network. The viscous flow of
alkali-alumino-silicate melts is constrained by oxygen exchange between polymeric units
(Poe et al., 1992). In this flow, structural relaxation may occur through the formation of
transient five-coordinated (Si or Al) species.
At T close to their glass transitions, liquid SiO2 and SiO2 –Na2O melts have
configurational entropies attributed to their bridging and non-bridging O atoms.
The decrease in the specific volume with the decrease in polymerizaton causes the
P-sensitivity of the viscosity of silicate melts. Using some assumptions, Bottinga and
Silicate Melts and Rocks 801

Ricket (1995) showed that the Kushiro discovery (1976, and subsequent papers) can be
explained with the Adam and Gibbs theory (J. Chem. Phys., 43, 139 –146, 1965) for the
P-sensitivity of polymerization and the change in appropriate configurational entropy.
For modelling the physical chemistry of the melts and for parametrizing melt
viscosities for modelling of igneous processes, complete viscosity – temperature
relationships are needed. This is all the more so because of the non-Arrhenian temperature
dependence of the viscosity of most melts. This non-Arrhenian temperature dependence
remains one of the main hindrances in obtaining a fully generalized model of melt
viscosities for petrological calculations.

11.3.7. Density and viscosity determination

Silicate melt densities at high pressure are usually estimated using molecular
dynamic simulations and elastic property measurements at low pressure. Dynamic
determinations using shock-wave techniques and static determinations using the “falling
sphere” technique (density þ viscosity) and the “sink”/“float” method (density) have been
carried out on silicate melts to determine the variation of density as a function of pressure.
In the falling sphere technique, the density is determined by measuring the distance
of sinking (or floating) of a crystalline or metal sphere as a function of time and is
calculated using Stoke’s Law. In such studies, spheres of such minerals as ruby, forsterite,
diamond, graphite, platinum, etc., have been used. A centrifuge can also be employed
to accelerate the Stokesian settling velocity by up to a factor of 1,000 – 1,500.
This accelerated falling sphere technique is used to provide data in the viscosity range
of 105 – 108 Pa s.
The experiments by Zhen – Ming Jun (1994) suggest an effective mantle viscosity to
be around 1012 Pa s (the ratio of the stress about 107 Pa to the strain rate is , 1025 s21),
which is some six orders of magnitude lower than that commonly assumed for the Earth’s
asthenosphere.
A drastic lowering of the viscosity of silicate rock by the presence of water has
important geodynamic consequences. A strong partitioning of water in melts migrating
upwards leaves behind a largely dehydrated residual solid. This dehydration
strengthening may be the reason for the narrowness of the upwelling zone beneath
ridges. Dehydrated viscous melt is sufficiently viscous to keep lateral corner-flow
gradients capable of focusing melt into a narrow ridge axis (Hirth and Kohlsted, 1996;
Braun et al., 2000).

11.3.8. Melt percolation

In the upper mantle, small amounts of melt may distribute so as to form an

inter-connected framework of tubules (such as along the triple junction where
olivine grains meet). However, to a certain extent, the melt equilibrates chemically
with the solid through which it migrates. The melt migrates only a short distance
by percolation before finding its way into broader channels where it can move
802 Chapter 11

rapidly upwards (Iwamori, 1993). However, when, under stress, the sample is stretched
a little, the melt redistributes, spreading to grain boundaries. This melt redistribution
greatly reduces the strength of the matrix, aiding the escape of the melt. A partial melt
channel with locally high strain rate develops a greatly increased ability for upwelling
of the material.
Partial melting and separation of melt through melt percolation led to thermal
expansion, contraction and volcanism, all of which are linked to plate tectonism. As the
less dense hot material rises and displaces the dense material downwards, the total
gravitational energy is lowered. Gravitational energy dominates in large-scale processes
within the Earth and other planets.
The effects of solid –solid phase transitions on melt productivity during isentropic
pressure-release melting were investigated by Ashimow (1995). The principles governing
these effects were developed by graphical construction in one- or two-component systems.
These constructions show that solid – solid transitions with positive Clapeyron slopes
diminish rather than enhance melt productivity.

11.3.9. Crystal – melt phase equilibria

Crystal – melt phase equilibria are also affected by pressure. For example, enstatite
(MgSiO3) melts incongruously to forsterite and liquid at # 0.3 GPa pressure but, at higher
pressure, it melts incongruently. Olivine (in basalt) is typically a liquidus phase at crustal
pressures but, at upper mantle pressure, pyroxene becomes a liquidus phase. Other pressure
effects include cation diffusion, trace-element partitioning in crystal – liquid and redox
ratios of iron (Mao et al., 1981).
The inter-tetrahedral angles in the three-dimensional network decrease by several
percent under pressure. A significant reduction in cavity volume will most likely affect the
viscosity diffusion property, etc., in addition to the density.

11.3.9.1. fO2 , fH2O and aH2O

Oxygen fugacities, fO2, of mantle-derived melts and mineral assemblages are
commonly considered to lie within approximately ^ 2– 3 log units of NNO buffer curve
but fugacities up to that of the H – M buffer have been estimated in some cases (e.g.,
Righter and Carmichael, 1993).
There is little quantitative documentation of aH2O for derived melts and mineral
assemblages. Bell (1994), however, has determined log aH2O in the range 22 to 23 based
on the H contents of a site of olivines from xenoliths (entrained in alkali basalts) from a
number of localities.
The low H2O activities may represent low wt% H2O contents in the co-existing
magmas under vapour-absent conditions. Under vapour-present conditions, low aH2O
would result if H2O represents only a small mole fraction of all species in a C – O – H fluid
(Popp et al., 1995).
Silicate Melts and Rocks 803

11.4. H2O in silicate melts

H2O dissolves both as OH and as molecular H2O in silicate melts (e.g., Stolper,
1982). Speciation of H2O in melts and glasses poses a problem. Dingwell and Webb (1990)
suggested that the H2O speciation in a glass represents the equilibrium at the bulk Tg of the
glass. Shen and Keppler (1995) have reported direct FTIR measurements of H2O
speciation in hydrous silicate melts to 1,0008C and 1 GPa in a DAC designed for FTIR
studies.
A peraluminious sodium silicate melt with 30– 40 wt% H2O shows the presence of
both OH and H2O. The standard enthalpy of the speciation reaction H2O þ O ¼ 2OH is
very different for the melt and glass phases. The temperature dependence of the H2O
speciation equilibrium can be described by two equations:
(a) for glass phase: DH ¼ 1:6 kJ=mol; and lnK ¼ 22:65 2 1:91 £ 102 KT 21
(b) for melt phase: DH ¼ 30:3 kJ=mol; and lnK ¼ 3:04 2 3:64 £ 103 KT 21
The intersection of these two equilibrium curves defines a glass-transformation
temperature of 3358C.
The water contents of glasses from MOR in close proximity to the Azores and Iceland
hotspots are enriched by factors of 2 –4 relative to the surrounding mantle (Kingsley and
Schitting, 1995). Hence, these two areas are referred to as mantle “wetspots”.

11.4.1. K2O –SiO2 – H2O system

The solubility and solubility mechanisms of H2O have been determined by Mysen
(1998) for melts in the system K2O –SiO2 – H2O from 0.8 to 2.0 GPa in the 700– 1,1008C
temperature range. For the most potassic composition studied, K2SiO5, nearly complete
miscibility between hydrous melt and silicate-saturated aqueous fluids were observed at
2.0 GPa. The H2O solubility and silicate solubility are strongly non-linear but show
positive functions of pressure. The temperature dependence becomes more pronounced
with increasing pressure and with increasing K/Si.
The experimental study by Mysen (1998) reveals that the in situ high-T/high-P
Raman spectroscopic information for potassium silicate melts is consistent with a
solution mechanism for H2O in potassium silicate melts, which can be schematically
summarized as:

K2 O·nSiO2 þ H2 O ) Q0 ðHÞ þ Q2 ðKÞ þ Q3 ðKÞ þ K· · ·OH ð11-7Þ

In this expression, Q0(H) denotes a unit with no bridging oxygens and where all
the NBOs are associated with Hþ (thus in effect producing a H4SiO4 species), the Q2(K)
and Q3(K) denote structural units with two and three bridging oxygens, respectively,
where the NBOs are associated with Kþ.
The silicate-solution mechanism in aqueous fluid can possibly be described by a
relation like that presented in equation (11-7).
The phase relations along the joins K2SiO4 – H2O and K2SiO9 –H2O under pressure
(Mysen,1998) show an immiscibility between potassium silicate melts and fluids. But the
804 Chapter 11

immiscibility gap shrinks rapidly with increasing pressure and with increasing K/Si ratio.
Strangely, however, the gap continues to exist at 2 GPa.

11.5. REE patterns

The depth constraints on the formation of tonalitic magmas in the continental crust
are provided by REE patterns of the synthetic melts calculated from the REE abundances
in metagabbro and metabasalt. The REE pattern of tonalites from active continental
margins and Archean TTG associations show low values in REE, with LaN (chondritic
normalized) of 10 –30 and YbN of 1 – 2. These values are reproduced at 1 –1.25 GPa
pressure from metagabbro, which displays a slightly low REE-enriched pattern with
LaN ¼ 8 and YbN ¼ 3 (see Springer and Seck, 1997).

11.5.1. Fe31 in glass

The content of Fe3þ in melts and glasses is a strong function of T, fO2 and composition.
At high values of fO2, Fe3þ is in tetrahedral coordination in many silicate melts. It has been
demonstrated by Virgo and Mysen (1985) that Fe3þ transforms gradually to octahedral
coordination, either when oxygen fugacity decreases at constant temperature or when
temperature increases at constant fO2 if, in the melt, the Fe3þ/SFe is less than 0.5.
Fe3þ/SFe decreases initially with increasing pressure in glasses quenched from
Na2O –Al2O3 – SiO2 –Fe –O melt (Fig. 11.3). In alumina-bearing systems, however, the
melt shows greater oxidation at higher pressures. The quenched melts (glasses) were
analysed using 57Fe Mössbauer spectroscopy by Brearly (1990), who observed pressure
changes of the spectral envelopes. The changes may result from an increased distortion of
Fe3þ (IV).

11.5.2. Partition coefficient in melt/solid

The partition behaviour of elements under pressure is controlled not only by P and T
but also by composition. The effect of temperature on the partition (or rather, distribution)
coefficient, D; is estimated by thermodynamic considerations (e.g., Murthy, 1992).
By assuming Nernst partitioning and neglecting the non-ideality of the system, the
equilibrium condition is expressed as:
mLX þ RT ln CXL ¼ mSX þ RT ln CXS ð11-8Þ
where R is gas constant, T temperature and mLX and mSX represent standard chemical
potentials of component X of the liquid phase and solid phase, respectively. Then the
partition coefficient is expressed as:
lnD ¼ ðmLX 2 mSX ÞRT ð11-9Þ
This equation predicts that the partition coefficient approaches 1 with increase in
temperature. It is therefore expected that the relative difference of the elements in D will
Silicate Melts and Rocks 805

Figure 11.3. Fe3þ/SFe of quenched glasses as a function of pressure and composition of starting material
(X ¼ mole fraction of acmite component of the starting material) at 1,4508C (Mysen and Virgo, 1978).

decrease with increase in temperature. Since the liquidus temperature increases with
increase in pressure, all the values should be approaching to 1 under high temperature if the
effects of temperature alone are considered.
The partitioning of selected trace elements between peridotite minerals (such as
olivines, orthopyroxene, clinopyroxene and phlogopite) and carbonatite melt at 1.8–
4.6 GPa pressure was investigated by Sweeney et al. (1995). They determined the partition
coefficients for trace-element partitioning between olivine– orthopyroxene –carbonatite
melt. Earlier (Bell et al., 1979; Heinz and Jeanloz, 1987), it had been observed that Fe
partitions into melt and the oxide structure relative to perovskite structure.
Between H2O-rich fluids and natural peridotite (olivine þ orthopyroxene þ
clinopyroxene þ accessory minerals), the partition coefficients of Rb, Nb, La, Sr, Sm,
Zr, Tm and Y were measured by Ayers et al. (1997) at 2.0 –3.0 GPa and 900– 1,1008C.
The stable sub-solidus assemblages observed were:
at 2:0G Pa ! spinel – lherzolite ð^rutileÞ
at 3:0G Pa ! garnet – peridotite þ zircon
806 Chapter 11

The high solubility of TiO2 is seen in spinel –lherzolite at high temperature.

Therefore, rutile should hardly be present at the source region of island-arc basalts during
melting.
Peridotite/fluid bulk-partition coefficients at 2.0 GPa range from , 0.1 for Rb to
, 100 for Tm. Fluid/melt partition coefficients (D fluid/melt) have been reported by Ayers
and Eggler (1995) for Ca, Mg, Na, K, Rb, Sr, La, Sm, Y and Tm at 1.5 and 2.0 GPa at
1,2508C. They reported that coefficient D fluid/melt ranges from 0.43 to 1.31 for these
elements (and Ti), which suggests a near-congruent dissolution of melt in fluid.

11.6. Rocks under pressure

11.6.1. Transformation under shock: pseudotachylites

Impacting of meteorites on planetary bodies creates enormous transient pressure on

rocks (or ices) which causes phase transition, deformation, melting and vaporization. Such
impacted regions are characterized by shattered cones and pseudotachylites (pulverized
rocks melted by impact or frictional melting).
At Sudbury, impact -structure pseudotachylites (B type) are seen. Coesite and
stishovite have been reported in the pseudotachylites in quartzites from the Vredefort
impact structure in South Africa (Martini, 1978).
In laboratory hyper-velocity impact experiments of Fiske et al. (1995), quartz was
shock-loaded from 42 to 56 GPa by using the 6.5 m two-stage light gas gun at Lawrence
Livermore National Laboratory. In this experiment, the melting of SiO2 provides a lower
limit on the temperature of the pseudotachylite (2,000 K). These temperatures are similar
to those required to reduce SiO2 in soils struck by lightning (Essene and Fisher, 1986).
During a meteorite impact, the compression and development of a hemi-spherical transient
crater cavity involves bulk plastic deformation for a time of 1023 –100 s. Plastic flow and
pseudotachylite formation are important processes in meteorite impacts. The central part
of large impact structures may have a substantial static thermal metamorphic overprint.
This explains the increase in crater melt with crater size. Such thermal structures have,
however, been attributed (wrongly?) by some to endogenic processes.

11.6.2. Terrigenous and pelagic sediments under subduction

Subduction of the continental crust along with the terrigeneous and pelagic
sedimentary plays a major role in the evolution of the crust – mantle geochemical system
(see also Section 2.10.2). The evidence of subducted sediments has been well observed in
many calc-alkaline magmas that erupted in island arcs and continental margins from
source regions at depths of 100 –150 km. The isotopic signatures of Pb, Sr and Nd in many
intra-plate magmas indicate that these were derived from subducted terrigeneous and
pelagic sediments (e.g., Tera et al., 1986).
Silicate Melts and Rocks 807

About 1.6 km3/yr of terrigenous and pelagic sediments are subducted into the
mantle (Von Huene and Scholl, 1991). Some are returned to the crust via underplating and
calcalkaline magmatism, while some go to greater depths to be incorporated into the
mantle. Subduction of terrigeneous materials may also occur along the zones of continental
collision.
Recent discoveries of coesite in high-grade metamorphic rocks indicate that large,
coherent volumes of quartzose-felspathic continental crust may be subducted to depths
exceeding 100 km (Schreyer et al., 1987). Also, the reports of inclusions of micro-
diamonds in garnets from these metamorphosed lithologies suggest that the subduction in
such cases reached depths of ,120 km (e.g., Sobolev and Shatsky, 1990). Evidently,
buoyancy forces at these depths become sufficient to inhibit the transportation of these
sediments to greater depths via entrainment by subducting slabs. The depth of subduction
is, therefore, delimited by the buoyancy of these materials. At shallow depths (,150 km),
terrigeneous material is buoyant relative to the surrounding mantle and opposes
entrainment by the subducting slab. In isolated cases, the entrainment continues and is
reflected by the occurrences of coesite and diamond in some materials which were later
pushed up to the surface (Irifune et al., 1994) (Fig. 11.4, from Liou et al., 1998; see inset).
The subducting slab includes crustal-hosted fragments and mantle wedge blocks. The VHP
and HP slabs returned to shallow depths after recrystallization within coesite or diamond
stability fields at depths .100 km due to slab breakoff and the buoyancy of a low-density
continental sheet.
However, small rafts of fragmented continental blocks on the oceanic plates may
be dragged deeper along the subduction zones. This process was very significant in the
early past.
The compositions of the argillaceous and siliceous facies of anhydrous pelagic
sediments (Table 11.2) are sufficiently close to that of the upper crust. Irifune et al. (1994)
experimented on a material of continental crust composition, similar to that of the average
upper continental crust. During partial melting of continental crust lithologies at relatively
low pressures (5 – 10 GPa), the early formed orthoclase, wadeite and K-hollandite are
eliminated near the solidus, whereas the stability field of Na-clinopyroxene extends to
temperatures well above the solidus. Liquidus in this pressure interval shows high K/Na
ratios and high SiO2 contents. At higher pressures (16 – 24 GPa), the stability fields of
K-hollandite and stishovite extend towards the liquidus. The resulting partial melts show
lower K/Na ratios and SiO2 contents.

11.6.2.1. Density change and buoyancy

At depths with pressures lower than 6 GPa, the major phases in the crustal
compositions are orthoclase, coesite and clinopyroxene. Near to 6.5 GPa, orthoclase
transforms to wadeite þ kyanite þ coesite and the density of the assemblage increases to
3.2 –3.3 g/cm3. Around 9 GPa, K-hollandite and stishovite become major stable phases. At
10 GPa, the density of the assemblage becomes , 3.7 –3.9 g/cm3 (compared with pyrolite
at , 3.4 g/cm3). The density of crustal compositions remains substantially higher that that
of pyrolite up to , 24 GPa, i.e., near the bottom of the transition zone (, 660 km). Below
this depth, however, the stated density relationship is reversed.
808 Chapter 11

Figure 11.4. A tectonic model and P – T time path for subduction and exhumation of crustals fragments from
continental lithosphere using for example the Dabie– Sulu collision zone. The required time between subduction
and exhumation is about 10 million years (modified after Liou et al., 1998, q 1998 Mineralogical Society
of America).

In many cases, buoyancy causes subducted sediments along with the slab materials
to return to the near-surface crustal environment. Evidence of such a history is offered by
the presence of coesite and diamond. In other cases, the geochemistry of some oceanic
basalts has offered evidence for the continentally derived rocks and sediments to have been

TABLE 11.2
Chemical composition of continental crust and pelagic sediments

Continental crust Pelagic sediments Argillaceous Siliceous facies

(Taylor and McLennan, 1985) clay facies (Chester, 1990) (Chester, 1990)

SiO2 66.0 (66.2) 60.50 70.80

TiO2 0.5 (0.6) 0.92 0.72
Al2O3 15.2 (15.9) 19.50 14.73
FeOa 4.5 (4.6) 8.17 6.38
MgO 2.2 (1.9) 1.50 1.51
CaO 4.2 (4.8) 4.18 2.71
Na2O 3.9 (2.7) 1.67 1.04
K2O 3.4 (3.3) 3.56 2.10

Values in parentheses represent the results of electron microprobe analyses of the starting material actually used in
the present study (FeO was replaced by CoO).
a
All iron has been calculated as FeO.
Silicate Melts and Rocks 809

transported down to depths of , 200 km. Once the critical depth of , 200 km is reached,
the subduction of the continentally derived material may attain the density (and loss of
buoyancy) for it to sink deep into the mantle.
If significant amounts of associated water are subducted to depths greater than
150 km, partial melting of the subducted slab would occur. The melt would react and
hybridize with the surrounding mantle, modifying the properties of the magmas which are
derived from the mantle. Several lines of evidence suggest that significant amounts of
water may be subducted to depths of 150 – 160 km (e.g., Thompson, 1992). Serpentinite
underlying oceanic crust may provide the most plausible sources of water during deep
subduction. H2O is thereby released and is added to that formed by the process of
successive dehydration of dense hydrated magnesium silicates (DHMS). This would
ascend and cause partial melting of the subducted oceanic crust along its overlying
continentally derived lithologies.

11.6.2.2. Potassium mobility in subduction pressures

The upper continental crustal rocks at low pressures (, 6– 8 GPa) and near to
solidus are seen to eliminate orthoclase and wadeite (Irifune et al., 1994). Thus, the partial
melts are expected to be K-rich but, because the partition coefficient of Na in crystal/liquid
is high (. , 2.5), Na-rich clinopyroxene exists throughout much of the melting interval.
Accordingly, melts formed by a modest degree of partial melting in the low-pressure
region have high K2O contents (up to 6%) and high K2O/Na2O ratios (, 2 – 3).
These results suggest that partial melts of subducted terrigenous lithologies
between depths of , 150– 250 km could be partially responsible for potassium-enrichment
processes, which are believed to be widespread in the sub-continental lithosphere. They
may also contribute to the systematic increase in K2O contents of calc-alkaline magmas.
This is observed to occur as the distances from their eruption centres to the underlying
Benioff –Wadati zones increase. Small degrees of partial melting with increasing depth
cause enrichment of K2O in the melts (Irifune et al., 1994).
However, at higher pressures between 10 and 21 GPa, the K-hollandite field
expands during partial melting while that of Na-clinopyroxene contracts. At 21 GPa
(/1,9008C), the partial melt contains only 1.4% K2O, compared with 7.1% Na2O, and
hollandite as the refractory residue continues to enlarge. The SiO2 content in the melt also
reduces (, 55% SiO2) through expansion of the stishovite field.

11.6.2.3. Lead paradox

Potassium-hollandite strongly hosts Pb (and La) but not U or REE (Kesson and
White, 1986). Consequently, the liquid after partial melting will be richer in the U/Pb ratio
when K-hollandite is present and lower in LREE/HREE, K/U and La/Nb ratios compared
with the initial compositions (Irifune et al., 1994). The capacity for retaining Pb by residual
K-hollandite during partial melting of subducted continentally derived lithologies at high
pressures is significant.
In the early history of the Earth (4.0 –2.0 b.y. ago), the subduction of
continentally derived lithologies might have occurred on a larger scale than exists at
present and these were carried to greater depths (300 – 600 km) where they
810 Chapter 11

experienced partial melting. The K-hollandite phase as residium or the megaliths

would retain Pb in its structure and concentrate at the boundary layer overlying the
660 km discontinuity. These would ultimately have sunk into the lower mantle
(Allègre and Turcotte, 1985).
The extraction of these lithologies from the crust to the upper mantle may have
caused the over-abundance of U in them compared with Pb. This is known as the ’lead
paradox’. Indeed, Class et al. (1993) found that magmas that erupted along the hot-spot
trail of the Ninety-East Ridge in the Indian Ocean were characterized by an exceptionally
high U/Pb ratio ðm , 30 – 60Þ: The lead paradox with reference to calc-alkali magmatism
has been discussed in Section 2.8.5.2.

11.6.2.4. Subducting slabs

In subducted slabs where the temperatures are lower than the average mantle
temperatures, volatile-bearing phases could be present. At depths of temperatures , 1,5008
C, the content of H2O in b-phase (wadsleyite) is suggested to be nearly as high as that
of nominally hydrous phases. And, since under hydrous conditions the stability of
b-phase þ stishovite is much increased, they are likely to be the major constituents of
the slabs sinking through the transition zone.
During subduction, as the slab warms up to sufficiently high temperatures,
hydrous b-phase þ stishovite would melt incongruently to garnet þ melt. This melt
may serve as a lubricant between the slab and the enclosing mantle. A sudden
movement of the slab through lubrication by the volatile-rich melt from the mantle
rock would cause a deep-focus earthquake. Such melting can also occur through the
introduction of volatiles to the transition zone from the lower mantle or from the
subducted slabs.
Oceanic lithosphere may show a trend of increasing olivine at depth,
corresponding to a change from basalt at the upper part to residual dunite at the
base. This subducting lithosphere would show a gradual change with depth from garnet
to b-phase/spinel. Such a change would account for the observed global presence of a
high-velocity gradient in the transition zone. Hence, the transition zone may serve as
the shallow graveyard (initial to the D” zone) for the subducting lithosphere (see also
Section 2.10).

11.6.3. Ultra-high-pressure metamorphism

Metamorphism occurring at pressures greater than , 2.5 GPa (, 80 – 90 km depth)

is designated as an ultra-high-pressure or very high-pressure (UHP or VHP) process
(see also Section 1.2.3.2).

11.6.3.1. Coesite –diamond

The rocks formed at a UHP of 3– 4 GPa (< 100– 120 km depths) and showing the
occurrence of coesite and diamond are believed to be the crustal material which has been
transported back to the surface (Xu et al., 1992; Okay et al., 1993).
Silicate Melts and Rocks 811

An appropriate chemistry through UHP at great depths can produce coesite plus
micro-diamond. In garnet –peridotite, the assemblage is of magnesite þ diopside ^ Ti-
clinohumite and in eclogite talc is formed. UHP regimes offer important clues towards an
understanding of subduction and continental collision. These help to bridge the gap in our
knowledge about the upper mantle and crustal processes. The VHP rocks reveal a complete
record of geodynamic pathways and offer the constraints for the mechanism of subduction
and tectonic exhumation.
However, coesite and diamond, as UHP minerals of mantle origin, have long been
recognized in kimberlite pipes and in meteorite craters. The discovery of coesite and micro-
diamonds in UHP crustal rocks has revolutionized our understanding of continental
collision zones and mantle dynamics attending subduction. In large UHP terranes,
widespread (but minor) occurrences of garnet peridotites are seen which may represent
fragments of mantle wedge overlying the subduction zones in convergent-plate boundaries.

11.6.3.2. Crustal metamorphic regimes

For HP and UHP metamorphism of the crustal rocks, the maximum temperatures
noted are in the range 750 – 8008C. Above these temperatures, granitic melt ^ migmatite
may be generated when H2O is available. At successive P and T, the assemblages that
appear are: granulite ! amphibolite and epidote amphibolite ! green-schist. While the
green-schist assemblage is stable at low pressure, blue-schist facies occur between 0.5 and
1.6 GPa and T , 400 –4508C. The metamorphic regimes along with geotherms of
, 58C/km (extremely high P=T) and 208C/km (ancient cratons) are shown in Fig. 11.5
(source Liou et al., 1998). Stabilities of diamond (Bundy 1980), coesite (Hemingway
et al., 1998), glaucophane (Holland, 1988), jadeite þ quartz (Holland, 1980), Al2SiO5
(Bohlen et al., 1991), paragonite (Holland, 1979) and aragonite (Hacker et al., 1992) and the
minimum melting of granitic and tonalite solidus (Huang and Wyllie, 1975) are also shown.

11.6.3.3. Hot and cold eclogites: collision/subduction zones

The VHP terranes can be classed as (a) ’hot eclogite’ formed at P greater than
coesite ! quartz transition and (b) ’cold eclogite’ formed at P lower than coesite ! quartz
transition. HP blue-schist and ’cold eclogite’ belts have been seen to occur in tectonic
contact with the coesite-bearing VHP belt. These segments were evidently formed from a
subduction zone during, or even prior to continent –continent collision (Liou et al., 1998).
The VHP terranes, manifesting solid-state crystallization in high P and moderate T
tectonic settings, are seen with major continental collision belts in Eurasia (and Africa) and
are confined to Alpine-type orogens. These are mostly exhumed units, show lithologies
mainly of eclogites and garnet peridotites (included in pods and slabs) and bear
geochemical characteristics which are sub-continental in nature. The occurrences are
commonly associated with late-stage granitic plutons. The VHP phases like micro-
diamonds are seen to occur within garnet and zircon while coesite occurs in garnet and
omphacite.

Sulawesi Indonesia subduction zone. Recently, coesite inclusions in zircon have been
reported from eclogitic rocks from Sulawesi Indonesia (Parkinson et al., 1998).
812 Chapter 11

Figure 11.5. P – T regimes corresponding to various metamorphic types: (1) Very high-P (VHP), (2) High-P, and
(3) Low-P. Geotherm of 58C/km and 208C/km are indicated. P – T boundaries of various metamorphic facies are
from Spear (1993) and subdivision of the eclogite field into amphibole eclogite, epidote eclogite, lawsonite
eclogite and dry eclogite are from Okamoto and Maruyama (published in 1999) (Source: Liou et al., 1998,
q 1998 Mineralogical Society of America).

The terranes show metamorphic phases which appear at depths of , 100 km (e.g.,
Coleman and Wang, 1995). Phase-equilibrium constraints and thermobarometric
calculations reveal that the peak temperatures were , 700 –9008C at confining pressures
greater than 2.8 –4.0 GPa. These conditions may reflect a low geotherm , 7– 88C/km
(present in subduction-zone environments).

11.6.3.4. Dabie –Sulu collision zone

The most recognized VHP terrane in the world occurs in the Dabie – Sulu collision
zone (210 – 240 Ma), in east central China. The UHP terranes, with areas of up to
Silicate Melts and Rocks 813

20,000 km2, consist mostly of crustal rocks. In these rocks, inclusions of high-pressure
minerals such as coesite and/or micro-diamond bear an imprint of pressure to over 3 GPa
during aborted subduction to depths as great as 135 km. This has been thoroughly
investigated by Liou and his co-workers. Their tectonic model and the P – T time path for
the terrane are illustrated in Fig. 11.7 (see caption).
In the Dabie – Sulu terrane, the pre-Cambrian protoliths (granitic, pelitic,
psammitic, carbonate and minor mafic – ultramafic rocks) were subjected to VHP
metamorphism (. 2.5 GPa) at mantle depths through subduction prior to and during
the Triassic collision of the Yangtze and Sino-Korean cratons. In this region, the rocks in
the Dabie Mountains show the presence of coesite and micro-diamond (?) in eclogites and
other metamorphosed crustal rocks (e.g., Liou et al., 1996). The model for the origins of
Dabie – Sulu peridotites in the collision between the Sino-Korean and Yangtze plates and
the cold subducting plate reaching UHP is schematically shown in Fig. 11.4 (read the
caption).
The Dabie – Sulu VHP rocks are unique in the occurrence of (Liou et al., 1998): (1)
abundant coesite and hydrous phases (such as talc, zoisite/epidote, nyböite and phengite) in
eclogite (Zhang et al., 1995a,c), (2) the world’s lowest d18O values (rutile shows 215 per
mil) for mineral separates from eclogites and meta-sediments (Zhang et al., 1998; Rumble
and Yui, 1998), (3) abundant garnet peridotites of mantle origin (Zhang and Liou, 1998)
and (4) abundant exsolution textures in VHP minerals from garnet peridotite and eclogite
(e.g., Zhang and Liou, 1997).
For such VHP complexes, a descent of ancient cold sialic crust overlying a
subducting lithosphere is indicated (Ernst and Peacosk, 1996). Through buoyancy, the
crust may decouple from the downgoing slab and experience adiabatic decompression
while traversing up through the P – T regime of granulite ! amphibolite facies (600 –
8008C, 0.3 –1.0 GPa). However, some abnormally high-P peridotites may be transported
by mantle convection from great depths to the subduction zone and be incorporated into
the subducting continental crust.
The North Dabie Complex (NDC) consists of granitic to monzonitic plutons
of Cretaceous age intruding into amphibolite-facies (ortho) gneisses. The mineral
assemblages and compositions point to an early eclogite-facies metamorphism
(, 800 –8208C).
The Ca –Na clinopyroxene present shows oriented quartz needles (, 20– 20 mm
wide, , 5 – 200 mm long) within it. This implies the prior existence of a non-stoichiometric
“supersilicic” omphacite stabilized at UHP (. 2.5 GPa) conditions, although no evidence
for coesite is noted. The SiO2 needles are interpreted by Tsai and Liou (2000) to be an
exsolution from a precursor, non-stoichiometric omphacitic clinopyroxene, which
contained excess silica at the peak metamorphic condition. The implication of excess
silica in clinopyroxene has been discussed in the Section “Supersilicic clinopyroxene” in
Chapter 6.

11.6.3.5. Alpe Arami UHP lherzolite

The garnet lherzolite from the Alpe Arami peridotite massif (400 £ 1,100 m2) of
the Central Alps may have been metamorphosed at much higher pressures of 10 –15 GPa
814 Chapter 11

(< 300 –450 km depths) (Dobrzhinetskaya et al., 1996). This implies that pieces of mantle-
transition zone (10 – 13 GPa) could be transported to the Earth’s surface.
The basis for considering this lherzolite to be of transition-zone origin lies in the
discovery of (i) a lattice preferred orientation (LPO) of olivine, (ii) a few mm-size FeTiO3
rods, topotactic with the host olivine, and (iii) high TiO2 (inferred) contents of olivine
(Dobrzhinetskaya et al., 1996). FeTiO3 rods in crystallographic structures are ilmenite and
the structures are intermediate between ilmenite and perovskite. These rods are
hypothesized to have exsolved (along [010] direction of host olivine) at 10 –15 GPa
pressure (< 300 –450 km) (Mehta et al., 1994), which is high enough for olivine to
transform from b- to g-phase and manifest a LPO of these phases (wadsleyite or
ringwoodite). The great abundance of FeTiO3 and spinel precipitates in first-generation
olivine in Alpe Arami have no known counterpart in any other peridotite massif or
xenolith.
However, commenting on the work of Dobrzhinetskaya et al. (1996), Hacker et al.
(1997) opined that the Chinese or Alpine lherzolite experienced maximum pressures
only . 4 – 5 GPa (i.e., not 10 –15 GPa). They suggested that the study needs some
additional (see Fig. 11.6, top center) evidence as is also required after the discovery
of some such high-pressure polymorphs as: TiO2 with a-PbO2 structure (stable at
5 –7 GPa), high-pressure C2=c clinopyroxene (6 GPa) and majorite garnet (7 GPa)
(Angel et al., 1992).

Figure 11.6. Pressure–temperature stability fields for Mg1.8Fe0.2SiO4 (Mehta et al., 1994) and FeTiO3 (Helfrich
et al., 1989). Conditions within the upper mantle range from the coldest subducting lithosphere to the
subcontinental upper mantle, as shown by dotted lines. Box shows equilibrium pressure and temperature of
silicate minerals in Chinese garnet lherzolite (Hacker et al., 1997).
Silicate Melts and Rocks 815

11.6.3.6. Exsolutions in VHP minerals

Exsolution textures are common in VHP minerals obtained from both eclogitic and
ultramafic rocks. These are:
(i) ilmenite exsolution: such as rods,
(ii) magnetite exsolution: such as plates in olivine and clinohumite,
(iii) quartz exsolution: such as rods in clinopyroxene omphacite,
(iv) monazite exsolution: such as lamellae in apatite, and
(v) rutile exsolution: such as needles in garnet.
Exsolution of lamellae in VHP minerals may have taken place during nearly
isothermal decompression, in contrast to exsolution with falling temperature in primary
igneous minerals (Liou et al., 1998).

Ilmenite rods in olivine

Alpe Armi massif in Central Alps. Micron-sized ilmenite rods in olivine have recently
been identified in garnet peridotites from several VHP terranes (Fig. 11.6, from Hacker
et al., 1997). TEM study of Alpe Arami garnet lherzolite revealed that the FeTiO3 rods in
olivine are topotactic with the host, parallel to its [010] direction and presumed to have
exsolved at 10 – 15 GPa (300 – 450 km) as perovskite, followed by variable conversion
to ilmenite. The unique preferred orientation of Alpe Arami olivine may have formed
during recrystallization of a wadsleyite- or ringwoodite-bearing protolith (Green and
Dobrzhinetskaya, 1997). Thus, it can be argued that the Alpe Arami complex was possibly
derived from the transition zone in the mantle.

Sulu garnet peridotites in Chijiadian (China). The olivine grains in the Chijiadian
lherzolite show remarkably homogeneous distribution of rod-like inclusions of titanates
(Fe0.82Mg0.15Mn0.03TiO3), which are more magnesian than the Alpe Arami iron titanates
(Fe0.94Mg0.06TiO3). The long axes of the inclusions are parallel to [010] of the host olivine
crystals.
TiO2 contents of olivine from the Alpe Arami and Sulu terrane are seen to be of
values , 300 ppm, which implies P – T conditions , 4 – 6 GPa and 780 –8208C (Nakajima
and Ogasawara, 1997).

Magnetite exsolutions. From some mafic-ultra-mafic rocks, magnetite rods/plates have

been reported to have exsolved in olivine and clinohumite. One such case is Dabie
harzburgite and garnet pyroxenite.
The topotaxial intergrowth between olivine and oriented magnetite lamellae is as
follows:

½220mag k½220ol ½111mag k½33 1

 ol ½111
 mag k½331ol ½242mag k½220
 ol

This offers evidence for an exsolution phenomenon involving the binary system
Fe3O4 – (Fe,Mg)2SiO4.
If the original (Fe,Mg)2SiO4 phase was actually b-phase (wadsleyite) with distorted
spinel structure, it would accommodate Fe3O4 as spinel solid solution along the binary join
816 Chapter 11

Fe3O4 – (Fe,Mg)2SiO4. During decompression, the transformation of wadsleyite (b-phase)

to olivine (a-phase), exsolution of excess Fe3O4 (magnetite) takes place bearing the
topotaxial relationship.
Under pressure, a significant amount of Fe3þ can be incorporated into a wadsleyite
structure through substitution of 2Fe3þ ¼ Fe2þ þ Si4þ. The difference in stability of Fe3þ
wadsleyite (5.0 – 6.0 GPa at 1,100 –1,2008C) is adequate to stabilize Fe3þ-enriched
(Mg,Fe)2SiO4 to lower pressure (shallower depth) compared with a Fe3þ-poor system.
Evidently, such a relationship would alter the depth of the ’410 km discontinuity’ in the
mantle (Woodland and Angel, 1998).

Silica rods. The nature of silica rods in clinopyroxenes has already been discussed in the
Section “Supersilicic clinopyroxene” in Chapter 6.

Monazite lamellae. Monazite lamellae in apatite have been discussed elsewhere.

11.6.4. Basalts and eclogites

The melting T of eclogite increases with P whereas potassic basalt shows this
characteristic only at 1.5 – 2.5 GPa and at . 3.0 GPa. However, between 2.5 and 3.0 GPa,
the melting T decreases with P. If the whole deeper mantle is composed of eclogite
(Anderson, 1979), then a sharp discontinuity would appear at 400 km depth (< 13 GPa).
The composition of the rocks in the lithosphere and the types of hydrous mineral
and their stable P – T conditions are important factors controlling the melt behaviour of
rocks. This may explain the partial melting of rocks and the origin of the low-velocity zone
in the deep lithosphere.
Eclogite with a jadeite-rich (Na – Al-rich) pyroxene shows a density of , 3.3–
3.8 g/cm3. This metamorphic equivalent of basalt is composed primarily of garnet and
omphacite. Eclogites such as from the Fransiscan tectonic block, experiencing retrograde
and prograde metamorphism to eclogite facies, consist of garnet, glaucophane, phengite,
albite, quartz and rutile, which record peak metamorphism at pressures , 1 GPa and
temperatures , 3008C (Oh and Liou, 1990). The other distinctive rock types associated
with these are blue-schists and garnet amphibolite. A typical P – T path for this complex is
shown in Fig. 11.7 (Source: Mysen et al., 1998) (see caption).
Dissolution of omphacite in the co-existing garnet is minimal at 13 GPa and is
complete at 17 GPa when eclogite transforms to garnetite. The formation of the pyroxene-
rich garnet could occur isobarically by a univariant transformation from pyrope-rich garnet
and omphacite at pressures between 13 and 17 GPa (Gasparik, 1996). In a pyrolite mantle,
the dissolution of pyroxene in garnet occurs over a wide range of pressures below 17 GPa
with pyroxene fully dissolved at 17 GPa. If the deeper part of the upper mantle is composed
of eclogite, discontinuity in seismic velocities at 400 km depth is expected to be sharp.
The stability field of garnet expands with increasing pressure from 13 to 17 GPa by
the dissolution of pyroxene in garnet, while the field contracts between 17 and 25 GPa by
the exsolution of CaSiO3 perovskite and MgSiO3 ilmenite. The compositions of the first
Silicate Melts and Rocks 817

Figure 11.7. Pressure–temperature path of in situ eclogite facies rocks from Jenner, Fransiscan Complex.
Modified after Oh and Liou (1990) with albite breakdown curve from Newton and Smith (1967) (Source Mysen
et al., 1998a, q 1998 Mineralogical Society of America).

garnet at 13 GPa and the garnet co-existing with CaSiO3 perovskite and MgSiO3 ilmenite
at 22 GPa are coincidentally identical and close to En51Di9Jd40.
Compositional gaps of variable widths could exist between the ternary and more
pyrope-rich garnets, between the Na-poor and Na-rich garnets and between the Ca-poor
and Ca-rich garnets. (Note: The stability field of garnet is seen to expand substantially by
addition of even a small amount of Na.) Such immiscibility could produce a sharp
discontinuity in an eclogite mantle at 400 km depth. However, a sharp 500 km
discontinuity, corresponding to the breakdown of the diopside omphacite to garnet and
CaSiO3 – perovskite, can only be present in a Ca-rich mantle (Gasparik, 1996).
Partial fusion experiments with basic granulites were performed at 0.5 –1.5 GPa
by Springer and Seck (1997). The melt compositions were seen to change as:
trondhjemitic ! tonalitic ! choritic, with an increasing degree of partial melting.
At 0.5 GPa, the crystalline residua with plagioclase and pyroxene are dominant.
At 1.5 GPa, garnet/pyroxene dominate. Melts from granulites match the major element
compositions of natural trondhjemites and tonalites. At 0.5 GPa, their Al2O3 content is
relatively low, similar to tonalites; at 1.5 GPa, Al2O3 is high due to the near absence of
plagioclase in the crystalline residua.

11.6.4.1. Dehydration melting of metabasalt at 0.8– 3.2 GPa

Partial melting experiments on amphibolites representing metamorphosed Archean
tholeiite (greenstone), high-alumina basalt, low-K tholeiite and alkali-basalt were carried
out by Rapp and Watson (1995). Silicic to intermediate liquids result from , 20 –40%
melting between 1,050 and 1,1008C, leaving a granulite residue at 0.8 GPa and garnet
granulite to eclogite residues at 1.12– 3.2 GPa.
818 Chapter 11

The experimental data suggest that the Archean TTG (tonalite-trondhjemite –

granodiorite) suite of rocks can be generated by 10 – 40% melting of partially dehydrated
metabasalt at P . the garnet phase boundary ($ 1.2 GPa) and T of 1,000 – 1,1008C.

11.6.5. MORB

Because MORB has high Si, Al, Fe and Na content, the minerals developed at
mantle pressures are substantially complicated in nature compared with those obtained
from transformed mantle peridotite, viz. perovskite, CaSiO3 –perovskite and Mg-wüstite
(Kesson et al., 1998). The high Al2O3 content of MORB also results in higher
majorite – perovskite transition pressure than required for peridotite transition. The zero-
pressure density of basaltic crust with perovskitic lithology is 4.23 g/cm3. Hirose et al.
(1999) calculated the zero-pressure densities at 24 and 26 GPa as 3.87 and 3.92 cm3,
respectively. These values are consistent with those obtained much earlier by Irifune
and Ringwood (1993). Up to 27 GPa, the sub-solidus phase relations are determined by
Hirose et al. (1999).
At 24 GPa (/2,023 K), the mineral assemblage is composed of majorite þ
stishovite þ CaSiO3 – perovskite. Along with this, an aluminous phase with Ca –ferrite
structure is noted. This observation is consistent with that recorded in earlier experiments
by Irifune and Ringwood (1993).
At 26 GPa (/. 2,473 K) a new Al –Ca phase is found. This is akin to the CAS phase
described by Irifune et al. (1994). A majorite –perovskite transformation is seen to start at
this pressure of 26 GPa (< 720 km depth) and 2,000 K (Hirose et al., 1999). The lithology
changes from garnetite to perovskitite , 26 GPa.
Under the 660 km discontinuity, if the slabs of basaltic lithosphere accumulate to
form a megalith of 60 km thickness, its transformation to denser perovskite lithology
would cause it to penetrate deep inside the mantle. This transition boundary has a positive
P – T slope, whereas the transition boundary in the underlying harzburgite mass has a
negative P – T slope (Irifune and Ringwood, 1987).
However, no other major phase transformations have been reported at higher
pressures up to 100 GPa (Kesson and Ringwood, 1994). This suggests that these phases
remain stable in the deep mantle except for majorite, which is fully transformed to
perovskite at P . 27 GPa. At this pressure (, 4,000 K), the partial melt generated should
have a compositional enrichment in MgO, FeO (with depletion in SiO2 þ Al2O3). Thus, the
solid residue in MORB composition at lower mantle pressure would become denser
because of its higher iron content.
At a depth of 1,500 km (< 64 GPa), the melting temperature of basalt is about
250 K lower than that of mantle peridotite (Zerr et al., 1998). Extrapolation to 135 GPa
yields a melting temperature of MORB , 4,000 K, i.e., at the CMB. The melting curve of
MORB extrapolated to the CMB is presented in Fig. 11.8.
Thus, if the temperature locally reaches 4,000 K in the D00 region, which may be a
graveyard for subducted lithosphere, the crustal material of the basaltic component would
partially melt. This melt at the base of the mantle can account for the recent observations of
the seismic anisotropy (Kendall and Silver, 1996) and anomalously slow P-wave velocities
Silicate Melts and Rocks 819

Figure 11.8. Melting curve of MORB extrapolated to the core –mantle boundary using the melting relationships
of Simon (S) and Kraut and Kennedy (KK). Open and solid circles represent melting temperatures of MORB and
MgSiO3, respectively (determined in a laser-heated diamond cell). Open squares represent melting temperatures
of MORB (determined in the multi-anvil apparatus). Melting curves of St, Ca-pv and Mg-pv are substantially
higher than that of MORB (from Hirose et al., 1999).

(Williams et al., 1996; Revenaugh and Mayer, 1997). However, under such a scenario,
the temperature of the outer core must be higher than the 4,000 K required for the melting
of MORB perovskite (Fig. 11.8; Hirose et al., 1999). The temperature difference over
the thermal boundary between core and mantle may reach 1,500 K and hot mantle
plumes, including partially molten slab materials, are likely to arise from this depth (Hirose
et al., 1999).

11.6.6. Komatiite, picrite and lherzolite: CaO –MgO(FeO) – SiO2 systems

The geochemistry of komatiites holds considerable clues to the scale of mass

transfer in the Earth (cf. Herzberg, 1995). High-pressure liquid-phase diagrams obtained
from multi-anvil experiments on komatiite and peridotite (e.g., Herzberg et al., 1990;
Herzberg and Zhang, 1996) in the 5 – 25 GPa pressure range offered important
information.
The CaO –MgO – FeO – Fe2O3 ^ Fe0 –Al2O3 – SiO2 (CMFS and CMFAS ^ Feo)
system represents komatiitic and picritic rock compositions. Komatiites record multiple
saturation at , 3 –10 GPa in the upper mantle (Herzberg and Zhang, 1996). The 3,500-Myr
Barberton komatiite has a range of CaO:Al2O3 that indicates multiple saturation pressures
of 8 – 11 (^ 2) GPa.
820 Chapter 11

Herzberg and Zhang (1997) reported the results of multi-anvil melting

experiments on a wide range of komatiite analogue mixed with compositions in the
system CaO – MgO –FeO – Fe2O3 ^ Fe0 –Al2O3 –SiO2 at 5 GPa. The liquidus crystal-
lization fields for olivine (ol), orthopyroxene (opx), clinopyroxene (cpx) and garnet
were mapped out, as were their intersections at various cotectic and invariant points.
The effect of FeO is to expand the liquidus crystallization phases: pyroxenes at the
expense of olivine and clinopyroxene at the expense of orthopyroxene (see also
Trønnes et al., 1992).
Herzberg and Zhang (1998) experimented on a range of compositions in the system
CaO – MgO – Al2O3 –SiO2. The liquidus crystallization fields for forsterite, orthopyroxene,
clinopyroxene and garnet have been mapped out at 10 GPa, as have been at their
intersections at various cotectics. It is observed that pressure reduces the content of Al2O3
and increases MgO and SiO2 contents in magmas formed by the melting of garnet
lherzolite with increasing pressure (Herzberg, 1992).
The viscous flow of alkali-alumino-silicate melts is constrained by oxygen
exchange between polymeric units (Poe et al., 1992). In this flow, structural relaxation may
occur through the formation of transient five-coordinated (Si or Al) species.
Progressive melting must take place because in the ascending plume the adiabatic
gradient has a smaller dT=dP than the solidus. The chemical composition of the upper
mantle is considered to be lherzolitic. High-pressure experimentation on garnet lherzolite
(liquid þ olivine þ orthopyroxene þ clinopyroxe þ garnet) showed that, during melting
on decompression, all clinopyroxene and garnet can be dissolved (e.g., Herzberg, 1995).
The liquids that erupt solidify to picrites and komatites.
Experiments in both CaO – MgO –Al2O3 –SiO2 and MgO – SiO2 demonstrate that
there is a maximum normative olivine content to liquids formed by the initial or advanced
melting of peridotite in the upper mantle and this occurs at 7– 8 GPa. During the ascent of a
peridotite in a plume, clinopyroxene and garnet are the early phases to melt out during
decompression. Advanced anhydrous melting will yield liquids with a residual harzburgite
mineralogy (L þ ol þ opx).
Komatiites and picrites of Cretaceous age are thought to have melted in a plume
that gave rise to the Carribean plateau (Storey et al., 1991). The parental komatiite
composition was dominated by harzburgite (L þ opx þ ol) (e.g., Herzberg, 1995).
A residual harzburgite signature for most komatites with Cretaceous and late-Archean
ages has been interpreted to have been formed by about 25 –60% anhydrous melting of
mantle peridotites and in plumes with potential temperatures that were 200 – 4008C higher
than those of present-day oceanic ridges (e.g., Walter, 1998; Herzberg and O’Hara,
unpublished work).

11.6.7. Garnet peridotites: “forbidden zone”

The P – T diagram shows the positions for garnet peridotite xenoliths from Lesotho
kimberlites (Carswell and Gibb, 1980). The estimated P and T for Lesotho garnet
peridotite xenoliths range from 870 to 1,4508C and from 26 to 56 kbar, respectively.
A small portion of the terranes in SW China is made up of garnet peridotites, studies on
Silicate Melts and Rocks 821

which have led to some exciting findings. The garnet peridotites of the Dabie – Sulu terrane
of east central China may have crystallized in the ’forbidden zone’ at depths of 185–
250 km. These peridotites originate in the mantle and their mineralogical characteristics
imply that they have experienced recrystallization at high temperatures (. 1,0008C). In
some cases, their crystallization ages are considerably older than their host country rocks
(Krogh and Carswell, 1995). Amongst all Eurasian garnet peridotites, P – T conditions
are seen to be highest in Sulu-Debie UHP terrane, indicating processing conditions of
750 –9508 and 4.0 – 6.7 GPa. Some of these P – T conditions lie within the forbidden zone
(Liou and Zhang, 1998).

11.6.7.1. Exsolutions
Garnets in peridotites from the Western Gneiss Region contain pyroxene needles,
indicating that exsolution (unmixing) has occurred after initial formation of a supersilicic
garnet at depths . 185 km (e.g., Terry et al., 1999).
The Alpe Arami garnet peridotite of northern Italy contains exsolution lamellae of
clinoenstatite within diopside, with crystallographic evidence implying initial formation of
depths . 250 km (Bozhilov et al., 1999).
In Chinese UHP rocks, the peridotites contain unusual FeTiO3 rods observed in
olivine (Debrzhinetskaya et al., 1996).

11.6.7.2. Emplacement of garnet peridotites

The mechanism for emplacement of deep, mantle-borne garnet peridotite at
the Earth’s surface may involve transportation by asthenopheric upwelling from depths
of about 135 km, tectonically inserted into subducted continental crust and then
exhumed.
These forbidden-zone garnet peridotites of Dabie – Sulu may have been placed in
the crust before subduction and subjected to in situ UHP metamorphism together with the
subducted slab. These bodies offer the first evidence of continental rocks being subducted
to depths of 200 km or more. The discovery of garnet peridotites from the forbidden zone
now provides a revolutionary new window into the subduction of continental margins, the
thermal structure of the subduction zones and the recycling of volatiles into the mantle
(Liou et al., 2000).
Beneath northeast Japan, a 130-million-year-old Pacific Plate is foundering at the
rate of 9.0 cm/year (Peacock and Wang, 1999). This could be a locale for the crust to be
subducted down to 200 km depth without being heated beyond 9008C. If the subduction to
2 km depth takes , 5 m.y., the rocks in the middle zone of the subducted slab will undergo
little heating as the diffusion distance for this time scale is 10– 15 km. These cold
subduction zones are clearly the sites of major recycling of H2O into the mantle.
This page is intentionally left blank
 823

Chapter 12
Simple Oxides and Carbonates

12.1. Dioxides: SiO2

Silicon dioxide (SiO2) has many polymorphs developed through different pressure,
temperature and chemical environments. The stability fields and crystal chemical
properties at high temperature and pressure of the SiO2 polymorphs (a-quartz, b-quartz,
cristobalite, tridymite, coesite and stishovite) melts, and glasses have been extensively
studied. These phases represent archetypal framework structures, which provide insights
into the intrinsic behaviour of the Si –O linkages. As already discussed in Section 3.2,
under AB2 structure, the quartz form of SiO2 is a very important piezoelectric material
used for transducers and frequency-control devices. Large crystals of quartz are grown
commercially in large pressure vessels from the system Na2O – SiO2 –H2O at about 1 kbar
and 5008C.
While reviewing the high-pressure behaviour of silica, Hemley et al. (1994) showed
evidence for extensive metastability in silica at high pressure in both the low-pressure,
tetrahedrally coordinated Si phases and higher pressure octahedrally coordinated phases.
This fact complicates the unambiguous identification of equilibrium phases and the
determination of thermodynamic P – T stability fields. These results confirm the propensity
of SiO2 to exhibit extensive metastability in both the low- and high-pressure phases.
A view of the structure of a-SiO2 called quartz is shown in Fig. 12.1. The
generalized P – T phase diagram of SiO2 is shown in Fig. 12.2. The solid lines are
equilibrium-phase boundaries and the dashed line is the stishovite melting line as
suggested by Zhang et al. (1993). The dotted lines are metastable extensions of the melting
lines (Hemley et al., 1994). The high-pressure form of SiO2 was identified in natural
samples from the meteor crater at Arizona.
The formation of probably disordered, metastable crystalline phases during room-
temperature compression seems to be a general feature on the pathway to pressure-induced
amorphization for most of the known SiO2 polymorphs. Fig. 12.3 (Gillet, 1996) shows how
in situ Raman spectroscopy reveals the passage of crystalline a-phase to another
crystalline polymorph (quartz II) and finally to an amorphous phase characterized by a
strong decrease in Raman intensity and the growth of broad, glassy bands similar to those
observed in silica glass compressed to the same pressure (Hemley et al., 1986, 1988). The
metastable transition and pressure-induced amorphization of quartz above 20 GPa have
been observed by Raman spectroscopy (Fig. 12.3) (Gillet, 1998). Up to 20 GPa, the
824 Chapter 12

Figure 12.1. Structure of a-quartz. Shaded circles: Silicon; Open circles: Oxygen.

Figure 12.2. P – T phase diagram for SiO2. The a-quartz-coesite (horizontal arrow) transition pressure boundary
is close to the metastable extensions of the equilibrium melting of these phases (Hemley et al., 1988, 1994).
Transitions to metastable crystalline phases (quartz II and coesite II) occur at similar pressure. The temperature-
induced amorphization of coesite (Richet 1988) is shown by the broken arrow (after Hemley et al., 1994).
Simple Oxides and Carbonates 825

Figure 12.3. Phase transition and pressure induced amorphization of quartz above 20 GPa observed by Raman
spectroscopy (Gillet et al., 1998, q 1998 Mineralogical Society of America).

spectrum is characteristic of quartz. Between 20 and 23 GPa, birefringent lamellae appear

throughout the crystal (Kingma et al., 1993a). Band broadening as well as the
disappearance of some bands is related to the transformation of quartz into a new variant
called quartz II. At 25 GPa, a major band can be observed in spectra recorded in some
places of the sample that resembles that of SiO2 glass compressed at similar pressures
(Hemley et al., 1986). The 25 GPa spectrum has been corrected for the strong fluorescence
background systematically observed at the onset of amorphization, which might be due to
the creation of defects within the sample. The spectrum of the decompressed material
indicates that the sample is predominately amorphous (Gillet, 1996).
At ambient conditions, the XRD peaks at 1.59, 2.61 and 3.07 Å are attributed to Si –
O, O –O and Si – Si pairs, respectively. Under compression to 28 GPa, the Si – O peak
increases from 1.59(1) to 1.64(1) Å and the Si –Si separation seems to increase. Studies of
the crystalline SiO2 phases show that the Si – O separation increases with the compression
of the Si – O – Si angle (Gibbs, 1983) and with increase in Si coordination. At this pressure,
Meade et al. (1992) found that the Si – O distance (1.64 Å) is significantly larger than the
value that would be expected for tetrahedrally coordinated SiO2 (1.57(2) Å). Between
826 Chapter 12

8 and 28 GPa, the coordination of Si increases and, at 42 GPa, the value is close to 6
(Meade et al., 1992).
Crystal to crystal (c ! c) phase transitions represent the first step in densification,
which is achieved by the closure of Si –O – Si angles. The bulk moduli of the polymorph
show a distinct correlation with the Si – O – Si angle change with pressure (see below).

SiO2 Bulk moduli kSi–O –Si anglel

polymorph K0 (GPa) change (8C/GPa)

Quartz 37 21.4(1)a
Cristobalite 11 25.7(2)b
Coesite 100 20.58(2)c
a
Glinnemann et al. (1992); b Downs and Palmer (1994); c Levin and Prewitt (1981).

Further closure of these angles cannot occur due to oxygen – oxygen and silicon–
oxygen repulsion, as revealed by the strong decrease in the pressure-induced frequency
shifts of Raman-active modes (Williams et al., 1993).
Coesite shows a bulk modulus of 100, which is quite large compared with alkali
felspars (see Section 7.1.2), although its overall angle bending is significantly greater
than for the alkali felspars (compare with Table 7.1). The elastic modulus of three
high-pressure phases of silica from LDA pseudopotential calculations are shown in
Fig. 12.4. (Karki et al., 1997c). The pressure required for the phase transition from
stishovite to the CaCl2 structure is in excellent agreement with experiments (Kingma
et al., 1995).

12.1.1. a-Quartz: structural change

a-Quartz structure has hexagonal D43 symmetry and is defined by two lattice
constants ðc; aÞ and four internal parameters ðu; x; y; zÞ: The parameter u and the set of
parameters x; y; z define the positions in the unit cell of the three Si and six O ions,
respectively.
The structural parameters derived from the model of the ideal bcc packing of
oxygen are x ¼ y ¼ 1=3; z ¼ 1=12 and c=a ¼ ð3=2Þ < 1:225: For ideally centred (SiO4)1/2
tetrahedra, u ¼ 5=12 < 0:417:
First-principles calculations of the electronic and structural properties of the
pressure-induced transformation of quartz show that the oxygen sublattice changes to a
body-centred cubic structure during the pressure range of 30 –60 GPa (Binggelli and
Chelikowski, 1991).
The formation of the cubic sublattice facilitates structural transformations involving
a change in Si coordination. In the oxygen bcc lattice, a small displacement of Si ions along
the empty channels of the structure is sufficient to transform a-quartz to a structure with
mixed 4-/6-fold coordination or, for a larger Si displacement, to a purely 6-fold coordination
structure. In molecular dynamics (MD), simulations such crystalline phase with mixed Si
coordination above the amorphization pressure is also noted (Tsuneyuki et al., 1989)
Simple Oxides and Carbonates 827

Figure 12.4. Elastic constants of three high-pressure phases of silica from LDA pseudopotential calculations
(Karki et al., 1997c). The pressure of the phase transition from stishovite to the CaCl2 structure is in excellent
agreement with experiment (Kingma et al., 1995). The predicted phase transition to the columbic structure is
consistent with some of the diffraction data reported by Kingma et al. (1996) (see also Teter et al. 1998).
Experimentally measured elastic constants of stishovite are indicated by symbol (Weidner et al., 1982a, q 1997
American Physical Society).

The compression/expansion in quartz is controlled dominantly by the Si – O –Si

angle (¼ 1448 at room P; T) changes. The strongest Raman band of a-quartz arises
from a tetrahedral network structure composed of six- and four-membered rings,
respectively (Sharma et al., 1981). The relatively large negative temperature
dependence of the ni (Si – O – Si) band at 464 cm21 in quartz indicates that this angle
increases with increasing temperature. The temperature and pressure dependence of the
strongest (A-type) Raman bands of the three SiO2 polymorphs are shown in Table 12.1
(Liu et al., 1997).
Compression to 15 GPa brings about significant distortions of the SiO4 tetrahedra
and large changes in the Si –O – Si linkages (Hazen et al., 1989). Indeed, there seem to be
severe steric constraints to the compression of the polymerized network structure of
a-quartz. A large increase in the density of planar defects with increasing pressure
828 Chapter 12

TABLE 12.1
Raman bands of SiO2 polymorphs (Liu et al., 1997)

ni cm21 ðdni =dPÞT 2ðdni =dTÞP

cm21/kbar cm21/8C

Quartz 464 0.85 0.012

Coesite 521 0.29 0.0037
Stishovite 753 0.34 0.016

(.15 GPa) indicates that compression-induced lattice distortions produce irreversible

strains in quartz at high pressures.
The generation of the amorphous phase results in extensive short-range
disordering of SiO2 within the glassy lamellae. The conversion from a crystalline to
an amorphous phase involves bond breaking with relatively small atomic displacements
but without long-range diffusion. In contrast, the misalignment of crystalline blocks
across the amorphous lamellae produces disorder over much greater length scales
(Kingma et al., 1993).
In their test materials of oxides, Kawai and Nishiyama (1974) noted drastic
electrical resistance changes from 105 to 1021 V at the 100 –200 GPa range. However,
they ruled out the possibility of insulator to metal transition for SiO2.
Liu (1974), however, observed that SiO2 in the presence of MgO is unstable in both
the orthorhombic (a-PbO2 type) and rutile structures.

12.1.1.1. Fracture strength

The fracture strength of quartz at ambient conditions is about 4 GPa (Paterson,
1978, p. 30). However, stress concentrations in localized areas such as fractures and crack
tips may be much higher. Amorphous SiO2 on a TEM scale has been found along
fractures that were formed during indentation deformation experiments that tested the
fracture toughness of natural quartz (Ferguson et al., 1987). During shock of quartz single
crystals, a luminescence coupled with extreme localized heating has been observed
(Brannon et al., 1983).

12.1.2. Stishovite

Stishovite is the highest-pressure polymorph of SiO2 and is stable at pressure in

excess of 10 GPa. It has a rutile structure with each Si atom bonded to six O atoms in an
octahedral configuration and each O atom bonded to three Si atoms. The density of
stishvoite (4.29 g/cm3) is ,46% greater than coesite, the densest of silica polymorphs with
tetrahedrally coordination of Si. The high density of stishovite is achieved mainly by
increasing the coordination of Si from four to six. In the mantle, it is the only form of free
silica that is produced from disproportionation reaction or from a-quartz (carried down by
a subducted slab).
Simple Oxides and Carbonates 829

12.1.2.1. Structure
Stishovite crystallizes with the tetragonal rutile-type (C4) structure, which can be
described by using two parameters: the ratio of cell edges, c=a; and the single-anion
positional parameter, u: Because of the greater compressibility of a; the ratio c=a increases
with increasing pressure but the x parameter of O does not show such a change.
However, rutile isomorphs do not conform to the inverse relationship between
temperature and pressure where the structural variation with increasing pressure is similar
to the variation with decreasing temperature (Hazen and Finger, 1982). The bonding in
rutile-type compounds is more covalent than in many other O-based structures.
Electrostatic forces may play a smaller role in determining details of the structure. The
bond covalency, non-spherical electron distribution and metal –metal interactions will
change differently with temperature and pressure, thus accounting for the violations of the
inverse relationship.
In stishovite, the SiO6 octahedron (Fig. 12.5) is not a regular polyhedron. The
shared octahedral edges subtend an angle of 818 (not 908) and the non-bonded O…O
separation in the shared edge (2.29 Å) is one of the shortest O…O separations known,
being considerably shorter than twice the traditional ionic radius of O (r ¼ 2:64 – 2:80 Å;
Shannon, 1976).
The inter-octahedral Si –O(3) – Si angle does not show any significant variation
with pressure and nor does the angle subtended by the shared octahedral edge such as the
O(1) –Si – O(2) angle. The longer axial Si – O bonds are nearly twice as compressible as the
shorter equatorial Si – O bonds. However, the shared octahedral edge is more compressible
than the O(1)…O(3) separation. The incompressibility of the O(1)…O(1) separation can
be explained by the nature of the strong Si…Si repulsive forces across the shared
octahedral edge.
Under pressure in the range 0– 5.2 GPa, no significant increase in the distortion of
the SiO6 octahedron is seen (Sugiyama et al., 1987; Ross et al., 1990). However, the
volume of the SiO6 octahedron is seen to decrease linearly with pressure (Fig. 12.6)
(Sugiyama et al., 1987; Ross et al., 1990).

Figure 12.5. One unit cell of the stishovite crystal structure at ambient conditions, with atoms labeled (Ross et al.,
1990, q 1990 Mineralogical Society of America).
830 Chapter 12

Figure 12.6. The volume of the SiO6 octahedron as a function pressure (Ross et al., 1990, q 1990, Mineralogical
Society of America).

The pressure-induced transition from stishovite to the CaCl2-type structure involves

an orthorhombic distortion of the rutile structure (e.g., Andrault et al., 1998). The transition
is driven by a shear ðc11 – c12 Þ instability, which in turn is coupled with a soft Raman mode,
yielding a pseudo-proper ferroelastic transition. The transition pressure is found to be near
50 GPa at room temperature (Kingma et al., 1995). The strongest Raman mode in the post-
stishovite phase (Kingma et al., 1995) would appear at 950 cm21 at 70 GPa. The strongest
peak of stishovite would appear at 801 cm21 in the spectra taken at 15 GPa (Serghiou et al.,
1998).
12.1.2.2. SiO6: densities
Silicon in octahedral coordination, as in stishovite, has been reported from a host
of silicate minerals produced under high pressure, such as from pyroxene (e.g., Angel
et al., 1988), garnet (Ringwood and Mayor, 1971), ilmenite (e.g., Kwai et al., 1974),
wadeite (e.g., Kinomura et al., 1975), hollandite (Ringwood et al., 1967), perovskite
(e.g., Liu, 1974) and others. Stishovite, because of its dense structure with SiO6 octahedra,
shows the highest density among a host of high-pressure phases (Table 12.2).
Under pressure, the volume of the SiO6 octahedron shows a linear decrease
(Fig. 12.6), as stated earlier.

12.1.2.3. Elastic moduli

With the breakthrough in synthesizing high-quality single crystal of stishovite,
a great deal of investigation of it has become possible. The single-crystal elastic
moduli of stishovite have been determined from Brillouin scattering experiments (Weidner
et al., 1982).
Determination of the bulk modulus from compression experiments is difficult
because of the presence of so many free variables in the P – V EOS (such as K 0 T and V0 ).
However, acoustic experiments determine elastic moduli directly and hence give reliable
values. The polyhedral bulk modulus obtained by fitting a Birch – Murnaghan EOS to the
Simple Oxides and Carbonates 831

TABLE 12.2
Densities of some HP mineral phases

Minerals Densities (g/cm3)

(Orthoclase) 2.6
Coesite 2.9
Wadeite (K2Si4O9) 3.1
Hollandite (KAlSi3O8) 3.9
Clinopyroxene 3.3
Kyanite 3.6
Garnet 3.7
CAS phase 3.8
NaAlSiO4 (calcium–ferrite structure) 3.9
CaSiO3 perovskite 4.2
Stishovite 4.3

data is 342 GPa, which is the largest value yet known among rutile-type oxides. The bulk
moduli of the octahedral units of GeO2, SnO2 and TiO2 for comparison are 270, 230 and
240 GPa, respectively (Hazen and Finger, 1982).
The acoustic value of 306 ^ 4 GPa for KT leads to the K 0 T value of 2.8 ^ 0.2
(Ross et al., 1990). Thus, the KT value for stishovite is substantially lower than the
values of other rutile oxides, TiO2, SnO2 and GeO2, which range between 5.5 and 6.8.
By high-pressure study to 16 GPa, Ross et al. (1990) determined the polyhedral bulk
modulus of SiO6 as 342 GPa, the largest known value among octahedral units of rutile-
type oxides.
Stishovite belongs to the new family of super-hard materials: the metastable
high-pressure phases of high-valence cation oxides with high bulk moduli induced by an
increase in coordination number.

12.1.2.4. Compressibility
Sugiyama et al.’s (1987) single-crystal XRD study of stishovite to 6 GPa pressure
revealed that the shared octahedral edge is more compressible than the unshared edges and
the longer Si – O bonds were less compressible than the shorter Si – O bonds. In addition,
the polyhedral bulk modulus of the Si –O octahedron (250 GPa) is less than the isothermal
bulk modulus (313 GPa).
The a-axis of stishovite is nearly twice as compressible as the c-axis. Therefore, the
c=a axial ratio increases with increasing pressure. The axial compressibilities ðbÞ have
been determined by Ross et al. (1990) as

ba ¼ 1:19 £ 1023 GPa21

and

bc ¼ 0:68 £ 1023 GPa21

The difference in a and c axial compressibilities can be broadly explained in terms
of the cation – cation and anion– anion electrostatic interactions across and along the shared
832 Chapter 12

edges between SiO6 octahedra. The structure is most incompressible along c because of the
repulsive force between the Si atoms in the edge-sharing octahedron of the structure.

12.1.3. Fluorite (CaCl2) structure

The densest form of BO2 compounds consists of a pressure-modified fluorite-type

structure but this structure can be attained by SiO2 only at very high pressure ($150 Pa).
A Rietveld structure analysis of stishovite (Andrault et al., 1998) with the angle-dispersive
XRD synchrotron source at the European Synchrotron Radiation Facility confirmed
a CaCl2 form of stishovite (distorted) at 54 ^ 1 GPa. At the 54-GPa transition, the
P-induced lattice modifications are similar to those found in a Landau-type temperature-
induced transition. Above this transition pressure, the critical temperature increases above
300 K and the lower entropy form becomes stable.
Several ab initio calculations on SiO2 for the pressure range 45 –200 GPa also
indicated a CaCl2 structure in post-stishovite distortion (Karki et al., 1997c). At lower-
mantle pressure (#130 GPa), a post-CaCl2 structure (a-PbO2 type) seems to be feasible.
(Note: At the CMB, the pressure is 135 GPa and the temperature ranges between 3,000
and 4,000 K.)
A single Birch – Murnaghan EOS fits the volumes of stishovite and a CaCl2
form showing that the tetragonal distortion occurs without a substantial change in
volume. Andrault et al. (1998), however, found no further phase transformation up to
120 GPa. The XRD spectra at 54.8 GPa indicated the onset of transformation from
stishovite to a CaCl2 distortion of stishovite. At 63 GPa, the diffraction peaks that
are located at <1.45 and 1.78 Å become doublets. The occurrence of the CaCl2-type
SiO2 polymorph at ,63 GPa agrees well with other workers (e.g., Kingma et al.,
1995). The Raman spectra for transition from rutile to CaCl2 transition and the
pressure dependence of Raman frequencies for stishovite showing the transition from
TiO2 to CaCl2 (Kingma et al., 1995) is shown in Figs. 12.7 and 12.8 (see caption)
respectively.
The experimental and calculated Bragg lines with cell parameters of SiO2
ðP42 =mnmÞ and CaCl2-type SiO2 ðPnnmÞ at 53.2 and 63 GPa are presented in Table 12.3
(Andrault et al., 1998).
No other spectral modifications were observed from .63 GPa to the maximum
pressure of 120 GPa. At the transition pressure of 54 GPa, between stishovite and CaCl2-
type SiO2, the polar Si –O bond becomes shorter than the equatorial bond and the oxygen
sublattice becomes more symmetrical. Above this pressure, the modification of a and b cell
parameters is achieved by a rapid rotation of the SiO6 octahedra in the ða; bÞ plane, which
is supported by the pressure evolution of the ðx; y; oÞ oxygen coordinates. The density of
SiO2 increases up to the pressures available at the depth of the CMB (,2,900 km).
The pressure evolution of c=a and c=b axial ratios (Fig. 12.9) shows the
transition of stishovite to the CaCl2 form with an increase of the CaCl2 distortion at
pressure up to 120 GPa but, strangely, above 45 GPa, the mean value (,0.651) between
the c=a-axes and the c=b-axes is seen to remain almost independent of pressure.
However, the pressure evolution of the a and b cell parameters (Fig. 12.9) is similar to
Simple Oxides and Carbonates 833

Figure 12.7. Raman spectra showing evidence for the rutile to CaCl2 transition in silica. Lorentzian fits to the
bands predicted by theory are shown as smooth traces. Initial frequency decrease (broadening) of the rutile B1g
mode with pressure followed by frequency increase (hardening) of the CaCl2 Ag mode with which the rutile B1g is
correlated. A broad band, which is not predicted by theory, was observed above 40 GPa (marked by asterick in
upper most panel). The new broad band is highly dependent on stress conditions; in nonhydrostatic experiments, a
broad feature is observed to split from the rutile to CaCl2 transition (from Kingma 1995, q 1995 Nature
Publisher Group).

that usually found for a Landau-type temperature-induced transition (Salje, 1990). For
the pressure evolution of the ða 2 bÞ=a-order parameter, which is related to the entropy
change, there might be an entropy gain through a more symmetrical oxygen sublattice
being attained.
All these suggest that the CaCl2 form is the lower entropy polymorph of SiO2. That is,
the CaCl2 form may as well be the low-temperature polymorph of stishovite. When pressure
reaches 54 GPa, the critical temperature of phase transition, Tc ðPÞ; would reach the tem-
perature of 300 K. Beyond 54 GPa, the gap between the experimental and critical tempera-
tures is seen to increase (Fig. 12.9). For such a second-order transition, this increasing gap
accounts for the increase in CaCl2 distortion. However, the volume change between
stishovite and CaCl2-type SiO2 is negligible. In Landau-type transitions, the energies
834 Chapter 12

Figure 12.8. Pressure dependence of Raman frequencies for stishovite showing the transition from rutile-to-
CaCl2 transition (Kingma et al., 1995). The lines are the results of independent calculations. The frequencies are
compared with those from PIB and LAPW calculations (modified from Gillet et al., 1998, q 1998 Mineralogical
Society of America).

involved in octahedral rotations are usually hundreds of joules per mole (de Ligny and
Richet, 1996).

12.1.3.1. EOS
The EOS of silica up to CMB pressure is presented in Fig. 12.10 where the phase
transition of stishovite to CaCl2-type structure is shown to occur at 54 GPa.
Andrault et al. (1998) calculated various third-order Birch –Murnaghan EOS for
silica for different sections of the compression curve (see Fig. 12.9). In the pressure range
of 53.2– 120 GPa, no significant variation in EOS has been noted.
The octahedral bulk moduli, determined from single-crystal data (Ross et al., 1990)
and the results obtained under pressures below 53.2 GPa or above 63 GPa, were found to
be significantly different in stishovite and CaCl2 forms of SiO2 (Table 12.4).
Simple Oxides and Carbonates 835

TABLE 12.3
Experimental (Exp.) and calculated (Calcd) dhkl Bragg lines with cell parameters of SiO2 ðP42 =mnmÞ and CaCl2-
type SiO2 ðPnnmÞ at 53.2 and 63 GPa, respectively (from Andrault et al., 1998a)

dhkl Cell parameters Stishovite CaCl2-type SiO2

Exp. Calcd Exp. Calcd

110 2.810 2.8111 2.789 2.792
a a
011 2.1622 2.160
b
101 2.143
a b
020 1.9878 2.000
b
200 1.948
111 1.898 1.8995 1.888 1.889
120 1.781 1.7779 1.778 1.780
210 1.750 1.752
121 1.464 1.4634 1.461 1.462
211 1.446 1.447
220 1.404 1.4056 1.394 1.396
a-axis (Å) 3.9755 (3) 3.9871 (7)
b-axis (Å) 3.9755 (3) 4.0008 (8)
c-axis (Å) 2.5767 (8) 2.5658 (2)
V (Å3) 40.724 (2) 40.005 (2)

V, volume; numbers in parentheses are uncertainties in measured parameters.

a
SiO2 lines superimposed with Pt lines.
b
Lines found with negligible intensity (Andrault et al., 1998).

Figure 12.9. Pressure-evolution of the unit cell parameters showing a rapid increase of the difference between a-
and b-axes above 54 GPa. The mean value between the a- and b-axes (as well as the c-axis) continues without
discontinuity the pressure evolution found for SiO2 at lower pressures. Symbols correspond to c/a for stishovite
(squares) and c=a (triangles), c=b (circles), and the mean value between c=a and c=b (black squares) for the
CaCl2 form of SiO2 (Andrault et al., 1998a).
836 Chapter 12

Figure 12.10. EOS of silica up to CMB pressure. Data are from Ross et al. (1990), Hemley et al. (1988) and
other studies.

TABLE 12.4
EOS data for silica and SiO6 octahedra. Errors are estimated to be 1 GPa, 0.05 and 0.005 Å3 for K; K 0 and V0 ;
respectively (Andrault et al., 1998)

Pmin (GPa) Pmax (GPa) K0 (GPa) K0 V0 (Å3)

Stishovite
0.0001a 306
0.0001 15 313 4b 46.591
0.0001 53.2 291 4.29b 46.629
CaCl2-type silica
63 120 282 4.29b 46.868
Silica
0.0001 120 289 4.41 46.635
0.0001 120 291 4.29b 46.630
SiO6 octahedra
0.0001 53.2 319 3.32 11.190
0.0001 53.2 303 4b 11.198
0.0001 120c 312 4.95 11.190
0.0001 120c 346 4 11.169
a
Brillouin scattering measurements (25). All other calculations are with the low-pressure single-crystal diffraction
data (20).
b
Fixed data between 15 and 63 GPa were excluded.
c
Experimental data between 15 and 63 GPa were excluded.
Simple Oxides and Carbonates 837

In stishovite, the SiO6 compression is high while the octahedral volume of SiO6
relaxes in the CaCl2 forms of SiO2. The integrated 1% volume variation of SiO6 octahedra
ðDVoct Þ corresponds to a compression work of ,35 kJ ðPDVoct Þ: During phase transition,
an energy exchange takes place between SiO6 octahedra and other polyhedra of the lattice.

12.1.3.2. Theoretical models

A hypothetical post-rutile phase of silica has been theoretically investigated by
many. Bukowinski and Wolf (1986) investigated the EOS and the dynamic and elastic
stability of SiO2 in fluorite structure using a first-principles band-structure calculation and
pair potentials derived from a modified electron gas (MEG) model. This fluorite structure
phase is unstable at pressure below 170 GPa and, therefore, is unquenchable. The
theoretical phonon spectrum also exhibits its dynamic instability below this pressure. This
dynamic instability may possibly make the fluorite phase a stable super-ionic conductor at
lower-mantle conditions. The relatively low-density and low-pressure dynamic instability
of the theoretical fluorite-structure silica is most likely due to strong oxygen – oxygen
repulsive interactions plus an inherent instability of the simple cubic oxygen sub-lattice
(Bukowinski and Wolf, 1986). The entropy increase caused by the positional disorder of
the oxygen ions may be sufficient to stabilize the fluorite phase (relative to stishovite) in the
lower mantle.
Experimentally, Liu et al. (1978) could quench a Fe2N phase of SiO2 from pressure
between 35 and 40 GPa. This phase has a niccolite (NiAs) structure with half the cations
randomly removed. The corresponding ordered structure shows a slightly distorted rutile
structure. However, among high-pressure rutile-type minerals, the unquenchability is very
common (Jamieson, 1977). The apparent unquenchability of high-pressure post-rutile
phases of silica suggests that these phases are dynamically or elastically unstable and
should hold a property that might characterize the fluorite phase (when that exists). The
APW zero-pressure EOS parameters and the thermal Debye temperature of fluorite (SiO2)
are remarkably similar to the experimental parameters of stishovite.
MEG calculation predicts that stishovite transforms to fluorite structure at 390 GPa
and decreases to ,9% by volume (Tossell, 1980). The static properties of this fluorite
phase were modelled by Bukowinski and Wolf (1986) with the augmented plane wave
(APW) method for band-structure calculations. Pair potentials obtained from the MEG
model were also used to complete its phonon spectrum as a function of density.
A sufficient number of augmented plane waves is used to keep convergence errors
in the pressure below 0.1 GPa. Eigen values are computed on a K space mesh
corresponding to 32 points in the Brillouin zone. A mesh of 256 points produces changes
in pressure of the order of 0.1 GPa, which is probably larger than the error inherent in the
approximate correction for the “muffin tin” potential.
From the computed lattice EOS, the values of zero-pressure density, bulk modulus,
K0 ; and its pressure derivative, K00 ; have been obtained by Bukowinski and Wolf (1986) as
4.33 g/cm3, 322 GPa and 4.39, respectively.
Bukowinsky and Wolf (1986) have used central pair potentials derived from the
MEG model to compute the phonon spectrum. At pressures below 170 GPa, the computed
phonon spectrum of fluorite-structured silica gives imaginary frequencies near the
838 Chapter 12

Figure 12.11. Phonon spectrum of SiO2 (fluor) at 170 GPa. Note the frequency acoustic branch at the Brillouin
zone boundary at X (1 /2 0 0) Mode frequencies at this point are imaginary (Bukowinski and Wolf, 1986, q 1987
American Geophysical Union).

Brillouin zone boundary point, X ð 12 0 0Þ: The phonon spectrum at the pressure where the
lattice just becomes stable is shown in Fig. 12.11. Phonons at symmetry-related X points
are dynamically unstable. The atomic motion corresponding to the unstable mode at X is
apparently consistent with a transformation toward hexagonally packed oxygen layers
(Bukowinski and Wolf, 1986). The dynamical instability at pressures encountered in the
mantle makes the flourite structure unquenchable but leads to a mechanism for super-ionic
conductivity, possibly by effectively melting the oxygen sublattice. A super-ionically
conductive phase of SiO2 (fluorite) could also be stable relative to stishovite at high
temperatures.
However, the dominant short-range interactions in SiO2 fluorite are among the
nearest-neighbour oxygen ions. It is this interaction which prevents the phase from having
a density higher than that of stishovite. If the repulsive interaction between Si and O ions in
the fluorite structure is much weaker than O –O repulsion, then the oxygens form a highly
unstable cubic sublattice. Therefore, the fluorite structure can only be brought into
stabilization by a strong Si – O repulsion.
Since Coulomb energy accounts for ,90% of the binding energy of a reasonable
ionic crystal and the standard free energy of stishovite is ,800 kJ/mol (Robie et al., 1978),
the standard free energy of the fluorite phase probably exceeds that of stishovite by about
40 kJ/mol. At a temperature of 3,500 K (; CMB temperature), the vibrational entropy
term may decrease the difference between the free energy of SiO2 (fluorite) and stishovite
by about 10 kJ/mol.

Configurational entropy: super-ionic conductivity. Additional entropy increase in SiO2

fluorite structure (a super-ionic conductor) is due to the disorder in the oxygen ion
occupancy at octahedral and tetrahedral sites. If the fraction of occupied octahedral site is
Simple Oxides and Carbonates 839

x; the entropy due to disorder is

S ¼ ð2 ln 2 2 2 ln x 2 ½2 2 x ln½2 2 x 2 ½1 2 x ln½1 2 xRÞ;

where R is the gas constant. S has a maximum value of 1:91r at x ¼ 0:67; which
corresponds to ,51 kJ/mol at 3,500 K. However, for fluorite structure superconductors, a
maximum expected value of x is ,0.2 (Hayes, 1978), which gives about 31 kJ/mol at the
same temperature. Fluorite structure SiO2, stable at the lowermost part of the mantle,
should show super-ionic conductivity. This bears significant relevance to the electrical
conductivity and rheology of the lower mantle.

12.1.3.3. Columbite (a-PbO2) structure: MD simulation

The possible high-pressure silica phases have been investigated with MD or lattice
dynamics (LD) simulations (Bolonoshko et al., 1996).
Bolonoshko et al. (1996) reported a new high-pressure silica phase, which they
denoted SBAD with space group Pnc2 from MD simulations. They considered the SBAD
structure to be intermediate between the a-PbO2 and baddeleyite structures. Later,
Kanzaki (1997) found that a simple cell transformation for the reported structural data of
the SBAD phase results in a structure that is virtually the same as the well-known a-PbO2-
type SiO2 structure with space group Pbcn. However, in their simulation, Bolonoshko and
Dubrovinsky (1997) started from ideal a-PbO2 at 10 GPa and obtained a structure which
distorted more with increasing pressure. At 100 GPa, they obtained a local minimum with
a 20 KJ/mol higher energy than stishovite with a configuration of atoms having low
symmetry.
The MD simulations by Kanzaki et al. (1997) and those of the first-principles
calculation of Karki et al. (1997) suggest that the a-PbO2-type SiO2 is a candidate for the
post-stishovite phase at the lower-mantle conditions.
Predictions based on LAPW calculations for the transformation of stishovite to
CaCl2 structure (e.g., Cohen, 1992) have been found to be valid experimentally (Kingma
et al., 1995). First-principles calculations predict a phase transition of SiO2 from CaCl2
structure to the columbite (a-PbO2) structure (see Fig. 12.4) (e.g., Karki et al., 1997c). This
ultra-high-pressure structure is built of SiO6 octahedra and differs from the lower pressure
polymorphs primarily in the kinking of the chains of octahedra. The structure with no kinks
is that of stishovite (with CaCl2 structure). The number of kinks increases the density and
the a-PbO2-type structure is the most dense in this series.
Tentative evidence for a-PbO2-type silica has been reported from SNC Shergotty
meteorite (Sharp et al., 1999). However, from this meteorite, Goresy et al. (1998) have
reported baddelyite-type SiO2 with Si in 7-fold coordination.
Teter et al. (1998) found that, at pressures .50 GPa, there exist a large number of
closely related and energetically competitive phases in which the oxygen anions show
close packing with different ordering of Si in the octahedral sites. Calculations show that
a-PbO2 structure is stable above 80 GPa (e.g., Karki et al., 1997c) but, at these pressures,
there should be extensive metastability of phases with similar bonding topology
(e.g., Teter et al., 1998).
840 Chapter 12

12.1.3.4. Seismic velocities: discontinuities

For SiO2, the density and wave velocities are much higher than those of the lower
mantle throughout its pressure regime (by as much as 20% in VS ) except in the immediate
vicinity of the stishovite-to-CaCl2 phase transition (47 GPa, 1,180 km depth). Near this
phase transition both VP and VS show large discontinuous changes (of ,20 and 60%,
respectively) which are associated with a pressure-induced shear instability. This indicates
that, if free silica exists in the lower mantle, it may lead to a globally observable seismic
discontinuity near a depth of 1,180 km (Cohen, 1992). The discontinuities in velocity are so
large that very small amounts of silica (as little as 2%) may be detectable (Karki et al.,
1997c).

12.1.4. Cristobalite

Cristobalite is the highest temperature polymorph of SiO2 and is found in diverse

conditions: in volcanic rocks, low-temperature veins, deep-sea cherts, meteorites, etc. It
has found wide application as refractory material and is used in furnace linings.
Cristobalite, like tridymite, is metastable but its reconstructive transformation to the
equilibrium polymorph, quartz, is kinetically hindered.
It is a tectosilicate and the idealized structure (b-cristobalite) is cubic, possessing a
sequence of silicate sheets of corner-sharing SiO4 tetrahedron arrayed in 6-fold rings
stacked in ABCABC… sequence parallel to a triad axis. With lowering of temperature,
cubic cristobalite undergoes displacive transition to a-cristobalite, a metastable tetragonal
phase. Its equilibrium-phase transition to tridymite or quartz is re-constructive.
In situations like meteorite impact or faulting, where transient high-pressure (i.e.,
stress) conditions prevail, low-pressure phases may be preserved. This metastable
transition involves crystalline – crystalline or crystalline– amorphous transformations
(Gratz et al., 1993; Kingma et al., 1993a,b).
Study of lattice dynamics of cristobalite helped towards understanding the phonon
behaviour of framework minerals employing the concept of “rigid-unit modes” (RUM).

12.1.4.1. Structure: phase transition

The structure comprises a three-dimensional network of corner-sharing SiO4
tetrahedra. Each O atom is a bridging O atom and so the structure is a fully polymerized
tetrahedral framework. The SiO4 tetrahedra are arrayed in six-membered rings within
close-packed layers and stacked parallel to [111] with three-layer ABCABC… repeat. In
the high-temperature form, this idealized cubic structure is analogous to that of diamond
with the C –C bond replaced by Si –O – Si. A space group Fd3m is inferred. On decreasing
temperature, there is a first-order displacive phase transition to a tetragonal form with
space group P41 21 2: Theoretically, this phase transition is due to condensation of a
low-lying rigid-unit mode (Giddy et al., 1993).
On increasing pressure, a phase transition is observed at Pc , 1:5 GPa, characterized
by the onset of twinning and the splitting of powder-diffraction lines (Palmer and Finger,
1994). This transition is reversible and first-order in character. The high-pressure phase is
referred to as cristobalite II, which is monoclinic. The transition (at P ¼ 3:1 GPa) from
Simple Oxides and Carbonates 841

tetragonal a-cristobalite to monoclinic cristobalite II involves a doubling of the unit-cell size

and must therefore be induced by a zone-boundary instability.
The dominant compression mechanism for cristobalite-I is the framework collapse
by “polyhedral tilting”, which is characterized by the decreasing Si –O –Si bond angle (and
the 420 cm21 mode). The cristobalite I ! II phase transition leads to a large volume
change accompanied by shear in the [101] plane of tetragonal phase (Palmer and Finger,
1994). The phase transition can be achieved by a shear of the unit cell with no breaking of
primary bonds. The shear causes a decrease in symmetry from tetragonal to monoclinic
with concomitant doubling of the unit-cell volume. In this sense, the phase transition
shows classical Landau first-order displacive behaviour. Preliminary Rietveld structure
refinement of the diffraction pattern indicates that a combination of polyhedral tilting
(decreasing Si – O –Si bond angle) and tetrahedral distortion is required. This transition
represents the “cross-over” between a compression mechanism dominated by polyhedral
tilting and bond-angle reduction and is dominated by bond-length compression and
polyhedral distortion (Palmer et al., 1994). The Raman spectra of cristobalite under
pressure are shown in Fig. 12.12 (see caption).

Figure 12.12. Pressure-evolution of the cristobalite Raman spectrum, showing low pressure region, spanning the
cristobalite I ! II phase transition at 1.2 GPa. The arrows point to new peaks, characteristic of cristobalite-II
phase. High-pressure region, showing evidence for a possible new “cristobalite III” phase above 13 GPa. Spectra
correspond to measurements on increasing pressure. Base lines have been subtracted from all spectra (Palmer
et al., 1994, q 1994 Springer-Verlag).
842 Chapter 12

The phase with highest possible symmetry is denoted by phase 0 then its subgroups
are numbered in succession with decreasing symmetry. The high-temperature, cubic


aristotype structure ðFd3mÞ; therefore, becomes cristobalite 0, the tetragonal a-phase
ðP41 21 2Þ is cristobalite I and the new high-pressure monoclinic phase is referred to as

and that of the a phase
cristobalite II. The symmetry of the cubic b-phase of silica is Fd 3m
is tetragonal.
Thus, the notations for cristobalite polymorphs, based on the super-group—
sub-group relationship given by Palmer and Finger (1994) are
Cristobalite 0: high-symmetry cubic phase: b-phase
Cristobalite I: tetragonal(a) P41 21 2 sub-group: a-phase or P
Cristobalite II: high-pressure monoclinic ðP21 Þ phase: B-phase
(Note: P21 is a sub-group of P41 21 2:)
There is an orientational relationship between the tetragonal and monoclinic
structures. An abrupt structural-phase transition is noted at P , 1:6 GPa. The unit cell
collapses parallel to a and b and the b angle increases dramatically from 90 to 91.68 at the
phase transition. Rather intriguingly, the c-axis, which is the most compressible axis in the
tetragonal phase, shows a remarkable increase or rebound at Pc ; almost as if a compressed
structural spring has been released. The cell parameters then follow a curved path of
decreasing slope (with continued compression), which must reflect a high compressibility
and high-pressure derivative of cristobalite II (Palmer and Finger, 1994).


In silica cristobalite, the phase transition from the cubic ðFd3mÞ b-phase to the
tetragonal a-phase ðP4321 2 or P41 21 2Þ occurs at ,493 K. From the high-temperature
XRD data, a structure of the cubic b-phase (C9 type) of SiO2 cristobalite was proposed
by Wyckoff (1964). Hatch and Ghose (1991) listed all the rotational modes involved in
going from the b-phase to the various domains of the a-phase and suggested that the cubic
b-phase is a dynamic average of microscopic domains of the tetragonal a-phase with
fluctuation of domain walls.
By repeated cycling through the phase transition, the co-existence of both low- and
high-P phases within the well-defined hysteresis interval ðPc ¼ 1:2 GPa and P0c ¼ 0:2Þ; a
characteristic first-order displacive transition is seen.
The monoclinic B cell is interpreted as a supercell comprising four unit cells of the
low-pressure tetragonal cell. The orientational relationship between cristobalite I
(tetragonal P) and cristobalite II (monoclinic B) can therefore be approximated by
0 1 0 10 1
a 2 0 0 a
B C B CB C
Bb C¼ B 0 1 0C B C ð12-1Þ
@ A @ A@ b A
c B 0 0 2 c P

The inferred space group of the monoclinic phase is P21 ; which is a sub-group of
P41 21 2: For this case, the unit-cell volume doubles at the phase transition. The driving order
parameter is related to an R-point zone-boundary instability (Stokes and Hatch, 1988).
The magic angle spinning nuclear magnetic resonance (MAS NMR) data on 29Si
17
and O in SiO2 as a function of temperature through phase transition have been reported
Simple Oxides and Carbonates 843

(Spearing et al., 1992). The spontaneous strain can be analysed in the context of
macroscopic Landau theory.

Crystal shear and deformation twinning: I ! II transition. In the tetragonal phase of

cristobalite, the extensional axis is parallel to ½101 in the (101) plane. This plane consists
of distorted six-membered tetrahedral rings which, in the high-temperature cubic phase,
becomes a pseudo-close-packed (111) plane. This direction is equivalent to ½1
12;
the slip
direction for the smallest possible “shear” of the {111} layers. The cristobalite
I ! cristobalite II phase transition can therefore be viewed as a shear of the crystal
structure. This involves the sliding of adjacent close-packed sheets against each other,
much like the martensitic-type transformation in close-packed metals.
In their single-crystal diffraction work on the high-pressure phase of cristobalite,
Palmer and Downs (1991) observed a split in single-crystal diffraction peaks (measured in
v and v=2u scans), which are indicative of deformation twinning. This deformation twin is
distinct from the tetragonal to monoclinic transformation twins. However, a co-existence
of both high- and low-pressure phases within the hysteresis region has been observed in
single-crystal X-ray experiments (Downs and Palmer, 1994) and in Raman spectroscopic
studies (Palmer et al., 1991).

Tetragonal ! monoclinic transition: strain tensor. In the cristobalite I ! II phase

transition (presuming a space-group change from P41 21 2 ! P21 Þ; the transition of zone-
boundary nature would allow one to expect a quadrupole coupling and e to vary as Q2 (in
lowest order) to the spontaneous strain. The symmetry of the spontaneous strain tensor is
determined by point-group relations. The individual components of the spontaneous strain
tensor may be calculated from the unit-cell parameters of cristobalite II. Those of
cristobalite I are extrapolated to the same pressure using the general equations of Schlenker
et al. (1978) (summarized in Carpenter, 1988). Non-symmetry-breaking strains ensb nsb
11 ; e22
sb sb
and e33 are very large in contrast to the smaller symmetry breaking strains e11 and e22 and
also the shear strain e13 :

12.1.4.2. Phase transition: symmetry change

At ambient conditions, cristobalite is tetragonal, with space group P41 21 2ðD44 Þ: The
group (422) representation is

optical
G422 ¼ 4A1 ðRÞ þ 4A2 ðIRÞ þ 5B1 ðRÞ þ 4B2 ðRÞ þ 8EðR; IRÞ

acoustic
G422 ¼ A2 þ E

where “R” and “1R” denote Raman-active and infrared active modes, respectively.
The inferred space group of cristobalite II is monoclinic P21 ; which is a sub-group
of P41 21 2: The unit-cell volume doubles at the phase transition, with the driving order
parameter related to an R-point zone-boundary instability (Stokes and Hatch, 1988).
844 Chapter 12

TABLE 12.5
Correlation table for zone-centre vibrational modes associated with the proposed 422 ! 2 phase transition in
cristobalite (Palmer et al., 1994)

422 ðD4 Þ 2 ðC2 Þ

A1 (R)
A2 (IR) A (R, IR)
B1 (R)
B2 (R) B (R, IR)
E (R, IR)

“R” and “IR” denote Raman-active infrared-active modes, respectively.

For the space group P21 ; there is one set of crystallographic sites and one can obtain
the group representation:
G2optical ¼ 35AðR; IRÞ þ 34BðR; IRÞ
G2acoustic ¼ A þ 2B
(all modes are Raman- and IR-active).
The correlation for zone-centre vibrational modes associated with the proposed
422 ! 2 phase transition in cristobalite are shown in Table 12.5.
Raman modes, which appear to vanish when the tetragonal phase appears on
decreasing pressure, need not be correlated to inactive-point modes in cristobalite II but
may become zone-boundary modes in cristobalite I.

12.1.4.3. Cristobalite III

At high pressure (.10 GPa), a phase (cristobalite III) appears which is
unquenchable. The phase is signified by the appearance of a new mode at ,470 cm21.
The transition from cristabalite II ! III is sluggish and offers scope for controversy. For
full evaluation of this phase, a wavelength dispersive synchrotron XRD of polycrystalline
samples (to avoid twinning problems) need to be employed. At higher pressures,
amorphization becomes evident with the appearance of broad bands above 600 cm21. It
should be noted here that amorphization of the lower-pressure phases of silica is associated
with, and is probably related to, the formation of metastable crystalline (or partially
crystalline) phases.

12.1.4.4. Raman study: I ! II transition

The most intense Raman modes of cristobalite (0) polymorph occur at 114, 231 and
420 cm21. These show marked increases in frequency with increasing pressure. Dramatic
discontinuities in frequencies under pressure correlate with the tetragonal – monoclinic
phase transition (Downs and Palmer, 1994).
The ambient-pressure Raman spectra distinctly distinguish tetragonal a-cristobalite
from a-quartz. The most intense peaks in the cristobalite spectrum are at 114, 231 and
420 cm21. Above 800 cm21, the Raman spectrum is characterized by a point of very weak
Simple Oxides and Carbonates 845

Si – O stretching modes close to 1,100 cm 21. The pressure evolution of the

cristobalite Raman spectrum is shown in Fig. 12.12 spanning the cristobalite I , II
phase transition.
At pressures between 1.03 and 1.41 GPa, an abrupt change in the Raman spectrum
is noted. This change correlates with the pressure for the tetragonal (I) –monoclinic (II)
phase transition in cristobalite (Down and Palmer, 1994; Palmer and Finger, 1994).
The kinetics of transformation can be investigated by repeated cycling of pressure.
On decreasing pressure from high-pressure cristobalite phase II, the phase transition is
observed as rapidly as is seen on increasing pressure but a substantial hysteresis is
encountered. The phase transition is characterized by the appearance of new Raman bands,
whose intensities increase with increasing pressure. The observed discontinuity in Raman-
mode frequencies and the large hysteresis observed on increasing and decreasing pressure
suggests that the phase transition is of first order. The transition is effected by a shear of the
unit cell with no breaking of the primary bonds. The shear causes a decrease in symmetry
from tetragonal to monoclinic with concomitant doubling of the unit-cell volume. Thus,
the phase transition follows the classical Landau first-order displacive behaviour. This
transformation requires a combination of polyhedral tilting (decreasing Si – O –Si bond
angle) and tetrahedral distortion. Raman measurements by Palmer et al. (1994) show a
possible co-existence of two phases, cristobalite I and II, within the hysteresis interval, as
is common in first-order phase transitions. Phase transition develops microstructures, often
through twinning, and the resulting domain-boundary strains may prolong the
metastability within the hysteresis interval.
The progressive decrease in mode intensities with repeated cycling through
transition indicates a phase-transition-induced microstructure, which leads the crystal to
become increasingly polycrystalline. On repeated cycling through the phase transition,
however, the overall intensities of the Raman modes decreased, which indicates possible
domain-boundary stresses at the phase transition.
The modes at 420 and 490 cm21 show coupled behaviour. Near to 2 – 4 GPa, the
modes are strongly mixed (mode hybridized). This requires the two modes to have the
same symmetry. In the pressure region of 4.4 GPa, nS (Si –O) for cristobalite and quartz are
nearly equal (Hemley, 1987) but they diverge above and below this pressure. At
P < 3:1 GPa, a distinct crossover in intensity is observed for the two modes at 420 and
490 cm21. The higher-frequency 420 cm21 mode appears to gain intensity at the expense
of its lower-frequency neighbour (Fig. 12.12). Evidence for an unusual hybridization of
modes at 490 –500 cm21 is noted by Palmer et al. (1994).
At pressures between 11.6 and 13.1 GPa, an abrupt change in Raman spectra is seen
with a broadening of the Raman modes. A new mode, close to 470 cm21, is sharp and
indicates another phase transition (cristobalite III). However, on decompression, this
merges with the broad glassy bands. Non-hydrostatic stresses developed at grain
boundaries and domain walls may contribute to the amorphization despite the quasi-
hydrostatic nature of the medium. The appearance of additional broad modes in the region
600 –700 cm21 is suggestive of amorphous silica bands. Kingma et al. (1993b) noted
earlier that such amorphization has to be driven by a structural instability.
846 Chapter 12

Figure 12.13. Correlation between the mean Si–O –Si bond angle and symmetric stretch frequency for
cristobalite and quartz. The cristobalite Raman frequencies are from Palmer et al. (1994), with structural
parameters for cristobalite-I taken from Downs and Palmer (1994). The cristobalite-II structural data wre obtained
by L. Finger (source: Palmer et al., 1994). The quartz data are from Hemley (1987) (Raman frequencies) and
Levien et al. (1980) (bond angles) (from Palmer et al., 1994, q 1994 Springer-Verlag).

nS(Si – O) and kSi– O – Sil. The 410 cm21 mode is due to a symmetric Si –O stretching
mode, nS (Si –O) (Sharma et al., 1981). This mode is highly pressure dependent in quartz
and coesite (Hemley, 1987). A correlation between its frequency and the smallest Si –O –
Si angle in the crystal structure has been noted by Sharma et al. (1981). The correlation
between the mean Si – O – Si bond angle and symmetric stretch frequency for cristobalite
and quartz is shown in Fig. 12.13 (Palmer et al., 1994). With increasing pressure, a steep
increase in nS (Si –O) coincides with a large decrease in the Si –O –Si bond angle.
Downs and Palmer (1994) found that closure of the Si – O –Si angle was correlated with a
very slight increase in the two Si –O bond distances, which could be explained by
increased ionic repulsion as the angle is closed.

12.1.5. a-quartz, coesite, stishovite and cristobalite

The thermodynamic properties of a-quartz, coesite and stishovite and their

equilibrium-phase relations at high pressures and high temperatures were determined by
Akaogi et al. (1995). The isobaric heat capacities were measured by differential scanning
calorimetry at 183– 703 K. The heat capacities and entropies were calculated using
Kieffer’s model. A plot of the normalized averaged kSi– O – Sil vs the normalized unit-cell
volumes of quartz, coesite and cristobalite show a linear relationship (Fig. 12.14).
Simple Oxides and Carbonates 847

Figure 12.14. Normalized unit cell volume V/Vo vs the normalized average Si–O–Si angle. kSi–O–Sil/kSi–O–
Sil0 for quartz, coesite, and cristobalite (from Prewitt and Downs, 1998, q 1998 Mineralogical Society of America).

Coesite is stable at P . 2:3 – 3:0 GPa (,100 km depth). It is seen to occur in UHP
rocks. Experiments at GL have revealed that, at P . 5 GPa and T , 1,370 K, hydrogen can
be incorporated into the coesite structure.

12.1.5.1. Coesite to quartz transformation kinetics

Coesite, a high-P polymorph of SiO2, is stable at P . 2:6 GPa at 7008C. It is
monoclinic, it is distinguished from quartz by its higher birefringence and it is biaxial
(þve). Synthetic (7 GPa and 6508C) coesite contains ,100 ppm H2O, documented by three
infrared sharp absorption peaks at 3,459, 3,524 and 3,572 cm21 associated with OH
stretching vibrations (Li et al., 1997).
Reversal experiments and thermal calculations of the a-quartz ! coesite phase
boundary have been carried out by many workers, e.g., Bose and Ganguly (1995b) and
Hemingway et al. (1998).
Mosenfelder and Bohlen (1997) studied the kinetics of coesite to quartz
transformation. Polycrystalline coesite aggregate was produced by devitrifying silica
glass cylinders containing 2,850H/106 Si at 1,0008C and 3 – 6 GPa for 24 h. Reaction
proceeded via grain-boundary nucleation and interface-controlled growth with character-
istic reaction remarkably similar to those seen in natural ultra-high-pressure rocks.
Extrapolation of growth rates suggests that coesite would not be preserved on geological
time scales if it reaches the quartz stability field at temperature of 375 – 4008C.
The survival of coesite has previously been linked to its inclusion in strong phases,
such as garnet, that can sustain a high internal P during decompression. Other factors that
848 Chapter 12

may be crucial are low fluid availability and the development of transformation stress,
which inhibits nucleation and growth.

12.1.6. Instability and ferroelastic transition

The pressure-induced amorphization of ice (Mishima et al., 1984), a-quartz and

materials in general (Richet and Gillet, 1997) is driven by an intrinsic instability in the
structure.
In the crystalline – crystalline transformation in quartz I ! II that precedes
complete amorphization, abrupt changes appear in the microstructure (macroscopic and
microscopic planar features) beginning at 18 GPa (Kingma et al., 1993). The instability
arises at the region where shortening of the nearest O – O distances occurs (Chelikowsky
et al., 1990) and one of the Born stability criteria ðB2 Þ (Born and Huang, 1954) is violated
at the pressure where the elastic modulus C33 suddenly decreases. However, the B2
instability is considered to be the result rather than the cause of transition. The first-
principles calculations suggest that the elastic instability associated with Born criterion B3
decreases with pressure and, in the case of quartz, becomes negative at ,30 GPa. (The
quartz I ! II transformation exhibits order –disorder character akin to a ! b transition.)
As discussed in Section 5.1.6, The Born stability criteria are the necessary
conditions to be met for a crystal to be mechanically stable. For a trigonal crystal, such as
a-quartz, the elastic constant matrix is positive as

B1 ¼ C11 2 lC12 l . 0

2
B2 ¼ ðC11 þ C12 ÞC33 2 2C13 .0

2
B3 ¼ ðC11 2 C12 ÞC44 2 2C14 .0

In their study up to 20 GPa, Gregoryanz et al. (2000) found that, initially, both B1
and B2 increase but B2 increases up to 17 GPa and then decreases (see also Fig. 5.2;
Tsuchiya et al., 2000). The linear behaviour and negative slope of the mode in the higher
pressure regime is indicative of proper ferroelastic behaviour. At pressures beyond
18 GPa, planer features (lamellae) develop. At ,39 GPa, B3 may become negative. The
calculations with pseudo-potentials give B3 ¼ 0 at 35 GPa (inter-atomic pair potentials at
22 GPa) (read also Section 5.1.6).
The elastic-phase transition is investigated employing soft-mode frequency, which
vanishes at transition.
Brillouin spectroscopy helps probe elastic waves propagating in a crystal, which
helps to determine the individual elastic modulii (see Zha et al., 1993). Gregoyanz et al.
(2000) report the first single-crystal Brillouin measurements (using Arþ-laser line of
l0 ¼ 514:5 nm and Ne and He as pressure medium) of a-quartz to above 20 GPa at
300 K. With increasing pressure, the frequencies of the slow transverse mode decrease
while those of longitudinal and fast transverse modes increase. The behaviour of single-
crystal elastic moduli, Cij ; with pressure shows that amorphization of a-quartz is driven by
Simple Oxides and Carbonates 849

an elastic instability. Their behaviour also shows that the transition to amorphous state is
ferroelastic in nature.
In general, for piezoelectric media, any calculation of elastic constants should take
into account the electrical field associated with the sound wave. Piezoelectricity modifies
the stress – strain relation as

Tij ¼ cijkl Skl 2 enij En ;

where Tij is the stress tensor, Skl the strain tensor, En the electric field and Cijkl the elastic
constants tensor, i.e., enij is the piezoelectric tensor.
All moduli ðcij Þ increase with pressure except c44 ; which steadily decreases. At high
pressures (39 GPa, e.g., beyond I ! II transition), c44 would vanish. An in situ high-
pressure study of cij and Born stability criteria of a material undergoing pressure-induced
amorphization, using a-quartz single crystal, revealed softening of c44 with pressure. This
indicates ferroelastic transition, which is triggered by a dynamic instability (Chaplot and
Sikka, 1993).

12.1.6.1. A new phase

Recently a new high-pressure phase of SiO2 has been prepared by Dubrovinsky
et al. (1997) by heating to .2000 K under the pressure of 68– 85 GPa. The high-P phase is
also identified using the first-principles total energy calculations. The structure (space
group Pnc2) is intermediate between a-PbO2 and ZrO2 structures. The X-ray diffraction
lines observed at 68 GPa and 85 GPa have been ascribed to a new SiO2 phase. On release
of P the Pnc2 phase reverts back to stishovite or CaCl2-type SiO2.

12.1.7. Amorphization experiments

A common mechanism seems to occur for solid-state amorphization of silicates in

static and shock high-pressure experiments, meteorite impact and deformation by tectonic
processes.
The formation, stability and physical properties of the amorphous forms of silica
are of special importance because of their technological applications and the role of glass
as archetypal non-crystalline material. Amorphous silica can be synthesized from a-quartz
by the application of high pressures (Hazen et al., 1989), comminution (Ray, 1923) and
shock compression (Gratz, 1984).
Formation of glass upon the grinding of quartz was noted in early experiments
(see review in Lin et al., 1975) but no studies on microstructures developed in quartz were
undertaken earlier than Kingma et al. (1993). In their experiments, Kingma et al. (1993)
compressed all together 10 polycrystalline and five single-crystal samples of a-quartz at
room temperature with Mao-Bell megabar-type diamond-anvil cells (detailed in Hemley
et al., 1988). Pressures were determined with the ruby-fluorescence method (Mao et al.,
1986). For the quasi-hydrostatic experiments, neon, which remains close to a hydrostatic
state over the pressure range of ,40 GPa (Bell and Mao, 1981), and argon, were used as
the pressure-transmitting media.
850 Chapter 12

Amorphization of a-quartz begins with the formation of crystallographically

controlled planar defects and is followed by growth of amorphous silicon dioxide at these
defect sites. This amorphous phase appears as glass lamellae, which are essentially formed
by solid-state transformation and not by melt quenching. The planar defects and
amorphous lamellae are seen to lie parallel to rational crystallographic planes, such as
{001}, {100}, {101}, {011};

and {111}
{102}, {201}
orientations (Kingma et al., 1993).
Similar alternating crystalline and non-crystalline bands have also been noted in anorthite
(CaAl2Si2O8) compressed to 22 GPa by William and Jeanloz (1989).
The stresses at crystalline –amorphous interfaces may have compressional, shear or
even tensile components. Additionally, even in hydrostatic medium, the formation of
multiple crystalline blocks will produce non-hydrostatic stress at the block interfaces. This
will in turn drive the breakdown of the parent crystalline form. The transformation-
associated displacements may produce heterogeneous stresses at the interfaces between the
crystalline and amorphous phases. This state of stress probably plays an important role in
the growth of the amorphous lamellae (Morris, 1992).
The amorphous state of quartz extends over a wide range of pressures and shear
stresses (Meade et al., 1992). Large shear stresses (1 –3 GPa) produce amorphous silica
at pressures and temperatures within the stability field of quartz (Dell’Angelo, 1991).
Shear stresses lower the transition pressure between the crystalline and amorphous state of
SiO2. Amorphization of geologic materials may also occur in subduction-zone
environments. The high solubility of amorphous phases may obscure their formation in
the geologic record.
Lamellar diaplectic glass (as well as melt glass) has been found in shock-
compressed SiO2 from both laboratory (Gratz, 1984) and meteorite-impacted samples
(Stoggler and Hornemann, 1972). From pseudotachylites in fault-gouges, non-crystalline
silicates have been reported (Yund et al., 1990). When a-quartz crystal is shock-loaded
above the Hugoniot elastic limit (12 GPa), it remains virtually strengthless on release from
high pressure. Presence of melt veins may be the principal cause for this loss of strength.

12.1.7.1. a ! c growth rate: magma viscosity

Growth rates of quartz from amorphous SiO2 were studied by Aziz et al. (1997) for
the pressure range up to 6 GPa. At 3 GPa, a sharp peak in crystal growth rate indicated a
minimum in viscosity. At .5 GPa, coesite nucleated.
The critical 3 GPa peak corresponds to a depth of ,100 km in the Earth. This
critical P may be related to the cut-off in subduction-related volcanism when descending
oceanic plates reach this depth. Magmas down to this depth may have a decreased viscosity
until the migration barrier disappears at 100 km and then suddenly becomes more viscous.

12.1.7.2. SiO2 glass

Synchrotron XRD. Meade et al. (1992) measured energy dispersive XRD spectra from
SiO2 glass in air and in the diamond-anvil cell at 8, 28 and 42 GPa (see Fig. 12.15a) at the
superconducting wiggler beam line of the National Synchrotron Light Source (X-17C).
The energy-dispersive diffraction spectra were measured at 5 – 7 scattering angles in the
range 6 , 2u , 338. For each measurement, the contribution from the diamond Compton
Simple Oxides and Carbonates 851

Figure 12.15. Effect of pressure on the local structure of SiO2 glass (Meade et al., 1992). (a) Average relation
function with increasing pressure obtained from the X-ray structure factor. (b) Bond length as a function of
pressure. The large circles correspond to the position of the peak (a) the small symbols are Si–O bond lengths
from crystalline SiO2 polymorphs (Meade et al., 1992).

scattering was subtracted and the intensities were normalized to the source spectrum of the
synchrotron. The individual measurements at each 2u were then combined and averaged to
form a single X-ray spectrum in the range 1 , Q , 16:5 Å21. The pressure was measured
with either the ruby fluorescence (Mao et al., 1978) or diamond Raman scales (Hanfland
and Syassen, 1985).

Medium-range order. The X-ray structure factor SðQÞ for SiO2 glass up to 42 GPa
shows large changes in the first sharp diffraction peak of SðQÞ with increasing
pressure (Fig. 12.15(a)). The largest changes in SðQÞ occur between 8 and 28 GPa.
852 Chapter 12

This, evidently, would reflect the pressure-induced structural transitions in the medium-
scale order (,4 – 10 Å) that is described by the first diffraction peak in SðQÞ (Elliot, 1991).
This first diffraction peak is a characteristic feature of the diffraction pattern for a large
class of network structural glasses which have medium-range order beyond nearest-
neighbour distances (,4.2 Å for SiO2 glass) (Elliot, 1991).
Above 28 GPa, the erosion of the first diffraction peak and the emergence of a
new peak at nearest-neighbour length scales (Q , 3:18 Å, d , 1:9 Å) are consistent with
a large decrease in the medium range ordering of the glass. The changes in bond length
kSi– Ol as a function of pressure is shown in Fig. 12.15b.

Raman spectrum. The Raman spectrum of SiO2 glass (Fig. 12.16 from Gillet, 1996) is
dominated at ambient conditions by a strong, polarized broad band centred near

Figure 12.16. Room pressure Raman spectra of various amorphous SiO2 samples. Intensity increases of the D2
band, attributed to the “ring-breathing” mode of three-membered siloxane rings, indicate increases in the
abundance of three-membered SiO4 rings. Spectra McMillan et al. (1994) for the glass (at 300 K) and melt
(at 1,950 K); McMillan et al. (1992) for a glass with a 20% densification. The intensities of all spectra have been
corrected for the frequency-and temperature-dependences (from Gillet, 1996, q 1996 Springer-Verlag).
Simple Oxides and Carbonates 853

440 cm21 (e.g., Sharma et al., 1981). This arises from oxygen symmetric stretching in
a line bisecting the Si –O – Si plane and its breadth indicates a broad distribution of
inter-tetrahedral angles. In addition, there are two sharp bands at 606 and 492 cm21,
called the “defect bands” D2 and D1 ; corresponding to the symmetric oxygen-
breathing vibrations of three- and four-membered siloxane rings, respectively (Sharma
et al., 1981).
On heating, the principal low-frequency band moves to a higher frequency
(490 cm21 at 2,000 K), indicating a decrease in the average O –Si –O angle (McMillan
et al., 1994; Ross et al., 1990). These changes are joined above Tg by a relative intensity
decrease of D2 and (to a lesser extent) D1 ; denoting the formation of three- and four-
membered SiO4 rings in the supercooled liquid. Such a change was also observed in SiO2
glasses quenched with various fictive temperatures (Geissberger and Galeener, 1983).
These observations are consistent with an endothermic formation enthalpy for these small
ring species.

Densification. The pressure-induced amorphized (PIA) phase of quartz has characteristics

identical to those of densified glass. This evidence indicates that the small rings have a
negative formation volume. Fused SiO2 undergoes an irreversible densification above
10 GPa. The changes seen in the Raman spectra suggest a decrease in the inter-tetrahedra
Si – O – Si bond angles (Hemley et al., 1986). The similarity of the Raman spectra of
pressure-amorphized quartz and that of densified silica suggests that the structure (in terms
of short- and medium-range order) of the former is closer to that of the densified silica
rather than to the fused silica.
At higher pressures, the signature of amorphization appears with broad bands
above 600 cm21. Shock-induced amorphization of cristobalite has been reported by Gratz
et al. (1993).

12.2. ZrO2 – SiO2: shocked

Zircon (ZrSiO4), commonly occurring as a primary accessory mineral in igneous

rocks (acidic), is also seen to occur in metamorphic and sedimentary rocks. It is the most
commonly used mineral in geologic age dating techniques, which measure the
uranium/thorium/lead nuclide ratios. Because of its resistance to chemical and physical
degradation, it is found useful to record the crustal processes. Use of a sensitive high-mass
resolution ion microprobe ((SHRIMP), allows one to measure the isotope ratios on areas as
small as 20 – 30 mm on a single zircon grain, thus providing age dates on separate zones.
Detrital zircons in a quartzite in Western Australia are dated at 4,100 –4,300 Ma, the oldest
terrestrial mineral ever found. The oldest zircons in the Solar System are found as rare
inclusions in meteorites and dated to be 4,560 Ma old (by SHRIMP). Zircon was seen to be
formed as a primary crystalline phase in Chernobyl lavas, the melted core of the reactor,
which exploded in the former Soviet Union.
Meteorite-impacted shocked zircons are seen to retain its U –Pb systematics. U – Pb
dating studies of shocked zircons in ejecta at the K – T boundary show a predominant age
854 Chapter 12

of 545 Ma. These zircon ages are in agreement with dates for shocked zircons from
Chicxulub crater and from Beloc, Haiti.
Zircons experimentally shocked to high pressures (59 GPa) showed no evidence for
fractionation or loss of lead. The zircons from Kokchetav massif (N. Kazakhstan), which
experienced an ultra-high-pressure metamorphic event at 530 Ma at depths of 125 km
(4 GPa and 900 – 1,0008C) to generate diamond-bearing gneisses, retained their original
isotopic composition.

12.3. TiO2

For detailed discussion on TiO2 and its polymorphic forms, refer to Section 3.2.2.

12.4. Simple monoxides

12.4.1. MgO, FeO, CoO, MnO and NiO

The lower mantle possibly contains simple oxides, which have been known to be
amenable to quantum mechanical calculations. Prototypal examples of a strong correlated
electron system can be seen in transition-metal (TM) monoxides. Although FeO, MnO,
NiO and CoO form simple NaCl (B1) structures, and seem like ordinary ionic insulators in
some ways, they are actually very complex and poorly understood through experiment or
by theory. FeO TM oxide is at the frontier of solid-state physics but is poorly understood.
FeO is the only monoxide that has been found to have a covalently and metallically bonded
NiAs structure (B8) at high temperature. This is in sharp contrast to CoO, SrO and BaO,
which transform from rock salt (B1) to CsCl (B2) structure.
Due to its high ionic conductivity and large elastic anisotropy at high-pressure
periclase (MgO), although a minor component (20%) of the Earth’s lower mantle, may
exert a strong influence on its physical and chemical properties.
The diffusion in MgO is influenced by planar defects such as grain boundaries and
point defects such as trivalent impurities have been investigated by Van Orman et al.
(2003). They determined the lattice and grain boundary diffusion coefficients for Mg, O
and Al at 2,273 K and up to 25 GPa. The results predict that in the deep lower mantle
when shear stress exceeds , 1 –10 MPa diffusion creep leads to dislocation creep. Across
the core – mantle boundary chemical exchange could have occurred with length scales of
, 1 – 10 km for lattice diffusion and 100 km for grain boundary diffusion. This may be
possible because the diffusion through periclase is fast enough.

12.4.1.1. MgO and CoO

For both MgO and CoO, the bonding, EOS and relative stabilities of the NaCl-type
(B1) and the CsCl-type (B2) structures have been calculated using both modified electron-
gas studies and first-principles band structure calculations (see also Fig. 12.21, for FeO).
Notwithstanding a large volume of studies, the questions regarding the degree of
non-stoichiometry, deviatoric stress, temperature on the phase boundaries and the EOS of
Simple Oxides and Carbonates 855

wüstite (FeO) and periclase (MgO) are not answered fully. The EOS for the cubic phase
has been studied by the static compression method (Liu and Liu, 1987; Fei and Mao, 1991)
and by shock-wave and ultrasonic experiments (Jackson et al., 1990).
High-pressure effects on MgO and CoO have been studied by Bukowinski (1985),
who calculated the lattice properties using the band structure and charge density obtained
from self-consistent augmented plane-wave calculations. A “muffin-tin” approximation to
the crystal potential in the self-consistent interactions and equal-sized atomic spheres with
radii equal to half the nearest neighbour distance have been employed.
The pressures involved in the B1 (NaCl-type) ! B2 (CsCl-type) phase transform-
ation for both MgO and CoO are calculated and compared with other theoretical
calculations and experimental values. The experimental transition pressure for CoO occurs
at ,60 GPa (Jeanloz et al., 1979) but, in the case of MgO, the ab initio pseudo-potential
calculations (Chang and Cohen, 1984) lead to the B1 ! B2 transformation at 10.50 GPa.
This transition pressure shows a much lower value of 20.5 GPa based on APW
calculations. The elastic-wave velocities of CoO in both B1 and B2 phases are
substantially lower than those inferred from seismology, thus limiting the presence of
CoO to small amounts in the lower mantle.

12.4.1.2. MnO and FeO

MnO is seen to undergo a first-order transition at ,90 GPa on shock compression
(Syono et al., 1998). Static compression studies reveal a distortion of the B1 phase at
,40 GPa followed by transitions at 90 and 120 GPa (Yagi et al., 1998). The 40 GPa
transition is similar to the rhombohedral distortion observed in FeO.
The higher pressure phase appears to be metallic with NiAs (B8) structure.
The intermediate pressure phase may be of mixed type, analogous to the polytype proposed
for FeO (Mazin et al., 1998). For NiO, no transition has been detected to 100 GPa (Shieh
et al., 2000).
Under high pressures, the TM-bearing oxides manifest high-spin ! low-spin
transition: e.g., MnO, CoO and Fe2O3 at 70 – 130 GPa and FeO at 25 –40 GPa pressures.
This aspect of magnetic collapse of FeO, MnO, CoO and NiO has been discussed in
Section 5.7.3.2. The magnetic moments M (in Bohr magnetons ðmB Þ) for anti-
ferromagnetic TM oxide in B1 structure as a function of pressure for these mono-oxides
are shown in Fig. 12.17.
FeO, MnO and CoO are regarded as wide-gap insulators but band-gap theory fails
for interpreting the properties of TM oxides. Consideration of the Kohn – Sham
eigenvalues as quasi-particle energies is unjust, although Kohn –Sham equations look
similar to the quasi-particle equations.
At room P; T; FeO shows cubic rock-salt (B1) structure. As temperature is lowered,
it passes TN at 200 K and becomes magnetically ordered and simultaneously assumes a
distorted rhombohedral structure. As pressure is increased, TN increases so that pressure
promotes the rhombohedral phase. (Note: Earlier it was thought that the rhombohedral
distortion was due entirely to magnetostriction.) At high temperatures, a conducting
hexagonal (NiAs structure) phase is found (Fei and Mao, 1994).
856 Chapter 12

A
MnO
4
FeO
CoO
3
NiO
M (µB)

0
0 100 200 300
Pressure (GPa)
Figure 12.17. Magnetic moments M (in Bohr magnetons, mB ) for antiferromagnetic transition metal oxide in the
B1 structure as a function of pressure. GGA results (open circles) are shown for MnO, FeO, CoO, and NiO; LDA
results (solid circles) are shown for FeO and CoO for comparison. Vertical lines denote transition pressures
(from Cohen et al., 1997).

LDA predicts that, above 100 GPa, the magnetic moment in FeO would collapse.
This is due to high-spin– low-spin transition, which can be either a continuous higher-order
phase transition or a first-order phase transition. With increasing pressure, the Neel
temperature (TN) initially increases and then decreases due to the decreasing local
moments. At ,100 GPa, the B1 phase appears to be insulating, not metallic.
Because the high-spin phase should be the high-entropy phase, the increasing
temperature should promote the high-spin magnetic phase rather than the low-spin non-
magnetic phase. Increasing temperature leads to disordering of the local moments.
Fei and Mao (1994) conducted experiments on FeO by using this high-temperature
cell, combined with in situ synchrotron XRD equipment. At room temperature, a transition
from the cubic (B1) to a rhombohedral phase was observed at 16 GPa under hydrostatic
condition (Yagi et al., 1985). The high P – T experiments in neon pressure showed a
transition boundary with a positive P – T slope with P ¼ 5:0 ^ 0:070T (P in GPa and T
in kelvin).
At about 70 GPa and 1,000 K, a triple point exists between the B1 (rock-salt)
structure, the rhombohedral phase with a distorted B1 and the B8 (NiAs) structure.
Tossell (1976) employed the MS-SCF-Xa scattered-wave cluster method to make
predictions regarding the electronic structures of iron-bearing oxide minerals at high
pressure. For the square planar FeO62 4 (seen in gillespite, BaFeSi4O10) and octahedral
FeO6 and FeO102 6 clusters at inter-atomic distances (representative of appropriate
minerals), the molecular-orbital energy levels in the valence region have been calculated.
The reduced inter-nuclear distances, affected by high pressure, were subsequently
considered in calculations.
In the square planar unit of FeO62
4 , with the oxygen ligand lying along the x- and
y-axes, the 3d orbitals of iron order as x2 2 y2 . xy . xz; yz . z2 : On reduction of the
Simple Oxides and Carbonates 857

inter-nuclear distances (2.03 ! 1.93 Å), the separation between dx2 2y2 and dz2 increases,
conforming to the inverse law ðR25 Þ of CFT. This primarily happens because of
high-spin ! low-spin transition, albeit this stops when R is reduced to the critical value
of 1.93 Å.
12.4.1.3. Normal and inverse NiAs structures
In condensed matter, the LDA method is a current effort and is important for
treatment of such systems as Mott insulators, e.g., FeO and MnO, in which electron
correlation causes the metal – insulator transition.
Calculations indicate a possible (unusual) polymorphism such as inverse NiAs
structure for FeO at extreme P; T condition. However, MnO above 90 GPa has been
discovered to show a normal NiAs structure (Kondo et al., 1998). (Note: The normal and
inverse NiAs-type structures are energetically close to each other.) CoO above 80 GPa
shows NaCl structure with the explanation of a magnetic collapse structure proposed by
Cohen et al. (1997).

12.4.1.4. FeO in D 00 zone

Because of its solid solution with magnesio-wüstite, FeO is believed to be abundant
in the lowermost mantle, i.e., at the D00 zone (Jeanloz and Lay, 1993). FeO may control the
electrical conductivity in this region, although FeO is a Mott insulator because of a strong
local correlation in electron motions. At low pressure Fe2þ ion is open shelled (while Mg2þ
is closed shelled). At high pressure, the ionic nature of Fe2þ changes and becomes more
covalent (Cohen et al., 1997). This change progresses more under high pressure and high
temperature when FeO undergoes phase transition to the metallic state, possibly assuming
a NiAs structure.

12.4.1.5. MgO
MgO is a simple oxide with NaCl structure stable up to megabar pressures and high
temperatures. Its elastic moduli have been measured experimentally under ambient
pressure up to 1,800 K. Chen et al. (1998) determined its elastic moduli at 8 GPa and
1,600 K. In the low-pressure region (, 8 GPa), the uniaxial stress ðt ¼ ðs3 2 s1 Þ GPa)
increases linearly with pressure as dt=dP < 0:5 – 0:85; at high pressure, it saturates to a
value of t 28.5 GPa (see Fig. 13 in Merkel et al., 2002). Elastic distortion of the lattice
suggests that non-hydrostatic stresses reach 8.5(^ 1) GPa and cause deformation of
50 –100%.
Stress hardening due to dislocation-density increase rises up to the pressure of
,10(^ 1) GPa, after which it stops or slows down drastically. Merkel et al. (2002) observe
a uniaxial stress component of ,8.6(^ 1) GPa at ,10(^ 1) GPa, implying that
sy $ 8:5ð^1Þ GPa at P , 10 GPa. This sy value possibly incorporates the effects of
stress hardening.
Radial X-ray diffractometry (Merkel et al., 2002) as well as Brillouin measurements
(Zha et al., 2000) on MgO indicate a decrease in anisotropy factor, A; from a value of 0.36
at ambient conditions to 20:05ð^0:10Þ at 45 GPa, leading to an almost elastically
isotropic material at deep-mantle pressures. However, with temperature, an increase of A is
858 Chapter 12

observed to be 0.77 at 1,800 K (Isaak et al., 1989). Thus, there is a competing effect
between pressure and temperature and ultimately, at the lower-mantle condition, single-
crystal elastic anisotropy of MgO would prevail.

(Mg,Fe)O: Fe on elastic anisotropy (A). In the lower mantle, Fe may be present in

(Mg,Fe)O magnesio-wustite phase to the extent of 10 –45 mol% FeO (e.g., Andrault et al.,
2001). The anisotropy factor A remains constant from the pure MgO phase to ,25% FeO
and then drops sharply with increasing iron content and becomes A , 0 (isotropic) at pure
wustite (FeO) composition. Thus, for magnesio-wustite, three competing factors seem
operative: pressure, temperature and Fe content. The value of A tends to decrease with
increasing pressure and Fe content, while it increases with temperature.

A of Mg-wustite. Single-crystal elastic anisotropy tends to decrease with pressure. At

,50 GPa, the material becomes almost elastically isotropic. The anisotropy factor A
decreases from ,0.115 at 670 km depth to ,0.050 at 1,050 km depth and then decreases
again to ,0.180 at 1,660 km depth.

Mg –O bonding. X-ray measurements of the electron distribution in MgO suggest a

structure that is closer to Mgþ þ –O2 rather than to Mgþ þ – O – , with the remaining
charge more or less uniformly distributed in the interstitial space. The difference in
charge density of MgO, generated with overlapping ions using a Gordon –Kim model
(PIB model) (Isaak et al., 1990) and calculated self-consistently using the linearized
argumented plane wave (LAPW) method (Mehl and Cohen, 1988), is shown in
Fig. 12.18. In Fig. 12.19, the band structure of MgO near zero pressure calculated by
using the LAPW method is compared with that determined from the potential generated
by the overlapping ion PIB charge density (Isaak et al., 1990; Mehl and Cohen, 1988).
MgO shows little or no covalent bonding and possibly its bonding is similar to that of
metal but without the large conductivity. The fact that the MgO valence bond width is
fairly large is an indication that the crystal field modifies appreciably the central
potential in the neighbourhood of the oxygen ion and that the overlap between the
Mg and O ions is significant.
Thus, MgO cannot be a good ionic solid. Indeed, its index of refraction is seen to
decrease as a function of density. This property is the opposite of what is manifested by
simple ionic compounds like alkali halides or simple oxides. The property therefore
violates the Cauchy relations.
The electronic structure of MgO cannot be adequately assessed at varying
pressures by use of such traditional parameters such as ionic radius, ionicity and ionic
polarizability. (Note: In simple oxides, the classical understanding of an anion is totally
inadequate.) The electronic structure of MgO has been extensively studied by optical
absorption (and reflection) spectroscopy, soft X-ray emission and absorption, and electron
spectroscopy.
Simple Oxides and Carbonates 859

Figure 12.18. Difference in charge density of MgO generated with overlapping ions using a Gordonkim
model (potential-induced breathing, or PIB model) (Isaak et al., 1990) and calculated self-consistently
using the linearized augmented plane wave (LAPW) method (Mehl et al., 1988). Contour interval:
0.005e/Bohr3 (see also Hemley and Cohen 1996) (Source: Stixrude et al., 1998, q 1998 Mineralogical Society
of America).

Thermal expansion. The stability of MgO over a wide range of temperature and pressure
conditions has allowed it to be used as an internal pressure standard for high P – T XRD
experiments (e.g., Utsumi et al., 1998).
Zhang (1999) carried out isobaric volume measurements for MgO at 2.6, 5.4 and
8.2 GPa at the temperature range 300– 1,073 K using a D1 A-type, large-volume apparatus
in conjunction with synchrotron X-ray powder diffraction.
At each pressure, the thermal expansion coefficient, aP;T ; is obtained from the
relation:
ð
VP;T ¼ VP;300 exp aP;T dT; ð12-2Þ

where VP;300 and VP;T are the unit-cell volumes at room and high temperatures,
respectively, for a given pressure.
Linear fit of the thermal expansion data over the experimental pressure range yields
the pressure derivative:

ðda=dPÞT ¼ 21:04ð8Þ £ 1026 GPa21 K21

860 Chapter 12

Figure 12.19. Band structure of MgO near zero pressure calculated using the LAPW method (lines) compared
with that determined from the potential generated by the overlapping ion PIB charge density (Isaak et al. 1990,
Mehl et al., 1988).

and the mean zero-pressure thermal expansion:

a0;T ¼ 4:09ð6Þ £ 1025 K21 :

The thermal and elastic properties of MgO obtained by different workers are set out
in Table 12.6.
TABLE 12.6
Thermal and elastic parameters of MgO

KT (GPa) aO;T a (K21 £ 1025) ðda=aPÞT ðdKT =dTÞP Reference

(GPa21 K21 £ 1026) (GPa K21)

155 (2)b 4.09 (5) 21.04 (8) 20.025 (3) Zhang (1999)
160 (2) 2 2 20.030 (3) Fei (1999)
153 (2) 4.21 (12)c 2 20.028 (4)c Utsumi et al. (1998)
161 (1) 2 2 20.029 (1) Isaak et al. (1989)
2 4.04 (6) 2 2 Suzuki (1975)
a
Representing average thermal expansion over the range 300 –1,073 K.
b
Obtained with K00 fixed at 4 in the data reduction.
c
Refitted using the data in the temperature range 300–1,073 K only for the sake of comparison.
Simple Oxides and Carbonates 861

12.4.1.6. Elasticity
MgO is a simple oxide with NaCl structure stable to high temperatures and megabar
pressures. The elastic moduli of MgO have been determined under both high pressure
(8 GPa) and temperature (1,600 K) (Chen et al., 1998). The elastic distortion of the lattice
suggests that non-hydrostatic stress reaches 8.5(^1) GPa and causes deformation of 50–
100%.
In the low-pressure (, 8 GPa) region, the uniaxial stress (t ¼ ðs3 2 s1 Þ GPa)
increases linearly with pressure as dt=dP ¼ 0:5 – 0:85 and at higher pressure it saturates to
a value of t , 8:5 GPa (see Fig. 13 in Merkel et al., 2002).
MgO with B1 structure is seen to be stable under static pressure up to at least
227 GPa (by DAC Study; Duffy et al., 1995) and up to 200 GPa under shear pressure
(shock-wave study; Vassiliou and Ahrens, 1981).
The temperature dependence of elastic properties has been studied experimentally
only at pressure ,3 GPa (e.g., Jackson and Niesler, 1982). The pressure dependence of
tðs3 2 s1 Þ for MgO is shown in Fig. 12.20.
Isaak et al. (1990) performed lattice dynamics on MgO as a function of lattice strain
beyond the normal quasi-harmonic approximation and studied the effects of temperature and
pressure on elasticity and the EOS. The results were then used to help understand the
increase in seismic parameter d ln Vs =d ln VP with depth in the Earth (Isaak et al., 1992).
By computing the stresses generated by small deformation of the equilibrium
primitive cell, Karki et al. (1997a) determined all three elastic constants, c11 ; c12 and c44 ;
which were used to calculate the bulk and shear moduli and thence the longitudinal and
shear velocities as a function of pressure.

Stress hardening. Stress hardening due to dislocation-density increase rises up to

the pressure of ,10ð^1Þ GPa, after which it stops or slows down drastically. Merkel
et al. (2000) observe a uniaxial stress component of ,8.5(^ 1) GPa at ,10(^ 1) GPa,
implying that sy $ 8:5 GPa at pressure .10 GPa. This sy value possibly incorporates
the effects of stress hardening.

Figure 12.20. The pressure dependence of t for MgO (Duffy et al., 1995a).
862 Chapter 12

12.4.1.7. B1 – B2 phase transition

In MgO, the B1 ! B2 phase transition is noted by different methods as

Method Authors Pressure (GPa)

First-principles pseudo-potential Chang and Cohen (1984) 1,050

LAPW Mehl and Cohen (1988) 510
Ab initio PIB Isaak et al. (1990) 486
Modified PIB Zhang and Bukowniski (1991) 580

To investigate the B1 – B2 phase transition in MgO, Karki et al. (1990) calculated

the enthalpy ðE þ PVÞ of the two-ion primitive cell of MgO in both B1 and B2 structures
as a function of pressure from 0 to 880 GPa. The intersection of the low curves occurs at
451 GPa, indicating the transition pressure, which has to be higher than other alkaline-
earth oxides because of the smaller ratio of cation to anion radii.
The transition pressure ðPt Þ; transition volume ðVt Þ and total energy difference
ðDEÞ for the B1 –B2 phase transition in MgO are shown in Table 12.7.

12.4.1.8. Elastic constants

For MgO, the MD simulation study obtains three isothermal elastic constants, c11 ;
c44 and c12 : Their temperature derivatives (Matsui et al., 2000) show an excellent
agreement with experiment (Isaak et al., 1989) and the calculated K and G values differ
from experiment by less that 1% (see Table 12.8).
Jamieson et al. (1982) have presented T – P – V based on shock-compression data of
MgO by Carter et al. (1971) combined with the Debye model. At low temperature, the
simulated T – P – V EOS listed in Table 12.8 (Matsui et al., 2000) agrees very well with the
EOS of Jamieson et al. (1982).

12.4.2. FeO at high P – T

FeO belongs to the group of highly correlated TM compounds and is expected

to show character between pure charge transfer and pure Mott –Hubbard insulators

TABLE 12.7
Transition pressure, transition volume and total energy difference for the B1–B2 phase transition in MgO

Pt (GPa) Vt ðB1Þ (Å3) Vt ðB2Þ (Å3) DE (eV)

Karki et al. (1997) 451 9.887 9.429 1.273

Cohen and Gordon (1976) 256
Chang and Cohen (1984) 1,050 6.94 6.61 1.506
Bukowinski (1985) 205
Causà et al. (1986) 220
Mehl and Cohen (1988) 510 9.03 8.61 1.27
Zhang and Bukowinski (1991) 580
Simple Oxides and Carbonates 863

TABLE 12.8
Observed and simulated isothermal elastic moduli of MgO and their temperature derivatives at 300 K and 0 GPa
(Matsui et al., 2000)

c11 c44 c12 KT G

Elastic constants (GPa)

Obsa 297 157 94 162 132
MD 294 157 94 161 131
Temperature derivatives of elastic constants (GPa/K)
Obsb 20.072 20.014 20.006 20.028 20.025
MD 20.070 20.014 20.007 20.028 20.024
a
From Isaak et al. (1989).
b
Adiabatic values from Chang and Barsch (1969), Spetzler (1970), Jackson and Niesler (1982) and Yoneda (1990).

(e.g., Bocquet et al., 1992). Accordingly, the band gap appears to be intermediate between
d – d and p – d gap (Saitoh et al., 1999).
The B1-phase is a paramagnetic insulator, which, at low T; undergoes transition to
an anti-ferromagnetic state. The RT transition to a rhombohedral phase occurs at 17 GPa
(Fig. 12.21) (e.g., Shu et al., 1998). The transition is driven by a soft c44 elastic constant
(Singh et al., 1998).
The NiAs hexagonal phase is related to the B1 and rhombohedral phases by a
martensitic-type sliding of the oxygen planes (Figs. 12.21 and 12.22). The inverse NiAs-
type structure is energetically close to normal NiAs type (e.g., Fang et al., 1999).
Theoretical study of the B1 to rhombohedral transition using DFT calculations
involving LDA shows an increase in charge between the iron atoms when the crystal
distorts; i.e., metal – metal bonding develops (Hemley and Cohen, 1996).
Elongation along [111] results in a “crystal-field” stabilization of a single spin-
down t2g level, whereas compression results in the stabilization of a doubly degenerate
pair, exactly the opposite electronic effect. FeO ðd 6 Þ has one electron in the spin-down
manifold, which means that crystal-field stabilization provided by [111] elongation is
favoured over compression. With further increase in pressure, a large decrease in magnetic
moment is predicted (Cohen et al., 1998). This seems to be the universal phenomena in
deep mantle (at CMB) involving TM-bearing minerals.
High P –T experiments demonstrate an enhanced solution of oxygen in molten iron
(Ringwood and Hibberson, 1991; Ito et al., 1995) and Ringwood (1977) suggested FeO as
a possible light-alloying component in the Earth’s core.
At ambient pressure, FeO assumes the cubic NaCl (B1) structure (Fig. 12.22),
which can be viewed as two inter-penetrating fcc sublattices of Fe and O. Some view the
structure as composed of close-packed planes stacked in the [111] direction, with the
stacking sequence AbCaAbCaBc where the lower-case letters stand for planes of Fe atoms
and the upper-case letters for planes of O atoms.
At low temperature, FeO displays a rhombohedral distortion (rhombohedral
angle a , 608), which increases with pressure (Yagi et al., 1985). The distortion
864 Chapter 12

Figure 12.21. P – T phase diagram of FeO (Fei and Mao, 1994) Data below 1,100 K are from resistance –heated
diamond-cell experiments. Three phases of FeO are identified by X-ray diffraction: NaCl (square),
rhombohedral (triangle) and B8 (solid circle). Data above 1,100 K are observations. The discontinuities in
density or electrical resistivity are obtained from shock-wave experiments (open circle; Jeanloz and
Ahrens, 1980) and laser-heated diamond cells (arrow, Knittle and Jeanloz, 1991). The dashed melting curve
is from a laser-heated diamond-cell study (Boehler, 1992a, q 1994 American Association for the Advancement
of Science).

occurs in a shortened Fe– Fe distance with increased metal –metal bonding under
pressure (Isaak et al., 1993). Direct electron hopping between Fe atoms contributes to
the binding energy of the crystal. The decrease in Fe – Fe distance and the increased
covalent character with increasing pressure cause an increasing rhombohedral
distortion leading to the B1 ! B8 (NiAs-type) transition. The B8 structure is derived
by distortion of the hexagonal closest-packed analogue of rock salt (B1) with one atom
located at [000] and the other at [2/3, 1/3, 1/4] under P63 =mmc symmetry. Fig. 12.22
depicts the structures of FeO phases and their bonding topologies. The XRD pattern of
Simple Oxides and Carbonates 865

Figure 12.22. Structure of FeO phases showing the similar bonding topologies.

FeO at 96 GPa and 800 K is presented in Table 12.9. Above 100 GPa, the hexagonal
NiAs(B8) structure is stable at RT but the rhombohedral phase is metastable (Fei and
Mao, 1994).
The maximum distortion observed by experiment is a < 53.88 (Yagi et al., 1985)
when a phase transition to a hexagonal structure of NiAs-type is observed (Fei and
Mao, 1994). The symmetry change is affected by a different stacking sequence of the
close-packed monolayers. The cell parameters of the B8 structure are a ¼ 2:574ð2Þ Å and
c ¼ 5:172ð4Þ Å.

TABLE 12.9
Observed and calculated XRD pattern of FeO at 96 GPa and 800 K (from Fei and Mao, 1994)

h k l dobs dcal a dobs 2 dcal lobs b

0 0 2 2.594 2.586 0.008 s

1 0 0 2.224 2.229 20.005 m
1 0 1 2.043 2.047 20.004 s
1 0 2 1.686 1.688 20.002 vs
0 0 4 1.293 1.293 0.000 mw
1 1 2 1.153 1.152 0.001 mw
a
Hexagonal unit cell: a ¼ 2:574ð2Þ Å and c ¼ 5.172/(4) Å, which gives c=a ¼ 2:01:
b
The relative intensities of the peaks are described as strong (s), medium (m), very strong (vs) and medium
weak (mw).
866 Chapter 12

12.4.2.1. Magnetic-phase transition

HS ! LS transition. By HP Mössbauer spectroscopy, HS to LS transitions in FeO have
been observed. Between 60 and 90 GPa, a diamagnetic (non-magnetic) LS state of Fe2þ is
manifested by a new quadrupole-split component in the Mössbauer spectrum. This
diamagnetic component increases with higher pressure at the expense of the magnetic
spectrum and, at 140 GPa, the Fe in FeO would be entirely in low-spin state. In the pressure
range 70– 90 GPa, the rhombohedral FeO is seen to transform to NiAs phase with LS state
(Pasternak et al., 1999).
The ground states of the HP phases of both FeO and MnO are anti-ferromagnetic
band insulators with the unusual inverse NiAs (iB8) structure (Fang et al., 1999).
The LS state of Fe2þ (d6 configuration) is characterized by a magnetic moment,
which shows a very clear spectral signature through the disappearance of the low-energy
satellite.

XES results. The spin state of Fe in FeO can be determined by X-ray spectroscopy using
the Fe Kb emission line. The high-spin state Fe is characterized by a main peak with an
energy of 7,058 eV and a satellite peak located at lower energy due to the 3p core –3d hole
exchange interaction in the emission final state. In low-spin Fe2þ(d6), the total magnetic
moment is zero and the resulting spectrum consists of a single narrow line.
Fe Kb X-ray emission spectra of FeO to 143 GPa and the XRD show it to be
rhombohedral (Badro et al., 1999). High-pressure Mössbauer spectra reveal a quadrupole
split component between 60 and 90 GPa and are assigned to a diamagnetic low-spin state
of Fe2þ (Pasternak et al., 1997).
High-resolution X-ray emission spectroscopic study of FeO under pressure up to
143 GPa (Badro et al., 1999) revealed the presence of a preserved magnetic state, at great
variance with the observation (i.e., yielding to a diamagnetic LS state) made by Mössbauer
spectroscopy. This observation is based on the spectral line shape of the Fe Kb emission
line. Up to the highest pressure, FeO remains a magnetic insulator.
There is an apparent discrepancy between the MS report and XES results. This is
primarily because of the difference in probing time scales. The lifetime of the 57Fe nuclear
resonance is on the order of 100 ns, while the Fe K shell core-hole lifetime is on the order
of a femtosecond. Hence, MS is sensitive to the magnetic order of the sample but cannot
distinguish a paramagnetic state from a diamagnetic state. On the other hand, XES exploits
the exchange interaction between the 3p-core hole and the 3d moment, which is insensitive
to the magnetic order of the sample but very sensitive to the local moment.

Néel temperature. The magnetic phase diagram of FeO constructed by Badro et al. (1999)
is shown in Fig. 12.23 (see caption). The figure shows an unusual closed-loop P – T anti-
ferromagnetic existence domain for FeO, and predicts a maximum Néel temperature TNmax
around 50 GPa. But the transition observed by MS at 90 GPa (RT ) is thought to arise
from a decrease in the Néel temperature ðTN Þ with pressure. At very high pressure, the anti-
ferromagnetic order ðTN ! 0Þ may be destroyed. At high P; T; the total magnetic moment
may vanish.
Simple Oxides and Carbonates 867

Figure 12.23. Magnetic phase diagram of FeO. The room pressure point is from McCammon (1992), the room
temperature at 15 GPa is from Nasu et al. (1986) and the very high pressure points are from Pasternak et al. (1997)
and Badro et al. (1999). A maximum Néel temperature TNmax is therefore expected in the 50 GPa, range, and is
denoted by the gray point. At very high pressure one can expect the destruction of the antiferromagnetic ðTN ! 0Þ;
and the phase diagram shows a closed-loop P – T antiferromagnetic domain of stability. At high pressure and
higher temperature, the total magnetic moment of iron may vanish (from Badro, et al., 1999).

Hemley et al. (2000) concluded that Fe2þ in FeO would be entirely in low-spin state
at 140 GPa and that there is a maximum TN (above 300 K) between 40 and 60 GPa, with
re-entrant behaviour back to the paramagnetic phase above 80 GPa (Badro et al., 1999).

Mott transition. A density discontinuity is seen to exist along the Hugoniot of FeO in the
shock-compression experiments (Yagi et al., 1988). The densification may be caused by a
phase transition from rock salt (B1) to CsCl (B2), from rock salt to NiAs structure, from a
spin-pairing transition or a Mott transition. The rock-salt (NaCl) structure of FeO under
pressure should experience metal – insulator transition before the high-spin– low-spin
transition. The density increase across the transition amounting to ,4% can be explained
by the c=a ratio. The NiAs structure shows the Fe – Fe distance to be shorter by ,8% than
that in the NaCl structure. The shorter Fe– Fe distance across the shared FeO6 octahedral
faces (in NiAs structure) can lead to metallic conductivity by electron delocalization. Loss
of local Fe moments can also enhance the metallization (Isaak et al., 1993). The NiAs
structure of FeO also allows the solubility of oxygen (in the form of FeO) in the molten
iron at high pressure.

12.4.2.2. NiAs phase

The phase diagram of FeO (Fig. 12.21) prepared by Fei and Mao (1994) represents
a density discontinuity at ,70 GPa (along the Hugoniot curve obtained from shock
compression experiment) allowing the transition from the rock-salt (B1) to the NiAs (B8)
structure.
The high-pressure phase of FeO is hexagonal with space group P63 =mmc and
cell parameters of a ¼ 2:574ð2Þ Å and c ¼ 5:172ð4Þ Å with one atom at [000] and the
868 Chapter 12

other at [2/31/314]. It can be considered as a distorted hexagonal closest-packed

analogue of rock salt. The NiAs structure of FeO consists of hexagonal close-packed
layers of O and Fe alternately stacked along the c-axis. The structure is closely derived
by variation in stacking along the body diagonal [111] direction (corresponding to the
c-axis in the NiAs structure) in the rock-salt structure (Jackson et al., 1990). The NiAs
structure can accommodate a greater degree of covalent and metallic bonding by
changing its c=a ratio whereby metal atoms are brought closer together and create more
distorted octahedrons. Under pressure, the c=a ratio of wüstite shows a dramatic change
at 20 GPa, when the phase changes from cubic to rhombohedral structure (Fig. 12.24
from Shu et al., 1998).

B8 and anti-B8 superlattice structure. B8 and anti-B8 superlattice structure FeO

assumes a NiAs-type structure, which is possibly a polytype of NiAs with anti-NiAs
structures with Fe on the Ni or As sites, respectively (Mazin et al., 1998). The NiAs part
is predicted to be metallic and the anti- NiAs part to be insulating. The NiAs polytypes
near the CMB would control the variations in the Earth’s magnetic field. However, the
experimental diffraction pattern represents a 5 : 5 layered sequence of B8 and anti-B8
structures. Such a polytype as this appears unique in the sense that anions and cations are
found substituting for each other. The stacking fault boundary between B8 and anti-B8 is
the rhombohedrally distorted B1 structure (Fig. 12.25). Thus, there can be continuous
transitions among B1, B8 and anti-B8 simply by changing the amount in each stacking
sequence. Thus, the three phases can join to a single coherent structure. B8 is metallic

Figure 12.24. The c=a ratio of wüstite as a function of pressure; solid square open circle, rhombohedral phase
(Shu et al. 1998, q 1998 E. Schweizerbart’sche Verlagsbuchhandlung, Stuttgart).
Simple Oxides and Carbonates 869

Figure 12.25. The 5 : 5 sequence of B8: and anti-B8 polytype of FeO. Normal and inverse NiAs superlattice
proposed for FeO (Cohen et al., 1998).

and anti-B8 is insulating (Mazin et al., 1998). This may be due to the fact that Fe– Fe
distances are very short in B8 while they are much longer in anti-B8. Again, experiments
show that the high-pressure hexagonal phase is metallic (Knittle and Jeanloz, 1986). A
stacking sequence of B8 and anti-B8 would have metallic and insulating layers, leading
to very anisotropic conductivity.

12.4.3. FexO

At high P – T; FexO has the NiAs-type hexagonal structure (B8) (Fei and Mao,
1994). The NiAs structure consists of hexagonally close-packed layers of oxygen and iron
alternately stacked along the c-axis. By changing the stacking of the NaCl-type cubic cell
along the body diagonal (111) direction, the c-axis of the NiAs structure is obtained.
The Fe– Fe distance in the NiAs structure is ,8% shorter than that in the NaCl structure.
This shortening may lead to metallization of FexO (Jackson et al., 1990), as manifested by
its reduced electrical resistivity (Knittle and Jeanloz, 1991).
At ambient pressure, FexO undergoes magnetic transition from paramagnetic to
anti-ferromagnetic state at 198 K. This is also confirmed by high-pressure 57Fe Mössbauer
spectroscopy (Zou et al., 1980). The anti-ferromagnetic phase has a rhombohedral cell
with the rhombohedral angle a ¼ 59:48 at 90 K. The high-pressure transition from cubic to
rhombohedral phase is an extension of the low-pressure transition caused by an increase in
Néel temperature with pressure (Yagi et al., 1985).
The effect of pressure on the composition of FexO in equilibrium with metallic Fe at
1,000 ^ 1008C is presented in Fig. 12.26. Three polymorphs in FexO have been identified as
the cubic NaCl-type (B1) and rhombohedral and hexagonal NiAs-type (B8) phases (Fei and
Mao, 1994). The latter two forms are not quenchable. The transition from cubic to
rhombohedral corresponds to a magnetic transition from paramagnetic to anti-ferromag-
netic state (Zou et al., 1980). The high-pressure transition of FexO to NiAs-type hexagonal
phase may manifest metallization (Knittle and Jeanloz, 1991; Fei and Mao, 1994). In the
870 Chapter 12

Figure 12.26. Effect of pressure on the composition of FexO in equilibrium with metallic Fe at 1,000 ^ 1008C.
The dotted line is from theoretical calculation (McCammon, 1993b; Fei and Mao, 1994).

NiAs structure, Ni is assigned to the position at the origin, resulting in nearest-neighbour

Ni –Ni separations (2.53 Å) that are quite a bit smaller than those for As – As (3.28 Å). The
high-pressure metallicity of FexO enhances the oxygen solubility (as FeO) in iron melt. This
explains how a light element like oxygen can be incorporated in the Earth’s core.
The hydrostatic compression data of FexO ð0:90 # x # 0:98Þ obtained by many
workers were plotted by Fei (1996). The least squares fits to the Birch – Munaghan EOS
yielded the best-fit results, KT0 ¼ 149 GPa and ðdKT0 =dPÞT ¼ 3:5: The KT0
and ðdKT0 =dPÞT values show a strong correlation. The results showed no systematic
effect of the degree of non-stoichiometry on the bulk modulus. The bulk moduli
determined by dynamic methods, including ultrasonic and shock-wave compression
(172 – 182 GPa) experiments seen to be significantly higher than those obtained by
static compression.
XRD experiment on the pressure dependence of the d-spacing of 200 peak of
cubic (corresponding to 102 in rhombohedral cell) indicated that the cubic phase is
more compressible than the rhombohedral phase. At transition pressure, the latter shows
higher density than that of the cubic phase. Upon heating, the cubic field enlarges and
it shows stability to higher pressures. XRD study revealed that the cubic to
rhombohedral transition at 16 GPa ðRTÞ is caused by distortion/elongation along a
body diagonal (111) of the NaCl-type cubic cell (Mao et al., 1993). This transition is
closely related to the observed paramagnetic to anti-ferromagnetic transition at low
temperature. As already stated, the rhombohedral distortion is very sensitive to the
sample stress environments.
The Néel temperature ðTN Þ of FexO decreases linearly as x increases from 0.90 to
0.95 but, from 0.95 to 0.98, there is a large discontinuity and TN increases with x
(McCammon, 1992). P
The Fe3þ= Fe ratio decreases with increasingP pressure to at least 18 GPa. The
g – 1 phase transition in Fe would favour a larger Fe3þ= Fe ratio at higher pressures and
lower temperatures but the transition boundary crosses the Earth’s geotherm only in the
Simple Oxides and Carbonates 871
P
deep mantle. In (Fe,Mg)O in equilibrium with metallic Fe, the maximum Fe3þ= Fe is
seen to be 0.05
P (McCammon, 1993). The electrical conductivity of (Fe,Mg)O increases
with Fe3þ= Fe—an observation which suggests that the lower-mantle conductivity is
dominated by (Fe,Mg)O (Wood and Nell, 1991; Li and Jeanloz, 1991).

12.4.3.1. Wüstite (Fe12xO)

Wüstite is a non-stoichiometric ferrous oxide with fcc structure of NaCl type.
Wüstite exhibits an irreversible change in lattice parameter above 12 – 15 GPa hydrostatic
pressure at room temperature (Jeanloz and Hazen, 1983). It is predominantly
paramagnetic. Below 198 K, it is anti-ferromagnetic but it becomes metallic at
,70 GPa. Charge imbalance due to cation deficiency is compensated by the presence of
Fe3þ, which, by neutron diffraction study has been proved to occur as small super-
paramagnetic islands of Fe3O4 in the wüstite structure. As stated earlier, wüstite may
be a principal phase in the lower mantle (Mao and Bell, 1977) or in the outer core
(Jeanloz et al., 1979).
In an experiment on Fe0.92O powder by Fei (1996), a transition from cubic NaCl
type (B1) to a rhombohedral phase was observed at 16 GPa (under neon pressure) and
300 K. The transition is marked by the splitting of 111, 220 and 222 XRD peaks of the
cubic phase. Indeed, the rhombohedral cell can be derived by stretching along the body
diagonal direction 111 of the cubic cell (Zou et al., 1980).
The rhombohedral phase was seen to transform back to the cubic phase at 21 GPa
and 400 K (Fig. 12.27). With increasing pressure at the same temperature (400 K), the
rhombohedral phase again appears from the cubic one. The transition from cubic to
rhombohedral phase is affected by the stress environment of the sample. The stress
environment changes with temperature. Reverse transitions were noted at 31 GPa/500 K

Figure 12.27. The cubic-rhombohedral transition determined by in situ X-ray diffraction measurements. The
arrows indicate the P – T path of the experiment. The solid and dashed lines represent phase boundaries under
hydrostatic and nonhydrostatic conditions, respectively. Experimental data (squares, cubic phase; and triangles,
rhombohedral phase) are shown for the Fe0.92O sample. Similar results were obtained when Fe0.98O was used as
the starting material (from Fei, 1996).
872 Chapter 12

and at 37 GPa/600 K when the cubic phase reappears. The transition boundary in Fig. 12.27
shows a positive P – T slope with the relation: PðGPaÞ ¼ 25:0 þ 0:070ð^0:003ÞTðKÞ
(Fei, 1996).
Wüstite is of great interest to condensed-matter scientists because of its electrical,
magnetic, structural and non-stoichiometric properties (e.g., Mao et al., 1996; Cohen
et al., 1997). At room P – T; wüstite crystallizes in the fcc B1-type cubic structure. At
16 GPa, a phase transition is reported (Yagi et al., 1986) when the 111, 220 and 311
XRD lines split, while the 200 line remains an unchanged singlet. At 18 GPa, a
transformation to four twin domains of a rhombohedral phase occurs when each of the
four body diagonals k111l of the original cubic crystal corresponds to a unique c-axis of
the rhombohedral phase. Shu et al. (1998) determined that the high-pressure phase is
unequivocally rhombohedral and the phase boundary exists reversibly at 18 GPa
(Fig. 12.21). Their study offers information about the nature of the transition, such as the
Landau like behaviour, the exponential change in axial ratio, the 4-fold twinning
mimicking the symmetry change and the development of internal strain in the absence
of external deviatoric stress.

12.4.3.2. Fe –FeO system

The effect of pressure on the composition of FexO in equilibrium with metallic Fe
at 1,000 ^ 1008C is presented in Fig. 12.26. The Fe –FexO equilibria at high P – T have
been modelled (Fei and Saxena, 1986, McCammon, 1993) but complexities in
quenching the thermal and pressure history make the interpretation of the results of
quenched experiments difficult. FexO is seen to exsolve iron or magnetite lamellae on
the scale of several tens of unit cells (Hazen and Jeanloz, 1984). FexO is unstable below
5708C and undergoes disproportionation as (Greenwood and Howe, 1972):

ð4z 2 3ÞFex O 2 ð4x 2 3ÞFex O þ ðz 2 xÞFe3 O4 ;

and 4FexO — (4z 2 3) Fe þ Fe3O4 (where z . x).

At high pressures (.10 GPa), the Fe –FexO phase is non-stoichiometric ðx – 1Þ:

The Fe content of FexO in equilibrium with metallic Fe increases with pressure until
reaching a plateau near Fe0.98O. The position of the Fe –FexO boundary at high pressures
and temperatures depends on the effects of Fe composition on the bulk modulus ðKT Þ of
FexO (McCammon and Liu, 1984). The KT values are estimated to range between ,150
and 154 GPa for x ¼ 0:90 – 0:95:

Melting point. The melting temperatures of iron and wustite at 16 GPa are found to be
1,945 ^ 208C and 1,875 ^ 208C, respectively, by Ringwood and Hibberson (1990). They
also found FeO solution to cause depression of the melting point of iron by 27.58 per wt%
FeO. 10% FeO is seen to cause a lowering of melting point of iron by 2758C (broadly
350 ^ 758C; Kato and Ringwood, 1989).
Simple Oxides and Carbonates 873

Thermodynamics. The thermodynamics of the Fe – O system under high pressure is

presumed to relate the large volume decrement for the magnetite to h-Fe3O4 transition
(e.g., Fabrichraya and Sundman, 1997).
The key thermodynamic parameters of wüstite, magnetite and hematite are shown
in Table 12.10.
Magnetite seems to be a stable phase in the Fe– O system only for pressures below
12 GPa; at ambient temperature h-Fe3O4 appears to be metastable.

Spin-transition depth. The HS ! LS PT curve of FeO crosses the mantle geotherm near
1,700 km depth. Majority of Fe2þ cations will be in low-spin state only when 1,700 km
depth is reached. However, it is assumed that FeO (LS) and FeO (HS) mix ideally.
A positive free energy of mixing will increase the transition depth still further. The high-
pressure phase change of FeO occurs near 70 GPa with a calculated Hugoniot temperature
of 1,200 K. At that temperature, the HS ! LS transition of FeO is predicted to occur
near 45 GPa.
High-spin FeO is believed to be in solid solution with MgO in the lower mantle. If
low-spin FeO is insoluble in MgO, then the HS ! LS transition may induce a breakdown
of (Fe,Mg)O:

ðMg12x ; Fex ÞO ! xFeOðLSÞ þ ð1 2 xÞMgO:

McCammon (1993) concluded that pressure reduces Fe3þ to Fe2þ in FexO wustite
equilibrated with Fe.

12.4.3.3. Fe –FeO 1 diluting elements: solid solution under P

The high-pressure changes in structure and bonding of FeO would allow it to form
solid solutions with FeS and Fe (hexagonal closest packing) under high P – T: This strongly
underlines the possibility of both oxygen and sulphur to be the major density diluting
elements present in the iron at the Earth’s core, which has been found to be 10% less dense
than pure iron at the pressure and temperature prevailing there (Birch, 1964). The other
suggested contending elements for bringing about this de-densification of the core are H,
C, Si and Mg (besides the stated oxygen and sulphur).

TABLE 12.10
Thermodynamic parameters for wüstite, magnetite and hematite (Fabrichaya and Sundman, 1997)

Compound Df Hm ð298Þ J/mol DSm ð298Þ J/(K mol) Df Gm ð198Þ J/mol Vm ð298Þ cm3/mol KTO GPa

FeO 2264.06a 60.45b 2243.52a 12.25c 150d

Fe3O4 21,115.37e 146.15e 21,012.53e 44.56 217
h-Fe3O4 – – 2945.79 41.89 202
Fe2O3 2826.23a 87.40a 2744.25a 30.30f 222 g

Note: All phases are treated as stoichiometric line compounds. KTO is assumed to equal 4.
a
Stølen and Grønvold (1996); b Stølen et al. (1996); c Haas and Hemingway (1992); d Fei (1996); e Grønvold et al.
(1993); f Morris et al. (1981); g Finger and Hazen (1980) and Staun-Olsen et al. (1991).
874 Chapter 12

12.5. Carbonates

The stability of carbonates under high pressure has aroused considerable

interest because they are the potential hosts for oxidized carbon in the Earth’s mantle
(e.g., Berg, 1986; Canil and Scarfe, 1990). In mantle-derived samples, the observed
occurrence of carbonates has lent credence to the concept that carbonates are the major
host for carbon in the Earth’s mantle (e.g., Biellmann et al., 1993a,b).
Of the natural carbonates, calcium and magnesium carbonates are the most
abundant (.90%). Studies on the CaO – MgO –CO2 system at high pressures and
temperatures have revealed that dolomite (CaMg(CO3)2) and huntite (CaMg3(CO3)4)
decompose to the two end-members aragonite (CaCO3) and magnesite (MgCO3) at
pressures greater than ,7 GPa and ,1,0008C (Liu and Lin, 1995). However, amongst all
carbonates, magnesite is probably the only stable carbonate to occur in the deep mantle
(Katsura et al., 1991; Redfern et al., 1993). Magnesite is also seen as an inclusion with
forsterite in a diamond from kimberlite (Wang et al., 1994).
Magnesite (MgCO3) and calcite (CaCO3) are isostructural and consist of layers of
Mg and/or Ca atoms that alternate layers of CO3 groups with the O atoms, corresponding
to an approximate hcp. In CO3, the three-oxygen coordination holds C atom at its central
position. The trigonal CO3 units lie parallel to the a – b-plane. Mg cations are octahedrally
coordinated to six O atoms, which, in turn, are collectively bonded to six different Mg
atoms and six different C atoms. Thus, the structure can be described, in terms of polyhedral
linkages, as consisting of corner-sharing octahedron and trigonal carbonate units.
Magnesite is stable even at a higher P than dolomite, since at high-P dolomite
breaks down in the presence of MgSiO3 to form magnesite and diopside. For these reasons,
a knowledge of the stability of magnesite at high P; T is important to understand the carbon
cycle on a global scale. The carbonates in equilibrium with silicates in peridotites,
carbonates and kimberlites change with increasing pressure (e.g., Eggler et al., 1976;
Wyllie and Huang, 1976).
The expected vibrational modes of calcite and magnesite with space group R3c
are
Gvib ¼ A1g ðRÞ þ 3A2g ðinactiveÞ þ 4Eg ðRÞ þ 2A1u ðinactiveÞ þ 4A2u ðIRÞ þ 5Eu ðIRÞ
Experimental studies have shown that aragonite, dolomite and magnesite are stable
under HP and VHP conditions. Field observations from various areas confirm that these
carbonate phases are present in the coesite and perhaps even in the diamond P – T fields.
Aragonite is seen to be stable at 28 GPa and ,1,0008C (Liu and Bassett, 1986) and at
36 GPa and ,1,8008C (Kraft et al., 1991) in the static HP experiments.

12.5.1. CaCO3 calcite ! aragonite polymorphism

The calcite and aragonite pair constitutes one of two most common polymorphous
mineral pairs in nature, the other being graphite and diamond. Aragonite is the high-
pressure polymorph of CaCO3, as diamond is for carbon.
There are more than 10 carbonates that exist with either calcite or
aragonite structure. The volumes of isostructural compounds are seen to be related to
Simple Oxides and Carbonates 875

Figure 12.28. The correlation between the ambient molar volume and the cation size of carbonates possessing
either the calcite or the aragonite structure. The values for calcite-type SrCO3 and BaCO3, shown by open
triangles, were estimated from high-temperature data of Chang (1965) by linear extrapolation. The molar volume
for the aragonite-type CdCO3 was obtained by Liu and Lin, (1997).

the cationic radii. Fig. 12.28 shows that correlation between cation size and molar volume
exists among carbonates having the calcite- or aragonite-type structure (Liu and Lin,
1997). The volume of the aragonite-type carbonates is generally 7 – 8% smaller than that of
the calcite-type carbonates. Since aragonite is the high-pressure polymorph of CaCO3,
other carbonates of calcite-like structure would transform to an orthorhombic aragonite-
like form under high pressure (or high temperature or both).
The compressiblities of carbonates with calcite and aragonite structure were
presented by Martens et al. (1982). They found that the bulk compressibilities of
carbonates (bMCO3) are related to their mean M– O bond length (dkM –Ol) by the linear
relationship:

ðbMCO3 ÞðGPa21 Þ ¼ 0:00097ðdkM2OlÞ3 A

3

which is ,60% greater than the compressibilities of the corresponding oxide polyhedra as
determined by Hazen and Prewitt (1977).
876 Chapter 12

However, the presence of non-hydrostatic stress may lead to erroneous

determinations of the EOS of a phase. This is more critical for carbonates, which show
large compressibilities.

12.5.1.1. Calcite, CaCO3

Calcite is the dominant carbonate acting as a buffer in the cycle of CO2 between the
atmosphere, hydrosphere and lithosphere. Besides aragonite and vaterite, the other forms
of CaCO3 are calcite II, III, IV, V and VI(?). At pressures above ,1.5 GPa, the
rhombohedral calcite structure transforms to a monoclinic polymorph via an elastic
instability (Singh and Kennedy, 1974; Barnett et al., 1985). However, a good number
of the high-pressure studies on calcite relate the transition of monoclinic calcite-II to
calcite-III of unknown structure.
By quenching CaCO3 from ,45 GPa and 1,7008C, calcite II can be obtained
(Biellmann et al., 1999). Calcite IV is calcite I with some disordering in the orientation of
the CO3 group (e.g., Redfern et al., 1989), while calcite V is a high-temperature
polymorph (Carlson, 1983). The stability field of calcite VI is unclear but it has been found
to form under pressures between 5.5 and 7.6 GPa in shock experiments (Vizgirda and
Ahrens, 1982).
In VHP areas, calcite phase occurs as an inversion product ofaragonite, which
inverts readily to calcite during decompression (e.g., Ernst, 1988). The stability of
calcite/aragonite under VHP conditions has been discussed by Wang and Liou (1993).
Gillet (1996) determined the DOS for calcite including lattice modes
(n , 400 cm21) and internal modes of CO22 3 groups. He showed that a decreases linearly
with pressure (between 105 Pa and 3 GPa) with temperatures as shown below:

da
At 300 K : ¼ 21:46 £ 1026 K21 =GPa
dP T

da
At 700 K : ¼ 21:94 £ 1026 K21 =GPa
dP T

da
At 1200 K : ¼ 22:43 £ 1026 K21 =GPa
dP T

The isothermal bulk modulus decreases linearly with temperature (2 0.011 GPa/K),
with no dependence upon pressure between 105 and 3 GPa and K 0 ð¼ dK=dPÞT decreases
from 5.37 to 4.72 between 300 and 700 K.

Calcite I ! II transition. For calcite, the unit-cell parameters have been measured
between room pressure and 1.5 GPa and the isothermal bulk modulus has been obtained
from the Birch– Murnaghan EOS. The cell parameters show a monotonic decrease
with pressure up to 1.435 GPa. Above this pressure, a transition to monoclinic calcite II
Simple Oxides and Carbonates 877

is observed when it twins and experiences a spontaneous strain with shear 113 :
This transition is ferroelastic in character, with softening of some elastic constants but
this transition is strongly of first order.
The transition of calcite-I to calcite-II results from two displacements. First, an 118
rotation in opposite directions of the adjacent CO3 groups along the c-axis and second,
small anti-parallel displacements of adjacent Ca ions. The CO3 group rotational
displacements are thought to drive the transition, just as they drive the high-T orientional
disordering transition in calcite-I (Redfern et al., 1989).

F-point brillouin zone. At ambient P – T; calcite displays strong neutron inelastic

scattering at the F-point of the Brilloiun zone, which should be a signature of the
high-pressure monoclinic calcite-II phase (Dove et al., 1992). The critical phonons at
the F-point of the Brillouin zone correspond to a distortion pattern akin to that of the
calcite-II polymorph.
In the transition from I ! II, the intensity of inelastic scattering becomes diffuse,
when monoclinic superlattice reflections appear and the critical F-point fluctuations
possibly gain importance. The diffuse intensity is associated with inelastic scattering at a
point in the reciprocal space ðX; Y; ZÞ with X ¼ 22:5; Y ¼ 0; Z ¼ 2 (; F-point in the
Brillouin zone), which becomes the zone centre (of Bragg reflection) in the monoclinic
structure.

12.5.1.2. Compressibility and bulk modulus

The cell parameters of carbonates show a smooth continuous decrease with
increasing pressure and show no evidence for a phase transition or change in compression
behaviour. The mean linear compressibilities of the axes are

a ¼ 1:88ð2Þ £ 1023 GPa21

c ¼ 4:22ð3Þ £ 1023 GPa21

Thus, c-axis is twice as compressible as the a-axis. The axial ratio, c=a; decreases with
pressure from 3.241 at ambient pressure to 3.188 at 6.88 GPa.
In Ca-bearing rhombohedral carbonates, dolomite and ankerite, c is approxi-
mately 3 times as compressible as a (Ross and Reeder, 1992; Martinez et al., 1996).
The axial ratio, c=a; decreases more rapidly with pressure in Ca-bearing
rhombohedral carbonates. The changes in a=a0 ; c=c0 and volume ratio ðV=V0 Þ of
magnesite as a function of pressure are shown in Fig. 12.29. The CO3 are
incompressible rigid units while the MgO6 octahedra show significant compression.
The linear compressibility of Mg – O bonds in magnesite is 3.0 £ 1023 GPa21.
The variation with pressure of Mg –O and C – O bond lengths in magnesite is shown
in Fig. 12.30.
Ross and Reeder (1992) found that MgO6 in dolomite is more compressible than
would be expected from high-pressure structural studies of Mg-bearing oxides and
silicates (e.g., Hazen and Finger, 1982). Redfern et al. (1993) suggested that KP
878 Chapter 12

Figure 12.29. The variation of a=a0 ; c=c0 ; and V=V0 of magnesite as a function of pressure with linear regressions
of the data are shown (Ross, 1997, q 1997 Mineralogical Society of America).

of MgO6 in magnesite should be higher than that of dolomite and closer to the values of
periclase (MgO) and monticellite ðKP ¼ 150 GPa; Sharp et al., 1987). This may be
because, in NaCl structure (e.g., in MgO), the metal – metal interactions may play a more
significant role than in the structures of rhombohedral carbonates, resulting in less
compressible octahedra.
Again, the volume compressibility is noted as bv ¼ 2bx þ bz ; as expected for a
hexagonal crystal. Singh and Kennedy (1974) obtained a value of bv ¼ 14:07 £
1023 GPa21.
The bulk modulus of calcite as determined by different workers is presented in
Table 12.11.

12.5.1.3. Oxy-anion –cation packing

Megaw (1973) considered that the array of oxygen atoms in calcite may be roughly
considered as corresponding to hexagonal packing of oxygens with the relationship 4a ¼
2c: The ratio ðtÞ ¼ 4a=2c; therefore, is a measure of the distortion of the oxy-anion –cation
Simple Oxides and Carbonates 879

Figure 12.30. Variation of (a) Mg –O bond lengths and (b) C–O bond lengths with pressure in magnesite (Ross,
1997, q 1997 Mineralogical Society of America).

packing away from the ideal hexagonal close-packed oxygen arrangement. At RT; the
values of t of different carbonates are noted as

Carbonates t values

Calcite, CaCO3 0.827

Dolomite, (Mg, Ca)(CO3)2 0.849
Magnesite, MgCO3 0.873

Thus, these carbonates are more expanded along c than along a compared with the
packing of spherical ions.
Evidently, for carbonates, the ratio tð¼ 4a=2cÞ of the compressibilities along a and c
increases with pressure towards the values of the Mg-rich end-member (Wunder, 1998).
The change in c=a ratio reflects the relative incompressibility of the C – O bonds compared
with the more compressible Ca – O bonds. The volume compressibilities follow the
relationship such that K0 V0 values remain constant (Anderson and Anderson, 1970).
880 Chapter 12

TABLE 12.11
Bulk modulus of calcite

Authors K0 (GPa) K00 Method

Bridgman (1925) 73.15 4.17 Piston cylinder

Peselnick and Robie (1963) 71.7 – Acoustic
Dandekar (1968) 73.3 – Acoustic
Humbert and Plicque (1972) 72.5 – Acoustic
Singh and Kennedy (1974) 71.1 4.15 Piston cylinder
Fiquet et al. (1994) 69.5 ^ 2.1 4 Powder ED XRD
Redfern and Angel (1999) 73.46 ^ 0.27 4 Single-crystal XRD

12.5.2. Mg-carbonates

Experimental studies also indicate that magnesite and dolomite should be the
principal carbonates under mantle conditions and may carry carbon in subducting plates
into the deep upper mantle (e.g., Gillet, 1993; Redfern et al., 1993). Magnesite is reported
from mantle-derived kimberlites and ultra-high-pressure metamorphics and eclogites (e.g.,
Zhang and Liou, 1994; Zhang et al., 1995).
Magnesite is seen as inclusions in garnet and clinopyroxene in both very high-
pressure (VHP) ultramafic and eclogites from the Dabie-Sulu terrane (e.g., Liou and
Zhang, 1996). Most magnesites from VHP ultra-mafics may have formed at P approaching
5 –6 GPa and 800 –9008C (Zhang and Liou, 1994).
Above 3.5 GPa, magnesite is seen to be the only stable carbonate in peridotite (Brey
et al., 1983) and it remains stable in the presence of (Mg, Fe)SiO3 perovskite and
magnesio-wüstite at 50 GPa in the temperature range 1,500 – 2,500 K.
Hence, until it was under extreme pressure, MgCO3 would not adopt the
aragonite structure. The calcite ! aragonite-type transition may not occur in MgCO3 at
pressures prevailing in the upper mantle. No phase transition in MgCO3 has been
observed up to 55 GPa (/,1,3008C; Katsura et al., 1991). However, the calcite !
aragonite-type transition may occur in MgCO3 at the P – T conditions of the lower
mantle (Liu and Lin, 1997). This high-pressure polymorphic transition of MgCO3 makes
magnesite the most likely stable carbonates phase in the Earth’s deep mantle (Redfern
et al., 1993).
Magnesite is known to be a stable phase up to a pressure of 50 GPa (in the presence
of silicates) (e.g., Biellmann et al., 1993) and retains its R3c
structure to 55 GPa
(hydrostatic) (e.g., Gillet, 1993).
Using a second-order Birch –Murnaghan EOS ðK00 ¼ 4Þ; the zero-pressure bulk
modulus of MgCO3 was obtained by different workers as shown in Table 12.12.
The adiabatic bulk modulus ðKS Þ values were determined as

KS (GPa) Authors

112 Christensen (1972)

113.8 Humbert and Plicque (1972)
Simple Oxides and Carbonates 881

TABLE 12.12
Bulk modulus of MgCO3 obtained by different workers

K0 (GPa) Authors Methods/Expt. conditions

142(9) Redfern et al. (1993) Multiple grains; methanol –ethanol (4 : 1);

DAC Decker NaCl EOS
138(3) Fiquest et al. (1994) Powder; silicon oil; Ruby fl.
111(1) Ross (1997) Single crystal; methanol–ethanol (4 : 1);
up to 7 GPa
103(1) Zhang et al. (1997) NaCl EOS; SAMS 85 cubic anvil
115(1) Fiquest and Reynard (1999) Third-order Eulerian EOS
108(2) Fiquest and Reynard (1999) 72 GPa, K0 ¼ 4:6; V0 ¼ 279ð1Þ Å3
110 Christensen (1972) Ultrasonic measurement; K0 ¼ 4:5ð1Þ

Note: Generally, values of K00 significantly lower than 4 have been inferred to be around 2.5 in all experiments
carried out earlier than 1999.

The EOS of magnesite derived by different workers are presented in Table 12.13.
Using single-crystal XRD, the unit-cell parameters of magnesite (MgCO3) have
been measured by Ross (1997) under pressures between 0 and 7 GPa. The main structural
change observed with increasing P is compression of the MgO octahedra while the
carbonate group (CO2 3 ) remained invariant through this pressure range. Magnesite is seen
to be stable in the presence of MgO under the P – T conditions of the lower mantle (Redfern
et al., 1993). In the lower mantle near 100 GPa (or less when iron carbonate is present),
MgCO3 decomposes to periclase (MgO) and CO2. The decarbonation of magnesite
through the reactions shown below has been experimentally determined in a number of
studies up to 4 GPa (Zhang et al., 1997).

MgCO3 ¼ MgO þ CO2

ðmagnesiteÞ ðpericlaseÞ

MgCO3 þ MgSiO3 ¼ Mg2 SiO4 þ CO2

ðmagnesiteÞ ðenstatiteÞ ðforsteriteÞ

TABLE 12.13
Equation of state parameters of magnesite (from Fiquet and Reynard, 1999)

Method K0 (GPa) K00 Reference

Acoustic 112 – Christensen (1972)

Acoustic 113.8 – Humbert and Plicque (1972)
X-ray: powder 142 (9) 4 Redfern et al. (1993b)
151 (7) 2.5
X-ray: powder 138 (3) 4 Fiquet et al. (1994)
156 (4) 2.5 (2)
X-ray: powder 103 (1) 4 Zhang et al. (1996b)
108 (8) 2.3
X-ray: single crystal 111 (1) 4 Ross (1997)
117 (3) 2.3 (7)
882 Chapter 12

Figure 12.31. Phase equilibrium calculations (shown by solid curves) and comparisons with the experimental
determinations for the reactions a magnesite ¼ periclase þ CO2 (after Zhang et al., 1996b).

The experimental results of Zhang et al. (1996b) are depicted in Fig. 12.31 (read the
caption). The above relations would determine the fate of carbonates buried deep in the
mantle in subduction zones and for the global carbon cycle.

structure at simultaneous
Gillet et al. (1991) showed that magnesite retains its R3c
high pressure (26 GPa) and high temperature (1,200 ^ 200 K). Zhang et al. (1996)
measured the molar volume of magnesite up to 9 GPa and 1,300 K and derived the EOS
parameters and the P; T derivative of KT :
Under pressure (RT), the XRD peaks of magnesite become broader and
asymmetrical. On heating, however, relaxation of the differential stress occurs and the
peaks become sharper. Full hydrostaticity in stress in magnesite is achieved by heating
above 873 K (Note: For NaCl, the differential stress becomes hydrostatic above 573 K.)
The P – V – T measurements of magnesite were carried out by Zhang et al. (1996) up to
8.6 GPa and 1,286 K using a DIA-type cubic anvil apparatus interfaced with synchrotron
XRD. Precise volumes were obtained at both room and high temperatures by the use of
data collected above 873 K on heating and during the entire cooling cycle. For the volume
data at room temperature, they obtained K0 ¼ 103ð1Þ GPa (see Table 12.13).

12.5.2.1. Dolomite stability at depths

The discovery of coesite inclusions in dolomite from VHP rocks from the Dabie
Mountains demonstates that continental sediments were subducted to pressures at depths
Simple Oxides and Carbonates 883

.100 km and that dolomite is a stable phase at mantle depths. Also, in these rocks
inclusions of calcite pseudomorphs after aragonite have been reported (e.g., Zhang and
Liou, 1996). The estimated P – T conditions for the peak metamorphism are 7608C and
.2.8 GPa, with XCO2 between 0.01 and 0.1.

12.6. Other carbonates

The calcite-type BaCO3 and SrCO3 do not exist at ambient conditions. For these,
the cation coordination number (CN) is 6 and for aragonite it is 9 and 9 is regarded as the
CN for SrCO3, BaCO3 and PbCO3 as well. Besides carbonates, some nitrate (e.g., KNO3)
and borates LnBO3 (Ln ¼ La, Pr and Nd) also possess the structure of aragonite (Liu and
Bassett, 1986).
The pressures required for the various aragonite-type carbonates to adopt the same
high-pressure phase are in the following order (Liu and Liu, 1997):
CdCO3 . CaCO3 . YbCO3 . EuCO3 < SrCO3 < SmCO3 . PbCO3 . BaCO3 :
The conditions of 4 GPa and 1,0008C are close to the triple point of disordered
calcite/aragonite/BaCO3 II.
CdCO3 (tavite) is known to have a calcite structure at pressures below 16 GPa.
The size of Cd2þ ion is only ,5% smaller than that of Ca2þ ion in a calcite-type structure
while Mn2þ is ,17% and Mg2þ ion is ,28% smaller than Ca2þ ion. At a pressure
of ,17 GPa (/,1,0008C), a transition of CdCO3 to aragonite structure was recovered by
Liu and Lin (1997) but this transition was not found in MnCO3 (rhodocrosite) in
the experiments up to 28 GPa (/,1,0008C).

12.7. CaO –MgO – SiO2 –CO2 system

High-pressure phase transformation of carbonates in the system CaO –MgO –

SiO2 –CO2 was studied by Liu et al. (1995). It was seen that at ,4– 26 GPa and ,1,0008C,
dolomite decomposes into aragonite þ magnesite in the P range 6– 7 GPa at ,1,0008C.
Huntite transforms to a new phase at P , 4 GPa and then decomposes to
dolomite þ magnesite at P . 6 GPa. The new phase with the huntite composition has an
orthorhombic cell in ambient conditions. The only silicates –spurrite and tilleyite –
decompose into their component silicates and carbonates at P , 4 GPa at ,1,0008C.
The results indicate that, at P . ,7 GPa and 1,0008C, aragonite and magnesite are
the only two carbonates which are stable in the entire CaO –MgO –SiO2 – CO2 system.

12.7.1. CaO –MgO – SiO2 –CO2 – H2O system: XCO2

In the CaO – MgO –SiO2 – CO2 – H2O system, the stability fields of Mg-silicate þ
Di þ En strongly depend on the XCO2 of the fluid phase. The occurrence of
dolomite/magnesite is highly dependent on XCO2. For magnesite-bearing ultramafic
884 Chapter 12

rocks, thermobarometry yields a high P of 4.5– 6.5 GPa at about 8008C (Zhang et al.,
1995). A petrogenetic grid for VHP metamorphism in this system has been proposed by
Ogasaware et al. (1998).
For the Dabie Sulu (China) eclogite at the estimated P – T conditions of peak
metamorphism (.2.8 GPa and 7608C), the stable co-existence of coesite (coe) þ dolomite
(do) þ omphacite (om) (XDi ¼ 0.5) requires 0:01 , XCO2 , 0:1: For this range of P – T –
XCO2 conditions, magnesite is stable with coesite. Wang and Liou (1993) estimated that, at
P ¼ 0:3 GPa and T ¼ 630 – 7608C, the XCO2 for the reaction, do þ 2coe ¼ di þ 2CO2, is
very low (,0.03).
 885

Chapter 13
Hydrous Minerals

13.1. Water in primary minerals

13.1.1. Introduction

The upper mantle is dominated by olivine, pyroxene and garnet phases but
substantial amounts of plagioclase felspar, kaersutitic amphibole, phlogopite mica and
many other as accessory minerals may exist. Many of these phases are volatile-rich,
providing host phases for H2O, CO2 and F-, Cl- and S-bearing species degassing from deep
within the Earth or returning via subduction processes (e.g., Thompson, 1992; Gillet, 1993;
Carroll and Holloway, 1994). In the upper mantle, the water contents in the reservoir
magmas, such as depleted, undepleted and enriched reservoir magmas, are estimated to
range from 100 to 500 ppm (Bell and Rossman, 1992). The water contents of mantle
xenoliths can be used to estimate the water contents of the mantle, presuming that no loss
of hydrogen occurred during decompression.
Presently, the geochemists and geophysicists are attempting to answer the
questions related to water in primary minerals at depths within the Earth. The questions
mainly arising are: (a) how much water can be stored at great depths?, (b) what are the
plausible sites for storing water?, (c) how does water affect the chemical and physical
(especially elastic) properties of minerals and rocks? and (d) can water be seismically
detected?
The sites for water at shallower depths are diverse and subtle. In the transition zone,
the most probable reservoirs for water are b and g polymorphs of olivine, which can hold
up to 3 wt% of water (Kohlstedt et al., 1996).
For different lattice defects, OH can occupy several types of sites in olivine
structure (Bai and Kohlstedt, 1993). OH can also be incorporated in olivine as a monolayer
of humite, as observed from Blue Park kimberlites (e.g., Kitamura et al., 1987). Point
defects and humite-like planar defects are common sites for OH (Libowitzsky and Beran,
1995).
The incorporation of water affects the elasticity of olivine to a lesser extent than
that of garnet or b-phase because the mechanisms of hydrogen incorporation in these
structures are different. In garnet, the O4H4 groups substitute for SiO4 tetrahedra, thereby
weakening the rigid framework (Knittle et al., 1992; O’Neill et al., 1993).
The compressibility of O4H4 is much higher compared with the incompressible
886 Chapter 13

SiO4 tetrahedra. At full saturation (3.3 wt% H2O), OH occupies 1/8 of all available anion
sites and 1/8 of the cation positions remain vacant.
Hydrous minerals such as micas, amphiboles, humites and high-pressure
synthesized hydrominerals are generally regarded as possible water reservoirs in the
mantle. But these minerals under mantle conditions require abundant incompatible
elements such as K, F, Ti, etc., and/or high water fugacity at high temperatures. However,
the mantle is generally poor in incompatible elements and also in water (0.01 –0.1 wt%
H2O), although these may be locally enriched in the mantle or in cold regions such as
subducting slabs (Bell, 1992; Thompson, 1992).
Hydrous species, even in trace amounts in mantle, can affect the melting and
rheology of mantle minerals. Hydrous minerals such as phlogopite mica and kaersutitic
amphibole can be stable under the P – T conditions of the upper mantle.
All the experimental results indicate that, away from regions of partial melting, the
upper mantle is roughly 10– 20% saturated with respect to water content. This observation,
together with the water-solubility data for stishovite and perovskite (Lu et al., 1994; Meade
et al., 1994), would lead one to conclude that the mantle has the capacity to store several
orders more water than is held in the hydrosphere. However, it should be emphasized here
that, for storing water, hydrous fluid phases need not be invoked — the water dissolved in
the anhydrous phases such as olivine, pyroxene (and garnet) in the mantle were adequate to
form the hydrosphere by magmatic outgassing.
Generally, garnet – peridotite olivines show a higher hydrogen content than spinel
peridotite olivines (e.g., Bell and Rossman, 1992). This may be due to difference of the
geotherm between localities such as continents and oceanic regions. The highest H content
at high-P and low-T regions would indicate the importance of olivine in subducting slabs
for a return circuit of water into the mantle.
“Nominally anhydrous” silicates such as olivines, pyroxenes and garnets contain
structural water, which rarely exceeds 0.1% by weight (e.g., Kohlstedt et al., 1996;
Rossman, 1996). An exception is grossular, Ca3Al2Si3O12, which, on hydration to a
composition of Ca3Al2(O4H4)3, can contain up to 28 wt% water (e.g., O’Neill et al., 1993).
The nominally anhydrous minerals contain hydrogen only at ppm level (Martin and
Donnay, 1972) but, considering the quantum of mass and volume of the nominally
anhydrous minerals in the mantle, the quantity of hydrogen or water would be enormously
large, especially under mantle high P, T conditions where the solubilities of OH (or water)
in these minerals are very high. Experiments have shown that, in the pressure range of
5 –10 GPa, olivine structure can accommodate 500 –1,000 ppm by weight of water.
Therefore, the total water content of the upper mantle could be incorporated in this mineral
alone. Even in the pyroxene structure, the solubility of water can be still higher (Bell and
Rossman, 1992). Hence, these minerals are often the most promising storage sites of
hydrogen in the mantle (Kurosawa et al., 1997). The bonds of hydrogen species have a
marked influence on mineral deformation (e.g., Mackwell et al., 1985; Karato et al., 1986)
and on electrical conductivity in the mantle (Karato, 1990) (see Section 13.2.2).
At P . 12 GPa, the nominally anhydrous phases such as clinopyroxene, garnet,
olivine and b-(Mg,Fe)2SiO4 dissolve significant amounts of H2O (Bell and Rossmann,
1992). Measured OH contents for natural olivines, pyroxenes and garnets, which constitute
Hydrous Minerals 887

.90% of peridotites sampled at the surface, range from 100 to 1,500 H/106 Si (e.g., Bai
and Kohlstedt, 1993). OH concentration of Ca-rich garnets may become high. Because
these phases are at least an order of magnitude more abundant than nominally hydrous
phases, the nominally anhydrous phases are expected to be the main carriers of H2O at
depth in excess of 300 km (Schmidt et al., 1998).
Water, along with volatiles composing aqueous fluids, play an important role in
eclogitization and retrograde amphibolitization. Fluids participate in many metamorphic
reactions and they act as catalysts (metasomatism) and control rheology. In very high-
pressure (VHP) regimes, deeply subducted packages of protoliths contain fluids in hydrous
and carbonate phases.
Besides amphiboles and micas, a number of well-known metamorphic phases have
been investigated under pressure because of their possible presence in the upper-mantle
region such as: talc, zoisite, lawsonite, chlorite, topaz and serpentine. Several of these
phases have stability fields that extend to depths of 150– 300 km. In subducted pelitic
sediments, topaz may appear along with lawsonite, zoisite and chlorite (Domanik and
Holloway, 1996). The hydrous Mg-silicates in the MgO – SiO2 – H2O ternary system are
shown in Fig. 13.1.
Lawsonite, ideally CaAl2Si2O7(OH)2·H2O, is commonly seen in blue-schist facies
rocks and has a large stability field within hydrous basalt. It also shows an unusually high
content (11 wt%) of water and high density (,3.1 g/cm3). The chemistry involves the
presence of both hydroxyl ions and water molecules in the lattice. Lawsonite seems
particularly efficient in transporting water to depths in subduction zones, even into the
transition zone. It does not break down until pressures of 12– 13.5 GPa (;360 –400 km

Figure 13.1. The hydrous magnesium silicate phases in the MgO– SiO2 –H2O system (Source: Prewitt and
Downs, 1998, q 1998 Mineralogical Society of America).
888 Chapter 13

depths) at temperatures of 800 – 9608C (Schmidt, 1995; Pawley, 1994). Indeed, seismic
evidence indicates that the blue-schist-rich assemblage persist to depths of 100 –250 km
within subducted slabs, as evinced from the thin (1 – 7 km thick), relatively low-velocity
(25 to 27%) layer near the top of northern Pacific subduction zones (Abers, 2000).
In both perovskite and stishovite, the coupling of hydrogen with aluminium
substitution for silicon for charge balance plays a key role in controlling the uptake of
hydrogen (e.g., Smyth et al., 1995; Navratosky, 1999). Stishovite may contain
,0.008 wt% H2O at 1,2008C and pressures corresponding to those at depths of 300 km
(Pawley et al., 1993). Mg-silicate perovskite synthesized under hydrothermal (H2O-
saturated) conditions at temperatures of 1,8308C and pressure of 27 GPa (;800 km depth)
can incorporate ,0.006 (^ 0.0015) wt% water (700 H atoms/106 Si atoms). This
comparatively small amount of water retained in nominally anhydrous perovskite could
produce a reservoir of water of ,12% of the mass of Earth’s hydrosphere (Meade et al.,
1994). Presence of Al3þ and increased pressures should each enhance the solubility of
water within silicate perovskite.

13.1.2. Hydrous minerals under pressure

In hydrous b-Mg2SiO4, the OH-bearing vacant Mg sites are probably compressible,

while O4H4 tetrahedra are very compressible. A compressional study of hydrous b (Yusa
and Inoue, 1997) indicates that linear compressibilities along the a- and c-axes are very
similar, while the linear compressibility along the b-axis is ,15% larger than that of
anhydrous b-Mg2SiO4. This suggests that the anisotropic axial compressibilities of
hydrous b-Mg2SiO4 might be controlled by the direction of the OH bond with a 5 –11%
overall decrease in bulk modulus (Yasa and Inoue, 1997).
XRD experiments on hydrous minerals document a loss of the crystalline
diffraction (i.e., amorphization) at high pressures, demonstrating that these materials
become disordered on scales of 103 pm or less (e.g., Hemley et al., 1988; Meade and
Jeanloz, 1990a; Kruger and Jeanloz, 1990; Williams et al., 1990). (Note: X-ray and Raman
scattering probe crystalline order over a range of length scales of ,103 pm for X-rays and
,103 –102 pm for Raman spectroscopy.)
In a high pressure study of hibschite (Ca3Al2(SiO4)1.5(O4 H4)1.5), Knittle et al.
(1992) found that both Raman active O – H stretching vibrations and the higher frequency
IR mode decrease in frequency with increasing pressure, as expected for increased
hydrogen bonding, but the lower-frequency IR band increased in frequency. It was
suggested that this could result from H· · ·H or H· · ·Ca2þ repulsive interactions (e.g.,
Williams, 1992).
Williams (1992) studied a natural chondrodite to ,9 GPa, showing the importance
of H-bonding in one of the O –H stretching vibrations, and used IR spectroscopy to
investigate phase changes following laser heating at pressures between 22 and 44 GPa.
Pawley et al. (1993) have shown that stishovite can accept hydrogen at defect sites
within its structure and the H content was shown to scale with Al substitution into the SiO2
phase (e.g., Smyth et al., 1995). This is consistent with dissolution of H into TiO2 rutile
(e.g., Swope et al., 1995). Many (e.g., Lu et al., 1994; Meade et al., 1994) have shown by
Hydrous Minerals 889

IR spectroscopy that (Mg,Fe)SiO3 perovskite in the lower-mantle condition can contain

trace amounts of H2O.
Both compressibilities and expansivities of many hydrous phases have been
measured by many and some recent workers are listed below:

Hydrous minerals Workers

Brucite, Mg(OH)2 Redfern and Wood (1992)

Diaspore, AlO(OH) Xu et al. (1994), Pawley et al. (1996)
Chondrodite Faust and Knittle (1994)
Lawsonite Comodi and Zanazzi (1996), Pawley et al. (1996), Holland et al. (1996)
Zoisite and clinozoisite Holland et al. (1996)
Paragonite Holland et al. (1996, 1997b)
Talc, phase A, and 10-Å phase Pawley et al. (1995)
Epidote Holland et al. (1996)

Volume measurements of hydrous phases are useful in determining the EOS of H2O
at high P and T.
EOS of H2O at high P can be constrained by measuring the P – T positions of
dehydration reactions in phase-equilibrium experiments (or by using in situ synchrotron
radiation). Then, by using the known thermodynamic parameters of the solid phases
involved, the unknown EOS of H2O can be derived (e.g., Johnson and Walker, 1993).
For the OH bands in hydrous phases, the correlation of pressure response dn=dP
with n is shown in Fig. 13.2 (Hofmeister et al., 1999) (see also Section 13.1.3.1).

13.1.2.1. Water in subducting slabs

The oceanic lithosphere is composed of basaltic volcanics overlain by pelitic and
carbonate sediments and underlain by gabbros, peridotitic mantle and plutonic cumulates.
Perhaps ,20% of the slab volume is made up of oceanic crust and sediments. These
lithologies constitute the prime carriers of H2O to the mantle through the process of
subduction of the lithospheric slab.
In the basaltic component, pargasitic amphibole may be the dominant H2O carrier
(e.g., Davies and Stevenson, 1992) whereas, in pelitic meta-sediments, that may be
phengitic white mica (Schmidt, 1996; Domanik and Holloway, 1996).
Hydrothermal reactions between ocean water and silicates transform the oceanic
crust and the uppermost mantle to form large volumes of hydrous minerals, e.g., talc,
chlorite and serpentine. These hydrous minerals are transported into the mantle through
subduction zones (Fig. 2.15). The stability of hydrous minerals in the Earth’s mantle is of
interest, particularly for an observation such as that minerals like lawsonite, talc and
antigorite would remain stable within cool regions of subducting oceanic lithosphere. Such
phases can carry water down to mantle depths exceeding 150 km. Ultimately, when these
break down, the released water enters the constitution of the dense hydrous magnesium
silicate (DHMS) phases within the slab and thus water is transported down to the depths of
the transition zone (400 –670 km). As the lithosphere sinks into the mantle, it is
compressed to pressures as high as 23 GPa and at low temperature (geologically). Some of
these hydrous minerals transform to metastable phases. Among these metastable phases
890 Chapter 13

Figure 13.2. Correlation of the pressure response dn=dP with n for the OH bands of hydrous phases. Various lines
show linear least squares fits of OH within a given mineral. The trend for phase B (solid line) also fits the other
DHMS. Small open circle: wadsleyite IR (Cynn and Hofmeister, 1994); large open circle: phase B IR (Hofmeister
et al., 1999); filled diamond: shy-B Raman (Hofmeister et al., 1999); filled downward-pointing triangle: phase A
Raman (Liu et al., 1997a); filled upward pointing triangle: chondrodite Raman (Hofmeister et al., 1999); open
upward-pointing triangle, chondrodite IR (Williams, 1992); open square: portlandite Ca(OH)2 Raman and IR
(Meade et al., 1992); filled square: brucite Mg(OH)2 Raman (Duffy et al., 1995) and IR (Kruger et al., 1989);
square with cross: approximate slope for gibbsite Al(OH)3 Raman (Huang et al., 1996); open diamond: talc
Raman (Holtz et al., 1993); X: mica Raman (Holtz et al., 1993); small filled circle: analcime Raman (Velde and
Besson, 1980). For the latter, pressure is probably not accurate þ ¼ OH in the Si site of garnet (Knitttle et al.,
1992) (from Hofmeister et al., 1999, q 1999 Mineralogical Society of America).

are amorphous forms. These “novel” transformations can be correlated with the shear
instabilities and large acoustic emissions, suggesting that they could provide a source for
deep seismicity in subduction zones.
The high equilibrium solubility of water in some mineral phases bears important
consequences for processes occurring in subduction zones. When hydrous fluids released
from the subducting slabs infiltrate the mantle wedge and the temperature exceeds the
water-saturated solidus, the generation of magmas in the subduction zone occurs. But
when the released fluid (mainly water) reacts with the anhydrous minerals, such as
olivines, pyroxenes and even garnets, a large quantity of the fluid becomes trapped inside
the crystalline mass. Furthermore, interaction of a subduction-zone fluid with anhydrous
minerals will change the composition of the fluid with a significant increase in halogens,
particularly chlorine (obtained from oceanic water).
The presence of water in mantle minerals would affect the mantle rheology
(Kohlstedt et al., 1996), melting temperature (Kawamoto et al., 1996) and electrical
conductivity (Li and Jeanloz, 1991). Water may broaden the transformation interval of
Hydrous Minerals 891

phase transitions, responsible for seismic discontinuities (Wood, 1995), and loss of water
may generate deep-focus earthquakes (Meade and Jeanloz, 1991). Recent measurements of
areas of low seismic velocity in the transition zone and lower mantle (e.g., Williams and
Garnero, 1996) offered evidence for water-induced partial melting of the deep mantle.
The b-phase of olivine has shown to have a high solubility of water in its structure.
Hence, the transition zone, rich in b-olivine, is suggested to be high in water content
(Young et al., 1993). Extrapolation of the thermodynamic model of Kohlstedt et al. (1996)
for the solubility of water in olivine into the b-stability field yields a partition coefficient
for water between the b- and a-phases as Db=a ¼ 20: Earlier, from samples containing
grains of both a and b, Young et al. (1993) obtained a much higher (double) value for
Db=a ð¼ 40Þ: Wood (1995) calculated the maximum concentration of water in olivine near
the 410-km discontinuity and, using Db=a ¼ 10; he found an upper limit to be about 3,400
H/106 Si.

Olivine fabric anisotropy. In olivine, dislocation creep and recrystallization are affected
by the presence of water. The deformation fabric (lattice-preferred orientation (LPO)) in
olivine is modified by the fugacity of water and stress. Jung and Karato (2001) identified
three types of olivine fabrics, A, B and C, as follows:
Type A: [100] axis is sub-parallel to shear direction and (100) plane is sub-parallel to
shear plane.
Type B: [001] axis is sub-parallel to the shear direction and (010) plane is sub-parallel to
shear plane.
Type C: [001] axis is sub-parallel to the shear direction and (100) plane is sub-parallel to
shear plane.
Type A dominates at low stress and low H2O content. Type C dominates at high
water content and modest stress and type B dominates at high water content and/or high
stress. In type B, diffusion creep plays an important role.
The [001] orientation sub-parallel to the flow direction in type C may correspond to
the contribution from the glide of dislocations with b ¼ ½001 (where b is the Burgers
vector of dislocation). The b ¼ ½001 dislocations are mostly straight screws, indicating a
high Peierls barrier. (Note: In plastic deformation, motion of dislocations occurs over a
high Peierls barrier.) Type-A fabric dominates under water-poor conditions. Both type-B
and type-C fabrics are found in the peridotites from Alpe Arami and Higashi-Akaishi-
Yama (Japan). (Note: Transition from one type of fabric to another will occur when the
strain rates of two different slip systems coincide.)
The peculiar anisotropy of the deep upper mantle beneath Hawaii (Montagner and
Guillot, 2000) can be attributed to high water content in the upwelling materials.

H incorporation mechanism. On the basis of hydrothermal experiments, it is proposed that

hydrogen becomes associated with negatively charged oxygen interstitials and its uptake is
governed by point-defect mechanisms. Thus, the hydrogen uptake is associated with oxygen
interstitial (Bai and Kohlstedt, 1992a,b; 1993). Assuming an average mantle geotherm (Ito
and Sato, 1992), the hydrogen content increases with depth. Hence, olivines in deep-mantle
rocks can release hydrogen species during slow ascent by mantle convection.
892 Chapter 13

Oxygen vacancies and protonation: perovskites. A cold subducting slab may bring
hydrous phases from the Earth’s crust through the transition zone into the lower mantle. If
the perovskite phase in the lower mantle surrounding the slab contains even small amounts
of oxygen vacancies, they could absorb water from the slab. These would be distributed
rapidly throughout the surrounding mantle by proton migration. The strong relaxation
around the defects and the process of proton hopping may lead to considerable softening of
modes with larger anharmonicity, larger thermal expansion, lower bulk modulus and easier
plastic flow than if there were no oxygen vacancies that could be hydrated (Navrotsky,
1999). The presence of vacancies and water in mantle perovskites can be summarized as
below:
(1) Significantly more aluminium and other trivalent cations would be incorporated into
silicate perovskite in the presence of water.
(2) The proton-conducting ceramic perovskites may serve as a model for the role of water
in aluminous silicate perovskites.
(3) Dense structures favour proton mobilization, as does the shortening of oxygen –
oxygen distances. High temperature opposes hydration while high pressure favours it.
The high density and short O – O distance in MgSiO3 would favour hydration and
proton migration. The incorporation of water is volumetrically favourable, reflecting
the high density of the perovskite with vacancies filled by hydroxyl groups.

Trivalent cations: coupling and diffusivity. Hydrogen incorporation may be coupled with
substitution mechanism, as exemplified by mantle garnets and pyroxenes, showing
correlation of hydrogen content with other substitutional and trace-element cations (Bell
and Rossman, 1992; Kurosawa et al., 1997). Kurosawa et al. (1997) determined the
concentrations of hydrogen and other trace elements in olivines from mantle xenoliths by
secondary ion mass spectrometry (SIMS). The hydrogen contents in olivines from mantle
xenoliths range from 10 to 60 ppm wt H2O.
On the basis of the coupling of hydrogen with trivalent cations, the model of
hydrogen uptake of mantle olivines is proposed (Kurosawa et al., 1997) as follows:

2½MeMe O6 x þ 1=2X2 O3 þ 1=2H2 O ¼ ½XMe OO  þ ½VMe ðOHÞO 0 þ 2MeO

where MeMe denotes a divalent metal ion (Mg2þ, Fe2þ) on a M site, OO represents an
oxygen ion on an oxygen site in the olivine structure, X2O3 denotes an oxide of trivalent
cations which is incorporated into the olivine with H2O and [XMeOO] denotes a positively
charged trivalent cation on a metal site. The term [VMe (OH)O]0 represents an association
of a doubly negatively charged metal vacancy with a positively charged hydrogen ion,
where the hydrogen bonds to one oxygen at the corner of a SiO4 tetrahedron adjacent to the
metal vacant site. The term MeO indicates a metal-oxide molecule released from olivine.
Because the contents of monovalent and trivalent cations in mantle olivines are
controlled by the P, T conditions, the hydrogen contents may vary the equilibrium pressure
and the pressure of the host rock. The hydrogen content increases with the pressure of the
host rock but decreases with increasing temperature. Trivalent cations in olivine show a
slow diffusivity (e.g., Jurewicz and Watson, 1988). Hence, for the establishment of
Hydrous Minerals 893

equilibrium, a stable condition is required with static temperature, pressure and chemical
environment of the mantle over a long period of time. Studies on xenolithic olivine (e.g.,
Lehman, 1983) reveal a coupling of hydrogen with trivalent cations whereby the diffusion
rate of the latter would control that of hydrogen, although hydrogen may possibly have a
higher diffusion rate.

13.1.3. H(D) – O bonds in hydroxides

The extent of hydrogen bonds in solid hydroxides can be assessed from the
observed interionic HðDÞ· · ·O distances, the OH or OD stretching frequencies and their
temperature dependence (Lutz et al., 1995). The OH(OD) stretching frequencies of
hydrogen-bonded water molecules and hydroxide ions correlate with Brown’s bond
valences of the internal OH bonds, which can be determined as the difference of the sum of
the inter-ionic HðDÞ· · ·O bond valences and the valence of the hydrogen atom (e.g.,
Brown, 1995; Lutz and Jung, 1997).
Following the procedure of Lutz et al. (1995), the bond valences of individual O – D
bonds can be given by

SOD ¼ exp½ðR1 2 rÞ=b

where r is the experimentally determined O – DðO· · ·DÞ distance, R1 ¼ 0.914 Å and

b ¼ 0.404 Å.
Two hydrogen atoms show different hydrogen-bond strengths. The OH(OD)
stretching vibrations observed at 3,400 cm21 (2,516 cm21) and 3,513 cm21 (2,599 cm21)
correspond to O(2) –H(1) (O(2) –D(1)) and O(4) –H(2) (O(4) –D(2)) bond strengths,
respectively. The bifurcated hydrogen bonds, in which one hydrogen coordinates to two
oxygen atoms, can be associated with low-frequency deviations in the relationship of OH
stretching frequency with O – H· · ·O bond length (Nakamoto et al., 1955). In simple
layered hydroxides, the effects of H bonding and H· · ·H repulsion are isolated into separate
sub-lattices (Parise et al., 1999). These two forces compete in water-rich compounds,
produced under high-pressure conditions.

13.1.3.1. OH bonds: dn/dP

In response to compression, the changes in hydrogen bonding in hydroxyl-bearing
minerals are noted from the complex response of peak parameters such as frequency,
width, height and area. Stretching frequencies ðni Þ depend not only on O – H· · ·O bond
length but also on the bond angle.
For M– O –H bending modes, if the site has axial symmetry about the O – H vector
(a linear O – H· · ·O configuration), then one doubly degenerate bending peak should exist
but the symmetry is lacking when the hydrogen bond is bent and then the out-of-plane
bending motions should have a different frequency.
For hydrous minerals under increasing pressure, narrow pressure intervals and
nearly hydrostatic environments provide details for the response of the OH vibrations. The
pressure response is indicative of hydrogen bonding. During compression, not only does
894 Chapter 13

the O –O bond distance become important but the bond angle also plays an additional role
in hydrogen bonding.
The correlation of pressure response dn=dP with n for the OH bands of hydrous
phases is shown in Fig. 13.2 (Hofmeister et al., 1999). The pressure derivatives of the
hydroxyl bands increase fairly linearly with frequency for several different phases.
For compressible phases such as micas, analcime, talc, brucite and portlandite, the
trends are steep, whereas, for incompressible phases such as phase B and wadsleyite, the
trends are shallow. This difference is connected with particular structural arrangements.
The locations of hydroxyl groups in layers of SiO4 tetrahedra suggest that the bulk moduli
of O – O bonds (being compressed) are similar to the polyhedral bulk modulus for IVSi at
300 GPa (Hazen and Finger, 1982).
Thus, the correlation is consistent with the dependence of frequency on OH· · ·O
distances. Inter-atomic distances change more rapidly with pressure in compressible solids
such as micas and hydroxides. The diagram in Fig. 13.2 offers a means of inferring bond
compressibilities of structurally related materials when a single-crystal refinement of one
substance at high pressure is given. The trends of most mineral groups intersect at dn=dP ¼
0:55 and nio ¼ 3;625 cm21 (Fig. 13.2). This frequency lies at the limit of hydrogen
bonding and the positive pressure shift at this frequency is in accord with compression
of the O –H bond (in the absence of hydrogen bonding) as the lattice contracts. Larger
and more positive shifts may in some cases arise from additional cation repulsion
(Hofmeister et al., 1999).

13.2. Geophysical effects of water

The presence of water is seen to have a pronounced effect on the weakening of

polycrystalline and single crystals of olivine (Mackwell et al., 1985; Karato et al., 1986)
and other phases. This is manifested by changes in physical properties such as creep rate,
electrical conductivity and seismic-wave velocity. As stated earlier, seismic and
rheological properties are affected by the presence of hydrous phases.
Hydrous magmas can result from dehydration melting reaction in the mantle and
may lead to explosive volcanism. Thus, earthquakes can be triggered by dehydration
reaction. The evolution of the Earth’s H2O budget over time depends on the ability of
mantle phases to store H2O.
For a convection mechanism to be operative in the mantle in solid state is a difficult
postulate (Gasparik, 1993) but the presence of melt would facilitate convection. Melting in
most of the upper mantle, including the asthenosphere, can occur only in the presence of
volatiles, in particular H2O. Again, the limited availability of volatiles would restrict the
convection in the upper mantle to narrow regions associated with mid-oceanic ridges,
subduction zones and hotspots.
Water can also affect seismic-wave velocity and attenuation (Karato, 1995). This is
because water increases the mobility of point defects and dislocations in olivine and thus
enhances anelastic relaxation processes with a resultant reduction in seismic velocity. It is
Hydrous Minerals 895

also postulated that the vibration of pinned dislocations in Mg-silicates can attenuate the
seismic waves (Mitra and Bhattacharya, 1982).

13.2.1. Creep rate

Water fugacity or OH concentration also controls the creep rate. Possibly, creep rate
increases nearly linearly with OH solubility, as observed in quartz aggregates (Kohlstedt
et al., 1995).
As depth increases from 50 to 400 km, the effect of increasing water solubility with
increasing pressure (under fixed f O2 and different stress) may cause an increase in creep
rate by a factor of ,10.
With OH solubility, the creep rate is mostly seen to increase, as does the water
weakening. The creep rate is approximately linearly dependent on water fugacity (i.e., OH
concentration) when the activation volume for creep is not too large (Kohlstedt et al.,
1995). Samples deformed at higher pressures are seen to be weaker than those deformed at
lower pressure (Borch and Green, 1989).

13.2.2. Electrical conductivity

The electrical conductivity of water increases dramatically at the mantle P – T

condition, consistent with the presence of proton carriers (Chau et al., 1999; Cavazzoni
et al., 1999). The presence of water would enhance the electrical conductivity of olivine.
An increase in electrical conductivity of the mantle is effected through two possible
mechanisms:
(1) A highly mobile Hþ ion would increase the electrical conductivity in the mantle
(Karato, 1990).
(2) The introduction of water-derived point defects will affect the concentration of other
point defects such as Mg vacancies and, therefore, will indirectly enhance electrical
conductivity (Bai and Kohlstedt, 1993).
Indeed, the long-period magneto-telluric data obtained by Lizarralde et al. (1995)
for the electrical structure beneath the eastern North Pacific indicate a contribution from
Hþ ions. The electrical conductivities observed are one to two orders of magnitude higher
than those expected for dry olivine at depth ranges of 100– 400 km.

13.3. H2O in the mantle and magmatic melt

Evidence is ample that a substantial amount of water is present in the mantle. Most
of this water is pristine and formed during the early stages of the Earth’s accretion. This
water has been released through time to form and replenish the oceans, rivers, glaciers and
other repositories of water. Groundwater, rainwater, seawater, glacier ice and extra-
terrestrial water each have a distinctive isotope ratio, which allows determination of the
water’s origin.
896 Chapter 13

As stated earlier, in the mantle a minute but significant hydrogen may be retained in
the nominally anhydrous minerals such as wadsleyite and pyroxene. Since the volume of
these phases is enormous compared with the total volume of the Earth, the amount of water
retained in the solid state in these could be several times that contained in the present
oceans.
Hydrous silicate such as amphibole, phlogopite, serpentine, talc and humites exist
stably up to depths of ,200 km. At greater depths, they decompose and water from their
structure is released.
A significant presence of H2O can affect the composition of partial melts (e.g.,
Johnson et al., 1991) and the physical properties of magmas such as density,
compressibility and viscosity (e.g., Schulze et al., 1996). These properties in turn govern
the dynamics of magma ascent and the style of eruption.
Experimental studies on the relationship between water activity in silicate melts,
am
H2 O , and the mole fraction ðXHm2 O Þ at H2O concentration less than several tens of mol%
consistently show the relationship (Burnham, 1975; Holtz et al., 1995) between them as:

am m
H2 O ¼ KðXH2 O Þ
2

The values of K as a function of T and P for felspar-composition melts have been

determined by Burnham (1994).
The activity of H2O in the magma, am 0
H2 O ; is simply the fugacity ratio: fH2 O =fH2 O :
The fugacity of H2O, fH2 O ; is derived from the calculated fH2 and dissociation constant of
H2O provided that the value of fO2 is known. In the H2O dissociation, the fH2 O is
proportional to ( fO2)1/2.
Results of high-pressure experimental work have provided a framework in which
investigators develop models that incorporate H2O, (OH)þ or Hþ as essential constituents
of the phase chemistry. High-pressure results show not only that Si coordination changes
from four to six but also that the structural elements are so reformed that they permit
hydrogen to be retained under extreme conditions.
In mid-oceanic ridge basalt (MORB), water is compatible to RE elements such as
Ce, La and Nd, which are more compatible than K. Mantle sources, depleted in
incompatible elements (termed as normal MORB (N-MORB)) may have water content of
the order of 80– 180 ppm. In MORB samples, a decoupling of H2O and K is often
observed.

13.3.1. H2O in plagioclase crystallization

High water pressure may cause plagioclase to crystallize at lower temperatures than
in dry conditions. The plagioclase commonly found in arc basalts (An90 – 95) is known to be
more An-rich than that from MORB. This disparity in An content can be attributed to a
higher water fugacity in arc basalts than in ridge basalts. The exchange equilibrium in the
crystal melt can be represented as:

KD ¼ ðCa=NaÞplag=ðCa=NaÞmelt
Hydrous Minerals 897

This KD values for dry melts at pressures below 2.0 GPa are always ,2.5. With increasing
water content in the melt, the KD value increases and reaches 5.5 for H2O-saturated liquids
at 0.2 GPa (Sisson and Grove, 1993). Thus, plagioclase co-existing with hydrous melts is
more An-rich than plagioclase co-existing with dry melts. An assemblage of high-An
plagioclase, Ca-rich clinopyroxene and hornblende would represent a phase equilibrium at
high water pressure. 4 –5 wt% of dissolved H2O is required to stabilize hydrous
hornblende in basaltic to dacitic melts.

13.4. MgO – SiO2 – H2O ternary system

The structure and stability of high-pressure hydrous silicate phases have received
considerable attention because of their importance in many geodynamical processes. For
this study, the most important system is MgO –SiO2 – H2O (MSH). The ternary diagram of
this system showing the compositions of hydrous magnesium silicates was shown in
Fig. 13.1. Those containing octahedral Si, such as phase B, superhydrous B (Shy-B) and
phase D, are called DHMSs. The nomenclature is based either on one prominent XRD line
or alphabetical scheme.
The phases described thus far are: 10-Å phase (Sclar et al., 1965a,b), 3.65-Å phase
(Sclar et al., 1967), A, B, C (Ringwood and Major, 1967b), D (Yamamoto and Akimoto,
1974; Liu, 1988), D0 (Liu, 1987), E (Kanzaki, 1989, 1991), F (Kanzaki, 1991; Kudoh et al.,
1995), G (Ohtani et al., 1997) and anhy-B (anhy-B) (Herzberg and Gasparik, 1989) and
Shy-B (Gasparik, 1990). These are discussed in Section 13.4.1. At pressure above 12 GPa,
the nominally anhydrous phases such as clinopyroxene, garnet, olivine and b-(Mg,Fe)2
SiO4 dissolve significant amounts of H2O (Bell and Rossman, 1992). Because these phases
are an order of magnitude more abundant than the hydrous phases, the nominally
anhydrous phases are expected to be the main carriers of H2O at depths in excess of
300 km. Along with many such others, the Al2O3 – SiO2 – H2O system has recently been
drawing much attention (Schmidt et al., 1998). In this system, phases like kaolinite and
pyrophyllite are stable only at crustal conditions (see Section 13.11 and Fig. 13.21).
The simplified MSH and FeO – MgO –SiO2 – H2O (FMSH) systems represent ,95–
96 wt% of the total oxide components of harzburgite to lherzolite mantle and therefore are
important for studies on the stability of H2O-bearing minerals in the mantle. The remaining
4 –6 wt% includes oxides such as Al2O3, CaO, Cr2O3, Na2O, NiO and TiO2.
Investigations on the MSH system at high pressure have revealed that several
H-bearing phases are stable at pressures to 20 GPa (see review of Prewitt and Finger,
1992). The ability of Si at four- and six-coordination to share corners and edges with Mg
octahedra plus the availability of H for local charge balance give rise to a wide variety of
stable and metastable structure types. The number of Mg atoms versus the number of Si
atoms based on 12 oxygen for dense hydrous and anhydrous magnesium silicate are plotted
in Fig. 13.3 (see the caption).
Gasparik (1993) investigated the MgO –SiO2 – H2O system at 16– 23.4 GPa and
located the low-pressure stability limit and dehydration of the assemblage superphase
B þ stishovite and he also determined the hydrous solidus at 22.3 GPa. His study indicated
that volatile-bearing phases can co-exist with stishovite at pressures above 13 GPa.
898 Chapter 13

Figure 13.3. The number of Mg atoms vs number of Si atoms based on 12 oxygens for dense hydrous and anhydrous
magnesium silicates. PhA: phase A; PhE: phase E; PhF: phase F; SubB: superhydrous B; PhB: phase B; HyBt:
hydrous beta (hydrous wadsleyite); Anh B: anhydrous B; Bt: Beta (anhydrous wadsleyite). Solid lines are of
C ¼ 9.5, 11.4 and 12.0 for 5.0H, 1.2H and 0.0H, respectively (Kudoh et al., 1996, q 1996 Springer-Verlag).

The assemblage superphase B þ stishovite was observed at 16 – 23.4 GPa and, at

16 GPa/1,1708C, this was replaced with phase E.
The microprobe analysis suggested that b-phase could have up to 7 wt% of
structural H2O at 16 GPa, while the co-existing clinoenstatite could have up to 3 wt%
H2O. The hydrous b-phase could also attain a higher iron content than indicated by
experiments in an anhydrous Mg2SiO4 – Fe2SiO4 system (e.g., Katsura and Ito, 1989) and
thus it would expand its stability field at the expense of spinel. Numerous experimental
studies have been made since the pioneering work of Ringwood and Major (1967) on the
phase relations and stability fields of DHMS phases in the MgO –SiO2 – H2O system. Of
the 12 known DHMS phases (Table 13.1), seven have been synthesized at #10 GPa.
At pressures below 7 GPa, the hydrated mantle assemblage could be: forsterite þ
enstatite þ talc. At this pressure and T , 6008C, talc will react with forsterite producing
phase A and enstatite. At 15 –16 GPa, phase A reacts with enstatite to produce hydrous
b-phase and stishovite. The hydrous b-phase could potentially contain more water than the
maximum allowable water content in a slab.

13.4.1. DHMS phases

These DHMS phases show stability fields corresponding to depths much greater
than 200 km. The structure of 10-Å phase is considered equivalent to that of talc
Hydrous Minerals 899

TABLE 13.1
List of DHMS and other hydrous phases in the mantle

Terminology Abbreviation Composition References

Phase A A Mg7Si2O8(OH)6 Horiuchi et al. (1979)

Phase B B Mg12Si4O19(OH)2 Finger et al. (1989)
Superhydrous Shy B Mg10Si3O14(OH)4 Pacalo and Parise (1992)
Clinohumite-OH CH Mg9Si4O16(OH)2 Yamamoto and Akimoto (1977)
Chondrodite-OH Chon Mg5Si2O8(OH)2 Yamamoto and Akimoto (1977)
Humite-OH Hum Mg7Si3O12(OH)2 Wunder et al. (1955)
Phase C C ? Ringwood and Major (1967)
Phase D D MgSiO2(OH)2 Liu (1987)
Phase E E Mg2.08Si1.16O2.8(OH)3.2 Kudoh et al. (1993)
Mg2.17Si1.01O2.38(OH)3.62 Kudoh et al. (1993)
10-Å phase 10 Å Mg3Si4O10(OH)2A0.65H2O Wunder and Schreyer (1992)
Mg3Si4O10(OH)2A1.00H2O Bauer and Sclar (1981)
Mg3Si4O10(OH)2A2.00H2O Yamamoto and Akimoto (1977)
3.65-Å phase 3.65 Å Mg1.4Si1.3A1.3(OH)8 Rice et al. (1989)
Hydrous wadsleyite Shy b Mg1.75SiO3.5(OH)0.5 Kudoh et al. (1996)

(e.g., Prewitt and Parise, 2000). Similarly, phase C may correspond to Shy-B, phase D to
chondrotite while phase E is described as a hydrous form of forsterite (Liu et al., 1997) and
may have a stability field that extends to about 1,1008C at 12– 15 GPa (Inoue, 1994),
suggesting it is another host of water in the subduction zone at depths greater that than of
lawsonite.
The stability field of phase A decreases relative to the pure MgO – SiO2 – H2O
system by the presence of Ca (by 70 –1208C at 8 GPa), Al (by 40 –808C) and Fe (by ,208C)
and is unaffected by CO2 content.
Phase D has the highest pressure stability amongst the DHMS phases and thus could
be the major phase for sequestering water to depths between 600 and 1,250 km.
Phase B and/or Shy-B can be formed from material with Mg/Si ratios in the 1.5– 2.0
range (e.g., Ohtani et al., 1995). Similarly, with that ratio close to 2, phase E may form
from olivine and water. Like phase E, phase D and Shy-B are stable at high P –T conditions
and are quenchable (Shieh et al., 2000).
Phases A, B and D have been observed to decompose or amorphize rapidly at
ambient pressure at 440, 300– 400 and 1008C, respectively (Liu et al., 1998a).
Most hydrous phases have bulk moduli that are generally within ,10% of those of
(Mg,Fe)2 SiO4 polymorphs with a similar stability field.
The presence of H in ringwoodite (estimated compositions of Mg1.89Si0.97O4H0.33,
or 2.2 wt% water) lowers its elastic moduli by ,10% and produces compressional and
shear-wave velocity decreases of 5.3 and 3.6%, respectively. It also becomes elastically
more anisotropic (Inoue et al., 1998).
The investigation on the crystal structures of phase B by Finger et al. (e.g., 1991)
reveals how hydrogen is incorporated into phase B.
900 Chapter 13

Shy-B is likely to be equivalent to phase C, reported first by Ringwood and Major

(1967). The crystal structure of phase D (Yang et al., 1997) seems to be the same as that of
phase G (Kudoh et al., 1997).
Phase E is characterized by long-range disorder (e.g., Kudoh et al., 1993) and it is
regarded as the hydrous form of forsterite (Liu et al., 1997). As already stated, these
are called “alphabet phases”. In this category, more phases are included which are 10 and
3.65-Å phases.
Thompson (1992) reviewed the stabilities of possible hydrous and anhydrous
phases present in the Earth. A list of DHMS is presented in Table 13.1.

13.4.1.1. MgO – SiO2 1 volatiles (H2O, F2, Cl2) system

The phase relations in the system MgO –SiO2 with low volatile contents of H2O, F2
and Cl2 were investigated at 10– 23.4 GPa and 850 –2,0508C using a split-sphere anvil
apparatus, USSA-2000 (e.g., Gasparik, 1993). The following phases were observed
(at the respective pressures):

Phase Composition Pressure

Superphase B Mg12Si4O19F2 17.8 GPa

Hydrous phase F Mg4Si7O16(OH)4 17.8 GPa
Phase A Mg7Si2O8[OH,F]6
Phase C Mg(72x)Si(2þx)O(4 þ x 2 y)[OH,F]4
Phase E Mg(22x)Si1þxO(4þx 2 y)[OH,F]2y

Dehydration produces b-Mg2SiO4, possibly containing up to 7 wt% of structural

H2O.
Volatile-bearing phases can only be present in the colder subducted slabs, not in the
rest of the transition zone. Hydrous b-phase þ stishovite are the most common
constituents of the slabs. Incongruent melting of the hydrous b-phase co-existing with
stishovite, and producing garnet and melt, could be the cause of deep-focus earthquakes.
The density, H content and stability of hydrous phases of the Earth’s mantle are listed
in Table 13.2.
The chemically more complex hydrous silicate phases that have been synthesized in
multi-anvil apparatus include:
(a) K-substituted K-richerite, K(KCa)Mg5Si8O22(OH)2 (at 15 GPa, 1,4008C)
(b) clinopyrobole, K0.96Ca1.56Na2.51Mg6.01Al1.12Si12O34(OH)2 (at 10 GPa, 1,2008C)
(c) hydrous phase (aenigmatite structure), Na2(Mg5.30Al0.48)Si5.91O18(OH)2 (at 10 GPa,
1,2508C)
(d) Na-phase X, (Na1.16K0.01)(Mg1.89Al0.14)Si2O7H0.65 (at 10 GPa, 1,2508C)
(e) K-phase X, K1.54Mg1.93Si1.89O7H1.04 (at 16 GPa, 1,3008C).
The reflectance spectrum of phase D shows that only SiO6 structural units (near
700 cm21) are present, with no evidence for SiO4 tetrahedron (expected between 1,000 and
1,500 cm21), which are typical of crustal and upper-mantle silicate minerals.
Vibrational spectra also reveal variable hydrogen bonds, with a low hydroxyl
frequency at ,2,850 cm21, corresponding to an O – H – O bond length of 2.67 Å
(Novak, 1974).
 Hydrous Minerals
TABLE 13.2
Selected hydrous phases in Earth’s mantle

Mineral name Formula Density H content Stability References

(wt%)

Phase A MgIV
7 Si2O8(OH)6 2.96 1.3 6–10 GPa, ,1,0008C Ringwood and Major (1967); Wunder (1998)
Phase B MgIV VI
74 Si2 Si6O38(OH)4 3.32–3.38 0.3 12–24 GPa, ,1,0008C Ringwood and Major (1967); Finger et al. (1989)
Shy-B MgIV VI
10 Si2 Si O14(OH)4 3.21–3.33 0.7 12–24 GPa, 1,3008C Pacalo and Parise (1992)
[IVþVI]
Phase E Mg2.3 Si1.25H2.4O6 2.78–2.92 1.3 12–15 GPa, 1,1008C Kanzaki (1989); Inoue (1994); Shieh et al. (2000b)
Phase D MgVISi22xH2þ4xOg6 3.50 1.1 4–50 GPa, ,2,4008C Liu (1987); Irifune et al. (1998); Shieh et al. (1998)
10-Å phase MgIV
3 Si4O10(OH)2H2O ,2.65 1.0 3–9 GPa, 5008C Sclar et al. (1965); Yamamoto and Akimoto (1977)
Brucite Mg(OH)2 2.37 3.4 .78 GPa, 1,3008C Duffy et al. (1991); Fei and Mao (1993);
Johnson and Walker (1993)
Serpentine MgIV
3 Si2O5(OH)4 2.55 1.5 5 GPa, 7008C Ulmer and Tromsdorff (1995); Wunder and
Schreyer (1997)
Norbergite MgIV
3 SiO4(OH)2 3.19 1.1 .5 GPa, 1,1008C Yamamoto and Akimoto (1977)
Chondrodite MgIV
5 Si2O8(OH)2 3.06–3.16 0.6 10 GPa, 1,1008C Akimoto and Akaogi (1980); Wunder (1998)
Humite MgIV
7 Si3O12(OH)2 3.10 0.6 .3 GPa, 9008C Liu (1993); Wunder et al. (1995)
Clinohumite MgIV
9 Si4O16(OH)2 3.14–3.26 0.3 .10 GPa, .1,0008C Yamamoto and Akimoto (1977); Kanzaki (1991);
Wunder (1998)
Lawsonite CaAl2(IVSi2O7)(OH)2H2O 3.09 1.3 2–11 GPa, ,8008C Pawley (1994); Schmidt (1995)
Phlogopite K2(Mg,Fe2þ)IV6 Si6Al2O20(H,F)2 2.78 0.3 .10 GPa, ,1,0008C Sudo and Tatsumi (1990); Liu (1993)
K-Richterite K2Ca(Mg,Fe2þ)IV 5 Si8O22(OH)2 3.01 0.2 5 GPa, 1,2008C Inoue et al. (1998)
Phase X K4MgIV8 Si8O25(OH)2 (?) 2.95–3.28 (0.3) 17 GPa, 1,6008C Luth (1997); Inoue et al. (1998)
Zoisite Ca2AlIV
3 Si3O12(OH) 3.15 0.2 4 GPa, 7008C Poli and Schmidt (1998)
Talc MgIV
3 Si4O10(OH)2 2.78 0.5 5 GPa, 7008C Yamamoto and Akimoto (1977)
Topaz-OH AlIV
2 SiO4(OH)2 3.37 1.1 11 GPa, 1,0008C Wunder et al. (1993a,b)
Diaspore AlOOH 2.38 1.7 6 GPa, 4008C Wunder et al. (1993a,b)
Phase “pi” AlIV
3 Si2O7(OH)3 3.23 1.0 6 GPa, 7008C Wunder et al. (1993a,b)
Phase “egg” AlVISiO3OH 3.84 0.8 .11 GPa, 1,3008C Schmidt et al. (1998)
Mg-pumpellyite Mg5AlIV5 Si6O21(OH)7 ,3.3 0.8 ,6 GPa, 7708C Artioli et al. (1999)
Wadsleyite I MgIV
7 Si4O14(OH)2 3.47 0.4 .17 GPa Smyth et al. (1997)
Wadsleyite II MgIV
7 Si4O14(OH)2 3.51 0.4 (0.2) .17 GPa, ,1,4008C Smyth and Kawamoto (1997)

901
902 Chapter 13

Superhy-B (Shy-B) is stable to the bottom of the transition zone and top of the
lower mantle, whereas phase D decomposes along slab geotherms at ,1,250-km depth.
This may define the lower depth limit for dense hydrous Mg silicates. The DHS phases
show different degrees of hydrogen bonding through an O – HO linkage (involving a
hydrogen bond/covalent bond) at ambient pressure but this is not valid at high pressure
when a symmetric hydrogen-bonded state with weakened OH covalent bond is
approached. The prototypical case is found in the symmetric hydrogen-bonded phase of
H2O at 60 GPa (Aoki et al., 1996). Compression of O –HO linkage gives hydrogen
bonding with distances of 2.38 – 2.40 Å.
Phase D exhibits the shortest hydrogen bonding r(O – HO) ¼ 2.67 Å. Under pressure,
weakened covalent OH bonds in phase D or other DHMS may give rise to large
anharmonic effects prior to melting, including super-ionic conductivity (e.g., as predicted
for sub-solidus H2O; Cavazonni et al., 1999). Such behaviour would give rise to entropic
stabilization of the solid as well as seismic attenuation (in the absence of free H2O-rich
fluid phase) (Hemley et al., 2000).
Pressure increases hydrogen bonding but some workers reported decreased
hydrogen bonding. These suggest hydrogen –hydrogen repulsion under high pressure.
Pressure-induced disordering in hydrous silicates may be associated with hydrogen
through sub-lattice amorphization or “melting” (e.g., Parise et al., 1998).

13.4.1.2. Halogens in DHMS phases

It is known that, with increase in halogen content, the thermal stability of DHMS
phases is notably enhanced. The substitution of fluorine for hydroxyl appears to increase
the stability of Shy-B and phase E by 1008C (Gasparik, 1993), The fluorinated versions of
these phases have stabilities that approach 1,5008C at 15 GPa (;mantle condition). The
upper-mantle micas and amphiboles typically show the pressure of 0.4– 1 wt% fluorine
(Smith and Dawson, 1981). Hence, comparable levels of halogens may be expected to
stabilize markedly the hydrous phases.

13.4.2. NMR spectroscopic study

Phillips et al. (1997) obtained NMR spectroscopic data on MSH phases such as A,
B, Shy-B and E, along with hydroxyl-chondrodite (Mg5Si2O8(OH)2) and forsterite
(Mg2SiO4, 95% 29Si-enriched). They collected the 1H SP MAS spectra with 908 pulses
from 2.5 to 4 ms; the spinning was between 8 and 15.2 kHz.
Phillips et al. (1997) measured the NMR spectra with chemagnetics CMX-300
spectrometer at Larmor frequencies of 59.6 MHz for 29Si and 300.1 MHz for 1H. Phase B
and Shy-B gave very narrow 29Si NMR peaks and displayed the most shielded SiVI
chemical shifts yet reported: 2170.4 ppm for B and 266.6 for Shy-B. The 1H NMR spectra
of B and Shy-B confirmed the presence of paired hydroxyls (from XRD). The chemical
shifts for both 29Si and 1H were referenced to tetramethylsilane (TMS), while
kaolinite was used as a secondary 29Si solid-state reference, which gives two peaks at
290.86(2) and 291.50(2) ppm from TMS. For a secondary 1H chemical shift, standard
Hydrous Minerals 903

reagent grade hydroxyl-apatite was used. The position of the hydroxyl resonance was taken
to þ 0.2 ppm from TMS.
The 29Si spectrum of each of these phases contains a single resonance, which is
consistent with their crystal structures. The NMR chemical shifts and structural parameters
for dense phases in the system MgO – SiO2 – H2O containing tetrahedral Si are given in
Table 13.3.

13.4.3. Choke point

The choke point refers to the maximum depth along the subduction zone, beyond
which the temperature and pressure conditions are such that no H2O-bearing phase would
be stable and H2O could be transported in hydrous phase to greater depths.
Wunder and Schreyer (1997) studied the high-pressure stability field of antigorite in
the MSH system and asserted that very cold subduction paths of 5008C at 5 GPa (160 km)
would be necessary to carry any H2O in ultra-mafic rocks deeper than this choke point of
completed dehydration of antigorite to forsterite plus enstatite.
Ulmer and Trommsdorff (1995) determined the choke point around 6 GPa and
6008C, where antigorite transforms to other H2O-bearing assemblages containing phase A
and possibly 10-Å phase. Later, Bose and Navrotsky (1998) found this choke point to be
located at 5.9 GPa and ,6208C.
Phase D passes the “choke” point in slabs (geotherms cool enough to avoid the free
H2O and melt fields), which would persist to a depth of ,1,250 km before dewatering and

TABLE 13.3
NMR parameters of MgO– SiO2 –H2O phases

Phase d 1H (ppm) d 29Si (^0.1 ppm) Si site Xa Reference

NMR Structure

Forsterite 261.8 20.0470 1.2 6

b-Mg2SiO4 279.0 20.1046 3 7
Condrodite 1.1(1) 261.8 20.0491 1 8
B 4.7(3), 3.3(3)a 264.0 Si (2) 20.0612 1 9
275.05 Si (4) 20.0956
275.85 Si (3) 20.1009
2170.4 Si (1) 20.0145
Shy-B 5.0(3), 3.4(3) 274.6 Si (2) 20.0974 1 10
2166.6 Si (1) 20.0215
A 263.9 Si (2) 20.0520 4 11
270.6 Si (1) 20.0742
MgSiO3 garnet 268.1 T (1) 20.0849 5 12
274.5 T (2) 20.0837
290.2 T (3) 20.0855

References: 1. Phillips et al. (1997); 2 Mägi et al. (1984); 3. Stebbins and Kanzaki (1991); 4. Kanzaki et al. (1992);
5. Phillips et al. (1992); 6. Birle et al. (1968); 7. Horiuchi and Sawamoto (1981); 8. Yamamoto (1977); 9. Finger
et al. (1991); 10. Pacalo and Parise (1991); 11. Horiuche et al. (1979); 12. Angel et al. (1989).
a
Isotropic shifts estimated from line-shape calculations and differ from peak position by 0.4 ppm.
904 Chapter 13

possibly defining the lower depth limit for stability of stoichiometric DHS (Williams and
Hemley, 2001).
The proposed hydrous phases, MgSiO2(OH)2 and MgSi(OH)6, isomorphous of
diaspore and stottite, respectively, are expected to occur in the Earth’s deep interior. Their
presence, integrated over the Earth’s volume, may account for the major repository of
water in the Earth.

13.4.4. Serpentine and phase A

13.4.4.1. Serpentine, Mg3SiO5(OH)4

Serpentine is a hydrated phyllosilicate composed of sheets of tetrahedra and
Mg(OH)6 octahedra. This is generated through the hydration of typical mantle rocks,
which can contain up to 13 wt% water. Its maximum thermal stability is between 650 and
7308C at 2– 4 GPa (Wunder and Schreyer, 1977). Hence, serpentine is capable of
transporting water within subducted slabs into the 150 – 200-km depth range. Its
dehydration from the perdotitic portion of the slab could contribute to arc volcanism
and hydrate the overlying basaltic crust, forming lawsonite.
The principal hydrous phases in H2O-saturated peridotite to the pressure of ,8 GPa
are serpentine, phase A, chlorite, talc and amphibole. In natural peridotites, serpentine
(antigorite) forms during low-grade metamorphic hydration, its maximum temperature
stability being 7208C at 2.1 GPa (Fig. 17 from Mysen et al., 1998).
Antigorite is the most important hydrous phase in ultra-mafites metamorphosed to
green-schist to amphibolite to eclogite facies conditions. There is field evidence (e.g., SE
Spain) for its breakdown in high-pressure fields to spinifex-textured olivine and
orthopyroxene (Trommsdorff et al., 1998).
Small amounts of substitutional Al and Fe in the pure Mg end-member (antigorite)
distort the crystal structure such that OH vibrations between 3,400 and 3,500 cm21 are
observed in addition to the normal band at 3,700 cm21 (Heller-Kallai et al., 1975; Velde,
1980). The breakdown of antigorite and the formation of post-antigorite hydrous phases
might control the transport of H2O at deeper parts of the subducting oceanic lithosphere.

13.4.4.2. Phase A (Mg7Si2O8(OH)6)

Phase A (Mg[6] [4]
7 Si2 O8(OH)6) was synthesized by Ringwood and Major (1967). The
upper thermal stability of phase A is limited by the equilibrium relation (Yamamoto and
Akimoto, 1977):

phase A ! 1 chondrodite 2 OH þ 2 brucite

Serpentine (antigorite) decomposes to talc þ olivine þ H2O as a function of

increasing temperature below 1.9 GPa (Evans et al., 1976) to orthopyroxene þ olivine þ
H2O between 1.9 and 6.2 GPa and to phase A þ orthopyroxene þ H2O at higher pressures
(Ulmer and Trommsdorff, 1995; Bose and Navrotsky, 1998).
In a subducted peridotite, phase A (,11.8 wt% H2O) replaces serpentine at
pressures between 6 and 7 GPa through a water-conserving reaction. Above 6.2 GPa,
Hydrous Minerals 905

phase A þ orthopyroxene decompose with increasing temperature to enstatite þ H2O.

This reaction has a moderate positive slope in P –T space (Luth, 1995).
In an experiment, Irifune (1996) observed that, upon compression at 28 GPa at
room T, serpentine did not change but, at P . 1.5 GPa, it changed to amorphous state in a
limited T interval of 200 –4008C. Above ,4008C, a rapid crystallization of high-P phase
started. Since the temperatures experienced by the subducted slabs are assumed to be
.4008C, amorphization of serpentine does not provide an appropriate mechanism for
deep-focus earthquakes.

(i) XRD observation. The crystal structure of serpentine’s non-hydrogenous component

was investigated by Horiuchi et al. (1979) using the single-crystal XRD method.
Kagi et al. (1999) collected neutron powder diffraction data of phase A at ambient
pressure and 3.2 GPa from the deuterated compound and the structure was refined using
the Rietveld method. The derived crystal structure implies that hydrogen atoms occupy
two distinct sites in phase A, both forming hydrogen bonds of different lengths with the
same oxygen atom.
In the pressure range of 6.6 –22.6 GPa, it is observed that the intensity of 111 line of
serpentine remains nearly constant till 9.7 GPa but, between 6.7 and 22.6 GPa, all other hkl
lines are gradually eliminated. These hkl lines (below 25 GPa) are fully recoverable on
quenching to ambient pressure. It should be noted here that the XRD patterns for pressure-
induced amorphous phases may be relatively sharp compared with those of conventional
melt quenched glasses (Meade, 1991).

(ii) IR bands. This picture is supported by IR spectra, which exhibit two absorption bands
at 3,400 and 3,513 cm21 corresponding to OH stretching vibration (Fig. 13.4a,b) and
proton NMR spectra, which display two peaks with equal intensities and isotropic
chemical shifts of 3.7 and 5 pp. Similar IR frequencies were earlier reported by Akaogi and
Akimoto (1986). These two frequencies suggest the existence of two different hydrogen
sites with different extents of hydrogen bonding in phase A. For phase A, Raman OH bands
depend linearly on P, with constant peak height and slopes similar to those observed for
weak stretching modes of trace H in phase B. This behaviour is common and is associated
with linear hydrogen bonds and unpaired OH2 groups. Raman frequencies of lattice modes
of phase A (and also of Shy-B) are seen to depend linearly on P.

(iii) Raman bands. At ambient pressure, the following three major Raman peaks are noted
in serpentine (Meade et al., 1992):
231 cm21 ! Aig octahedral mode
388 cm21 ! Eg tetrahedral vibrations
689 cm21 ! Aig tetrahedral vibrations
220 cm21 ! internal vibrations of MgO6 octahedra.
Above 10.4 GPa, the ,220-cm21 band disappears and two bands at ,145 and
205 cm21 appear. Up to 25 GPa, the two tetrahedral modes (388 and 689 cm21) shift to
higher frequencies. Between 25 and 27.5 GPa, the Raman peaks abruptly disappear.
Decompression brings back only a marginal recrystallization. This observation contrasts
906 Chapter 13

Figure 13.4. IR spectra of phase A. (a) Hydrated sample for NMR measurements. (b) Deuterated sample (Kagi
et al., 2000, q 2000 Springer-Verlag).

with X-ray observation, in which the diffraction intensity diminishes continuously between
6.6 and 22.6 GPa and is recoverable.

(iv) Amorphization. Up to a compressive pressure of 28 GPa at RT, serpentine does not

change but, at P .28 GPa, it is seen to change to the amorphous state in a limited
temperature interval between 200 and 4008C. Above ,4008C, a rapid crystallization of
high-P phase starts. Since the temperature of the subduction slabs are presumed to be
.4008C, amorphization of serpentine does not provide an appropriate mechanism for
deep-focus earthquakes (Irifune, 1996). In Fig. 13.5, the acoustic emission observed
Hydrous Minerals 907

Figure 13.5. Distribution of earthquakes with depth in the mantle (Frolich, 1989). Also shown is the range of
pressures at which acoustic emission is observed during stable (dehydration) and metastable (amorphization)
phase transitions in serpentine (Irifune, 1996).

during stable (dehydration) and metastable (amorphization) phase transitions in serpentine

and the distribution of earthquakes in the mantle depths are shown.

13.4.4.3. Chrysotile transformations

Chrysotile, another species of serpentine, has been observed to transform from a
pressure range of 0.05 GPa to as high as 6 GPa as:
5 chrysotile ¼ 6 forsterite þ talc þ 9H2 O
Below 4 –5 GPa, the talc þ forsterite stability field is limited by the reaction:
forsterite þ talc ¼ 5 enstatite þ H2 O
Aluminium expands the stability field of serpentine. At high pressures, various
parageneses are observed along the serpentine breakdown curve and beyond. The
following reaction products are observed in varying fields (Mysen et al., 1998):
talc þ H2O ¼ 10-Å phase
5 chrysotile ¼ 6 forsterite þ 10-Å phase þ 7H2O
11 chrysotile ¼ 3 phase A þ 4 1-Å phase þ H2O
3 chrysotile þ phase A ¼ 8 forsterite þ 9H2O
5 phase A þ 3 10-Å phase ¼ 22 forsterite þ 24H2O
forsterite þ 10-Å phase ¼ 5 enstatite þ 3H2O
phase A þ 5 10-Å phase ¼ 22 enstatite þ 18H2O
phase A þ 3 enstatite ¼ 5 forsterite þ 3H2O
908 Chapter 13

Evans et al. (1976) reported experimental results that delimit the breakdown of
antigorite to forsterite þ talc þ H2O to 15 GPa:
antigorite ¼ 18 forsterite þ 4 talc þ 27H2 O

13.4.4.4. Talc and phase A

The equilibrium to 4 GPa of the reaction talc ¼ enstatite þ quartz/coesite þ H2O
has been studied by Bose and Ganguli (1995). The compressibility data for talc have been
retrieved from the experimental phase-equilibrium data and several other equilibria on the
MSH system involving talc, antigorite and the DHMS Mg7Si2O8(OH)6, commonly
referred to as phase A. In cold oceanic slabs (.50 m.y. with subduction velocity
.10 cm/yr), antigorite will transform to the DHMS phase through a vapour-conserved
reaction at a depth of ,200 km. Phase A will then serve as a carrier of water into the deeper
mantle.

Compressibility and expansivity. The thermal expansivity of talc and phase A and the
compressibility of talc, phase A and 10-Å phase, Mg7Si2O8(OH)2, x H2O, were measured
by Powley et al. (1995) using powder XRD.
Compressibility measurements of talc, 10-Å phase and phase A were made at P
,6.05, 8.52 and 9.85 GPa, respectively. The data for 10 Å are consistent with a positive
slope for its dehydration reaction making the 10-Å phase a good candidate for H2O storage
in subducting slabs. The measurements of the thermal expansivity and compressibility of
phase A allow its enthalpy of formation and entropy to be derived from the results of phase-
equilibrium experiments on phase A.

13.4.4.5. Discussion
The term “amorphous” is used by X-ray workers when no measurable XRD pattern
is seen in a sample but it may, in fact, possess strong ordering (i.e., non-amorphous) at a
scale of ,103 pm or less, which can be detected by Raman spectroscopy.
By combining the two methods, a study on serpentine shows that there are at least
two transformations to metastable high-pressure phases (cf. Hemley et al., 1989). At
22.6 GPa, the XRD lines disappear, suggesting a phase which is marginally crystalline
over the length scale of 103 pm. The complete loss of Raman scattering above 27.5 GPa
suggests the amorphous phase is completely disordered.
In serpentine (as also in Ca(OH)2, portlandite crystals), some portions of layered
SiO4 tetrahedra and Mg(OH)6 octahedra may be retained to 25.0 GPa because sharp
tetrahedral and octahedral vibrations are observed up to this pressure. Thus, high-pressure
(,25 GPa) serpentine samples that are X-ray amorphous may still show significant
ordering on the scale of the Raman measurements (,100 –1,000 pm).
Non-hydrostatic stresses promote the gradual disordering as the shear stresses are
seen to increase the line widths of XRD and spectroscopic peaks. The results indicate that
metastable transitions in serpentine (and Ca(OH)2, portlandite) occur through the discrete
transformation of crystalline to amorphous material. The mechanism involved resembles
equilibrium-phase transitions rather than gradual disordering of the crystal structure.
Hydrous Minerals 909

Again, the RT reversibility would imply that the transformations do not involve long-range
thermal diffusion. Thus, the crystal ! amorphous (c ! a) transition would require only
small atomic displacements, possibly of the order of a unit cell or less, and accomplished
by simple shear translations between layers. The acoustic emissions during c ! a
transition in serpentine (Meade and Jeanloz, 1991) would be analogous to the acoustic
emissions that are generated by phase transformations between crystalline phases at high
pressures (Meade and Jeanloz, 1989).

13.4.4.6. 10-Å phase, Mg3Si4O10(OH)2, n H2O

The high-pressure stability of talc and 10-Å phase has been investigated by Pawley
and Wood (1995) using the bulk composition: Mg3Si4O10(OH)2 þ H2O at 2.9 –6.8 GPa
(650 – 8208C). Three different compositions with n ¼ 0:65; 1.00 and 2.00 have been
described by different authors (Table 13.1). By a study of the reaction 10-Å
phase ! talc þ water in the range 3 –7 GPa and 200 – 7008C, Wunder and Schreyer
(1992) found the 10-Å phase broke down. Nevertheless, the lower-pressure stability of the
10-Å phase is defined by its dehydration to form talc þ water at 4 –5 GPa and 600– 7008C
(Powley and Wood, 1995).
A solution to this discrepancy was later attempted by Wunder and Schreyer (1997)
during further investigation on the MgO – SiO2 –H2O system, with special reference to
antigorite stability under high pressures. However, uncertainty still remains as to the
stability of 10-Å phase in the MSH system.

13.4.4.7. 3.65-Å phase

Sclar and Monzenti (1972) described the synthesis of the 3.65-Å phase at pressures
above 9 GPa and T , 5008C. They opined that, at lower pressures, this phase will break
down into the 10-Å phase. Using energy dispersive analysis, Rice et al. (1989) determined
a Mg/Si atomic ratio of 1.077 and proposed a composition of Mg1.4Si1.3A[6] 1.3(OH)8. This
would result in a substantial substitution of 2Mg2þ for Si4þ þ A (Sclar, 1990).

13.4.5. Anhy-B

Anhy-B of composition (Mg0.88Fe0.12)14Si5O24 has been synthesized at 15 GPa and

1,8008C by Hazen et al. (1992). In this, Si occurs in both 4- and 6-fold coordination and is
distinguished by Si octahedra that share all 12 edges with Mg –Fe octahedra to form a
unique 13-cation cluster (Finger et al., 1991). Extensive Mg –Fe ordering is seen to exist in
anhy-B. Two types of (010) layers are present. Olivine-type layers contain (a) Mg-rich M2,
M5 and M6 octahedra and (b) SiO6 octahedra, which are surrounded by Mg – Fe octahedra.
The Fe-enriched M3 octahedron alone shares edges with two octahedral Si sites
(see Fig. 13.6). Other ordered high-pressure phases, such as Mg12Mn2Si5O24 or even
Mg12Ca2Si5O24, should be considered as possible compositional variants.
A coupled substitution for octahedral (M4) Si by Al and Mg may accomplish size
compensation and the aluminous phase [6](Mg13Al2)[4]Si4O24 may adopt the anhy-B
structure. The stability of the anhy-B is significantly altered by the substitution of Fe
910 Chapter 13

Figure 13.6. In the anhydrous phase B structure two types of (010) layers reveal a possible rationalization for
extensive Mg– Fe ordering. Olivine-type layers (a) contain M2, M5, and M6 octahedra, all enriched in Mg. The
(b) layers, on the other hand, contain Si octahedra surrounded by Mg –Fe octahedra. The Fe-rich site, denoted M3,
is the only octahedron to share edges with two octahedral Si sites (in black). This concentration of Fe creates two
chemically distinct kinds of edge-sharing octahedral strips parallel to a. Predominantly Mg sites form strips with
equally sized M6 or M4 octahedra. Enclosed by these chains are strips in which smaller Si and larger (Mg,Fe)
octahedra alternate. The concentration of larger Fe cations in the latter strips compensates for the smaller Si
octahedra (Hazen et al., 1992, q 1992 Mineralogical Society of America).
Hydrous Minerals 911

for Mg. In such a condition, the stability field of anhy-B may expand to approach that of the
orthosilicates, stable near the P – T conditions of the transition zone.

13.4.5.1. Octahedral sites: M3 site

In anhy-B, Fe is seen to be concentrated in the M3 site — the only octahedron to
share two edges with the Si octahedra. The average distribution coefficient between M3
and the other five octahedral sites in the 12% Fe sample is 6.6 (in the 7% Fe sample it is
greater than 9; Hazen et al., 1992). The KD between M3 and M4 is 8.5, although these two
octahedra share edges.
The Si octahedron is significantly smaller (polyhedral volume ¼ 7.85 Å3) than the
average Mg octahedra (<12 Å3). Si octahedra and M3 octahedra form edge-sharing strips
parallel to the a-crystallographic axis. These strips share edges with similar octahedral
chains composed entirely of M4, the octahedron with the least Fe (Fig. 13.6; Hazen et al.,
1992). The distinctive ordering behaviour of M3 compared with the other five Mg
octahedra suggests that this site may provide a repository of other divalent cations larger
than Mg.

13.4.5.2. Phase B
Phase B, Mg[4] [6]
12 Si3 Si1O19(OH)2, the Mg-richest of all known DHMS phases
(Fig. 13.7), was first synthesized by Ringwood and Major (1967) at pressures above
10 GPa.
According to the synthesis experiments by Akaogi and Akimoto (1980), the
stability of phase B is limited by the equilibria:
Phase B ! phase A þ forsterite þ periclase
and
Phase B ! chondrodite-OH þ forsterite þ periclase
Phase B might break down to an anhydrous structural analogue (anhy-B, Mg[6] [6] [4]
14 Si Si4 O24)
[6] [6] [4]
and/or to the superhy-B Mg10 Si Si2 O14(OH)4 at pressures .16 GPa.
The phase B structure is similar to anhydrous B structure (Fig. 13.6), which has an
orthorhombic (Pmcb) structure. Based on the close packing of O atoms, the structure
contains a six-layer (b ¼ 14.2 Å) stacking sequence. Layers are stacked in 2 –1 –2 – 2– 1– 2
sequence, in which each silicon octahedron shares all 12 edges with adjacent magnesium
octahedra. The hy-B is proposed by Finger et al. (1989) as the major repository for H2O in
the Earth’s mantle.

IR study. For phase B, n of the most intense OH peak decreases to a minimum at 5 GPa
and then rises to a broad maximum ,35 GPa, whereas n of the other intense peak decreases
to a broad minimum ,30 GPa.
The changes in O –O bond length and H – O –O angle calculated from the trends of
OH frequency with increasing pressure are consistent with the relatively incompressible
layer of SiO4 tetrahedra. Widths, areas and heights of the hydroxyl peaks also increase
912 Chapter 13

Figure 13.7. (a,b) Polyhedral projection of the structure of VIMgVI IV

12 Si1 SiO3O19(OH)2, Phase B. The structure has
a two-layer form. There is a “sheared” relationship between adjacent strips (Finger et al., 1987).

with P (whereas areas of the lattice modes remain constant), with H1 being much more
affected.

13.4.5.3. NMR study

The octahedral Si site in phase B shows an unusual coordination environment in
which all the 12 polyhedral edges are shared with MgO6 octahedra (Finger et al., 1991).
This accounts for the chemical shift that is identical for both condrodite and forsterite.
The synthetic phase B of Phillips et al. (1997) gives 29Si SP and 29Si {1H} CP-MAS
spectra with well-resolved peaks for SiIV at 261.8, 264.0, 275.0 and 275.8 ppm, plus a
peak at 2170.4 ppm that is attributed to SiVI (see Section 13.4.2).
Hydrous Minerals 913

In phase B, the crystal structure contains four inequivalent Si positions with equal
multiplicity factor: three SiIV and one SiVI (Finger et al., 1991). Interestingly, the SiVI of
phase B shares one corner with an additional Mg octahedron and shy-B shares two,
whereas their relative shieldings are reversed from the trends observed for SiIV, in which
addition of corner-shared Mg octahedra increases the shielding. It should be remembered
here that structural differences beyond the first coordination sphere affect the shielding
only indirectly, through changes induced in the electronic structure of the Si – O bond
(e.g., Tossell, 1993).
The small line-widths (0.2 – 0.3 ppm FWHM) observed by Phillips et al. (1997)
indicate high ordering on a local scale. The peaks at 275.0 and 275.8 ppm are prominent
at short tc, whereas those at 264.0 and 2170.4 grow gradually with increasing tc, reflecting
a wide range of cross-relaxation times (Table 13.4).
For different Si sites in a phase lacking molecular re-orientation, the rate
of polarization transfer from the 1H spin system to 29Si (T 21 SiH), dependsPlinearly upon
M2(Si –H) (Mehring, 1983). Owing to the dependence of M2(Si – H) on [di(Si –H)]26,
T 21
SiH also correlates with distance from Si to the nearest H (d1(Si – H)). These assignments
are consistent with empirical correlations of chemical shift and crystal structure.
The peaks at þ5.1 and þ3.0 ppm are assigned to phase B in whose structure there
are two inequivalent H positions. The hydroxyl configuration consists of paired OH groups
oriented such that both H are directed towards a third oxygen and lie just off the respective
O – O vectors ð, H – O…O # 58Þ: The H of each hydroxyl pair is crystallographically
distinct and the structure refinement of Finger et al. (1991) gives 1.93 Å for the distance
between them. This short d(H –H) results in strong homonuclear dipolar coupling and is
responsible for the large breadth of the spinning side-band envelopes. Eckert et al. (1988)
showed a good correlation of the hydrogen-bond distance with 1H chemical shift over a
wide variety of structure types.
The dipolar coupling constant determined by Phillips et al. (1997) for phase B
corresponds to d(H –H) ¼ 1.86(2) Å, somewhat shorter than the 1.93(6) Å distance
between the H positions obtained from XRD study (Finger et al., 1991). However, these H
positions give d(O –H) values (0.82 and 0.92 Å) indicative of some apparent X-ray
shortening. Adjustment of the H positions along the respective O – H vectors such that

TABLE 13.4
Assignment of 29Si NMR peaks of phase B based on comparison of 29Si (1H) cross-relaxation times (TSiH) and the
second moment of the dipolar coupling, which correlates with the distance from Si to the nearest H (d1(Si–H))
(from Phillips et al., 1997)

Si position d 29Si (ppm) TSiH (ms) M2(Si–H)a ( £104 Hz2 ) d1(Si–H)a (Å)

Si (1) 2170.4 27 (3) 7.4 4.4

Si (2) 264.0 40 (8) 3.8 4.8
Si (3) 275.8 7.5 (7) 26 3.1
Si (4) 275.0 2.9 (3) 66 2.65
a
Obtained from structural data of Finger et al. (1991).
914 Chapter 13

d(O –H) equals the standard neutron diffraction value of 0.96 Å (Chiari and Ferraris, 1982).
This shortens d(H– H) to 1.84 Å, in good agreement with NMR data (Phillips et al., 1997).
The Raman spectra of phase B and anhy-B (discussed below) are shown in Fig. 13.8.
The peaks in the phase B spectrum, which arise from OH-stretching vibrations, appear at
3.44 and 3.356 cm21 frequencies similar to those seen in IR absorption spectrum.

Phase B and Shy-B. Phase B (Mg12SiVISiIV VI IV

3 O19(OH)2) and Shy-B (Mg10Si Si2 O14
23
(OH)4) are DHMS with densities of 3.368 and 3.327 g cm , respectively (e.g., Pacalo and
Parise, 1992). The Shy-B (Shy-B) belongs to the space group P21 mn (e.g., Hofmeister
et al., 1999).
An assemblage of Shy-B þ stishovite is stable between 15 and 24 GPa at
temperatures of 800 –1,4008C. Shy-B is also stable in the mantle assemblages (e.g.,
Frost, 1999) and is the only hydrous phase stable at pressures between 17 and 25 GPa in
H2O-undersaturated (4 wt% H2O) harzburgite compositions. Shy-B has a higher Mg/Si
ratio and greater proportion of octahedral Si relative to anhy-B.

Figure 13.8. Raman spectra of phase B (top curve) and anhydrous phase B (bottom curve) measured with
457.9-nm laser excitation (,25 mW) and 5-cm21 resolution. The peaks in the phase B spectrum, which arise from
OH-stretching vibrations, appear at 3,414 and 3,356 cm21 frequencies similar to those found in the infrared
absorption spectrum (Finger et al., 1989, q McMillan Magazine Ltd. 1989).
Hydrous Minerals 915

Single crystals of Shy-B and phase E were synthesized by Crichton et al.

(1999) from a sample of ringwoodite plus a molar mixture of brucite and quartz (2 : 1)
held at 18.0 GPa and 1,1508C. Phase E crystals were deep blue in colour, whereas
the Shy-B crystals were clear, suggesting that the bulk of the iron partitioned into
phase E.

Anhydrous B and Shy-B. Under increasing pressure, the volumes of anhy-B and Shy-B
show a smooth decrease. Third-order Birch– Murnaghan EOS were fit to the P – V data
of these two phases and the data obtained are presented in Table 13.5 (from Crichton
et al., 1999). Shy-B is ,6% more compressible than anhy-B. The Grüneisen parameter
of Shy-B is greater than observed in other silicates. The unit-cell parameters of anhy-B
and Shy-B decrease smoothly with increasing pressure but show significant curvature.
The K a0 near 6 in both phases can be attributed to the variation of the a- and c-axes with
pressure.
The isotropic aggregate properties for Shy-B are shown in Table 13.7.

Structure. The structures of these two phases consist of two types of layers: (i) an “O – T”
layer containing octahedrally coordinated Mg (and Fe) and tetrahedrally coordinated Si
and (ii) an “O” layer containing both Mg and Si in octahedral coordination that can be
described as a defect rock-salt layer where vacancies are created by Si substituting for Mg
(Finger et al., 1991).
There are two “O – T” layers for every “O” layer in anhy-B and Shy-B and there
is a six-layer repeat (e.g., O, O – T, O, O –T, O – T and O) along the b-axis. Hydrogen
atoms in Shy-B are located within the O –T layers and share bonds with oxygen atoms
belonging to three different Mg octahedra. The six-coordinated Mg and Si atoms in the
O layer are sandwiched between the octahedral chains of the O –T layers. In both
structures, the octahedrally coordinated Si atoms share all 12 edges with neighbouring
MgO6 octahedra. The O – T layer of anhy-B is similar to the (100) layer of forsterite
(Crichton et al., 1999).
In anhy-B, there are different linkages between polyhedra so that the SiO4
tetrahedra share corners with MgO6 octahedra whereas the other tetrahedron shares two
edges with Mg octahedron, in contrast to three basal SiO4 edges shared with octahedra in
the olivine structure (Finger et al., 1991). The O – T layer in Shy-B is related to the (010)

TABLE 13.5
Elastic parameters of anhy-B and Shy-B

Anhy-B Shy-B
3 3
V0 (Å ) 838.86 ^ 0.04 Å 624.71 ^ 0.03
KTO (GPa) 151.5 ^ 0.9 GPa 142.6 ^ 0.8
K0 5.5 ^ 0.3 5.8 ^ 0.2
x2W 1.5 1.1
DPmax ¼ jPobs 2 Pcalc jmax 0.023 GPa 0.023 GPa
916 Chapter 13

TABLE 13.6
Pressure dependence of unit-cell parameter reflected in bulk moduli of anhy-B and Shy-B (Crichton et al., 1999)

Anhy-B Shy-B

Ka 148 ^ 1 GPa 135 ^ 1 GPa

K 0a 5.9 ^ 0.4 6.4 ^ 0.4
Kb 175 ^ 4 GPa 146 ^ 3 GPa
K 0b 3.4 ^ 1.2 5.2 ^ 0.9
Kc 137 ^ 3 148 ^ 3
Kc0 6.9 ^ 0.9 6.4 ^ 0.9

layer in anhy-B by crystallographic shear along the [001] direction of anhy-B (Pacalo and
Parise, 1992).
Incorporation of water in Shy-B and corresponding structural rearrangements in the
O – T and O layers relative to anhy-B have a profound effect on the EOS and overall
compression of the structure. The isothermal bulk modulus of Shy-B is approximately 6%
lower than anhy-B and both structures have K 0 values significantly greater than 4 (about 6).
The rigid edge-shearing chains of MgO6 and SiO6 octahedra in the O layer control
compressibility along the a- and c-axes in both structures (Crichton et al., 1999).
g-Olivine (spinel) phase with stishovite surrounded by hy-B was investigated by
Kohtstedt et al. (1996). The FTIR (unpolarized) spectra exhibit strong absorption peaks
near 3,408, 3,364, 3,357 and 3,267 cm21. The first two of these peaks and such peaks near
3,410 and 3,350 cm21 were reported to be arising from hy-B (Finger et al., 1989; McMillan
et al., 1991). The Raman spectra of phase B and anhy-B with 457.9-nm laser excitation
(,25 mW) show similar OH vibrations near 3,414 and 3,356 cm21 (see Fig. 13.8).

NMR study. Shy-B appears to give two poorly resolved 1H MAS peaks, consistent with the
presence of two distinct hydrogen pairs in the P21 mn crystal structure. Analysis of its spin-
echo spectrum gives d(H –H) ¼ 1.83(3) Å, slightly shorter than for phase B.
In Shy-B, the SiIV : SIVI ratio is 2 : 1, which is common for such a structure. The
Si : SIVI intensities decrease with increasing contact time (8.5 : 1, 5 : 1 and 2.6 : 1 for
IV

contact times of 2, 5 and 20 ms, respectively). With M2(Si – H) of 100 £ 104 Hz2 for SiIV

TABLE 13.7
Isotropic aggregate properties for Shy-B (from Pacalo and Weidner, 1995)

Reuss HS2 HSþ Voigt VRH

K (GPa) 153.9 154.0 154.1 154.2 154.0 (4.2)

m (GPa) 96.4 96.9 97.0 97.6 97.0 (0.7)
V (km s21) 9.21 9.23 9.23 9.24 9.23 (0.08)
V PS (km s21) 5.38 5.40 5.40 5.42 5.40 (0.02)

Note: The calculated density is r ¼ 3:327 g cm23. Linear compressibilities are ba ¼ 0:230; bb ¼ 0:224 and
bc ¼ 0:195 (102 GPa)21.
Hydrous Minerals 917

and 15 £ 104 Hz2 for SiVI, calculated from the atom positions of Pacalo and Parise (1992),
the measured cross-relaxation times fall along the same trend as those for phase B.
The SiVI chemical shift observed by Phillips et al. (1997) for Shy-B was 3.8 ppm
less shielded than that of phase B; and this is the most positive chemical shift for SiVI yet
reported. Shy-B, A, and E all show peaks at 274.6 and 2166.6. Kanzaki et al. (1992)
reported for phase A narrow peaks at 264.0 and 270.6 ppm, and for the E broad peak
centred near 274.6 ppm.
The similarity of CP dynamics displayed by the Shy-B peak at 274.6 ppm and for
phase B at 275.0 ppm (both resonances display TSiH of ,2.5 ms and T1P ðHÞ p TSiH )
makes their CP-MAS intensities approximately proportional to the concentration at contact
times satisfying TSiH p tc # T1P ðHÞ:

Element partitioning in perovskites and Shy-B. The partitioning of Si, Mg and Ca in

major concentrations and Al, Ti, Sc and Sm in minor concentrations among MgSiO3
perovskite, CaSiO3 perovskite, Shy-B [Mg10Si3O14(OH,F)34], and melt containing H2O and
F, were determined by Gasparik and Darke (1995) at 23 GPa and 1,500 –1,6008C.
For Sc, the MgSiO3 perovskite is seen as the main repository, while Ti and Sm are
primarily concentrated in CaSiO3 perovskite. Superphase B excludes all cations other than
Mg and Si. When the melt is rich in volatiles, Ti and Al become compatible in solid
perovskite.
Volatile-bearing melt is conspicuous by its high Mg content and the low content of
all other cations. This observation raises the possibility that the Mg/Si ratio of the Earth’s
upper mantle, which is related to that of most classes of primitive meteorites, was
established by upward segregation of Mg-rich volatile-bearing melts from the lower to the
upper mantle. The most plausible scenario involves degassing of the lower mantle by
removal of 5% of the melt produced by a low degree of partial melting.

13.4.6. Phase D (MgSi2H2O6)

A humite mineral, hydroxyl condrodite, has been described (by Yamamoto and
Akimoto, 1977) as phase D. Liu (1987) synthesized it from serpentine under P . 22 GPa
and laser heating in a diamond-anvil cell. Phase D is the highest pressure DHMS, being
stable between 17 and ,50 GPa and at temperatures up to 1,4008C (Frost and Fei, 1998;
Shieh et al., 1998).
In phase D, all Si are in octahedral coordination. Its density is 3.5 g cm23, the
highest of all DHMS phases. It may be an important mineral within the subducting
lithosphere in the transition zone and lower mantle. The structure consists of alternating
layers of SiO6 and MgO6 octahedra stacked along the c-axis. One in three of the
octahedra in the SiO6 layer and two out of three octahedral positions in the MgO6 layer are
vacant. OH bonds occur between adjacent octahedra in the MgO6 layer (Fig. 13.9)
(Yang et al., 1997).
Phase D has only SiO6 structural units, manifested by the presence of a 700-cm21
band in IR reflectance spectrum, with an absence of bands between 1,000 and
1,500 cm21, typical for SiO4 tetrahedra (present in crustal and upper-mantle minerals).
918 Chapter 13

Figure 13.9. Crystal structure of phase D projected along c. Shaded and unshaded octahedra represent SiO6 and
MgO6 octahedra, respectively. Large spheres represent Mg and small ones H. (Yang et al., 1997d, q 1997
Mineralogical Society of America).

In addition, a low hydroxyl frequency is present at ,2,850 cm21, corresponding to an

OH· · · bond length of 2.67 Å (Novak, 1974); this is reconfirmed by X-ray structure
refinement (Yaug et al., 1997). The degree of compression of the short oxygen – oxygen
distance in phase D (rðO – H· · ·OÞ ¼ 2:67 Å) is estimated to decrease to 2.54 Å at
30 GPa and to 2.51 Å at 50 GPa. This suggests the existence of very anharmonic and
possibly diffusive behaviour of the protons at high P –T conditions prior to breakdown
(Williams and Hemley, 2001).
Phase D is seen to be stable to lower-mantle conditions (Ohtani et al., 2000) and is
reported to co-exist with silicate perovskite and stishovite (Li and Jeanloz, 1991). It is seen
to decompose above 50 GPa and 2,100 K, releasing H2O above this P and T. Thus, if
Shy-B is stable to pressures at the bottom of the transition zone and top of the lower
mantle, it may also be stable to temperatures near those of the geotherm in the deeper
portion of the transition zone (e.g., Gasparik and Drake, 1995).
The calculated O – (H) – O distance is consistent with the frequency of the OH
stretching vibration (2,850 cm21) of phase D (Frost and Fei, 1998). The hydrogen bonding
in phase D displaces the major OH stretching vibration to a lower wave number than in
other DHMS phases.
Phase D was reported by Liu (1986, 1987) as the breakdown product of serpentine
at 22 GPa and 1,0008C. Yang et al. (1997) synthesized phase D at 20 GPa and 1,2008C.
The ideal formula of it is MgSi 2H2O6 with cell parameters a ¼ 4:7453ð4Þ Å,
c ¼ 4:3450ð5Þ Å and S.G. P31m:
Frost and Fei (1998) determined the stability of phase D in a Mg2SiO4 þ 20.5 wt%
H2O composition between 16 and 25 GPa at 900 –1,4008C. Phase D co-exists with Shy-B
and a Mg-rich liquid to temperatures of 1,0008C at 17 GPa and 1,4008C at 26 GPa. The
experiment confirmed that phase D is stable to 50 GPa at 9308C.
Hydrous Minerals 919

Figure 13.10. The stability of phase D determined from both multi-anvil (solid circles) and DAC experiments
(shaded circles). Triangles mark conditions where phase D is not observed. An average mantle adiabat, assuming
whole mantle convection (Jeanloz and Morris, 1986), and an estimate of the range of temperatures within the top
10 km of the subducting slab (Irifune et al., 1996) are also shown (Frost and Fei, 1998, q 1998 American
Geophysical Union).

The HP and HT stability of phase D makes it most suitable to be stable within

subducting lithosphere, in the transition zone and in the lower mantle. A positive P – T
stability slope of phase D suggests that it may be stable at temperatures much greater than
1,4008C in the lower mantle. The high density (3.50 g cm23) and HP stability make phase
D the prime phase to be stable in the transition zone and lower mantle. The stability of
phase D in the subducting slab with respect to the transition zone and mantle adiabat
is shown in Fig. 13.10 (Frost and Fei, 1998).

13.4.6.1. Structure
The structure consists of an hexagonal closest-packed array of O atoms. In its
structure, Si occupies the octahedral sites in a layer similar to that of brucite, but one in
every three octahedra is vacant. The MgO6 octahedra are located above and below each
vacant octahedral site. Between SiO6 octahedral layers, all O –H bonding occurs. This is
reportedly the only hydrous Mg-silicate phase in which all Si are octahedrally coordinated.
The MgO6 and SiO6 octahedra occur in two separate layers stacked along c. The MgO6
octahedra are located above each vacant octahedron in the SiO6 octahedral layers. Owing
to edge sharing, the SiO6 octahedron is considerably distorted in terms of octahedral angle
variance (OAC ¼ 29.2; Robinson et al., 1971). The powder XRD data for phase D
obtained by Yang et al. (1997) are presented in Table 13.8.
920 Chapter 13

TABLE 13.8
Powder diffraction data for phase D

h k I Liu (1987) Yang et al. Ohtani et al. Frost and Fei

(1997) (1997) (1998)
d I d I d I d I

0 0 1 4.3 15 4.345 23 4.341 32

1 0 0 4.110 16 4.131 10
1 0 1 3.00 100 2.986 100 3.00 s 2.992 100
1 1 0 2.373 10 2.38 vw 2.385 6
0 0 2 2.173 0
1 1 1}
2 21 1 2.082 82 2.09 s 2.090 53
2 0 0 2.055 5
1 0 2 1.921 0
2 0 1 1.858 26 1.87 m 1.865 12
1 1 2 1.61 80 1.602 76 1.61 s 1.605 36
2 21 2}
2 1 0 1.553 0
2 0 2 1.493 2 1.51 vvw 1.487 7
2 1 1 1.496 20 1.463 25 1.47 vw 1.465 10
3 21 1}
0 0 3 1.448 0
3 0 0 1.390 20 1.370 33 1.35 vvw 1.372 14

Note: The data were indexed on a trigonal unit cell.

Synthesis (Frost and Fei, 1999). A mixture of reagent-grade SiO2 and Mg(OH)2 (molar
ratio 3 : 2) and phase D (with stishovite) was subjected to 17 GPa and 9008C in multi-anvil
apparatus by Frost and Fei (1999). Powder XRD analysis of the run product showed it to
comprise phase D with stishovite. Electron microprobe analysis of phase D in the run
product showed the composition to be Mg1.11Si1.6H3.4O6 where the H2O content is
estimated from the loss in the microprobe analysis totals with an uncertainty of the order of
0.3 formula units (2 wt%). A rhenium gasket was compressed between 500-mm flat-
diamond anvils. A neon-gas pressure medium was loaded into the sample chamber at
200 MPa using a high-pressure gas loading device (Jephcoat et al., 1987). Energy-
dispersive XRD measurements were made using polychromatic wiggler synchrotron
radiation at the X17 C beam line of the National Synchrotron Light Source, Brookhaven
National Laboratory. A solid-state germanium detector collected the diffracted X-rays at a
fixed 2u angle of 12.9918 (^ 0.0058). The angle was calibrated using a platinum and an
Al2O3 standard.
Diffraction data were collected during compression of the sample at room
temperature. The unit-cell parameters of phase D and gold were determined using a full-
pattern le Bail refinement, performed by using the GSAS structure-analysis software
(Larson and Von Dreele, 1986). Full structure models were used in the initial refinements
of each pattern, then the le Bail method was employed in the final stages to account for the
Hydrous Minerals 921

effects of preferred orientation. The pressure was calibrated from the unit cell of gold using
the EOS of Heinz and Jeanloz (1989). Uncertainties in the pressure were determined by
propagating the standard deviations in the measured unit-cell parameters for gold through
this EOS. Typically, this certainty is ^ 0.3 GPa (at 2 GPa) at 30 GPa.
A least squares fit of the Birch –Murnaghan third-order EOS was used to extract the
EOS parameters of phase D. The fit was weighted by the standard deviations in the unit-
cell volumes, including the uncertainty in the 2u angle, and by the previously described
uncertainty in the pressure.

13.4.6.2. Density and bulk modulus

The calculated density of phase D seems to be highest amongst other high-pressure
hydrous magnesium silicates known so far:

Phases Density (calc.) (g cm23) Authors

A 2.96 Horiuchi et al. (1979)

B 3.37 Finger et al. (1991)
B (superhydrous) 3.21 Pacalo and Parise (1992)
D 3.50 Yang et al. (1997)
E 2.82 Kudoh et al. (1993)
F (?) 2.83 Kudoh et al. (1995)

The high density of phase D is primarily a consequence of all Si being in

octahedral coordination. The bulk moduli of these phases (Kudoh et al., 1995) are
tabulated below:

Phases Bulk modulus Authors

A 97 Kudoh et al. (1995)

B 163 Kudoh et al. (1995)
B (superhydrous) 143 Kudoh et al. (1995)
D 200 Frost and Fei (1999)
E 112 Bass et al. (1991)
F (?) 143 Kudoh et al. (1995)

Again, for phase D, the bulk modulus is also the highest amongst the DHMS phases.
Given the large bulk modulus, high density, large H content and stability at high P – T
phase D comes out as a prime candidate for H2O storage in the transition zone and lower
mantle.
Phase D shows significant high-pressure stability at high temperature and its bulk
modulus is the highest of the DHMS. Because all its Si are in octahedral coordination
contributing to stiff elastic properties (especially along the a-direction), this hydrous phase
seems to continue its stable existence in the lower mantle. In the lower-mantle assemblage,
the dehydration may take place as:

0:49MgO þ Mg1:11 Si1:6 H3:4 O6 ¼ 1:6MgSiO3 þ 1:7H2 O

922 Chapter 13

TABLE 13.9
Calculated volumes of lower-mantle phases (from Frost and Fei, 1999)

K0t K0 V0 cm3 mol21 V, 30 GPa, 300 K V, 30 GPa, 1,200 K

Phase D 166 4.1 51.58 44.99 45.93

MgO 160.3 4.13 11.25 9.78 9.99
Perovskite 261 4.0 24.5 22.28 22.51
H2O 23.7 4.15 12.3 7.68 8.99
DV 3.1 21.1 0.5

Thermophysical data for MgO and H2O are from Fei and Mao (1993) and for perovskite from Mao et al. (1991). The
expansivity of brucite (Fei and Mao, 1993) provides a conservative estimation for the expansivity of phase D and
the EOS of Brodholt and Wood (1993) is used to calculate the volume of H2O at high pressure and temperature.

Table 13.9 (Frost and Fei, 1999) shows a calculation of the volume change of this
reaction at ambient conditions at 30 GPa, 300 K and at 30 GPa, 1,200 K.
However, calculations show that, at 1,200 K, a lower-mantle assemblage contain-
ing phase D is marginally denser than an assemblage containing a free fluid phase.

13.4.6.3. Anisotropic compressibility

Layered minerals are reported to manifest anisotropic compression arising from the
difference in compressibility within and between layers, reflecting different bond strengths.
In phase D, this arises due to the presence of strong SiO6 octahedra within layers but
weaker MgO6 octahedra and OH bonds between layers.
The unit-cell parameters of phase D display anisotropic compression behaviour
(Fig. 13.11a,b). The c-axis is more compressible than the a-axis with the c/a ratio
decreasing by 1.5% to 20 GPa. A weighted least squares fit to the axial compression data
yielded Ka0 ¼ 641(37) GPa, showing the c-axis to be twice as compressible as the a-axis.
Brucite also displays more pronounced anisotropic compression with c-axis
direction 5-fold as compressible as a (Fei and Mao, 1993). The c/a ratio of brucite also
becomes pressure independent at very similar conditions (,25 GPa) to phase D (Frost and
Fei, 1999).

13.4.7. Phase E (Mg2.08Si1.6H3.2O6)

Above pressures of 13 GPa and 1,0008C, Kanzaki (1991) discovered a new phase of
hydrous Mg-silicate, which was named phase E in accordance with other such alphabetical
phases studied earlier by Ringwood and Major (1967). This high-pressure phase E may or
may not be a component of the mantle but this represents a fascinating new class of
discovered materials with a highly compliant structure.
Phase E was synthesized in a uniaxial split-sphere (multi-anvil apparatus) using a
stoichiometric mixture of high-purity SiO2 and 2Mg(OH)2. Phase E is non-stoichiometric
and has a layer-type structure similar to phase D. Under pressure, the c/a ratio of phase E
remains unchanged while, for phase D, a negative slope is reported (Frost and Fei, 1998).
Phase E was also reported as a HP break-down sub-solidus phase in water-saturated
Hydrous Minerals 923

Figure 13.11. ((a) and (b)) The variation with pressure of a V=V0 and b a=a0 ; c=c0 for phase D. The curves show
the least squares fit of the Birch–Murnaghan equation of state.

peridotite KLB-1 (e.g., Kawamoto and Holloway, 1997). Thus, phase E could be a
potential carrier of water in the subducted oceanic lithosphere between approximately 350
and 500 km in a sufficiently cold subducting regime in which serpentine forms DHMS
phases (cf. Irifune et al., 1998).
In situ HP XRD study of phase E by Shieh et al. (2000) up to 14.5 GPa shows that
phase E is stable over this pressure range at room temperature. With r ¼ 2:92 g cm23 and
KTO ¼ 93 ^ 4 GPa, this phase has the lowest density and bulk modulus of the DHMS
reported to date (Table 2, Fig. 6; Shieh et al., 2000). A plot of the normalized pressure (F)
as a function of Eulerian strain ( f) (Birch, 1978) (see Sections 2.6.1.2 and 5.9.6) for phase
E is given in Fig. 13.12.
The X-ray photographs of the phase showed systematic absence for reflections for
hkl with 2h þ k þ l ¼ 3n; corresponding to a rhombohedral lattice of space group R3; R 3;



R3m or R 3m: The diffuse scattering, and hence the short-range ordering pattern, obeys the
3
2=m symmetry of the sub-structure. The phase E does not possess a long-range
superstructure (hence rather disordered) but does have short-range order.
In the structure, both Mg and Si are distributed statistically throughout an
essentially close-packed oxygen net and all the three cation positions Mg1, Mg2 and Si are
partially occupied. The two synthesized samples of phase E, noted as E1 and E2, showed a
compositional relationship (Kudoh et al., 1993):

E1 þ 0:42H2 O þ 0:09MgO ¼ E2 þ 0:15SiO2 þ 0:10O2

924 Chapter 13

Figure 13.12. Normalized pressure ðFÞ as a function of Eulerian strain ðf Þ for phase E (Shieh et al., 2000, q 2000
Mineralogical Society of America).

In these phases, all Si is proved to be in isolated tetrahedral coordination, as determined by

employing 29Si magic-angle spinning (MAS) and cross-polarization MAS NMR
techniques (see Section 13.2.2).
All the crystal fragments of these samples produced diffuse bands of intensity
between the Bragg positions in SAED patterns. The observed three-dimensional walls of
diffuse scattering are characteristic of short-range order (SRO) (Cowley, 1981, Chapter
17). These have been observed in a large number of materials, including Cu –Au, Cu – Pd
and Cu –Pt alloys.

13.4.7.1. Related phases

Several new phases related to phase E have been synthesized by Prewitt and his
colleagues at GL, which can be distinguished by their characteristic colour, as shown
below:

Phase S.G. Composition Synthesis condition (P, T )

E

R 3m Mg2.40Si1.31O6H2.36 16 GPa, 1,0008C
Brown P 63 =mmc Mg2.29Fe3þ
0.60Si1.01O6H2.15 14 GPa, 1,4008C
Green P61 22; P65 22 Mg2.13Fe3þ
0.59Si0.87O6H2.52 14 GPa, 1,4008C
Dark green R32; R3m; R 3m
Mg1.96Fe3þ
0.87Si0.63O6H3.05 14 GPa, 1,4008C

NMR spectroscopy: phases A and E. There is close agreement between the 29Si NMR
data for phases A and E provided by Kanzaki et al. (1992) and Phillips et al. (1997): the
29
Si chemical shifts of 263.9 and 270.6 ppm, respectively, for Si(2) and Si(1) sites of
phase A and a broad peak at 275.7 ppm in 29Si-enriched phase E. A broad line is expected
from phase E for the disorder and the structural geometry of the brucite-like layers on
Hydrous Minerals 925

which its structure is based (Kudoh et al., 1993). Phase A may contain strongly-coupled
hydroxyl pairs similar to those in B and Shy-B.

13.4.7.2. Phase F(?)

A new hydrous magnesian silicate reportedly synthesized at 17 GPa (1,0008C) and
called phase F by Kudoh et al. (1995) was later withdrawn by them (1997) because of an
inherent error detected later.

13.4.8. Phase G and other MSH phases

In an experiment on the diopside –jadeite join at 1,5008C and 22 GPa in the

presence of H2O, a new Na-bearing hydrous phase, Na7(Ca,Mg)3AlSi5O9(OH)18, appeared
in place of pyroxene (Gasparik, 1996). This was called phase G (corresponding to an
equimolar mixture of enstatite, diopside and jadeite) and was considered to have the
composition Na(Ca,Mg)4AlSi6O8. Its fully hydrated analogue is called superhydrous
phase G, which is considered to be of composition Na13(Ca,Mg)2AlSi4O36 plus hydroxyl
component. Hydrous melts at these high pressures are seen to be typically high in Fe3þ
contents (Inoue, 1994).

13.5. Humite group minerals

Humites, Mg2SiO4·Mg(F,OH)2, are possibly formed as a secondary product

through hydration of olivine (unlike b- and g-phases which are primary minerals;
Thompson, 1992) and serve as important transporters of water in cooler slabs, perhaps
down to the transition zone. Evidently, humite-group minerals can be the abundant hosts
for water in the upper mantle. Hydration of olivine to clinohumite is seen to lower the
elastic (bulk and shear) moduli (Fritzer and Bass, 1997).
Humite minerals show strong structural similarities with olivine. Although the
structures of humite-group minerals were generally viewed as alternating olivine-like
Mg2SiO3(OH,F) layers and brucite Mg(OH,F)O layers (e.g., Williams, 1992) but the more
valid structure is considered to be composed of two OH groups incorporated into the
brucite Mg(OH)6 octahedral layers linked through strong SiO4 tetrahedra. Hence, there are
no brucite sheets in the humites. The key structural features of both humites and olivine are
zigzag chains of edge-sharing M2þO4(O,OH,F)2 octahedra aligned along the a-axis and
linked laterally by SiO4 tetrahedra.

13.5.1. Clinohumite and chondrodite

The most probable hosts for water of the humite group minerals are clinohumite,
4Mg2SiO4·Mg(OH)2, and chondrodite, 2Mg2SiO4·Mg(OH)2. These minerals bear strong
chemical and structural similarities to olivine, from which they can form by direct
hydration in water-rich environments, e.g., by absorption of H2O from dehydrating
minerals or crystallization from low-temperature water-saturated melts. Alternatively,
926 Chapter 13

chondrodite may form from reactions such as:

phase A ðMg7 Si2 O8 ðOHÞ6 Þ þ olivine ðMg2 SiO4 Þ ¼ chondrodite

or olivine þ water ¼ pyroxene ðMgSiO3 Þ þ chondrodite

depending on the quantity of water available.
Chondrodite is monoclinic, with S.G. P 21 =c: The unit-cell parameters of a natural
chondrodite obtained by Sinogeikin and Bass (1999) are:
a ¼ 7.866(5) Å
b ¼ 4.725(3) Å
c ¼ 10.262(5) Å
V0 ¼ 360.5(5) Å3
b ¼ 1098080 (2)
density r ¼ 3.227(10) g cm23
The idealized structure of olivine, clinohumite and chondrodite can be shown in
planes normal to the b-axis and the coordinated system can be rotated so that the crystal
axes of olivine are aligned with analogous crystal-structural features of chondrodite
(Ribbe, 1980).
Chondrodite is shown to be stable to pressures up to 12 – 17 GPa but only at
temperatures below 1,2008C (Burnley and Navrotsky, 1996). Its presence is, therefore,
likely to be restricted to low-temperature regions such as the “seismic lid” — the seismic
lithosphere characterized by high seismic velocity — of the upper mantle, the cold
subducting slabs and mantle wedges above slabs.
Titano-clinohumite and titano-chondrodite are seen in mantle-derived kimberlite
rocks (Aoki et al., 1976). These are believed to have crystallized from hydrous kimberlite
magma at ,100– 120-km depth and temperature ranging between 900 and 1,0008C.

13.5.2. Elastic properties

The best-fit single-crystal elastic moduli of chondrodite, with uncertainties

calculated from the scatter of the acoustic velocities about the model values are given in
Table 13.10 (Sinogeikin and Bass, 1999) and olivine (Kumazawa and Anderson, 1969) for
comparison.
The b-axis direction (constant C22, Table 13.10) is the stiffest in olivine and humite-
group minerals. The longitudinal constant, C22, is seen to decrease from olivine to
chondrodite, while other longitudinal constants, C11 and C33, are less affected. In olivine,
every eighth tetrahedral site is occupied by Si, whereas only one tenth of the tetrahedral
sites are occupied by Si in chondrodite. Therefore, the number of strong columns per unit
volume is less in humites. However, the elasticity of the main structural blocks,
M2þO5(OH,F) and M2þO4(OH,F)2 octahedra, are likely to be similar to the elasticity of
M2þO6 octahedra. It may be that replacement of much or all of the upper-mantle olivine
with chondrodite or clinohumite would not be obvious as a seismic low-velocity zone
(LVZ). (Seismic-wave attenuation (or Q 21) may prove to be a more sensitive indicator of
water content in the mantle.)
Hydrous Minerals 927

TABLE 13.10
Single-crystal elastic properties of humites and olivine

Cij (GPa) Chondrodite Clinohumite Olivines

(Sinogeikin and Bass, 1999) (Kumazawa and Anderson, 1969) (Kumazawa and Anderson, 1969)

C11(C33) 213.4(15) 212(2) 235.1

C22(C11) 275.3(15) 296(2) 323.7
C33(C22) 198.4(12) 191(2) 197.6
C44(C66) 69.7(6) 72.0(8) 79.04
C55(C44) 72.1(9) 65(1) 64.6
C66(C55) 75.2(7) 74.3(8) 78.05
C12(C13) 69.6(32) 66(3) 71.6
C13(C23) 59.0(24) 80(4) 75.6
C23(C12) 66.6(25) 72(7) 66.4
C15(C14) 7.2(10) 20.3(6)
C25(C24) 21.7(12) 1(2)
C35(C34) 22.6(8) 2(1)
C46(C56) 20.7(4) 20.6(7)

13.5.3. Clinohumite-OH and chondrodite-OH

The H2O pressure and T stability fields of clinohumite – OH (Mg9Si4O16(OH)2) and

chondrodite – OH (Mg5Si2O8(OH)2) and phase A have been determined by Wunder (1998)
in reversed equilibrium experiments up to 10 GPa within the system MgO – SiO2 –H2O.
These two monoclinic OH end-members of humite group clinohumite – OH
(Mg9Si4O16(OH)2) were first synthesized by Yamamoto and Akimoto (1977). The stability
fields of these two phases seem to overlap to a large extent. At high pressures, the stability
of chondrodite –OH gets delimited by the equilibrium relations:

forsterite þ phase A ! chondrodite – OH

3 phase B ! 1 phase A þ 10 forsterite þ 9 periclase

1phase B ! 1 chondrodite – OH þ 2 forsterite þ 3 pericalse

The 1H MAS-NMR spectrum of synthetic hydroxyl chondrodite contains a single

Lorentzian-shaped peak at þ11.1(1) ppm with FWHM ¼ 1.4 ppm at a MAS rate of
12 kHz. This indicates that the H occurs at a single well-defined structural position.
The 29Si{1H}CP-MAS intensity of chondrodite varies with contact time when
relaxation times are TSiH ¼ 1.1(3) ms and T1P(H) ¼ 15(1) ms.

13.6. MgO – Na2O – SiO2 –H2O system: hydrated aenigmatites

Aenigmatite-group minerals occur commonly in volcanic plutonic, metamorphic

and metasomatic rocks and also in meteorites (e.g., Deer et al., 1997). These can be
928 Chapter 13

presented by a general formula, A2B6T6O20, where A cations are 7- or 8-fold coordinated

Na and Ca, and B cations are octahedrally coordinated Si, Al, B and Be.
In the aenigmatite-type structure, there are two crystallographically distinct A sites
(M8 and M9), seven B sites (M1 –M7) and six T sites (T1 –T6). Several high-temperature
synthetic phases have been reported to possess aenigmatic-type structure. A review of the
minerals of the aenigmatite – rhönite group is available (Kunzmann, 1999).
Recently, some compounds of the system MgO – Na2O – SiO2(^H2O) showing
aenigmatite-type structure have been synthesized; e.g., Na2Mg4þxFe2þ 222xSi6þxO20
(with x ¼ 0.25 –0.60 at 13– 14 GPa and 1,450 –1,6008C; Gasparik et al., 1999) and
Na2Mg6Si6O18(OH)2 (at 10 GPa and 1,2508C; Yang and Konzett, 2000).
The high-temperature sodic aenigmatites that have recently been synthesized and
investigated are listed as:
Na2(Fe5Ti)[Si6O18]O2 (aenigmatite; Cannillo et al., 1971)
Na2(Mg4Cr2)[Si6O18]O2 (krinovite; Bonaccorsi et al., 1989)
Na2Mg4þxFe3þ 222xSi6þxO20 (x ¼ 0.25– 0.60; Gasparik et al., 1999)
Na2(Mg,Fe)6(Ge,Fe)6O20 (NaMgFe – germanate; Barbier, 1995)

13.6.1. Hydrated-Na – aenigmatite: crystal structure

A myriad of structures with MgO – Na2O – SiO2 composition exist because of the
small sizes of Mg and Si and trade-offs in octahedral and tetrahedral coordination for both.
These form a host of DHMS minerals.
The crystal structure of Na2Mg6Si6O18(OH)2, synthesized by Yang and Konzett
(2000), is characterized by two types of polyhedral layers stacking alternately along [011]:
one layer is formed by slabs of B-octahedra (M3 –M7) linked by bands of A-polyhedra
(M8 and M9) and the other by open-branched single tetrahedral chains (Lieban, 1985)
connected by isolated B-octahedra (M1 and M2) (Fig. 13.13; Yang and Konzett, 2000).
The structure can be described as an ordered 1:1 intergrowth of spinel-type and pyroxene-
type structural units. The spinel-type slabs contain M1, M2 and M7 octahedra plus T5 and
T6 tetrahedra and the pyroxene-type slabs contain M3 –M6, M8 and M9 octahedra plus
T1 –T4 tetrahedra (Barbier, 1995). The pyroxene – spinel interface provides the locations
for Hþ (Fig. 13.13). However, the population of the O4 and O14 sites in the natural
minerals could occur through vacancies in the Na site, or as:

Fe3þ þ O22 ¼ Fe2þ þ OH2 or

Al3þ þ OH2 ¼ Si4þ þ O22 etc:

13.7. CaO –Al2O3 – SiO2 –H2O system

13.7.1. Zoisite and clinozoisite, Ca2Al2 (Al12pFep) (O/OH/Si2O7/SiO4)

Zoisite is a common mineral in eclogite and amphibolite facies rocks derived from
metamorphism of basaltic rocks (e.g., Franz and Selverstone, 1992). Zoisite contains
Hydrous Minerals 929

Figure 13.13. The crystal structure of Na2Mg6Si6O18(OH)2. All octahedra are occupied by Mg and tetrahedra by
Si. The large spheres represent Na cations and small black ones represent H. The spinel and pyroxene-type slabs
are labelled as “Sp” and “Py”, respectively (Yang and Konzett, 2000, q 2000 Mineralogical Society of America).

chains of edge-sharing AlO6 octahedra running parallel to b. In these chains, M1 octahedra

share three edges with neighbouring octahedra and M2 octahedra share two edges. The
chains are cross-linked by corner-sharing Si2O7 tetrahedral pairs and single SiO4
tetrahedra parallel to c. Ca occupies large, irregular cavities between chains. H is bonded to
an O of the octahedral chain and also may be hydrogen-bonded to an O of the adjacent
chain (Dollase, 1968). Schmidt and Poli (1994) experimentally determined its high-
pressure stability at <6 –7 GPa and ,1,0008C, indicating that it is capable of transporting
H2O to a depth well beyond the source of subduction-zone volcanism.
Zoisite and clinozoisite are Fe-poor members of the epidote group and are seen in
epidote – amphibolite facies. They also occur in both hydrothermal systems and high-P
eclogite facies terrain. These two, together with other hydrous phases such as antigorite,
lawsonite, chloritoid, talc, staurolite and phengite, constitute a good assemblage of
minerals for the transport of H2O to pressures higher than that of the amphibolite stability
field. They crystallize either in orthorhombic (zoisite, space group Pnma) or in monoclinic
(clinozoisite, space group P21 =m) form with the formula Ca2Al3FepSi3O12OH, where p is
,0.04 (Kvick et al., 1988). Zoisite shows a more restricted chemical range than
clinozoisite. The position of the polymorphic zoisite – clinozoisite reaction can provide a
useful indicator of P –T conditions attending metamorphism.
The epidote structure essentially consists of chains of edge-sharing octahedra
parallel to the b-axis. The chains are connected by both single tetrahedra (SiO4) and pairs
of tetrahedra (Si2O7). Although the M1 and M2 octahedra are completely occupied by Al
atoms, the main substitution of Fe for Al involves the M3 octahedral site, which is
distinctly different from M1 and M2.
930 Chapter 13

The H atom is uniquely located on the O 10 atom (O10 – H ¼ 0.69 Å). It is involved
in a bent hydrogen bond with the O4 atom (O4 –O10 ¼ 2.89 Å), as shown by a neutron
diffraction study at 15 K (Kvich et al., 1988).
In epidote structure, there is a closed sequence of building blocks along the a-axis.
The blocks are linked by tetrahedral corner sharing along the c-axis by means of variations
in the Si 1– O9 – Si 2 angle, which changes from 162.98 at 0.5 kbar to 158.18 at 42 kbar
(Comodi and Zanazzi, 1998).
A displacement by 1/4[001] on (001) planes between clinozoisite unit-cell modules
yields the zoisite structure. The polytypes can be interchanged by introducing a shear
between the various stacking modules.

13.7.1.1. Compressibility
Pressure is always seen to increase with temperature as the sample expands. The
isothermal bulk modulus, KT, and its pressure derivative, K 0 , are commonly derived by
fitting the Murnaghan equation to compressibility data
0
VP;T =V0;T ¼ ð1 þ K 0 =KT ·PÞ21=K

where VP;T is the unit-cell volume at pressure, P (in GPa) and temperature, T (in K) and
V0,T is the volume at 0 GPa and temperature T.
From least-squares fit of their experimental data on zoisite, Pawley et al. (1998)
obtained K303 ¼ 127(4) GPa and V0,303 ¼ 909.0(7) Å3. This value of V0,303 was used to
obtain Vp,303/V0,303. Linear fits of the individual cell parameters relative to their values at
0 GPa (Fig. 13.14) gave compressibilities:

ba ¼ 0:0016ð1Þ GPa21 ; bb ¼ 0:0026ð1Þ GPa21 ; bc ¼ 0:0030ð1ÞGPa21

[ b c . bb . ba

This relation holds for thermal expansivity as well. However, the bulk modulus obtained
by Holland et al. (1996) was much higher: 279(9) GPa (for K 0 ¼ 4).
Comodi and Zanazzi (1997) measured the compressibility of zoisite and
clinozoisite using single-crystal XRD in a DAC up to ,5 GPa. The unit-cell parameters
were seen to vary linearly in anisotropic pattern. The average compressibility coefficients
obtained by linear regression are shown in Table 13.11.
The compression mechanism includes both shrinking of the polyhedra (i.e.,
octahedra and Ca polyhedra) and tilting of the Si2O7 group, with reduction of the Si – O –Si
angle. A geometric stability of clinozoisite can be constructed on a P – T gradient.
Assuming that the derivatives da=dP and db=dT are zero, that the av value is
3.0 £ 1025 K21 and that the bv value measured for clinozoisite is 7.7 £ 1024 kbar21, then
the calculated P=T gradient is 38 bar K21, close to the 34 bar K21 measured for lawsonite
(Comodi and Zanazzi, 1996). The geothermal gradient is 20 bar K21 (Angel et al., 1988).
Therefore, clinozoisite should remain structurally stable with geothermal gradients of
108 km21, endorsing the experimental petrology.
Hydrous Minerals 931

Figure 13.14. Compressibility of zoisite at <303 K. Error bars are 95% confidence limits on the cell refinements.
Curve through volume data is the least squares fit to the Murnaghan equation; the other lines are linear fits to the
individual cell parameters. Data points are normalized to 0 GPa values obtained from the fits (Pawley et al., 1998,
q 1998 Mineralogical Society of America).

A single-crystal of zoisite was investigated under pressure by Winkler et al. (1989)

using infrared spectroscopy. A band related to the OH-stretching vibration displayed a
linear shift from 3,170 cm21 at 1 bar to 2,795 cm21 at 11.6 GPa (Fig. 13.15; Winkler et al.,
1989). The half-band width increased linearly with respect to pressure from 60 cm21 at
1 bar to 500 cm21 at 11.6 GPa.

13.7.1.2. Thermal expansivity

Pawley et al. (1999, 1998) measured the thermal expansivity of zoisite. The volumes
measured up to 7508C were linear in temperature. Their data were fit by the equation:

V0;T =V0;298 ¼ 1 þ aðT 2 298Þ

Giving a constant value for the coefficient of thermal expansion a of 3.86(5) £ 1025 K21,
the derivation of the dKT =dT is obtained, as discussed below.

TABLE 13.11
Linear compressibility coefficients for clinozoisite and zoisite (Comodi and Zanazzi, 1997)

Cell-edge Principal-strain Orientation of

compressibilities coefficients ( £ 1024 kbar21) principal coeffici-
( £ 1024 kbar21) ents (angle (8) with
respect to axes)
ba bb bc b1 b2 b3 b1 b2 b3

Clinozoisite 2.1(1) 2.8(1) 3.3(1) 2.0(1) 2.7(1) 3.3(2) 12 ;b 12

Zoisite 2.3(2) 2.9(1) 3.7(2) 2.3(2) 2.9(1) 3.7(2) ;a ;b ;c
932 Chapter 13

Figure 13.15. Position of the nOH band in cm21 as a function of pressure in kbar. The shift is linear over the whole
range measured (Winkler et al., 1989, q 1989 Springer-Verlag).

13.7.1.3. dKT =dT and Anderson – Grüneisen parameter

A linear fit of the bulk moduli with temperature gave values for the bulk
modulus at 298 K, K298 ¼ 125ð3Þ GPa and its variation with temperature, dKT =dT ¼
20:029ð6Þ GPa K21.
The Anderson – Grüneisen parameter for zoisite at 298 K, d298 ; can be determined
from the relationship:

dT ¼ 2ð1=adT ÞðdKT =dTÞP

From Pawley et al. (1996, 1998), a ¼ 3:86 £ 1025 K 21 , K298 ¼ 125 GPa and
dKT =dT ¼ 20:02 GPa K21. Therefore, d298 ¼ 6:0:
However, for all minerals in the revised thermodynamic data and for computer
programme THERMOCALC (Holland and Powell, 1990; Holland et al., 1996), an average
value of Anderson – Grüneisen parameter d298 is taken as 7.0.

13.7.1.4. Stability
The maximum pressure stability (at 25 GPa) of zoisite is represented by the
reaction:

zoisite ¼ grossular þ kyanite þ coesite þ H2 O:

Under a typical continental shield geotherm of <108C km21, zoisite’s volume

increases with depth. But under typical subduction-zone geotherms, which are
,88C km21, its volume decreases with depth. For example, P –T paths presented by
Peacock et al. (1994) for the top of subducting slabs have gradients of < 100 K GPa21
between 2 and 5 GPa, equivalent to < 38C km21. Therefore, in a typical subduction zone,
Hydrous Minerals 933

zoisite becomes more dense as subduction proceeds, helping it to stabilize to high

pressures (Pawley et al., 1998).

13.7.1.5. Zoisite and lawsonite

The volumetric behaviour of zoisite (Ca2Al3(SiO4)(Si2O7)O(OH)) and lawsonite
(CaAl2(Si2O7)(OH)2·H2O) is of considerable interest because these are stable to pressures
of ,7 GPa and more (e.g., Poli and Schmidt, 1995). It contains less water and serves as a
minor contributor to the water budget in the slab.
However, a significant difference in the values of the bulk modulus, KT, is seen in
those determined by different workers:

Authors Zoisite Lawsonite Method

Holland et al. (1996) 279 ^ 9 GPa 191 ^ 5 GPa Powder diffraction

synchrotron source
Pawley et al. (1998) 127 ^ 4 GPa (V0 ¼ 909 Å3) Single-crystal XRD
Comodi and 102 ^ 6.5 GPa 98.2 ^ 2 GPa
Zanazzi (1996, 1997a)
Daniel et al. (1998) 107.0 ^ 3 GPa
Grevel et al. (2000) 125.1 ^ 2.1 GPa 106.7 ^ 1.3 (GPa) XRD synchrotron source
(V0 ¼ 901.0 Å3) (V0 ¼ 674.5 Å3)

In their experiments, Grevel et al. (2000) used a MAX 80 cubic cell anvil H-P
apparatus, installed at beamline F.2 of the DORIS III storage ring of HASYLAB
(Hemburg, Germany). The powder samples of lawsonite (synthesized at 6008C, 4 GPa,
82 h) and zoisite (synthesized at 8008C, 3 GPa, 72 h) were mixed with vaseline to ensure
hydrostatic pressure transmitting conditions and placed in direct contact with pallets of
powdered NaCl or NaCl-BN, which served as the pressure calibrant (Decker, 1971). The
measurements were carried out in energy-dispersive mode using a white synchrotron beam
of 100 £ 100 mm2 dimension at a fixed 2u angle determined from the diffraction pattern of
NaCl at ambient conditions.
To describe the P – V –T behaviour of lawsonite, Gravel et al. (2000) determined the
thermal expansivity at ambient pressure, aT;0 from the relation:
aT;0 ¼ a þ bT

By fixing K 0T;0 ¼ 4 and KT;0 ¼ 106:7 GPa, their experimental data yielded the values of
these parameters as:

a ¼ 3:953 ^ 0:588 £ 1025 =KT;0

b ¼ 20:363 ^ 1:146 £ 1028 =K 2

and
ðdKT =dTÞP ¼ 20:0153 ^ 0:006 GPa=KTO
934 Chapter 13

They simultaneously fitted their P –V – T data and the data of Comodi and Zanazzi
(1996, 1997a) to the function proposed by Berman (1988) and obtained the following
relation:
0
VP;T ðZ0 Þ ¼ V298 ðZ0 Þ½1 þ 2:74 £ 1025 =KðT 2 298:15 KÞ 2 0:69 £ 1026 =barðP 2 1Þ
(with V0298(Z0) ¼ 135.65 cm3 mol21; Smith et al., 1987) and
0
VP;T ðLWÞ ¼ V298 ðLWÞ½1 þ 3:19 £ 1025 =KðT 2 298:15 KÞ 2 0:85 £ 1026 =barðP 2 1Þ
(with V0298(LW) ¼ 101.55 cm3 mol21; Grevel et al., 2000).

Figure 13.16. Stability fields of lawsonite in CASH and MORB þ H2O system. Thin lines delineate the stability
fields of lawsonite in the CASH system (Pawley 1994; Schmidt, 1995) and the quartz, coesite, and stishovite
stability field (Zhang et al., 1996). The thick lines delineate the stability field of lawsonite in the MORB þ H2O
system below 6 GPa (Poli and Schmidt 1995) and the thin dashed lines delineate the stability field of lawsonite
above 6 GPa from this study. Lws: lawsonite; zo: zoisite; qtz: quartz; ky: kayanite; coe: coesite; grs: grossular;
toz: topaz; stish: stishovite; egg: phase egg; dsp: diaspore (Okamoto and Maruyama, 1999, q 1999 Mineralogical
Society of America).
Hydrous Minerals 935

In their experiments, Grevel et al. (2000) observed that, at high pressures

(6.6 ^ 0.3 GPa), the reaction occurs as

3 lawsonite ðlwÞ ¼ grossular ðgrÞ þ 2 kyanite ðkyÞ þ coesiteðcsÞ þ 6 H2 O

whereas the upper pressure-stability limit of zoisite is determined to be above 4 GPa from
the relation;

6 zoisite ðzoÞ ¼ 4 grossular ðgrÞ þ 5 kyanite ðkyÞ þ coesite ðcsÞ þ 3 H2 O

The stability fields of lawsonite in CASH and MORB þ H2O system are shown in
Fig. 13.16 (see caption) (Okamoto and Maruyama, 1999). For further studies on lawsonite,
see Section 13.8.3.

13.7.1.6. Subducting andesitic rocks

Poli and Schmidt (1995) showed that, in the subduction of andesitic rocks,
amphibole disappears between 2.4 and 2.6 GPa, forming lawsonite at T , 6308C. Zoisite
and clinozoisite are expected to carry H2O beyond the amphibole stability field, with
geothermal gradients higher than those possible for lawsonite, to depths of 100 –120 km,
where epidote breaks down (Comodi and Zanazzi, 1997). To determine H2O storage and
release in subduction zones, it is necessary to know the stability of these hydrous phases
within the slab and mantle wedge.
The calculated volume-expansivity-to-compressibility ratio of 38 bar K21
(Comodi and Zanazzi, 1997) indicates that the cell volume of clinozoisite
remains unchanged with geothermal gradients of ,108C km21. The crystallographic
data (see Section 13.7.1.1) support the results of experimental petrology in
indicating that epidote is a good candidate for transporting H2O in down-going
subducting slabs.

13.8. CaO –Al2O3 – SiO2 –H2O system

13.8.1. Amphiboles

The introduction of hydrous phases, both amphiboles and mica, into a refractory
mantle causes mantle metasomatism (Harte, 1983). This would account for the
re-enrichment in the major elements such as Ti, Fe, K and P, which were lost in earlier
melting events. The chemical variability of amphibole in mantle samples has been
summarized by many (viz. Wilkinson and LeMaitre, 1987). Amphiboles have been
widely recognized in mantle xenoliths that are entrained in both alkali basalts
and kimberlites.
In water-saturated peridotite, amphibole near the water-saturated solidus
(,1,0008C at 2 – 3 GPa) is pargasitic. Near the peridotite solidus, amphibole is paragasitic
but in LT – HP metamorphic peridotite it is aluminous tremolite. The decomposition
936 Chapter 13

pressures for different species of amphiboles at ,1,0008C have been determined as

(Source: Mysen et al., 1998):

Species Decomposition pressure

Pargasitic 2–3 GPa

Ca-amphibole (in harzburgite) 2.2 GPa
Amphibole (in lherzolite) 2.5– 28.0 GPa
Amphibole (in pyrolite) 2.8– 3.0 GPa

At high pressure, cummingtonite exhibits phase transition with symmetry changes

at the OH sites (Yang et al., 1997).
K-amphibole is stable to depths of 450 km (,15 GPa) at temperatures below
1,2008C. Alkali- rich hydrous phases exist at pressures of 10– 17 GPa and temperatures up
to 1,2508C (for further details, see Yang and Konzett, 2000).
In sub-continental mantle above dehydrating subducting slabs, free water can occur
at P – T above the stability field of non-alkalic amphiboles (.,3 GPa, ,90-km depth)
(e.g., Watson et al., 1990). In shallow upper mantle, if free water is present, there would be
a significant magneto-telluric signature.

13.8.1.1. Kaersutitic amphibole: oxidation –hydrogenation reactions

Kaersutite is a Ti-rich amphibole with the formula Ca2(Na,K)(Mg,Fe2þ,Fe3þ)4
(Ti,Al)(Si6Al2O22(O,OH,F)2. In its structure, there are two tetrahedral sites (T1 and T2)
containing Si and Al and five octahedral sites (two M1, two M2 and one M3 site)
containing Fe3þ, Fe2þ and Mg in the M1 and M3 sites, and Ti and Al in the M2 site. There
are three O sites (O1, O2 and O3) with 2(OH,O,F) in the O3 site associated with the M3
octahedral site (Pachar et al., 1989).
Alkali basalt associations are generally characterized by Ti-rich hornblendes such
as titanian pargasite or kaersutite. Kaersutitic amphiboles also occur in magmatic
inclusions in SNC-type meteorites, thought to have originated from the planet Mars
(Harvey and McSween, 1992).
Kaersutite amphiboles have high Fe3þ/Fe2þ ratios, as well as variable H2Oþ
contents, such that the sum of (OH þ F þ Cl) is generally less than the ideal 2.0
atoms per formula unit (apfu). On the basis of the chemistry of 90 natural
kaersutitic amphiboles, Virgo et al. (1994) have concluded that Ti is primarily
accommodated by a multi-variate substitution that involves Ti and OH in 1 : 21
proportions.
To quantify the variation of Fe oxy-component content in a titanian pargasite
megacryst amphibole from Vulcan’s Throne (Arizona), Poop et al. (1995) carried out
experiments from 1 atm to 1 GPa (500 –1,2008C) and fH2 from that of the IQF buffer to air.
It is seen that the Fe3þ/Fetot ratio of the amphibole is controlled by T, P and fH2 in
the substitution mechanism:
Fe2þ þ OH2 ¼ Fe3þ þ O22 þ 1=2H ð13-1Þ
Hydrous Minerals 937

The oxidation –dehydrogenation reaction is written for a pair of Fe end-member

calcic amphiboles as:

Ca2 Fe2þ 2þ 3þ
5 Si8 O22 ðOHÞ2 ¼ Ca2 Fe3 Fe2 Si8 O24 þ H2 O ð13-2Þ
The reaction represents equilibrium between a single amphibole phase and a fluid phase,
for which the equilibrium constant is given by:

fH2 aprod aprod aprod

Fe3þ Fe2þ ½
K¼ ð13-3Þ
areact areact
Fe2þ OH
where the a terms represent the activities of Fe3þ and Fe2þ in the product and reactant
amphibole, the activity of OH in the reactant amphibole and the activity of 03
crystallographic sites from which the H has been lost in the product is designated as A to
indicate an H vacancy. (Note: Equilibrium refers to chemical equilibrium within a single
mineral phase that accommodates both the reactant and product amphibole species.)
K changes. The variation in log K as a function of pressure is defined by (e.g., Eugster and
Skippen, 1967):
DVs0
D log K ¼ 2 DP ð13-4Þ
2:303 RT
where DV0s is the change in the standard state volume of the solid phases in reaction (13-2).
The variation in log K for reaction (13-4) vs P (kbar) for 9008C amphibole
experimental products is shown by Popp et al. (1995). In the random mixing model, it is
assumed that Fe3þ and Fe2þ ions mix randomly on the five M1, M2 and M3
crystallographic sites, which leads to the equilibrium:
ðxFe3þ Þ2 ðX½ Þ2
K ¼ fH2 ð28:94Þ ð13-5Þ
ðXFe2þ Þ2 ðXOH Þ2
From the expression of equilibrium constant it follows that, at any given
temperature and pressure, the proportion of the O3 sites occupied by OH and A exerts
control over the Fe3þ/Fe2þ ratio. For a given Fe3þ/Fe2þ ratio, OH content varies in relation
to both the Ti content and Fetot content.
The values of K are seen to be related to the P –T conditions as:
log K ¼ 4:25 2 4;363=TðKÞ þ 0:11ðP 2 1ÞðkbarÞ
for a random mixing model, and
log K ¼ 5:29 2 5;903=TðKÞ þ 0:13ðP 2 1ÞðkbarÞ
for a non-random mixing model.
The log K values provide a means to predict equilibrium values of fH2 from the
Fe3þ/Fetot ratios of the amphibole. Virgo et al. (1994) estimated that the sum of
OH þ Fe3þ þ Ti equals 2.0 pfu. The theoretical slope for the pressure dependence can be
calculated from equation (13-4).
938 Chapter 13

The composition of kaersutitic amphibole can be a quite sensitive indicator of the

fH2 in the environment of formation, provided the temperature and pressure are known.
Each of the mixing models provides a reasonable estimate of the fH2 at which kaersutitic
amphiboles have equilibrated.

13.8.1.2. Kaersutite in SNC meteorites: Martian H2O

To characterize the accretional and degassing history of the planet Mars, one needs
information on the H2O abundance in the Martian interior. The lower estimates are based
on solar elemental abundance ratios in SNC (shergotites, nakhlites and chassignites)
meteorites and in carbonaceous chondrites. The SNCs are regarded as of Martian origin
(e.g., Bogard and Johnson, 1983) and are seen to contain kaersutitic amphibole
(co-existing with glass) in association with pyroxene and olivine.
Amphiboles that are water-saturated, i.e., 2(OH) pfu ,2 wt% H2O, would most
likely crystallize from a magma that contains several wt% H2O. A Martian mantle source
of such magma would probably contain several thousand ppm H2O or several times the
concentration estimated for the interior of the Earth (Jambon and Zimmermann, 1990).
The kaersutites in Chassigny (C) and Zagami (Z) SNC meteorites were found to
contain H2O amounting to only 0.1– 0.2 wt% (Watson et al., 1994), corresponding to only
5 –10% of the H2O content of kaersutite with 2(OH) pfu as shown in Table 13.12 (Mysen
et al., 1998). H2O of Martian mantle as a function of pressure of kaersutite crystallization
has been presented in Fig. 13.17 (Mysen et al., 1998).
The study on H diffusion in minerals (Dyar et al., 1993) showed that the OH content
of kaersutite does not change significantly during transport from the Earth’s mantle to the
surface. The amphibole in SNCs must have experienced shock pressure during the impact
which ejected these to be transported from Mars to Earth. Shock experiments on
hornblende to ,30 GPa have been carried out by Minitti and Ratherford (1998).

TABLE 13.12
Composition of Vulcan’s Throne kaersutite compared with kaersutite from SNC meteorites (Mysen et al., 1998)

VTa Cb Zc

SiO2 40.32 39.49 35.15

TiO2 3.84 7.00 8.94
Al2O3 15.35 14.22 15.13
FeOd 7.70 10.47 16.18
MgO 14.53 11.53 6.40
CaO 10.46 11.80 11.48
Na2O 2.74 2.99 2.38
K2O 1.60 0.33 0.20
H2O 1.27 0.1 –0.2e 0.1 –0.2e
F 0.08 0.08
a
VT ¼ Vulcan’s Throne, Arizona, Fe3þ/(Fe3þ þ Fe2þ) ¼ 0.317 (from Popp et al., 1995a,b).
b
C ¼ Chassigny (from Johnson et al., 1991); c Z ¼ Zagami (from Treiman, 1985).
d
Total iron as FeO; e H2O concentration range from Watson et al. (1994).
Hydrous Minerals 939

Figure 13.17. H2O of Martian mantle determined on the basis of the water contents in Chassigny and Zagami
kaersutite amphiboles for 1,0008C as a function of pressure of kaersutite crystallization in the inclusions
(Mysen et al., 1998b, q 1998 Mineralogical Society of America).

The f O2 during equilibration of these SNC samples is reported to be near that of the
quartz –fayalite –magnetite (QFM) buffer (McSween, 1994; Johnson et al., 1991). The
experimental relationship between f O2 and T for the QFM buffer is given by Chou (1987).
When f O2 is higher than that of the QFM buffer, a lower f O2 is observed. The model
developed by Mysen et al. (1998) predicts quantitatively the relationships between iron
oxidation and hydrogen deficiency in kaersutite. The H2O content of the magma from
which the kaersutites in SNC meteorites could have crystallized is in the range 100–
1,000 ppm. This amount of H2O would lead to an estimated water content of 1 –35 ppm for
a Martian mantle that could have been the source rock for such magmas. The results of
Mysen et al. (1998) suggest that the source rock of the magma that gave rise to SNC
meteorites of age #1.3 Ga could be considerably drier than the terrestrial mantle that melts
to yield primitive MORB on Earth even today.

13.8.2. High Fe31 content and the aH2 O

In amphiboles at any given buffer, the Fe3þ/Fetot in amphibole increases as aH2 O

decreases. Thus, amphiboles with high Fe3þ/Fe2þ ratios can be stable at fO2 equivalent
relatively reducing buffer if the activity of H2O is low.
Even though the Fe3þ/Fetot values of the amphiboles in the Dish Hill xenolith
assemblage are quite high (in the range 0.70 –0.91), the fO2 of the samples lies within
approximately 1 log unit of the FMQ buffer. The relatively low fH2 required for the high
Fe3þ contents reflects relatively low H2O activities in the environment.
940 Chapter 13

13.8.3. Lawsonite, CaAl2Si2O7·H2O

Lawsonite is chemically similar to CaAl2Si2O8-anorthite with the addition of two

water molecules but has substantially higher density (3.09 g cm23) compared with that
of anorthite (2.76 g cm23). In contrast to the tetrahedral coordination of Al in anorthite,
in lawsonite it is in octahedral coordination. Structurally, lawsonite is an orthorhombic
sorosilicate with S.G. ¼ Cmcm and contains total water 11.5 wt%. The manner of
holding water in its structure is novel (Baur, 1978); cavities are formed by rings of two
Al octahedra and two Si2O7 groups, each containing an isolated water molecule and a
calcium atom (Fig. 13.18). The hydroxyl units are bound to the edge-sharing
Al octahedra.
Lawsonite is a water-rich mineral common in metabasalts and metagreywackes
which have encountered high-P and low-T conditions. It is common in blue-schist terrain,
presenting P – T conditions of 0.3 –1.2 GPa and 500 –770 K (e.g., Evans and Brown, 1986).
Lawsonite is thus an important blue-schist facies mineral. In a subducting slab, the
temperature at a depth of 350 km is lower than 1,200 K (Peacock, 1990; Furukawa, 1993).
Such a low temperature would allow lawsonite to remain in the down-going slab up to its
maximum pressure stability. Hence, lawsonite containing about 11 wt% H2O would be

Figure 13.18. Crystal structure of lawsonite in the Cmcm phase (After Baur, 1978).
Hydrous Minerals 941

Figure 13.19. Ambient pressure and temperature absorption spectra showing H2O, OH, and Si2O7 regions
(Scott and Williams, 1999, q 1999 Springer-Verlag).

able to store water down to depths of about 250 km and contribute water for recycling in
the deep mantle.
At ambient pressures and below 273 K, the structure of lawsonite changes to
the Pmcn(S. G) and then to P21 Cn symmetry below 150 K. The first transition is a
result of decreased thermal liberation of water molecule, with the molecule slightly
canting and allowing cooperative hydrogen bonding to form. This bonding strengthens
at lower temperature (150 K) transition when further canting of the water molecule
occurs and the crystal symmetry is reduced by the movement of hydrogen atoms
out of the [010] plane (Libowitzky and Armbruster, 1995). The infrared-active
hydroxyl stretching and bending vibrations (Fig. 13.19) shift smoothly across these
transitions, indicating that they are of dynamic order –disorder type (Libowitzky and
Rossman, 1996).
DAC experiments in the CaO – Al2O3 – SiO2 – H2O (CASH) system show that
lawsonite is stable up to ,12.0 GPa (at #1,230 K; Schmidt, 1995). The maximum stability
of lawsonite in MORB ranges to 8.5 GPa and 1,100 K (Schmidt and Poli, 1998).
The XRD lines of lawsonite at a pressure of 6.88 GPa are presented in Table 13.13.
The crystallographic parameters noted are:


a ¼ 5:767 ^ 0:005 A


b ¼ 8:663 ^ 0:010 A


c ¼ 12:977 ^ 0:010 A

and

3
V ¼ 648:3 ^ 1:0 A
942 Chapter 13

TABLE 13.13
Representative refinement of lawsonite x-ray lines at P ¼ 6.88 GPa and T ¼ 7008C (Schmidt and Poli, 1998)

hkl dobs dcalc dobs 2 dcalc

11 0 4.804 4.800 0.004

11 1 4.501 4.502 20.001
02 1 4.115 4.109 0.006
02 2 3.609 3.603 0.006
11 3 3.213 3.214 20.001
20 0 2.886 2.883 0.003
11 4 2.692 2.688 0.004
20 2 2.633 2.635 20.002
02 4 2.594 2.597 20.003
13 0 2.594 2.582 0.012
22 0 2.401 2.400 0.001
13 2 2.401 2.399 0.002
22 1 2.364 2.360 0.004
11 5 2.293 2.283 0.010
22 2 2.255 2.251 0.004
02 5 2.228 2.226 0.002
22 3 2.099 2.099 0.000
04 3 1.939 1.937 0.002
20 6 1.732 1.730 0.002
11 7 1.732 1.729 0.003
02 7 1.702 1.704 20.002
24 2 1.672 1.673 20.001
31 4 1.624 1.624 0.000
00 8 1.624 1.622 0.002
22 6 1.608 1.607 0.001
24 3 1.608 1.608 0.000
33 0 1.608 1.600 0.008
04 6 1.528 1.530 20.002
24 4 1.528 1.528 0.000
40 0 1.438 1.442 20.004
24 5 1.438 1.441 20.003
13 8 1.372 1.374 20.002

Notes: a ¼ 5.767 ^ 0.005 Å, b ¼ 8.663 ^ 0.010 Å, c ¼ 12.977 ^ 0.010 Å and V ¼ 648.3 ^ 1.0 Å3.

13.8.3.1. K0 and a values

For lawsonite, the isobaric thermal-expansion coefficient, a, and isothermal bulk
modulus, K0, have been determined by different workers as:

a ¼ 3.16(5) £ 1025 K21 Pawley et al. (1996)

a ¼ 3.13(9) £ 1025 K21
a ¼ 3.22(5) £ 1025 K21
K0 ¼ 191(5) GPa (set K 0 ¼ 4)
K0 ¼ 96(2) GPa (set K 0 ¼ 4)
K0 ¼ 124(18) GPa
V0 ¼ 675.6(2) Å3
Comodi and Zanazzi (1996)
Le Cléac’h and Gillet (1990)
Holland et al. (1996)
Comodi and Zanazzi (1996)
Daniel et al. (1999)
Hydrous Minerals 943

13.8.3.2. Compressibility
The deformation of lawsonite is linear up to 8 GPa. The mean axial compressibility
coefficients ðbÞ of lawsonite are determined as (Daniel et al., 1999):

ba ¼ 2:56ð4Þ £ 1025 GPa21

bb ¼ 2:43ð4Þ £ 1025 GPa21

bc ¼ 2:17ð3Þ £ 1025 GPa21

The bulk modulus, calculated as the reciprocal of the cell-volume linear compressibility, is
obtained as:
b21 ¼ 110:0ð40Þ GPa Comodi and Zanazzi ð1996Þ

b21 ¼ 142:0ð24Þ GPa Daniel et al: ð1999Þ

where the deviatoric stresses are moderate, (s1 2 s2) , 0.13 GPa (Meng et al., 1993), the
linear compressibility of lawsonite rises from 142 GPa under fully hydrostatic conditions
up to 156.7(44) GPa while the value of K0 increases up to 138.0(29) GPa. This indicates
that, for lawsonite, small deviatoric stresses can have intrinsic strong effects on its
compressibility.
The EOS of lawsonite calculated via linear and Birch –Murnaghan formalisms
indicate that the incompressibility of lawsonite displays non-linear variations with
increasing temperature and reaches a minimal value at 486 K (Table 13.14; Daniel et al.,
1999). A linear regression of the data yields the values of ba ; bb ; bc ; b and b21 ; while an
adjustment to a second-order Birch – Murnaghan EOS provides values for KT;0 and the
associated VT;0 (Table 13.14)
The reference unit-cell volume VT;0 is calculated from
ð T 
VT;0 ¼ V298;0 exp aT;0 dT ð13-6Þ
298

TABLE 13.14
Axial compressibilities and bulk modulus of lawsonite obtained at various temperatures

Temperature ba bb bc b b 21 KT,0 VT,0

(K) (1023 GPa21) (1023 GPa21) (1023 GPa21) (1023 GPa21) (GPa) (GPa) (Å3)

298 2.37(4) 2.15(4) 1.96(18) 6.38(18) 156.7(44) 137.8(47) 675.7(2)

486 2.37(4) 2.94(10) 2.32(8) 7.81(20) 128.0(33) 109.3(42) 680.4(8)
662 2.52(5) 2.64(6) 1.97(4) 6.94(15) 144.1(31) 118.9(68) 683.4(14)
759 2.13(5) 2.49(9) 2.00(8) 6.49(19) 154.1(45) 137.5(67) 684.6(8)

Source: Daniel et al. (1999).

944 Chapter 13

where aT;0 is the thermal expansion at T and ambient pressure, and aT;0 can be expressed
as a constant a0 (Comodi and Zanazzi, 1996; Holland et al., 1996). Meng et al. (1993)
showed that deviatoric stresses decrease rapidly upon heating. But the non-linear
behaviour of the bulk modulus of lawsonite with increasing temperature may be of
structural origin.
The compression at 298 K is almost isotropic and is mostly due to compression of
the distorted CaO6 octahedra and of the large cavities in which H2O is located and, to a
lesser extent, by the compression of the AlO6 octahedra. The thermal expansion coefficient
of the Al polyhedra at 0.1 MPa is about half that of the Ca polyhedra (Comodi and
Zanazzi, 1996).
At temperatures higher than 486 K, the bulk modulus of lawsonite starts increasing
and with increasing temperature all the a-, b- and c-axes become less compressible. As this
behaviour is unknown in the anhydrous minerals studied in situ at high pressure and high
temperature, Daniel et al. (1999) propose that the whole structure of lawsonite stiffens
when pressurized at high pressure. This is because the large cavities in which H2O is
located become less compressible at high temperature and also because of the
strengthening of hydrogen bonds (Libowitsky and Armbruster, 1995). However, this
postulation can only be verified by proton NMR studies on lawsonite or its related hydrous
minerals.

13.8.3.3. IR study: Grüneisen parameter and thermal expansion

The infrared spectra of lawsonite have been characterized by Scott and Williams
(1999) to pressures up to 20 GPa and 300 K. At ambient pressure and temperature, the
spectrum of lawsonite contains bands of H2O, OH and Si2O7 (Fig. 13.19).
Under pressure, the asymmetric and symmetric stretching and bending vibrations of
the Si2O7 groups (at zero-pressure frequencies between 600 and 1,000 cm21) increase in
frequency at rates between 3.6 and 5.90 cm21. All silicate modes appear to shift
continuously with pressure to 20 GPa.
The band assignments in the infrared spectrum of lawsonite between 500 and
4,000 cm21 are presented in Table 13.15 (Scott and Williams, 1999). Under compression,

TABLE 13.15
Band assignments of infrared spectra of lawsonite and the Grüneisen parameter (from Scott and Williams, 1999)

v0 (cm21) Assignment dv=dP (cm21 GPa21) gK(191 GPa) gK(110 GPa)

3,556 O –H stretch of OH 29.60 20.52 20.30

3,242 O –H stretch of OH 11.01 0.65 0.37
2,948 O –H stretch of H2O 0.83 0.05 0.03
1,627 H –O –H bend of H2O 0.26 0.03 0.02
990 SiO4 stretch 5.91 1.14 0.66
930 SiO4 stretch 5.48 1.13 0.65
878 SiO4 stretch 4.89 1.06 0.61
606 O –Si–O and Si–O –Si bend 3.64 1.15 0.66
Mean mode Grüneisen parameters for observed vibrational modes 0.65 0.34
Thermodynamic Grüneisen parameters 2.21 1.27
Hydrous Minerals 945

Figure 13.20. Change in n O–H, n H2O and d H2O bands of lawsonite under compression and decompression
(Scott and Williams, 1999, q 1999 Springer-Verlag).

the hydroxyl stretching bands show dramatic shifts in both frequency and amplitude. In
contrast, the bending vibration of the water molecule (d H2O) remains nearly unchanged at
high pressure. On decompression, the positions of n OH, n H2O and d H2O remain identical
(Fig. 13.20).
In Table 13.15 are presented the summary of observed vibrational modes and their
corresponding assignments. Grüneisen parameters are calculated for mode shifts below
8 GPa, frequency shifts with pressure and different values for mode Grüneisen parameters.
A discontinuity in mode shift is observed for both O – H stretches of hydroxyl units at
8 –9 GPa. Band assignments in Table 13.15 are after Le Clec’h and Gillet (1990) and
Libowitzky and Rossman (1996).
However, mode shifts in the hydroxyl-stretching region provide an insight into
pressure-induced changes in the hydrogen bonding in lawsonite (Table 13.15). Because of
the extremely low mode Grüneisen parameters of vibrations of the water molecule
946 Chapter 13

compared with the silicate and hydroxyl vibrations (Table 13.15), under compression, the
water molecules respond little compared with the alumino-silicate framework.
The results of the experimental study of by Scott and Williams (1999) show that, in
lawsonite structure, the volumetric compactions of ,11% appear to have little effect on the
hydrogen bonding of the H2O molecule. This is manifested by the small shifts in stretching
and bending vibrations of this unit under compression. Thus, the cavities in the framework
of the lawsonite structure appear to provide a largely unaltered site in which H2O can
reside up to ultra-high pressure. However, the hydroxyl units are affected by pressure,
notably through enhanced hydrogen bonding, and this contributes to the thermodynamic
stability of lawsonite.
The spectra of the silicate stretching and bending regions taken on decompression
are identical to those on compression. This depicts the reversibility.
The two Grüneisen parameters ðgi Þ for each of the vibrational modes bear the
relationship
 
K dv
gi ¼ ð13-7Þ
v0 dP 0
and the two available values of the bulk modulus of lawsonite are 191 and 110 GPa
(Comodi and Zanazzi, 1996; Holland et al., 1996). The average of all the individual
Grüneisen parameters are recorded as 0.65 and 0.34 for the corresponding bulk moduli of
191 and 110 GPa (see Table 13.15). For comparison, the corresponding thermodynamic
Grüneisen parameters are calculated for the two cases using the formula:
aK
g¼ ð13-8Þ
rCP
In this calculation, the thermal expansion ðaÞ is derived as:

a ¼ 3:16ð^0:05Þ £ 1025 K21

using Pawley et al. (1996) and the heat capacity value is:

CP ¼ 66:35 cal K21 mol21

which is obtained from Perkins et al. (1980). The density (r) of lawsonite is then taken as

r ¼ 9:83 £ 103 mol m23

13.9. MgO – Al2O3 –SiO2 – H2O system

Extensive metastability has been known in the MgO –Al2O3 – SiO2 –H2O system,
even at low pressures. The 10-Å phase has been investigated for its (meta)stability because
of its low P – T stability range. Talc is seen to react with excess water to form the 10-Å
phase at pressures in excess of 5 GPa.
However, the large stability field of lawsonite in basaltic compositions may obviate
the necessity of invoking the presence of the 10-Å phase, phase A or talc within subducted
Hydrous Minerals 947

basaltic crust. Talc and phase A may be relatively unimportant for water transport in the
upper mantle. In a sufficiently cold subduction zone, the nominally hydrous phases might
be carried to depths well beyond their formal thermodynamic stability fields (e.g., Scott
and Williams, 1999; Daniel et al., 2000).
Phase relations in MgO – Al2O3 – SiO2 – H2O system provide a basis to examine the
stability relationship of aluminium-bearing hydrous phases in mantle composition during
subduction. Chlorite is the major aluminous hydrated phase in harzburgite mantle. In this
MASH system, chlorite has a large stability field and it breaks down at pressures
exceeding 7 GPa.
Fockenberg (1995) reported an experiment at 5.4 GPa and 7008C containing the
assemblage chlorite þ forsterite þ pumpellyite þ diaspore þ enstatite near the final
breakdown of chlorite. He also reported an experiment at 5.5 GPa and 6408C where
chlorite þ enstatite þ pumpellyite were observed.
Chlorite contains ,20 – 25% of the total H2O (2 – 2.5 wt% H2O) stored in a hydrated
lherzolite to harzburgite composition in the stability field of forsterite þ antigorite þ
chlorite.

13.9.1. Muscovite

Comodi and Zanazzi (1995) determined the compressibility and structural

variations of two 2M1 muscovites, under pressures between 0.1 and 3.5 GPa, by single-
crystal XRD using Merrill –Bassett DAC. Structural refinement showed that the different
compressibility was largely due to the partial substitution of Na for K in the inter-layer
region. Isothermal bulk moduli (assuming K 0 ¼ 4) were determined as:

Na muscovite GPa ba : bb : bc

Na-poor muscovite 49(^3) 1 : 1.15 : 3.95

Na-rich muscovite 54(^3) 1 : 1.19 : 3.46

At 5 GPa, Mg in the mica decreases with increasing temperature. The micas becomes
less K-rich and more Si- and Mg-rich between 5 and 7.5 GPa at 1,3008C. The low total for the
mica at 7.5 GPa may indicate excess protons in the structure (e.g., Robert et al., 1995).
13.9.1.1. Phlogopite
Philogopite shows lower compressibility than muscovite. The reason lies in the
greater compressibility of the dioctahedral layer with respect to that of the trioctahedral
layer. Phlogopite seems to have a large thermal stability field at modest pressures, with a
reported maximum temperature of existence between 1,180 and 1,3008C near 4 GPa in
periodite assemblages (e.g., Olafsson and Eggler, 1983).
Sudo and Ratsumi (1990) studied the phase relations of phlogopite þ diopside at
pressures between 5 and 13 GPa and at temperatures between 1,000 and 1,3008C. With
increasing pressure, they found the following assemblages:
5– 5.9 GPa: phlogopite þ diopside
6.5 –7.4 GPa: phlogopite þ diopside þ garnet
9.3 –11.1 GPa: phlogopite þ diopside þ garnet þ forsterite þ vapour
948 Chapter 13

The co-existence of clinopyroxene with amphibole suggests that the breakdown

reaction of the phlogopite þ diopside is divariant in nature.
Because of its stability at pressures .10GPa (Sudo and Tatsumi, 1990), phlogopite
may be regarded as a likely candidate for the storage of H2O at depths.

13.9.1.2. Phase X
Luth (1997) reported an unknown K-rich phase from the breakdown of phlogopite at
pressure .11 GPa and 1,6008C. This phase, called phase X, showed high thermal stability,
making it a potential host for K and H2O at these pressures. Its chemistry shows it is high in
K2O (10 –15 wt%) and MgO (16 – 30 wt%) and low in Al2O3. It contains ,16 cations per
220 atoms.
Trønnes (1990) experimentally studied the low-Al, high-K amphiboles present in
subducted lithosphere at depth from 200 to 400 km. The chemical composition in wt% of
the phase corresponded to:
K2O ¼ 17.6
MgO ¼ 29.7
Al2O3 ¼ 2.9
SiO2 ¼ 47.7
H2O ¼ 2.0
Inoue et al. (1995) reported similar findings for phase X formed by the
decomposition of K-rich amphibole. They found that this phase had a molar K : Mg/Si
ratio of ,1.2/2 and reported a SIMS measurement of 1.60 wt% H2O.
The stability of phase X needs further study.

13.10. K2O –MgO – Al2O3 – SiO2 – H2O system

Experimental investigations on this pelitic system, K2O –MgO – Al2O3 – TiO2 –

SiO2 –P2O5 –H2O, have revealed possible occurrences of several OH-bearing phases such
as the following:

Mineral name Chemical formula Author

Mg Al-pumpellyite (Mg5Al5Si6O21(OH)7) Fockenberg (1998)

Mg-carpholite (MgAl2Si2O6(OH)4)
Piezotite or phase Pi (Al3Si2O7(OH)) Wunder and Schryer (1992)
(3.5–5.5 GPa/6508C)
K-cymrite (KAlSi3O8(H2O)1.0 – 0.9) Massonne (1995)
OH-topaz (Al2Si4(OH)2) Wunder et al. (1993)
Phase A (Mg7Si2O8(OH)6) Wunder et al. (1996)
Lawsonite (CaAl2Si2O8(OH)2) Poli and Schmidt (1995)

13.10.1. Cold geotherms

In the pelitic system, K2O – MgO –Al2O3 – TiO2 –SiO2 – P2O5 – H2O, the stabilities
of VHP phases require abnormally low geothermal gradients, ,5 – 78C km21, which can
Hydrous Minerals 949

only be attained by the subduction of old, cold, oceanic crust capping the lithosphere with
or without pelagic sediments or an ancient continent (Schreyer, 1988). Fig. 13.16 also
shows that lawsonite may persist up to at least 6 GPa if the temperature is lower than
6508C. P – T regimes for the eclogite facies can be sub-divided into amphibole eclogite,
lawsonite eclogite and dry eclogite.
Many of these OH-bearing phases should occur at 500– 7508C near 4– 10 GPa but
only along extremely cold geotherms, i.e., considerably less than 58C km21. This P – T
field (4 –10 GPa/,58C km21) is regarded as the “forbidden zone” in the mantle. Peacock
(1990) has shown that rocks might be subducted along subduction-zone gradients as low as
38C km21. Such low geotherms are confined only to the inner regions of the rapidly buried
lithosphere in long-lasting subduction zones (e.g., Wunder et al., 1993). Mg-pumpellyite,
Mg5Al5Si6O21(OH)7, is stable to pressures of 5 –6 GPa and temperatures of 680– 7708C
(Artioli et al., 1999; Fockenberg, 1998).

13.11. Al2O3 – SiO2 – H2O (ASH) system

In the system Al2O3 –SiO2 – H2O (Fig. 13.21) a number of phases exist. This system
is particularly relevant for subducted pelitic sediments and perhaps aluminous basalts. At
low temperatures (,4008C), diaspore, AlOOH, is stable to pressures near 6 GPa and
another phase, “pi”, Al3Si2O7(OH)3, continues to exist to a higher temperature of 7008C.
Diaspore and phase pi react with SiO2 at 6 GPa to form hydroxylated topaz, Al2SiO4(OH)2
(Wunder et al., 1993a,b).
At higher pressures, OH-topaz reacts with SiO2 to form phase egg (named after
Eggleton et al., 1978). In this phase of composition AlSO3, OH silicon is entirely in
octahedral coordination. Phase egg and phengitic micas together are the main phases in
subducted sediments. Humite groups (generalized formula Mg2SiO4.Mg(OH,F)2 with
n ¼ 2 for chondrodite and n ¼ 4 for clinohumite) show stability fields below ,13 GPa

Figure 13.21. The major phase in the Al2O3 –SiO2 –H2O system.
950 Chapter 13

and temperatures below 1,0008C (e.g., Kawamoto et al., 1996). These are structurally
similar to olivine, intercalated with hydrous defects. Liu (1993) proposed that the
reaction 5Mg2SiO4 þ H2O ¼ MgSiO3 (enstatite) þ Mg9SiO4O18H2 (clinohumite) could
produce a “water line” in the mantle defining the lower depth limit for the low-velocity
zone, the top of which is defined by the stability fields of amphiboles or phlogopite
(Liu, 1993).
In crustal conditions, Al2O3 – SiO2 – H2O system kaolinite and pyrophillite constitute
stable phases. The water-saturated system of Al2O3 –SiO2 – H2O (see Fig. 13.21), moving
towards higher pressures from the crustal condition, initially produces phase pi ¼ Al3Si2O7
(OH) (Wunder et al., 1993a) and then topaz-OH (¼ Al2SiO4 (OH); Wunder et al., 1993b).
But phase egg þ stishovite þ diaspore are co-linear and thus do not have a stability field in a
three-component system. Schmidt et al. (1998) synthesized these from a mixture of
cristobalite, Al2O3 and Al(OH)3, at ,12 GPa and 7008C.

13.11.1. AlSiO3OH, “phase egg”

A phase of composition AlSiO3OH, with Al : Si ¼ 1 : 1, was first synthesized by

Eggleton et al. (1978) and was termed “phase egg” (Schmidt, 1995). In this phase, both
Al and Si are octahedrally coordinated. The crystal structure of phase egg is shown in
Fig. 13.22.
The highest pressure at which phase egg was reported is 17.7 GPa (/1,2008C) and
the highest temperature is 1,3008C (/15.5 GPa) (Irifune et al., 1995). This phase has
features common with stishovite structure.
The unit cell of phase egg contains four Al and four Si atoms. For an Al : Si: H ratio
of 1 : 1: 1, phase egg has a stoichiometric water content of 7.50 wt%. In the studies carried
out by Schmidt et al. (1998), the XRD pattern showed systematic absences which are
consistent with space group P21 =n: A peak with 0.8 e Å23 was found in the position

Figure 13.22. Crystal structure of AlSiO3OH. (a) The structural module in a–c projection. (b) The linkage of
these units. The diagrams were produced with program DRAWxtl (Finger and Kroeker, 1997). Dark octahedra
contain Al, and the position of the H atom is indicated by white spheres (Schmidt et al., 1998, q 1998
Mineralogical Society of America).
Hydrous Minerals 951

expected for the H atom. In the final refinement, background parameters were varied along
with the structural parameters of AlSiO3OH (Schmidt et al., 1998).
The crystal structure of the synthesized phase was studied by high-resolution X-ray
powder diffraction using Siemens D5000 powder diffractometer with flat-plate geometry
(Table 13.16). The space group determined was P21 =n (from systematic absences), Z ¼ 4;
V0 ¼ 212.99(1) Å3 and the zero-pressure density was 3.74 g cm23.
The fundamental module of the structure consists of edge-shared Si octahedra
linked to an Al2O10 dimer (Fig. 13.22a,b; after Schmidt et al., 1998). The structure
resembles that of stishovite, wherein edge-linked octahedra form columns of corner-linked
octahedra. In phase egg, H atoms lie in the openings between columns. This efficient
packing of H stabilizes this phase (relative to SiO2, Al2O3, H2O and other hydroxyl-
bearing minerals).
This phase, which has features akin to stishovite structure, occurs above 11 GPa and
7008C. The phase AlSiO3·OH forms from topaz-OH under increasing pressure and persists
to more than 17.7 GPa and 1,3008C. The highest pressure at which phase egg was reported
is 17.7 GPa at 1,2008C and the highest temperature is 1,3008C at 15.5 GPa (Irifune et al.,
1995).
Phase egg has been synthesized in the chemical system Al2O3 –SiO2 – H2O and
CaO – Al2O3 – SiO2 –H2O. The stability of the hydrous high-pressure silicates, such as
topaz-OH and phase egg, are identical in these systems. By conducting experiments at
1,2008C and 13.5 GPa, Irifune et al. (1995) obtained phase egg þ corundum.
The crystal structure of phase egg is related in the reactions: kyanite ¼ stishovite þ
corundum (Schmidt et al., 1997), topaz-OH ¼ kyanite þ vapour (Wunder et al., 1993a,b)
and diaspore ¼ corundum þ vapour (Fockenberg et al., 1996) and the following reaction
have also been experimentally determined (Schmidt et al., 1998):

2 AlSiO3 OH ! 1 topaz-OH þ 1 stishovite ð13-9aÞ

2 AlSiO3 OH þ 1 corundum þ 1 H2 O ! 2 topaz-OH ð13-9bÞ

and

1 diaspore þ 1 stishovite ¼ 1 AlSiO3 OH ð13-9cÞ

The extrapolation of dP=dT slopes of phase egg and topaz-OH-bearing reactions

suggest that topaz-OH decomposes within the stability field at kyanite. Phase egg forms
from topaz-OH and stishovite according to reaction (13-9a). This reaction has a flat slope of
,45 MPa K21 in P– T space. Over a limited pressure range (max. 1 GPa), topaz-OH co-
exists with phase egg until topaz-OH decomposes according to reaction (13-9b) at
temperatures above 9508C. Below approximately 9508C, topaz-OH decomposes through:

1 phase egg þ 1 diaspore ¼ 1 topaz-OH ð13-9dÞ

At lower temperatures, phase egg forms around 7008C according to reaction (13-9c).
952 Chapter 13

TABLE 13.16
Diffraction patterns for AlSiO3OH (d in Å; from Schmidt et al., 1999)

hkl Schmidt et al. (1998) Eggleton et al. (1978)a

d (calc.) d (obs.) I d I

1 0 1
5.3339 5.3295 6 5.34 50

110 3.6949 3.6945 13
011 3.6670 3.6665 100 3.67 100
200 3.5338 3.5334 1 3.54 10
002 3.4390 ,1 3.44 10
1 1 1
3.3638 3.3633 2 3.36 20
210 2.7390 2.7386 1 2.74 10
012 2.6941 2.6938 1 2.695 10
2 1 1
2.6497 2.6495 13 2.649 40
112 2.6190 2.6188 67 2.616 100
211 2.4511 2.4509 16 2.450 70
111 2.4267 2.4264 ,1
3 0 1
2.3358 2.3356 3 2.334 30
202 2.3022 2.3029 10 2.301 60
1 0 3
2.2807 2.2805 1
120 2.0721 2.0716 2
310 2.0699 2.0696 8
021 2.0671 2.0670 ,1
3 1 1
2.0562 2.0561 39 2.054 90
212 2.033 2.0331 46 2.031 100
013 2.0267 2.0265 3
1 1 3
2.0184 2.0182 25 2.014 60
312 1.8834 1.8833 9 1.881 60
2 1 3
1.8650 1.8648 ,1
220 1.8475 1.8474 ,1
022 1.8336 1.8334 ,1 1.832 10
2 2 1
1.8194 1.8193 ,1
1 2 2
1.8094 1.8092 8 1.807 60
3 0 3
1.7779 1.7778 2 1.775 30
400 1.7669 1.7666 ,1
221 1.7501 1.7508 ,1
122 1.7421 1.7421 ,1
2 2 2
1.6819 1.6820 ,1
312 1.6807 1.6806 2
4 0 2
1.6741 1.6740 ,1 1.678 20
213 1.6676 1.6675 5 1.666 40
2 0 4
1.6435 1.6432 ,1 1.641 10
320 1.5958 1.5949 ,1
3 2 1
1.5887 1.5886 33 1.587 100
222 1.5781 1.5780 26 1.574 90
1 2 3
1.5711 1.5710 32 1.570 100
411 1.5454 1.5453 2 1.544 20
303 1.5367 1.5365 ,1
114 1.5153 1.5152 17 1.514 70
Hydrous Minerals 953

TABLE 13.16 (continued)

hkl Schmidt et al. (1998) Eggleton et al. (1978)a
d (calc.) d (obs.) I d I

123 1.5055 – 2
3 2 2
1.5048 1.5050 4 1.502 20
402 1.4859 1.4857 ,1
204 1.4643 1.4642 ,1
4 1 3
1.4261 1.4260 11 1.425 60
3 1 4
1.4149 1.4150 2
031 1.4140 1.4142 1 1.413 30
1 3 1
1.3946 1.3943 1
1 0 5
1.3889 1.3988 1
223 1.3877 1.3874 3 1.387 40
4 2 1
1.3732 1.3737 ,1 1.372 100
501 1.3464 1.3460 16 1.346 80
510 1.3439 1.3436 5
4 0 4
1.3334 1.3332 14
1 3 2
1.3227 1.3224 3 1.322 20
5 1 2
1.3151 1
105 13148 1.3145 11 1.314 60
2 2 4
1.3096 1
231 1.2994 1.2992 10 1.298 50
a
Lines seen by Eggleton et al. at d ¼ 3.98 and 2.129 Å (both with I ¼ 30) and d ¼ 1.630 Å with I ¼ 10 belong to
diaspore.

13.11.1.1. Discussion
High-pressure hydrous aluminosilicates, such as phase egg and topaz-OH, are
potential carriers of water at large depths. As long as water is present, topaz-OH should
occur in eclogites. This is also seen in peraluminous basalts, in most greywackes and in
metapelites. In pressure experiments on metapelites, topaz-OH is encountered in the
pressure range 6– 11 GPa with the corresponding temperatures of 700– 9008C.
Phase egg with an Al : Si ¼ 1 : 1 is expected to occur in a wider range of bulk
compositions than topaz-OH (Al : Si ¼ 2 : 1). Hence, phase egg could also appear in meta-
aluminous bulk composition such as MORB and can significantly modify the water budget
at depth. Phase egg could be the carrier of water into the deep Earth owing to its reasonable
chemical composition (unlike most of the previously known high-pressure hydrous
silicates, which have a very high Mg : Si ratio) and to its high-pressure and high-
temperature occurrence (Schmidt et al., 1998).
In the chemical systems Al2O3 – SiO2 – H2O and CaO – Al2O3 –SiO2 – H2O, the
stability of the hydrous high-pressure silicates, such as topaz-OH and phase egg, are
identical for these systems because no phases in the ASH system have a significant Ca
solubility. The experiments at 1,2008C and 13.5 GPa by Irifune et al. (1995) yielded phase
egg þ corundum.
954 Chapter 13

Schmidt et al. (1998) performed Schreinemaker’s analysis of the Al2O3 – SiO2 –

H2O system considering the following phases: phase egg, topaz-OH, stishovite, corundum,
kyanite, diaspore and vapour. They predicted that phase egg would decompose with
pressure to an unknown, possibly hydrous aluminosilicate.

13.11.1.2. Phase egg in subduction zone

The P – T conditions for the occurrence of topaz-OH and phase egg are
characteristic of subduction zones. At pressures of 15 GPa, phase egg was synthesized
up to 1,3008C and the maximum temperature stability might be as much as 1,500– 1,6008C
(Schmidt et al., 1998).
Phase egg could be a carrier of water into the deep Earth, owing to its reasonable
chemical composition and its high-pressure and high-temperature occurrence. Potentially,
phase egg may replace topaz-OH or kyanite in subducted crustal-bulk compositions and
may transport some water into the deep Earth. Thus, phase egg and topaz-OH are potential
carriers of water to great depths. As long as water is present, topaz-OH should occur in
eclogites saturated with respect to kyanite, i.e., in peraluminous basalts, in most
greywackes and in metapelites. Phase egg could also appear in meta-aluminous bulk
compositions such as MORBs.

13.12. Clay minerals

The data of Wu et al. (1997) on montmorillonite under high H2O pressure show
that, although pressure above the H2O liquid – vapour (L – V) curve does not have much
effect on dehydration temperature, it has a pronounced effect on the rehydration hysteresis.
They suggested that, initially, montmorillonite in hydration state with the smaller d value
would swell as H2O molecules rearranged themselves in layers. This stage is followed by a
process in which molecules diffuse into the enlarged inter-layer space. This led to a
positive DV, which requires extra activation energy when the sample is under pressure, and
this results in a pressure-induced hysteresis.
The 001 peak broadening during a smectite hydration or dehydration is due to the
stratification of layers of different thicknesses or hydration states. A study by Wu et al.
(1997) revealed that, in the dehydration and re-hydration processes between 19 and 15 Å
states, there are two states: homogeneous dehydration (between 15 and 17.5 Å) and inter-
stratification (between ,17.5 and 19 Å) (see Fig. 13.23).
Starting from the 19-Å hydration state, the sharp peak of the basal spacing
001 decreased to ,17.5 Å. When the basal spacing dropped from ,17.5 to 15 Å,
the 001 peak first broadened and then narrowed. If the peak broadening is due to
inter-stratification of the third and second hydration states, then the narrow
peak width between 19 and ,17.5 Å indicates that the d values of layers are nearly
identical.
Besides H2O, the species of inter-layer cations play an important role in the
swelling property of the clay (e.g., McBride, 1994). The primary factors are the size and
charge of the cation.
Hydrous Minerals 955

Figure 13.23. Ionic potentials (charge/radius) of Naþ, Ca2þ and Mg2þ cations vs dehydration temperature for
SWy21 montmorillonite exchanged with those cations. Pairs of connected dots represent the range over which
dehydration occurs (Wu et al., 1997, q 1997 Mineralogical Society of America).

13.12.1. Structural disorder

Structural disorders and orders of various kinds seen in phyllosilicates such as clay
minerals include: thermal disorder, cation disorder, long- and short-range order, layer-
stacking disorder, order –disorder in mixed-layers, lattice disorder, etc. (Brindley, 1984).
Disorder due to layer stacking is common in clay structures. Such disorders commonly
arise from the interaction of material between layers, such as variable hydration from layer
to layer and many forms of inter-stratified structures.
Raman spectroscopy has proved most useful for the determination of, and order –
disorder relationships in, the clay group of minerals, particularly kaolin structure (Frost
and van der Gaast, 1997).

13.12.2. 19- and 15-Å hydrate

The 19-Å andydrate is seem to be stable up to 200– 3008C under pressures along the
geotherm. This has important implications for the clay mineralogy of sedimentary basins
and petroleum generation.
In a sedimentary basin with K and other available cations, chemical alteration
converts montmorillonite to illite and releases H2O. However, this reaction is kinetically
favoured below 2008C. The results of Wu et al. (1997) indicate that, for any H2O-releasing
reaction at shallow depths (,5 km), the 19-Å hydrate can release much more H2O. This
H2O contributes significantly in the surface hydrodynamics, e.g., fluid migration in porous
layers and over-pressure build-up in closed systems.
Clays in the 15-Å hydration state are stable up to 450 –6008C at pressures above the
H2O liquid – vapour curve. If the environment is relatively free of ions that might break
956 Chapter 13

down montmorillonite, it can be transported down the subduction zone until H2O is
released. The H2O released from the subduction zone through dehydration also causes
partial melting of the overlying mantle. These melts are the sources of igneous intrusions
and volcanic activities along the plate boundaries.
Smectite clays are very ductile and, if they escape any chemical breakdown during
the process of subduction, they can remain stable to considerable crustal depths and may
serve as good seals for the oil reservoirs.

13.12.3. Interlayer cations

Huang et al. (1994) showed that Na-exchanged montmorillonite dehydrates

from the 19-Å to the 15-Å state at 330– 3858C along the ,0.80 g cm23 isochore.
The dehydration temperatures for other interlayer cations are shown in Table 13.17
(Wu et al., 1997).
The ratio of charge to ionic radius (ionic potential) vs the dehydration temperature
is plotted in Fig. 13.23. For Mg, Ca and Na ions, the dehydration temperatures for the
conversions 19 Å ! 15 Å and 15 Å ! 12.5 Å have opposite trends. The former decreases
with increasing ionic potential whereas the latter increases. The different trends indicate
differences in inter-molecular forces.
The long-range electrostatic attractive force between the negatively charged silicate
sheets and the positively charged cations is the dominant force binding the layers together.
The cations, in addition to providing the cohesive force, affect the configuration of the
inter-layer H2O molecules. The radius of Mg2þ ion is smaller than that of Ca2þ and also
Mg2þ has a higher charge than Naþ ion. Mg2þ strongly favours 6-fold coordination in most
silicates and oxides and favours 6-fold coordination with H2O molecules as well. This 6-
fold coordination controls the spacing to 15 Å and, hence, the Mg-exchanged
montmorillomite should have a 15-Å hydrate that is stable over a larger
temperature range than those of the Ca- and Na-exchange montmorillomite, as shown
in Table 13.17.

13.12.4. Kaolinite: Raman study

The application of Raman microscopy for studying the intercalated kaolinites has
proved to be most useful (e.g., Frost et al., 1997).
TABLE 13.17
Summary of dehydration temperatures (8C) for various interlayer cations

Cation 19– 15 Å 15–12.5 Å

þ
Na 330– 385 485–500
Ca2þ 260– 350 –
Mg2þ 200– 250 590–605

Source: Wu et al. (1997).

Hydrous Minerals 957

Raman microscopy has been successfully employed by Frost and ven der Gaast
(1997) for the determination of the structural elements and order –disorder in kaolinite. On
the basis of the ratio of the intensities of the two types of inner-surface hydroxyls, they
classified kaolinite at 3,685 and 3,695 cm21. A relationship between the degree of defect
structures and the intensity of the 3,685-cm21 band was found. In kaolinite, the kindred of
order –disorder is measured in terms of the Hinckley index (Hinckley, 1963). A
linear relationship between this index and the ratio of 3,685- and 3,695-cm21 bands
was found.
Frost et al. (1998) studied the effect of pressure (2 GPa/2008C) on the
intercalation of an ordered kaolinite with potassium acetate (KCH3COO). In addition
to the normal kaolinite bands, new Raman bands at 3,590, 3,603 and 3,609 cm21 were
seen to appear. These new bands are possibly due to the inner-surface hydroxyls
hydrogen-bonded to the acetate. Frost et al. (1998) proposed that, under pressure, the
intercalated ordered kaolinite becomes disordered because the hydrogen bonds between
adjacent layers are broken to create space for the intercalating agent between the layers.
More defects are also thereby created.

13.12.5. Chlorite

In hydrated peridotite, chlorite is compositionally near clinochlore. At pressures

above 2.1 GPa, the chlorite stability in peridotite is limited by the reaction:

chlorite þ orthopyroxene ¼ olivine þ garnet þ H2 O

In average mantle compositions, the talc (4.7 wt% H2O) þ olivine phase assemblage
decomposes to enstatite þ H2O at 6908C between 1 and 2 GPa (Ulmer and Trommsdorff,
1995).

13.13. Hydrous oxides

Hydrous oxide minerals may have an important influence on the properties of the
Earth’s interior, even if present in relatively small amounts. A full understanding of the
structure and stability of such phases requires very detailed characterization, including
complete spectroscopic studies as well as XRD.

13.13.1. Hydrous silica: Shergotty and LM

In Shergotty meteorite, silica occurs as an accessory phase associated with

felspathic glass, which was formed at an estimated pressure of 35 – 40 GPa (Malavergne
et al., 2001). The silica phase was found to be hydrous (30 ^ 3 to 81 ^ 15 ppm H2O) and
the hydrogen isotope signature is extra-terrestrial (Boctor et al., 2000). Coesite and
stishovite can accommodate structurally bound hydrogen. In coesite, hydroxyl solubility
increases with pressure (Koch-Muller et al., 2001) while in stishovite, its increase is
dependent on aluminium content (Powley et al., 1993).
958 Chapter 13

If post-stishovite phases contain structurally bound H, they may be candidates for

the storage of minor amounts of water in the lower mantle.

13.13.2. AlO(OH), diaspore

The Raman spectra of diaspore (AlO(OH)) were studied at #10 and 25 GPa at room
temperature by Huang et al. (1995). The vibrational frequencies of diaspore were found to
increase linearly with pressure. Seven vibrational modes showed a change in slope and the
break-points range from 7 to 15 GPa. The new values for the slope of these modes
range from 0.21 to 2.74 cm21 GPa21. No sign of dehydration in diaspore was found
up to 25 GPa.
The Raman spectra of diaspore (AlO(OH)) were studied by Huang et al. (1995)
using a methanol –ethanol medium at #10 and 25 GPa in a DAC at RT. A total of 16
Raman-active vibrational modes were observed during the compression.
The vibrational frequencies of diaspore were found to increase linearly with P;
seven vibrational modes showed a change in slope and break points range from 7 to
15 GPa. The new values for the slope of these modes range from 0.21 to 2.74 cm21 GPa21.
No sign of dehydration in diaspore was formed up to 25 GPa.
At pressures above 136 GPa, Al2O3 (in ruby) is reduced in the presence of hydrogen
to from diaspore (AlOOH) and aluminium. Ruoff and Vanderburgh (1993) determined the
bulk modulus of diaspore as K0 ¼ 170.3(^ 0.6) GPa, which results in a large negative DV
for the reaction:
Al2 O3 þ 0:75H2 ¼ 1:5 AlOOH þ 0:5 Al ð13-10Þ
It is concluded by Ruoff and Vanderburgh (1993) that the large and positive Gibbs
free energy of the reaction DG decreased rapidly with increasing pressure and eventually
became negative at 136 GPa (see Table 13.18, col. 4 and Fig. 13.24).
Later, Mao et al. (1994) studied the effects of pressure on lattice parameters
and molar bulk volume of diaspore to 25.5 GPa at 300 K by polycrystalline XRD
techniques. The pressure – volume data fit with a second-order Birch – Murnaghan EOS.
The relation

P ¼ 1:5K0 ½ðV0 =VÞ7=3 2 ðV0 =VÞ5=3 

yielded K0 ¼ 167.5 GPa and V0 ¼ 117:85 Å3 (presuming K 00 fixed at 4.0). Because powder
diffraction is relatively insensitive to minor transitions in this pressure range, there could
be the possibility of a change in space-group symmetry, the formation of super lattice or
second-order phase transition.
The Gibbs free energy for the reaction at ambient condition, DG0 ¼ 201:1 kJ, is a
very large positive number. The change of the Gibbs free energy of reaction at high
pressure (DG P) and constant temperature can be calculated by integrating the difference of
the volume of the products VP and the volume of the reactants Vr:
ð
DGP ¼ DG0 þ ðVP 2 Vr ÞdP ð13-11Þ
Hydrous Minerals 959

The results obtained by Mao et al. (1994) of the volume change of diaspore, VP – Vr
and DG P as a function of pressure is listed in Table 13.18. VP 2 Vr is seen to be mostly
positive above 10 GPa upto 130 GPa. DG P increases with pressure and remains a large
positive number.

13.13.3. Mg(OH)2, brucite

For the study of the behaviour of hydrous minerals in the Earth’s upper mantle, the
high-pressure behaviour of brucite (Mg(OH)2) is of interest, the more so for knowing the
dehydration reactions at high pressure and compression-induced amorphization. Besides
Mg(OH)2, a large number of magnesian silicates containing structurally bound OH (Finger
and Prewitt, 1989; Kanzaki, 1991) are supposed to be present in the Earth’s interior. Of
these OH-bearing phases, brucite is chemically and structurally the most simple and is a
useful prototype for hydrous and layered minerals to be studied under high pressure.
Brucite maintains a crystal structure under high pressure (Catti et al., 1995) while
portlandite (Ca(OH)2), which is iso-structural with brucite, undergoes pressure-induced
amorphization (Meade et al., 1992).

TABLE 13.18
Pressure volume (cm3 mol21), volume of reaction (13-11) and Gibbs free energy of reaction (Mao et al., 1994)

P (GPa) K0 ¼ 85 GPa,a K 00 ¼ 5:2;a DG P (kJ)a K0 ¼ 167 GPa,b K0 ¼ 4:0;b DG P (kJ)b

1.5VALOOH VP 2 Vr 1.5VALOOH VP 2 Vr

0.0 26.64 218,513.69 201.100 26.619 218,513.71 201.10

10.0 24.29 20.89 163.966 25.229 0.094 168.10
20.0 22.78 20.87 155.553 24.254 0.504 171.55
30.0 21.68 21.02 146.120 23.227 0.527 176.70
40.0 20.81 21.15 135.234 22.459 0.499 181.83
50.0 20.10 21.26 123.173 21.794 0.434 186.50
60.0 19.49 21.34 110.181 21.209 0.379 190.56
70.0 18.97 21.40 96.492 20.687 0.317 194.04
80.0 18.52 21.44 82.278 20.217 0.257 196.91
90.0 18.11 21.48 67.672 19.789 0.199 199.19
100.0 17.74 21.50 52.774 19.398 0.158 200.98
110.0 17.41 21.52 37.662 19.037 0.107 202.30
120.0 17.11 21.53 22.398 18.704 0.064 203.16
130.0 16.83 21.54 7.033 18.393 0.023 203.62
140.0 16.57 21.55 28.398 18.103 20.017 203.29
150.0 16.33 21.55 223.863 17.831 20.049 203.29
160.0 16.10 21.55 239.337 17.576 20.074 202.68
170.0 15.89 21.55 254.800 17.334 20.106 201.78
180.0 15.69 21.54 270.237 17.106 20.124 200.63
190.0 15.50 21.54 285.362 16.890 20.150 199.26
200.0 15.33 21.53 2100.977 16.684 20.176 195.88
a b
Ruoff and Vanderborgh (1993); Mao et al. (1994).
960 Chapter 13

Figure 13.24. The Gibbs free energy calculated on the basis of Mao et al., (1994) as a function of pressure. Ruof
and Vanderburgh (1993) curve, which decreases rapidly and crosses zero at 136 GPa, is plotted for comparison.
(Mao et al., 1994, q 1994 Elsevier Science Ltd.).

The EOS of brucite under high static pressures (Duffy et al., 1995a) and shock
compression (Duffy et al., 1991) have been determined while its structure and bonding
have been theoretically investigated using Hartree –Fock approximation (Sherman, 1991;
D’Arco et al., 1994).
High-pressure phase transition in Mg(OH)2 has been investigated by Raman
spectroscopy to pressures of 36.6 GPa under non-hydrostatic conditions and to 19.7 GPa
under quasi-hydrostatic conditions (Duffy et al., 1995b,c).
Portlandite (Ca(OH)2) (an isomorph of brucite) has been seen to amorphize at
11 GPa at RT (Meade and Jeanloz, 1990). The amorphization of Ca(OH)2 is regarded as a
result of a frustrated phase transition. However, such a phase transition has not been noted
in brucite by static or shock-compression experiments aided by infrared spectroscopy over
a broad P –T range.
Brucite crystallizes in the trigonal CdI2 layered structure ðP3m1Þ in which each
Mg ion is surrounded by a distorted octahedron of O atoms (Fig. 13.25). The Mg ions
lie in planes with the O ions above and below them in a sandwich arrangement. The
brucite layers are stacked along the c direction and held together by weak inter-layer
dipole forces. Thus, brucite is composed of two-dimensional sheets of Mg(OH)6
octahedra (brucite layer) along the c-axis. The OH dipole is oriented out of the brucite
layer and parallel to the c-axis. The OH ions lie alternately above and below the
neighbouring layers. The original angle between the OH dipole and the oxygen of the
 is calculated as 133.898 from neutron diffraction results
adjacent layer ðO – H· · ·OÞ
(Catti et al., 1995).
The hydroxyl ions are positioned along c on 3-fold sites, above and below
the octahedra. The brucite layers are held together by weak inter-layer dipole forces.
Above 4 GPa pressure, brucite shows new peaks in Raman spectra, implying a
structural change (Duffy et al., 1995b). To ascertain reliable, high P –T thermodynamic
properties of H2O from the dehydration reaction of brucite– periclase (e.g., Johnson and
Walker, 1993), an accurate EOS of brucite is required.
Hydrous Minerals 961

Figure 13.25. Crystal structure of brucite. The large spheres are O and intermediate spheres are Mg atoms, and the
small spheres are H atoms. The O –H bonds are also shown (modified after Duffy et al., 1995a).

13.13.3.1. XRD study

Duffy et al. (1995c) studied a single-crystal brucite at 14 GPa in a quasi-hydrostatic
pressure medium using a diamond-anvil cell and energy-dispersive synchrotron XRD.
The unit-cell parameters a and c show pressure dependence as:
a ¼ 3:145 2 0:0106 P

c ¼ 4:769 2 0:0613 P þ 0:0021 P2 ðP in GPaÞ

The c parameter decreases much more rapidly than a and b, manifesting a marked
non-linear pressure dependence, as displayed by c/a ratio in Fig. 13.26.
The c/a ratio shows pressure dependence as:

c=a ¼ 1:516 2 0:0143 P þ 0:0006 P2

Figure 13.26 shows that this ratio initially decreases strongly with pressure and
becomes nearly pressure independent above 8 GPa.
The parameters of a third-order Birch– Murnaghan EOS are fitted to the data of
Duffy et al. (1995b) and the obtained values are K0T ¼ 42ð2Þ GPa and ðdK0T =dPÞ ¼
5:7ð5Þ: Catti et al. (1995a), however, obtained K0T ¼ 39ð1Þ GPa and K 00T ¼ 7:6ð7Þ
(Table 13.19).
962 Chapter 13

Figure 13.26. Pressure dependence of the c/a ratio in brucite (Duffy et al., 1995b,c).

The EOS parameters obtained by different methods are presented in Table 13.19.
However, non-hydrostatic stresses may significantly affect EOS determinations,
even for apparently low-strength materials like brucite. A mixture of 50% NaCl may
reduce the effects of deviatoric stresses at low pressures. Pressures determined by neon
EOS in neon medium experiments are believed to be more reliable than those from
the ruby.

13.13.3.2. IR study
The inter-layer proton transfer in brucite under pressure was investigated by
Shinoda and Aikawa (1998) using polarized IR absorption spectra of a single crystal of
brucite. Measurements were made by Fourier transform polarized micro-spectroscopy
with DAC.
Under 2.9 GPa pressure, the observed absorption peak at 3,645 cm21 is due to a
secondary OH dipole, which is found to be oriented 1368 to the c-axis at 5.3 GPa. Abrupt
onset of the secondary peak and its reverse pleochroism suggest that the process of

TABLE 13.19
Equation of state parameters of brucite (Mg(OH)2)

Study Method K0T (GPa) K 00T

Duffy et al. (1995a) Single-crystal X-ray diffraction 42(2) 5.7(5)

Saxena (1989) Optimized phase equilibrium data 57.1 4.7
Duffy et al. (1991) Shock compression 51(4) 4.6(4)
Fei and Mao (1993) Powder X-ray diffraction 54(2) 4.7(2)
Catti et al. (1995) Powder X-ray diffraction 39(1) 7.6(7)
Hydrous Minerals 963

Figure 13.27. IR absorption spectra of brucite as a function of pressure (Lu et al., 1999, q 1999 The Clay
Minerals Society).

secondary OH dipole formation is due to proton transfer between layers in brucite. The
secondary OH dipole implies a new site of proton in brucite under pressure (Shinoda and
Aikawa, 1998).
Goncharov et al. (1999) reported IR absorption spectra of brucite single crystals
with the synchrotron technique. Measurements were carried out at U2B NSLS beamline
using a FT-IR spectrometer and DAC. The IR spectrum of brucite in the mid-IR consists of
an O – H stretching fundamental and combination bands involving this mode and low-
frequency lattice vibrations. The IR absorption spectra of brucite as a function of pressure
are presented in Fig. 13.27.
At 3.7 GPa, a new band appears at lower frequencies and gains intensity and softens
with further increase in pressure. The initial O –H fundamental broadens under pressure
and gradually loses intensity. On pressure release, the initial IR spectrum is restored but
with some hysteresis and the appearance of a narrow band in the vicinity of the main
fundamental. This may be due to a phase transition involving displacement of hydrogen
atoms from their original axial sites (e.g., Duffy et al., 1995a).

13.13.3.3. Raman study

The Raman spectra of brucite exhibit a number of unexpected features at
high pressures. A new line appears on the high-frequency side of the Eg ðTÞ mode.
964 Chapter 13

Figure 13.28. Raman spectra of brucite under quasi-hydrostatic conditions to 10 GPa. (a) External modes,
(b) internal modes. New Raman peaks appear beginning at 4.4 GPa. An expanded version of the 360 cm21 region
is shown above each trace in (a). The inset in (b) shows the appearance of the high-frequency side band at 4.4 GPa
(From Duffy et al., 1995b, q 1995 Mineralogical Society of America).

With pressure, the Eg ðTÞ peak decreases in amplitude and the new peak increases such that
they are of roughly equal intensity near 20 GPa. The other lattice vibration, A1g ; becomes
weak and broad at high pressure and is hardly detectable above 20 GPa.
At high pressures, the O –H stretching vibration weakens and broadens. The peak
near 3,600 cm21 increases in intensity with pressure but the other two modes remain weak.
The frequencies of the lattice vibrations increase with pressure whereas the O –H stretch
exhibits a negative frequency shift.
The appearance of new bands in the Raman spectrum of brucite indicates a
pressure-induced structural change (Fig. 13.28), either through a reduction in symmetry
Hydrous Minerals 965

or an increase in the number of atoms in the unit cell. The new peak near the Eg ðTÞ
mode extrapolates to 360 cm21 at ambient pressure, close to the Eu ðTÞ infrared-active
mode at 361 cm21. Two of the three new Raman lattice vibrations correspond
to frequencies of infrared fundamental modes which occur at 461 and 3,688 cm1
(Dawson et al., 1973).
The low-frequency modes of brucite display the characteristics of resonant
interaction: at higher pressures, the intensity shifts from one mode to the other and the
intensities are approximately equal at the point of closest approach.
The pressure dependence of the intensity ratio is approximately exponential, as
expected for a resonance interaction (Hanson and Jones, 1981). In the absence of
resonance, perturbation theory can be used to determine the positions of the peaks.
The relationship between the perturbed (y þ and y 2) and unperturbed (y a and y b) frequencies
is given by:
ðy þ and y 2 Þ2 ¼ ðy a and y b Þ2þ þ 4d2
where d is a coupling constant that is assumed to be pressure independent and is given by the
minimum value of the half-separation (Lewis and Sherman, 1979). The unperturbed
frequencies can be obtained by solving the above equation for y a and y b.

13.14. Portlandite (Ca(OH)2)

Portlandite (Ca(OH)2) is iso-structural ðS:G: ¼ P3m1Þ with brucite (Mg(OH)2) and

is formed by stacking (001) layers of edge-sharing CaO6 octahedra.
High-pressure XRD experiments on a powder sample up to 376 GPa (Meade and
Jeanloz, 1990) led to amorphization between 10.7 and 15.4 GPa. Later, a high-pressure
neutron powder diffraction study was carried out using the ISIS (UK) facility up
to10.9 GPa by means of a Paris-Edinburgh cell installed on the POLARIS diffractometer
(Pavese et al., 1997).
The variation of the cell parameters a and c with pressure P (in GPa) is seen to bear
the following relationship:
a ¼ 3:580ð2Þ 2 0:019ð1ÞP þ 0:0006ð1ÞP2 and
2
c ¼ 4:900ð4Þ 2 0:076ð3ÞP þ 0:0026ð2ÞP
As expected for sheet structure, the c trend is much steeper than that of a, confirming that
the strain mainly affects the [001] direction by a shortening the inter-layer distance
between (001) octahedral sheets.
Portlandite (Ca(OH)2) is seen to be softer than brucite (Mg(OH)2) (K0 ¼ 41ð^2Þ;
K0 ;a ¼ 313; K0 ;c ¼ 57 GPa) and the replacement Ca/Mg mainly affects the elastic
properties in the (001) plane rather than along the [001] direction.
This page is intentionally left blank
 967

Chapter 14
Iron and Siderophile Elements: The Earth’s Core

14.1. Introduction

If we presume an approximate homogeneity in the element distribution in the Solar

System, the core has to be composed essentially of iron. In addition, iron is believed to be
the main component of the Earth’s core on the basis of a combination of evidence from
cosmochemistry (e.g., Anders and Ebihara, 1982), geomagnetism (e.g., Jacobs, 1987),
seismology and high-pressure experimentation (e.g., Brown and McQueen, 1986). With
the availability of high-resolution data from seismic arrays and from the studies of the
inverse problem of the Earth’s free oscillations and seismic travel times, attempts to learn
about the Earth’s core have been intensified. The other impelling reason for the belief
about what constitutes the Earth’s core is that the volume of the mantle is much greater
than of the core and, as iron is depleted relative to silicon in the Earth’s mantle, it has had to
be greatly enriched in the Earth’s core. The inner core is presumed to be pure iron, while
the outer core is considered to be a mixture of iron and some solute.
At core pressure, however, silicates could also transform to metallic conductor but
the density change would be small compared with what is required for the core – mantle
boundary. Also, a silicate core cannot possibly duplicate the known sound velocity of
the core.
To understand the properties and processes of the Earth’s deep interior, its
temperature profile, chemical composition, energy balance, dynamics and geomagnetism,
an examination of iron at high P – T becomes important.
Electronic character of iron. Iron is the most stable nucleus in the periodic table. An
iron atom in its free state has the electronic configuration of 3d64s2 in its ground state. For
the incompletely filled 3d82n4sn (where n takes values 0, 1 or 2), the environment around
the atom controls the value of n. A decrease in inter-atomic spacing will cause a decrease
in n and a pressure-induced transition ultimately leads to a 3d8 configuration. At core
pressure, iron atoms may undergo transition to the 3d8 state.
Iron atoms under pressure attain anisotropic character. In 3d10 configuration, the 3d
orbitals provide the spherical symmetry but the partly filled 3d6 orbitals (of Fe2þ ion) would
adopt a lower symmetry. Under pressure, states with an increase in electron density are
favoured in the electronic continuum, resulting from a combination of some 3d-orbitals.
(Note: g-iron would show 3d electronic states more degenerate with lower electron density
when the atoms adopt a mmm point symmetry and iron symmetry in 1-hcp form is 6m2:)

968 Chapter 14

From a quantum mechanical point of view, a study of iron under pressure is

essentially the solving of a many-body problem, wherein the solution lies in the reduction
of the many-electron Schrodinger equation to an approximate equation involving a more
manageable number of degrees of freedom. The 4s band becomes empty at approximately
4- to 3-fold compression, which is known through using the spherical Wigner-Seitz
approximation that presumes the atomic shell to be degenerate at the centre of the Brillouin
zone). In dealing with crystalline iron, the total electronic wave-function is expressed in
terms of single-electron eigen states.
To achieve good accuracy for high-pressure iron, the choice of orbitals to be treated
as valence orbitals is crucial. At ambient pressure, reasonable accuracy can be obtained by
treating all atomic states up to and including 3p as core states. The 3p states respond
significantly at high pressures. Raman studies under pressure have shown splitting of
Raman lines. This pressure dependence indicates that iron has anti-ferromagnetic
correlations below 60 GPa.

14.1.1. Theories of iron under pressure

Density functional theory (DFT) (see Section 4.5.1.3) helps us understand the
physics of iron at the P – T of the Earth’s core. LAPW calculations and the tight binding
method (parametric extensions of first-principles) allow one to predict the physical
behaviour of iron at extreme conditions. These theoretical calculations coupled with
experimental results constrain the composition of core, its thermal state, the crystalline
structure and the origin of its anisotropy.
DFT eliminates the bcc phase of iron as a likely constituent of the inner core
because of its larger volume with respect to hcp and fcc phases and the highly unfavourable
energetic state of bcc at high pressure. The bcc phase is also mechanically unstable as it
violates the Born stability criterion (see Section 5.1.6) above 150 GPa. LAPW calculations
of the total energy of iron in GGA and LDA approximations are shown in Fig. 14.1
(Stixrude et al., 1996).
The first-principles calculations show that hcp and fcc phases are non-magnetic at
high pressures (in great contrast to the familiar low-pressure phase). This means not only
that the net magnetic moment is zero but also that magnetic moments do not exist, even
locally, as the spin pairing is complete. Above its Curie temperature, the net magnetic
moment may be zero but local moments (although disordered) remain virtually
undiminished in magnitude from their values in their ferromagnetic, low-temperature
state. This discussion is elaborated further as follows.
At high temperature, the entropy of orientational disorder of magnetic moments
becomes unfavourable and above the Curie temperature this leads to a vanishing net
magnetic moment. However, local magnetic moments may survive. At core pressure, even
the local magnetic moments themselves vanish.
GGA calculations at zero pressure of the ferromagnetic state give the theoretical
magnetic moment as 2:174mB (close to the experimental value of 2:12mB ). The magnetic
moment of ferromagenetic bcc iron vanishes at high pressure. The density at which the
moment approaches zero corresponds to that of the inner core (where V < 48 Bohr3).
Iron and Siderophile Elements: The Earth’s Core 969

Figure 14.1. LAPW calculations of the total energy of iron in GGA (top) and LDA (bottom) approximations.
Results for ferromagnetic bcc (filled squares) non-magnetic fcc (open squares) and hcp (circles) are shown.
Results for both ideal (open circles) and minimum energy (filled circles) c=a ratios are shown for the hcp phase.
The inset shows the magnetic stabilization energy of bcc: the difference in total energy between non-magnetic and
ferromagnetic states (from Stixrude et al., 1994, q 1994 Physical Society of America).

The dependence of the magnetic moment on volume is quasi-linear between V ¼ 60

and 90 Bohr3 but then begins to decrease very rapidly at smaller volumes (Stixrude
et al., 1998).
By applying the solid-state theoretical method, however, one can predict the
elasticity of iron at high pressure. The calculations are commonly based on a Slater –
Koster total energy, tight-binding Hamiltonian (Cohen et al., 1994) and also on LAPW
calculations. The latter calculations are parameter-free and are completely independent of
experimental data.

14.1.1.1. Energy bands and electron transitions at core

Bukowinski (1976) outlined the basis and results of the theoretical calculations on
the electronic structure of iron lattices at core pressure. An outline of his model is
presented below.
970 Chapter 14

The calculations of Bukowinski (1976) indicate that the form of the 3d and 4s bands
of fcc iron is very stable to at least 2-fold compression (s ¼ 5:3138 Bohr units). The 4s
band goes above the Fermi level at approximately 4-fold compression (relative to fcc iron
at 0 K). While this transition is continuous in terms of the fractional occupation of the 4s-
derived band, it causes a discontinuous change in the topology of the Fermi surface. The
angular momentum characters of the conduction electrons have the symmetry of the lattice
and, as such, have contributions from several angular moments. Just after the transition
(when 4s band becomes empty), the band loses most of its s-character and a hole state
develops around the centre of the Brillouin zone corresponding to this band. (Note: A
portion of the band in the neighbourhood of the point is above the Fermi level, i.e., empty.)
Even if the core is assumed to be as hot as 10,000 K, the value of kT would still be
significantly smaller than the width of the occupied conduction band of iron. On this basis,
Bukowinski (1976) therefore concluded that there can be no electronic transitions within
the Earth’s core.

14.1.1.2. Phase predictions from theoretical calculations

Functional theory in the generalized gradient approximation (GGA) yields a good
description of the static EOS and the low-temperature phase diagram of iron (Stixrude
et al., 1994). Wasserman et al. (1996) investigated the thermoelastic properties of close-
packed phases of iron at pressures up to 400 GPa and temperature to 6,000 K using a tight
binding total-energy (TBTE) method and the cell model of the vibrational partition
function.
In the cell model (“particle in a cell” method, each atom is confined in the
Weigner – Seitz cell formed by its nearest neighbours. This model includes anharmonic
terms and can be applied at temperatures below the melting temperature (where collective
motions and diffusion become important) and also above the Debye temperature (where
vibrational states are fully populated). Compared with molecular dynamic simulations, the
cell model permits very efficient computation, thereby allowing investigations over a wide
range of thermoelastic parameters (over the entire range of temperatures and pressures
relevant to Earth at its depths).
Local density approximation (LDA) fails to predict the correct ground state of iron,
incorrectly finding that the hcp phase has a lower total energy than the bcc. The generalized
gradient approximation correctly recovers the bcc ground state. Electronic exchange
correlation is treated using the GGA of Perdew et al. (1992). It has been shown that the
GGA reproduces the density of the hcp structure over the whole high-pressure range within
, 6%. The equilibrium density and magnetic properties of the bcc structure of iron at
ambient pressure are also accurately described. Moreover, it accurately predicts the
pressure of the phase transition from (a-) bcc to hcp (1-) phase near 11 GPa (Stixrude et al.,
1994) (Fig. 14.2). This is an important result because the energetics is particularly subtle in
the case of this transition since it involves a ferromagnetic and a non-magnetic phase.
These calculations find that the hcp phase is the stable low-temperature phase of iron at
pressures beyond 11 GPa, in excellent agreement with the experiment of Mao et al. (1990).
The room temperature P – V EOS for iron (see Mao et al., 1990) compares well with the
GGA calculations (Stixrude et al., 1991) (see Fig. 14.1).
Iron and Siderophile Elements: The Earth’s Core 971

Figure 14.2. Room temperature pressure–volume equation of state for iron (see Mao et al., 1990) compared with
the theoretical calculations with the generalized gradient approximation (Stixrude et al., 1994, q 1994 Physical
Society of America).

P-wave anisotropy. The agreement between the first-principles theory and

seismological observation strongly supports the hypothesis that the P-wave travel-time
anomalies are caused by elastic anisotropy in the inner core and the region is composed of
hcp iron.
The elastic and also the P-wave anisotropy of hcp correspond to the anisotropy of
the inner core. An aggregate in which the c-axis of the constituent hcp crystals are nearly
aligned (at # 10% angle) with the spin-axis can account for 60% of the variance in BC –DF
travel time anomalies. Of course, a possibility exists that the inner core is composed of a
different hexagonal phase (e.g., dhcp), which may be similar to hcp elastically and
energetically (see Section 14.2.5).

14.1.1.3. First-principles approximation: bcc and hcp

There are extensive first-principles calculations on crystalline iron over the pressure
range from zero up to the Earth’s core values (Stixrude et al., 1994; Söderlind et al., 1996).
De Wijs et al. (1998) used first-principles molecular dynamics simulations based on ultra-
soft pseudo-potentials of the Vanderbilt type (1990), which allow a significant reduction in
the computational effort with no loss of accuracy.
Wijs et al. (1998) calculated the values of the atomic volume and bulk modulus of
the bcc and hcp phases of iron and the magnetic moment of the bcc phase at ambient
pressure. The obtained values are compared in Table 14.1 with those obtained by
experimental (Ahrens, 1995; Lonzarich, 1980), pseudo-potential and all-electron
calculations (Stixrude et al., 1994).
First-principles calculations show that it is highly unlikely that bcc will reappear as
a stable phase at extreme pressures and temperatures, independent of the form of the
972 Chapter 14

TABLE 14.1
Comparison of pseudo-potential calculations with all-electron calcul ations and experimental values of bcc and
hcp iron

Phase Property All electron calculations Pseudo-potential Experimental values

(de Wijs et al., 1998) calculations (Ahrens, 1995;
(Stixrude et al., 1994) Lonzarich, 1980)

bcc V0 (Å3) 11.4 11.55 11.80

bcc K (GPa) 189 176 162–176
bcc m (mB per atom) 2.17 2.25 2.12
hcp V0 (Å3) 10.2 10.4 11.2
hcp K (GPa) 291 290 208

V0 ; volume; K; bulk modulus; m; magnetic moment.

exchange-correlation potential. The reason is that bcc is found to undergo an elastic

instability with respect to a tetragonal strain at high pressure. At pressures beyond
150 GPa, the bcc structure will spontaneously distort (Stixrude and Cohen, 1995).

14.1.2. fcc and hcp phases

The two close-packed crystalline structures, fcc and hcp, are common in nature:
25% of the elements in the periodic table crystallize in fcc structure and 20% in hcp.
Thermodynamic equilibrium requires that crystallizing atomic or molecular fluids adopt
the structure which would have the lowest Gibb’s free energy under external conditions of
constant temperature and pressure.
For both the fcc and hcp crystal structure (Fig. 14.3), all the primary cells are
regular dodecahedral in shape, each of the 12 faces representing the perpendicular plane
bisects the line between the first-neighbour sites (see Fig. 14.4). The fcc and hcp lattice

Figure 14.3. (a) Plan view of the hard sphere ABCABC… stacking characteristic of the close-packed plane in the
fcc lattice. (b) Plan view of the ABABAB… stacking of close-packed planes leading to the hcp structure.
Iron and Siderophile Elements: The Earth’s Core 973

Figure 14.4. Iron hcp lattice projected on the 001 plane. One can observe the families of (100) dense atomic
p
layers, which show spacial configuration, separated in two sub-layers at a 3=6 from each other (Andrault et al.,
2000, q 2000 Mineralogical Society of America).

differ from each other only in the third and successive neighbours (see Fig. 14.3). The
coordination numbers of the first five coordination shells in the fcc and hcp lattice structure
are shown in Table 14.2 below.
The inner core is likely composed of a close-packed structure, either hcp or a
similar hexagonal or nearly hexagonal phase, or fcc (Stixrude et al., 1998). The hcp phase
is seen to be the low-temperature phase stable from 11 GPa to beyond the inner-
core pressure. The fcc and hcp have identical close-packed volumes and very similar

TABLE 14.2
Properties of the first five coordination shells

Neighbour
First Second Third Fourth Fifth

fcc
Numbers of neighbours 12 p6 p24 p12 p24
Separation/d0 1 2 3 4 5
hcp
Numbers of neighbours 12 p6 p 2 p18 p 12
Separation/d0 1 2 8=3 3 11=3

Note: In the single-occupancy cell model (the SO model) at intermediate densities around the vicinity of the phase
transition, collisions can occur between third neighbours in the fcc structure and up to fifth neighbours in the hcp
structure.
974 Chapter 14

EOS but, because of structural differences, they might have different thermodynamic
properties and stabilities. Woodcock (1997) found that fcc phase is more stable by around
0:005RT per mole (R, universal gas constant) and the entropy difference is , 0:005 R for
all temperatures up to the melting point.
The calculation of the stability relation of the fcc and hcp phases of hard spheres is a
long-standing problem in statistical physics. The fcc phase is the more stable crystal phase
for hard spheres (Woodcock, 1997). The total energy for iron is shown in Fig. 14.5 (read
the caption). The fcc crystal is indeed more stable than the hcp crystal. The free-energy
difference between hcp and fcc at melting is DF 0 ¼ 0:00087 ^ 20 and, at close packing,
DF 0 ¼ 0:00094 ^ 30: fcc shows lower free energy at close packing. The change in
Helmholtz free-energy difference between close packing and the melting volume amounts
to only 0:0003ð1ÞNKB T: Woodcock (1997) reports a small positive difference in the
reduced Gibbs free energy, which is equivalent to a difference in the reduced Helmholtz
free energy, DF 0 ð; ðFhcp 2 Ffcc Þ=RTÞ ¼ 0:0005; at the melting density. He reported a
substantial area of pressure difference (DP) between the fcc and hcp single-occupancy cell
models, which arises because of the difference in the order – disorder transition pressures.

Figure 14.5. Total energies calculated for iron. Inset shows the energy of the nonmagnetic bcc and fcc phases
relative to that of the hcp phase. Energy differences are expressed as temperature per atom. The inner core
pressure and the range of proposed temperatures are shown. The bcc–hcp energy difference is much larger than
the available thermal energy, suggesting that the inner core cannot be made of bcc iron (1 Rydberg ¼ 13.6 eV)
(modified from Bukowinski 1994, and Cohen et al., 1997).
Iron and Siderophile Elements: The Earth’s Core 975

This pressure difference implies an entropy difference at constant volume, which equals
the Helmholtz free-energy difference for hard spheres.

14.1.2.1. P – r relationship
The fcc solidus is controlled by an EOS which must satisfy the known
thermodynamic data of the fcc phase, the shock-wave data (e.g., from the derived
Hugoniot) and the melting theory.
Bullen (1968) found a linearity between pressure ðPÞ and K for both the lower
mantle and the outer core where the temperature increases with pressure (Note: In an
isothermal situation, this EOS does not work.] The K and P relationship can be obtained
(exploring Hugoniot) as:
K ¼ K0 þ K 00 P ¼ 160 þ 4P ð14-1Þ
The generalized value, where the Hugoniot crosses the liquidus, is: Tm ¼ 5,400 ^ 400 K
at P ¼ 243 ^ 2 GPa, with the mid-point value of q ¼ 1:7: For fcc phase, q ¼ 1:7 is
analogous to the solution for q ¼ 1:62 for hcp phase (Jeanloz, 1979). KP of fcc solidus is
less than the KP of hcp solidus.
For the hcp phase boundary, the K – P linear relation is found to be (Anderson,
1986):
K ¼ ðK0 þ K 0o PÞ ¼ 173 þ 5:3P ð14-2Þ
and

r=r0 ¼ ½1 þ 5:3P=173ð1=5:3Þ ð14-3Þ

and the density of the integrating constant is:

r0 ð0:750Þ ¼ 8:20 g=cm3 ð14-4Þ

The three parameters r0 ; K0 and K 00 are determined by experiment. At absolute zero, the
values of r0 ; K0 and K 00 have been determined by Anderson (1986) as:
r0 ¼ 8:29
K0 ¼ 175 GPa
and
K 00 ¼ 4:4
and K 00 is dimensionless (K 00 is the second derivative of K 0 with respect to P at zero
pressure).
The density – pressure trajectory is fixed when the EOS has been selected.
Using relations (14-3) and (14-4), the equation for density for the hcp phase at the
phase boundary can generally be obtained as:

r=8:20 ¼ ½1 þ 5:3P=173ð1=5:3Þ
976 Chapter 14

The P – r equation for the fcc solidus is:

r=r0 ¼ ð1 þ 4P=160Þ0:25 ðsince KS ¼ rðdP=drÞÞ ð14-5Þ

where r0 is the uncompressed density of the fcc solidus (at 1,990 K).
P is related to K and V as:

dP ¼ 2ðK=VÞdV
P P
At the triple point DV ¼ 0; DS ¼ 0 and dT=dP ¼ 2DS=DV:
As pressure increases, DV should diminish.
The variation of elastic constants ðcij Þ of fcc and hcp as a function of pressure is
presented in Fig. 14.6a –c.

Figure 14.6. Elastic constants of fcc (a) and hcp (b) iron as a function of pressure (from Stixrude et al., 1998).
(c) Elasticity tensor of hcp iron as a function of pressure at ambient temperature (Mao, et al., 1997).
Iron and Siderophile Elements: The Earth’s Core 977

14.2. Iron core

The mass of the core amounts to , 32.5% of the Earth It is the dominant repository
of siderophile elements in the Earth but is mainly constituted of Fe – Ni alloy. Assuming
that the bulk Earth has a chondritic Fe/Ni ratio (i.e., ,17) and the mantle has a ratio of 31.9,
then the core should have an Fe/Ni ratio of ,16.
A limited amount of Si (#5%) may have entered the core (O’Neill, 1991). As a
result of the observed depletion of V, Cr and Mn in the silicate Earth, these elements are
assumed to be enriched at the core (McDonough and Sun, 1995). Ringwood et al. (1990)
have argued for high-pressure metal –silicate equilibrium in the Earth’s mantle during core
formation. Cs and possibly other lighter alkali metals were partially sequestered into
the core.
Under high pressure, Ti, Cr, Mn, Co and P are seen to manifest sidero-
phile tendencies and hence these may be present as sulphides at core pressure. But
titanium is never seen as a siderophile element in the Earth nor in meteorites.
However, some astrophysicists speculated the presence of titanium sulphide at the
galactic centre.
Theoretical studies of the Earth’s core are carried out based on the high-
resolution data from seismic arrays and on the studies of the inverse problem of
the Earth’s free oscillations and seismic travel times. However, these studies only
delineated the mechanical properties of the core, not the electronic structure of iron.
Yet solutions of the inverse problem have yielded some anomalous properties
of the core, viz. extremely high Poisson ratio for the inner core and the possibility of
a sub-adiabatic temperature gradient in the outer core (Higgins and Kennedy, 1971).
From high-resolution data obtained from seismic arrays and from studies of the
inverse problem for the Earth’s free oscillations and seismic travel times, the theoretical
studies of the Earth’s core have gathered strength. Knowledge of an apparent sub-adiabatic
temperature gradient in the outer core and an extremely high Poisson’s ratio for the inner
core has become useful. Theories of viscoelasticity and metallic pseudo-potentials suggest
that possibly a highly viscous dense inner core may transmit shear waves at seismic
frequencies. Thus, the two types of observations that have led to speculation of the
structure of the core are: (i) the travel time and waveforms of body waves (P- and S-waves)
that traverse the core and (ii) the frequencies of free-oscillation modes of the Earth that are
sensitive to core structure. The travel times and free-oscillation frequencies depend on the
density and the elastic constants of the Earth’s interior. From seismological results, the
density and the elastic-wave velocity of the spherically symmetric structure of the core
have been derived.
Hemley et al. (2000) studied the shear modulus, ðGÞ; crystal elasticity tensor ðCij Þ;
aggregate Vp and Vs and orientation dependence of Vp and Vs of iron with the radial XRD
technique (Section 4.4.7.1). The aggregate velocities, Vp and Vs, are presented in
Fig. 14.12, along with shock-wave data at high temperatures.
Adopting the assumption that the core temperature is 6,000 K and 2dVp =dT ¼
2:9 £ 1024 km s=K (Brown and McQueen, 1986), the velocities obtained at core conditions
978 Chapter 14

seem to be higher than those obtained from seismological data. This offers an insight into
the presence of light elements in the inner core.
An insight into Earth’s core requires a knowledge of the behaviour of iron at
pressures up to 360 GPa (3.6 Mbar) and temperature up to 7,000 K. Advances in laser-
heated diamond-anvil cells and the use of synchrotron radiation sources have made such
studies possible. The very high-pressure experiments require a fine incident X-ray beam
(8.9 m2 or less) of l ¼ 0:4 Å or less, usually from a synchrotron radiation facility (e.g.,
ESRF, Grenoble, France; see Dubroviousky et al., 1999). Again, shock-wave studies have
helped to determine the high-temperature isotherm for the 1-phase to 200 GPa but, because
the Hugonoit intersects the phase boundary of the melt – solid state, useful data on iron
cannot be obtained by shock studies.

14.2.1. Core iron

Iron in the Earth’s inner core is important to the dynamics of the outer core,
the behaviour of the magnetic field and the thermal state of the planet. Direct measure-
ment of properties at core pressures i.e., , 350 GPa and temperature (, 6,000 K) is
not yet feasible but calculations based on first-principles theory can estimate these
constants.
Cohen (2000) has found a stable anti-ferromagnetic hcp iron which is more stable
than non-magnetic hcp mass (GL Yearbook, 00/01, p. 25). He has discovered an even more
stable non-collinear magnetic structure, which can account for the newly discovered
superconductivity in hcp iron. In this, the magnetic moments of iron atoms in different
layers are aligned in opposite directions (anti-ferromagnetic moments). The moment in
each layer is rotated by 908 from the previous one, giving a non-collinear structure.
However, collinear magnetism is frustrated in hcp structure because it is impossible to
make all of the moments anti-ferromagnetic with respect to one another (GL Yearbook,
00-01, p. 25).

14.2.2. Anisotropy

The solid inner core of the Earth is a sphere of iron alloy with a radius of 1,220 km,
which is one-third of the size of Earth’s liquid core. One of its remarkable properties is its
anisotropy, because of which seismic waves propagate faster parallel to the Earth’s
rotation axis than perpendicular to it. This (3 – 4%) anisotropy in seismic velocities is
probably due to the prevailing orientation of iron crystals that constitute the inner core. The
mechanism for this orientation may be related to magnetic and dynamical processes. This
is further discussed in Section 14.2.10.4.
The faster speed along the rotation axis, observed by seismology, can be effected by
30% alignment of hcp grains along the rotation axis. Computation of elastic constants
along the principal crystal axes of the hcp (non-magnetic) iron has revealed thermal
dependence, such that the direction of the fastest sound speed reverses at higher
temperature.
Iron and Siderophile Elements: The Earth’s Core 979

14.2.3. EOS and melting

Phase relations, thermodynamic and thermoelastic properties of iron have been

investigated by different methods: XRD, visual observation of melting, spectro-
radiometric measurements of temperature, etc.
For high-pressure 1-Fe (or g-Fe) phase, it becomes scarcely possible to quench such
a phase at low pressure and temperature (Saxena et al., 1995). Therefore, to measure the
EOS, one has to measure the unit-cell volume accurately by XRD over a wide range of
temperature and pressure. Using XRD under pressures up to 68 GPa, Dubrovinsky et al.
(1998) found the following P – V – T data for 1-iron:

Iron 1-Fe
Isothermal bulk modulus K300,1 (GPa) 164 (3)
Pressure derivative K300,1 5.36 (16)
Temperature derivative ðdKT:1 =dTÞP (GPa/K) 20.043 (3)
Molar volume V300,1 (cm3/mol) 6.76 (2)

Isobaric thermal expansion for 1-Fe at 1 atm (0.101 MPa) is given by (K21):

sT ¼ 5:7ð4Þ1025 þ 4:2ð4Þ1029 T 2 0:17ð7Þ=T 2

The melting point of iron rises from 1,5358C at 1 atm to , 1,7008C at 5 MPa. The variation
of melting temperature with pressure is given by:

dT
¼ DV=DS
dP

where DS and DV are the entropy and volume changes associated with melting.
Melting on iron Hugoniot at 243 GPa was noted by Brown and McQueen (1980).
Shocked iron is melted at pressures above 243 GPa and since, Hugoniot temperatures
overlap core temperatures in this regime, direct comparisons between the Earth and shock-
wave data require small thermal corrections. However, Hugoniot data agree well with
static data extending to 300 GPa. A plot of sound velocity against pressure for pure iron
reveals discontinuities due to hcp –fcc and fcc –liquid transition (Fig. 14.7, read the
caption).
A discontinuity in sound velocity offers evidence of melting (see Fig. 14.7).
A second discontinuity at 200 GPa is interpreted as resulting from the solid –solid 1 – g
transition. The longitudinal wave velocity increases smoothly with compression until a
pressure of 200 GPa, where it decreases by several % over a narrow pressure range
(, 5 GPa) before increasing smoothly again upon further compression. At , 243 GPa,
melting is identified with a second decrease in longitudinal wave velocity.
The experimental EOS for phases of iron at P ¼ 0 and T ¼ 300 K are shown in
Table 14.3. Using these values of av ; g and T; the isothermal bulk modulus is found to be
84.7 GPa at 1,181 K. The change in density, dr=dT; was found by Anderson and Ahrens
(1994) to be 2 0.64985 £ 1023 g/cm3 K.
980 Chapter 14

Figure 14.7. Sound velocity VP versus pressure for pure iron (Brown and McQueen, 1982, 1986). The first peak is
due to hcp–fcc transition, and the second peak is at the fcc–liquid transition. PREM refers to the model of the
Dziewonski and Anderson (1981) seismic model.

The value of the Grüneisen parameter for liquid iron is about 1.35 at
r ¼ 12:6 g/cm3 (Brown and McQueen, 1986). This value may be computed throughout
the core.
The slope of the solid –solid phase boundaries for monatomic solids is empirically
found to be nearly linear in P – T space, except when magnetization is involved. The
dT=dP derivative will be sensitive to the magnetic term in the Helmholtz energy because
the d phase is non-magnetic and the solid – solid boundaries above 2,000 K are linear in
T – P space.

14.2.4. Density deficit

As stated earlier, the density of the iron core, derived from seismic measurements,
is , 8 – 10% less than the density of pure iron obtained from theory or HP measurements.
The significantly abundant light elements in the (outer) core are presumed to be H,
C, O, Mg, Si and S (Poirier, 1994). Stixrude et al. (1998) have shown stabilization of iron
hydrides and an insulator –metal transition of MgO at high pressure.
The 10% density deficit of the core is fully accounted for by silicon, sulphur and
oxygen. A small amount of carbon (, 1%) could also be present in the core (Wood, 1993).
Following Poirier (1994), the composition of the alloys giving a core density deficit of 10%
 Iron and Siderophile Elements: The Earth’s Core
TABLE 14.3
Experimental EOS parameters for phases of iron at P ¼ 0 and T ¼ 300 K

Iron phase Tm0 (K) r0 (g/cm3) KTO (GPa) K00 g q dKT/dT dT av (1025 K 21) Source
(GPa/K)

a 7.783 166.6 5.29 0.8 Guinan and Beshers (1968)

a 7.783 166.6 5.79 Rotter and Smith (1966)
a 7.85(1) 162.8(8) 1.69(5) 20.034 5.8 3.6(1) Isaak and Masuda (1995)
a 1.66 0.6 Boehler and Ramakrishanan (1980)
1 8.28 178.2 5.15 Brown and McQueen (1986)
1 8.30 164.8(3.6) Mao et al. (1990) (B-M EOS fit)
5.33(9)
1 8.30 160.2(2.1) Mao et al. (1990) (Vinet EOS fit)
5.82(8)
1 8.28 193 4.29 Jephcoat et al. (1986)
1 (200 GPa) 0.91 Duffy and Ahrens (1993)
1 1,600 8.30 162.3 5.74 1.9 1.0 20.010 Anderson and Isaak (2000)
gi 155 5.5 5.0 Boehler et al. (1990) and Stassis
(1994)
gi 1,810 8.00 155 5.0 1.4 1.5 20.02 Anderson and Isaak (2000)
Liquid (high T) 7.0 136 5 Fillipov et al. (1966)
Liquid (1,811 K) 7.019 109.7 4.66 1.74 9.2 Anderson and Ahrens (1994)

981
982 Chapter 14

can be found from the equation:

X  rFe 
fx 2 1 ¼ 0:11 ð14-6Þ
rx

where fx and rx are the mass fraction and density at core conditions of element x and rFe is
the density of iron at core conditions.
Silicon can enter metallic iron only under very reducing conditions. O’Neill (1991)
has shown that 5% silicon in the core of the proto-Earth would release enough O to oxidize
12% Fe to FeO in the proto-mantle. He has decided that all of the Si in the core entered
during differentiation. If the Earth was formed in very reducing conditions in the beginning
(as were the E chondrites), then , 4% Si would have already been present in the metal
inventory. Only 3.7% silicon would have to be obtained by reduction of silicates, which
would yield , 8% FeO (very close to the proportion of FeO þ TiO2 þ MnO þ NiO
observed in primary mantle). At low pressure, the Fe– Si – S phase diagram shows a large
immiscibility gap in the liquid state which, however, closes at higher pressures (and/or
temperature).
The EOS of possible core phases that have been determined by different workers
are cited below.

EOS of core elements Authors

1-Fe Brown and McQueen (1982)
FeO, CaO Jeanloz and Ahrens (1980)
FeS Brown et al. (1984)
Fe–Si alloy Balchan and Cowan (1966)
FeH Badding et al. (1992)

Okuchi (1997) determined the density reduction of the outer core effected by the
presence of hydrogen, sulphur and carbon as:

Ratio Density reduction (%)

H/Fe ( ¼ 0.41) 5.5
H/S 1.1
H/C 2.2–2.7

The total density reduction with these elemental proportions is 9%, which
corresponds to the deficit for the outer core. If the H/Fe ratio reaches the value of 0.34, it
will reconcile two-thirds of the inner-core deficit (Williams and Hemley, 2000). On this
basis, the density profile through the inner core is compatible with the measured EOS of FeH
core (Bedding et al., 1992). The core formation in the presence of various C – H –O species
along with Fe3C has been discussed by many (e.g., Jana and Walker, 1999). The number of
moles of hydrogen that may be present in the outer core far exceeds those of other possible
lighter alloying components (e.g., Williams, 1998). The equilibrium solubility of hydrogen
at high temperature and higher pressures is likely to vary. It is, however, assumed that core
Iron and Siderophile Elements: The Earth’s Core 983

formation occurred in relatively oxidizing conditions: the core becomes progressively more
oxidized as S is incorporated and iron is oxidized and goes into silicates (e.g., Wood, 1997).
At the top of the core the temperature range from 3,900 – 4,500 K, with a temperature drop of
, 500 K across D00 and the CMB (Okuchi, 1998; Williams, 1998).

14.2.5. Iron phases

So far five phases of iron have been known: a-(body-centred cubic, bcc), g (face-
centred cubic, fcc), d (bcc), 1 (hexagonal close packing, hcp) and b (ambiguous
structure) (see Fig. 14.8). Low-pressure iron a-phase has a bcc structure while, at the
Earth’s inner core, iron is presumed to possess hcp. At moderate P, T conditions, a-bcc
phase transforms to g-fcc phase which, at higher pressure (.13 GPa; Boehler et al.,
1990), transforms to 1-hcp phase. The hcp-phase shows stability beyond 300 GPa at
high temperatures (300 K; Vocadlo et al., 2000).
A planar view of the hard sphere ABCABC… stacking characteristic of the close-
packed plane in the fcc lattice, and that of ABABAB… stacking of close-packed planes
leading to hcp structure, are shown in Fig. 14.3. Figure 14.9 presents the phase diagram of
iron. The phase boundaries between a (bcc), g (fcc), d (bcc) and 1 (hcp) are presumed to

Figure 14.8. Phase diagram of iron from DAC experiments. The melting band (darker shading) contains data to
40 GPa (Yoo et al., 1992), 50 GPa (Yoo et al., 1996), 75 GPa (Shen et al., 1998) (except their highest-pressure data
point), 150 GPa (Saxena et al., 1994), and 200 GPa (Boehler, 1993), including a new shock melting point of
preheated iron at 71 GPa (triangle) (Ahrens et al., 2000) and the latest measurements (dots) using modified
techniques. The location of the g – 1 phase boundary (lighter shading) shows significant uncertainty
(Source: Boehler, 2000, q 2000 American Geophysical Union).
984 Chapter 14

exist below 50 GPa. The location of the g – 1 – liquid triple point is the starting point of the
most important branch of the iron melting curve. The triple point is placed somewhere
between 70 and 100 GPa and 1-iron is seen to be the stable phase. The energy differences
between close-packed high-pressure phases of iron may be insignificant. Distortion from
hcp to orthorhombic and a change in stacking order in hcp ! dhcp transformation involve
only very minor structural changes. The characteristic features of four major iron phases
are as below:
a-Fe (bcc) phase: ferromagnetic, stable at ambient conditions and over a limited P – T
(up to 10 GPa and 1,200 K) range.
g-Fe (fcc) phase: exists at higher temperature than a (up to melting) and to higher
pressures.
1-Fe (hcp) phase: high-pressure phase, stable to at least 300 GPa at RT (Mao et al., 1990).
b-Fe (dhcp) phase: the occurrence of b iron polymorph was first reported in 1993 by
Boehler and the co-workers of Saxena.
In their studies in pressures between 40 and 150 GPa, Yoo et al. (1995) assigned a
phase as 1-Fe. This was observed in the portion of the stability field previously assigned to
g-Fe, covering a part of the field for b-Fe (see Fig. 14.8). 1-Fe exhibits a wide stability field
that increases with pressure and recent shock-wave data are consistent with this proposal
(Nguyen and Holmes, 1998). Within the stability field of Fe, some other phases such as
dhcp (Saxena et al., 1997) and orthorhombic b form (Andrault et al., 1997) have been
identified (Fig. 14.9) but diffraction lines of such phases may be produced only when the
transformation of the phases is complete (Hemley et al., 2000). As stated earlier, the
persistence of hcp phase to high P – T conditions is endorsed by first-principles calculations
(e.g., Laio et al., 2000).
At high temperature (3,000 K), the hcp phase (1-Fe) is seen to be stable from
50 GPa to at least 110 GPa. The wide stability field of 1-Fe is suggestive of this polymorph
being the only stable phase at the Earth’s core.

Figure 14.9. The iron phase diagram revisited (modified after Saxena et al., 1995) (read the text).
Iron and Siderophile Elements: The Earth’s Core 985

14.2.5.1. Stability of bcc and fcc phases

The bcc phase owes its stability at low pressures to its large magnetic
moment. The first-principles calculations indicate that the bcc phase is unstable at
high pressure but the pair-potential study by Matsui and Anderson (1997) indicates
that bcc phase may reappear as a stable phase at very high temperature and pressure.
At high pressure, the bcc phase is elastically unstable while the fcc phase is
elastically stable. In the presence of infinitesimal thermal fluctuations, the bcc lattice will
undergo a spontaneous distortion to the fcc structure. At low pressure, the total energy
presented as a function of c=a ratio displays two local minima, one corresponding to the
bcc structure, and the other to the fcc structure. The fcc structure can be derived from the
bcc structure in one of the two ways: (1) by increasing c=a ratio or (2) by decreasing c=a
and b=a ratio. By decreasing c=a below 1a, structure (tetragonal; Söderlind et al., 1996) is
generated which is intermediate between bcc and fcc.
For iron, transformation from bcc phase to fcc phase was observed at the following
pressure and temperature conditions (Zhang and Guyot, 1999) (Table 14.4):
Upon cooling, the reversibility of the transformation was observed to depend on
pressure and also on the transformation temperature. Reversal of the transformation of iron
was found to be sensitive to temperature. Hysteresis of the transformation increased from
258C at 3.8 GPa to 1008C at 7.0 GPa, primarily because the bcc –fcc phase boundary has a
negative Clayperon slope. The experimental results of the bcc ! fcc phase transformation
of iron is explained by Zhang and Guyot (1999).
In their experiments, Zhang and Guyot (1999) observed that, in the pressure range
8.2 –7.1 GPa, the fcc phase was first transformed in the second cooling cycle to the hcp
phase at about 3008C and 7.1 GPa. The bcc phase was not observed until the temperature
reached 1008C and 7.2 GPa. The relative stability of the bcc and hcp phases in this
cooling cycle suggests that these pressure and temperature conditions are probably near
the bcc – hcp phase boundary. Upon quenching to room temperature, the three
polymorphs of iron co-existed in the pressure range 5.3– 7.1 GPa. The phase diagram
of iron from diamond-cell experiments is presented in Figs. 14.8 and 14.9.
The iron used by man is the a-phase with bcc structure. There is also a low-P, high-
T d phase, which is bcc but, unlike a, is non-magnetic. At the core P, T conditions, iron
may have several structural states: the hcp or 1-phase, the high-T fcc or gI-phase and
possibly a new b-phase.

TABLE 14.4
Fe: bcc ! fcc phase transition

Pressure (GPa) Temperature (8C)

3.8 650
4.9 600
6.3 550
7.3 525
7.7 500
8.2 500
986 Chapter 14

In the P, Tphase diagram in Fig. 14.8, it is seen that, well above 5.2 GPa, liquid iron
is in equilibrium with gI (fcc) phase. The gI – d –liquid triple point is at 1,990 K and
5.2 GPa (Strong et al., 1973). (Note. At the triple point, the sum of the volume changes and
the sum of the entropy changes are zero.)
The dT=dP for the 1 –gI boundary has been determined by some workers as stated
below:

dT/dP (K/GPa) Authors

31 Liu (1975)
20 Takahashi and Bassett (1964)
59 Anderson and Isaak (2000)

In iron, the phase boundaries between a – 1 and a – gI are slightly curved. This
curvature arises because the magnetization of bcc iron decreases as the temperature
increases and finally disappears at Curie temperature (Tc ¼ 1,043 K; Kittel, 1956, p. 407).
It is assumed that all phases of iron at higher pressures are non-magnetic and, therefore,
solid –solid boundaries above 2,000 K are linear in T – P space. However, the melt
boundaries of iron are not linear.
At low pressures, the relative stability of these phases can be explained as a result of
competition between magnetic and non-magnetic contributions to the internal energy,
differences in vibrational and magnetic entropy and the differences in volume (Moroni
et al., 1996).

14.2.5.2. b-Fe (dhcp) phase

Boehler and the co-workers of Saxena (1993) proposed b-iron to occur above
70 –80 GPa (Boehler, 1993) and 30 –40 GPa (Saxena et al., 1993). The crystallographic
structure of b-iron was then proposed to be dhcp, in which the hcp cell is doubled along the
c-axis (Saxena et al., 1995) (Fig. 14.10; after Andrault et al., 2000; read the caption).
Yoo et al. (1995) studied the phase diagram of iron in pressures up to 150 GPa and
3,500 K by a combined laser-heated diamond-anvil cell and XRD. At moderate
temperature, they found that, below 40 GPa, a phase of iron formed, X-ray study of
which showed a diffraction pattern that could be indexed as a double hcp (dhcp) structure
i.e., an ABAC stacking sequence. Saxena et al. (1995) also reported such a dhcp phase at
pressures below 40 GPa. In addition, two new phases have been postulated such as a dhcp
(Dubrovinsky et al., 1997) and an orthorhombically distorted hcp structure (Andrault et al.,
1997). Saxena et al. (1995) obtained the fifth polymorph b-iron by heating iron under
pressure in a DAC with Nd-YAG laser. A plot of the laser power against temperature
showed a change in slope signifying a phase transition on melting. Again, X-ray
synchrotron experiments with in situ laser heating of iron in a diamond-anvil cell show that
the high-pressure 1 phase (hcp) transforms to phase b, possibly a polytype dhcp.
The required pressure is about 38 GPa at temperatures between 1,200 and 1,500 K
(Fig. 14.8). Table 14.5 shows the calculated diffraction pattern of a sample of dhcp iron.
However, many of the peaks correspond to the 1-hcp phase.
Iron and Siderophile Elements: The Earth’s Core 987

Figure 14.10. Structure model for Pbcm iron (Andrault et al., 2000). Iron is located on mirror plane, perpendicular
to the c-axis. Arrows represent a slight movement of the atoms away from the particular position. No. 1 atom is not
located at (0.25, 0, 1/4) position, but rather at (0.22, 0.04, 1/4). Drawing the same lattice using atom 1 as a (0, 0, 0)
position leads to diagram usually drawn for 1-iron reported by Andrault et al., 1997) (From Andrault et al., 2000,
q 2000 Mineralogical Society of America).

Using electrically heated DAC and XRD, Dubrovinsky et al. (1998) reported
transformation to the b-phase of iron at high-T above 35 GPa.
The phase transformation of hcp ! b phase has been determined to lie in the
pressure range between 135 and 300 GPa. At 300(20) GPa, hcp Fe is transformed to
b-phase at a temperature of 1,350 K. The transition boundary shows a very small negative
slope in the P – T field. The fcc – hcp boundary may be considered as the fcc-b-hcp phase
boundary (Shen et al., 1998) as well.
Although b-phase of iron (dhcp) is considered likely to be present at the core, the
1-phase (hcp) is still recognized as the suitable high-pressure phase of Fe at the Earth’s
core and all geophysical models are based on the properties of 1-iron.
The space group for dhcp lattice is P63 =mmc; with two independent atoms in the
(0,0,0) and (1/3, 2/3, 3/4) positions (see Wyckoff, 1964). This model leads to the absence
of odd l for all hckl Bragg lines. However, using angle-dispersive XRD patterns,
Andrault et al. (1997, 2000) proposed the structure to be orthorhombic (Sp. Gr. Pbcm)
(see Fig. 14.10). By XRD study, Yoo et al. (1995) reported a new polymorph between
988 Chapter 14

TABLE 14.5
Calculated XRD pattern for dhcp (a ¼ 2.427 Å, c ¼ 7.666 Å, Mo Ka1 radiation, P ¼ 35 – 40 GPa, T ¼ 300 K).
I, relative intensity (Saxena et al., 1995)

hkl d (Å) (%)

100 2.1018 8
101 2.0270 45
004 1.9165 33
102 1.8429 100
103 1.6233 12
104 1.4162 5
105 1.2387 9
110 1.2135 21
106 1.0918 16
021 1.0412 4
114 1.0253 22
202 1.0135 12
023 0.9791 3
107 0.9712 3
008 0.9582 3
025 0.8668 2
206 0.8116 5

15 and 40 GPa. This phase is metastable and is related to non-hydrostatic stresses. Later
(1996), they observed a splitting of the 100-hcp line compatible with both dhcp and Pbcm
structural models.
The volume differences between the 1 and b and the b and g polymorphs are
determined as 1.4 and 1.8%, respectively, indicating positive Clapeyron slopes between
these polymorphs in the P – T phase diagram. All the three polymorphs, 1, g and b, have a
similar bulk modulus between 30 and 60 GPa.
The heating techniques and observations on the b-phase by various workers are
summarized in Table 14.6 (Saxena and Dubrovinsky, 2000).

14.2.5.3. 1 ! Pbcm iron transformation

For the phase transformation of 1-iron to Pbcm; the main changes are the doubling
of the lattice volume and the breakdown of laws related to hexagonal c-faces. The
Pbcm110, 020 and 002 layers correspond to the 100 and 002 dense layers of the hcp
structure. Thus, the orthorhombic lattice produces a more symmetrical set of dense layers
than found in hcp.
Pbcm iron (Fig. 14.10) occurs through a slight modification in the ABA stacking.
The hcp ! Pbcmtransformation may occur by a shift of the hcp-AB layers relative to each
other (Andrault et al., 1997). This shift will cause the dense 100-layer to split into two close
atomic sub-layers. A vibration dynamics of these double 100-hcp layers will bring about the
transformation to orthorhombic lattice. If the gap between these two sub-layers in a
particular direction disappears, the 100 layer becomes similar to that along the 001

TABLE 14.6
Observation of the b-phase in various studies
Reference
Saxena et al. (1993)
Boehler (1993)
Saxena et al. (1995)
Yoo et al. (1995, 1996)
Saxena et al. (1996)
Funamori et al. (1997)
Dubrovinsky et al. (1998)
Andrault et al. (1998)
Dubrovinsky et al. (1999)
Saxena and Dubrovinsky (2000)
Iron and Siderophile Elements: The Earth’s Core
Technique of heating Observation
Laser heating, phase recognized using power –temperature relation b-Phase inferred over a broad field of P and
T; no structural data
Noted a similar behaviour Phase assigned the dhcp structure
Laser-heated quenched samples studied with Synchrotron X-ray
Laser-heated quenched samples dhcp phase; partly overlapping the fcc field
Laser-heated quenched samples studied under pressure b-phase between P of 40 –60 GPa at T
between 1,450 and 2,150 K
In situ X-ray in multi-anvil apparatus No b-phase found to a P of 35 GPa at T up to
1,600 K
Electrically heated ion wire b-phase at P ¼ 57 GPa, T ¼ 1,715 K and at
P ¼ 68 GPa, T ¼ 1,325 K. Conversion back
to hcp at low T
Laser heated sample with in situ X-ray study Assigned an orthrohombic structure to a new
phase with several common peaks with dhcp
Externally heated sample with in situ X-ray study b-phase appeared P ¼ 301 GPa and
T ¼ 1,350 K
Iron powder heated in an externally heated cell b-phase formed at T ¼ 1,550 and
P ¼ 37 GPa
989
990 Chapter 14

direction. If the shift of the AB layers between each other is affected by T and P, ortho-
rhombic iron will adopt cell parameters in accordance with the amount of displacement.
For the stability field for Pbcm iron, sufficient thermal energy needs to be provided
so that the energy barrier between 1 and g polymorphs is overcome. To achieve this, the
temperature required is nearly the melting point for a particular P. At melting T; a rapid
time-dependent variation of the black-body emission signifies the onset of melting with the
resulting mixture of solid and liquid iron phases. The 100 line of Pbcm orthorhombic
lattice is observed to be present.

14.2.5.4. Stability
Andrault et al. (2000) have observed that the g-iron stability field does not extend
above 55– 60 GPa and a positive Clapeyron slope exists between 1- and b-iron.
New synchrotron XRD data confirm the transformation of the hcp Fe to the Pbcm
orthorhombic lattice at high P and T. Pbcm polymorph is metastable as it arises from the
gliding of the same dense atomic layers which are involved in T-induced 1-hcp to g-fcc and
1-hcp to b-Pbcm transformations. Pbcm should possibly be observed at moderate T as an
intermediate structure before the phase transformation occurs for 1-hcp to g-fcc. But a
sufficient thermal energy needs to be provided to iron so that energy barriers between 1 and
g polymorphs are overcome. The space group Pbcm is extracted from the systematic
absence of the 010, 001 and 011 reflections. This Pbcm lattice usually back-transforms to
1-hcp iron during the T-quench.
Andrault et al. (2000) proposed that the formation of a set of most similar dense

020 and 002 Pbcm-layers) can be related to an entropy increase along
layers (the 110, 110;
the T-induced 1 ! Pbcm phase transformation.
At moderate T, the orthorhombic Pbcm phase is observed at P below 30 GPa
(a P range where only 1-hcp and g-fcc are known to be stable; Funamori et al., 1996). The
g-form needs ABA to ABC modification for the dense layer stacking. High-T should relax
stresses that might be encountered upon compression in the DAC and that may favour the
occurrence of a distorted iron form. At the melting T (2,400 K for 55 GPa), i.e., with
vanishing deviatoric stress, a rapid time-dependent variation of black-body emission
(due to the onset of melting) occurs. This results in a mixture of solid and liquid iron phases
of varying proportions. However, the 100 line of Pbcm orthorhombic lattice remains. This
also indicates that g-iron stability field does not extend beyond 55 – 60 GPa.
With increasing P, the volume difference between all iron polymorphs decreases.
Hence, the Claypeyron slope between 1- and b-forms will decrease with P and becomes
nearly horizontal in the P – T diagram at around 330 GPa. Beyond this pressure range, if no
other new phase of iron is stable, then Pbcm iron should be the only pure form of iron to
occur at the inner-core condition.

14.2.5.5. Thermal Grüneisen parameter

The thermal Grüneisen parameter, gth, can be used to characterize and extra-
polate the thermophysical properties of materials to high pressures and temperatures
Iron and Siderophile Elements: The Earth’s Core 991

Anderson (1967) found the relation between the Grüneisen parameter and volume as:

gth ¼ g0 ðV=V0 Þq ð14-7Þ

where q is a parameter. For iron, the values for q and g0 at different temperatures as found
by different workers are shown in Table 14.7.
Using the above relation of Mao et al. (1990) and the EOS of 1-Fe, Merkel et al.
(2000) obtained

g0 ¼ 1:68ð^0:20Þ

and

q ¼ 0:7ð^0:5Þ

when q ¼ 1; g0 becomes 1.81(^ 0.03).

The gi of E2g mode provides a good approximation for kgil.
Dubrovinsky et al. (2000) calculated the Grüneisen parameter at the Earth’s core
conditions at different radii using the temperature points from Stacy (1995) as:

r (km) T (K) g
1,400 4,902 1.28(15)
1,400 – 1.27
2,400 4,500 1.33(15)
2,400 – 1.30
3,400 3,847 1.44
3,400 – 1.34

14.2.5.6. Vibrational modes

The cubic fcc lattice has three acoustic modes whereas a hcp lattice has three
Raman modes and three acoustic modes These respond differently to both T and P
(in Debye to Einstein models). Fcc iron has four equivalent (111) layers whereas hcp has a
single (001) dense atomic layer. The atomic density along the three (100)-hcp layers is
also high. It is likely that phonons should travel more easily in the most symmetric g-(111)

TABLE 14.7
Variation of the thermal Grüneisen parameter of iron with volume as described by equation (14-7)

Phase Pressure range (GPa) g0 q Reference

a-iron 0–10 1.66 0.6 Boehler and Ramakrishnan (1980)

1-iron 15–100 2.2 1.0 McQueen et al. (1970)
1-iron 15–100 2.2 1.62 Jeanloz (1979)
Liquid iron 243 1.59(9) Brown and McQueen (1989)
1-iron 0–240 1.7 0.7 Anderson (1998)
1-iron 0–300 1.78(6) 0.69(10) Dubrovinsky et al. (2000)
1-iron 15–152 1.68(2) 0.7(5) Merkel et al. (2000)
992 Chapter 14

than in 1-(100) layers. This would also explain a lower specific entropy for the hcp lattice
relative to the high-temperature g-form.

14.2.6. Phase boundaries and the triple points

Well above 5.2 GPa pressure, the fcc phase appears to be in equilibrium with liquid
iron at high T (Bundy, 1965) There are four solid phases and one liquid phase around the
triple points (tp). The slope of the phase boundary ðDP=DTÞ at tp should satisfy the
boundary condition that the sum of the volume changes and the sum of the entropy changes
should both equal zero. The phase boundaries and their structures among the four phases
other than b-phase have been determined. The (1-) hcp phase is found to be stable at
3,000 K to pressure of the Earth’s core (Mao et al., 1990). The bcc – fcc – hcp occurs at
750 K and 11 GPa (Bundy, 1965). The b– g –liquid triple point is located around 55 GPa
and 2,400 K. This gives a positive Clapeyron slope between 1- and b-iron and also the
stability of 1-iron at 100 GPa and room T. The P – T stability field of Pbcm begins at
, 35 GPa (Saxena et al., 1995). In short, the Pbcm iron is stable above 35 GPa and 1,500 K
and the (g – b –liquid iron) triple point is located at about 55 GPa and 2,400 K.
The PREM velocity profile (Fig. 14.7) of the outer core lies substantially below
both fcc iron and hcp iron, supporting the fact that the outer core is not compositionally
pure iron but is diluted with lighter elements. Therefore, the triple point of hcp – fcc –liquid
phases must lie at pressure below the inner core – outer core boundary pressure.

14.2.6.1. a – 1 – g triple point and 1 – g transition

The constraints for the phase diagram of iron are shown in Fig. 14.8. The
fcc – d– liquid tp is at 1,990 K and 5.2 GPa (Strong et al., 1973). The experimental values
of the a– 1 –g triple point have been determined by different workers as:

a– 1 – g triple point Authors

11 GPa/750 K Bundy (1965)
11.6 GPa/810 K Boehler (1986)
8.3 GPa/713 K Akimoto et al. (1987)

The slope for 1 – g boundary is estimated by Liu (1975) to be 318/GPa. He showed

that a straight extrapolation of the slope 308/GPa would place tp at around 93.5 GPa,
corresponding to 1-iron at the inner core.
The slope of the g – 1 transition line has been determined as:

a– 1 – g triple point Authors

35 K/GPa Mao et al. (1987)
28.6 K/GPa Boehler (1986)

In the bcc ! fcc phase transition at 750 K and 11 GPa, the volume of fcc
decreases by 2 0.21 cm3/mol (Mao et al., 1967). Because r0 for fcc iron is between those
for bcc iron and hcp iron, the other physical properties of fcc iron should also be between
these two phases.
Iron and Siderophile Elements: The Earth’s Core 993

The hcp phase (1-iron) is stable above 280 GPa and, therefore, the inner core
becomes its host. The hcp phase is known to show an increase in Hugoniot (which is the
pressure density trace under shock pressure). From Hugoniot data, Brown and McQueen
(1986) estimated the liquid transition to lie between 5,000 and 6,000 K. The hcp – fcc
boundary was determined by Brown and McQueen (1982) to occur at 200 ^ 2 GPa and
4,400 ^ 300 K. The slope of the fcc – hcp phase boundary ðdT=dPÞ is between 20 and
308/GPa.
The hcp iron, showing a larger value of PP ; should naturally have a larger Poisson
ratio value ðsÞ: For iron at high pressure, the Poisson’s ratio for hcp iron extrapolates into
the seismically determined values of s for the inner core. But s for fcc iron is too high for
the inner core.
Saxena and Dubrovinsky (2000) have delineated the phase boundaries between
b-hcp and located the triple points b –fcc – hcp and b –fcc – melt. The hcp is seen to be
more stable than dhcp at core conditions.
Brown and McQueen (1986) proposed a solid –solid phase transition at , 200 GPa
and melting at 243 GPa at temperature of 4,400 and 5,500 K, respectively. Anderson and
Duba (1997) used Lindemann’s law with various Grüneisen values to extend the melting
curve to the inner-core pressures. On that basis, one could consider the co-existence of
solid Fe and melt with melting starting at 200 GPa and completing at 243 GPa.

Fcc –dhcp (or hcp) – melt triple point. Based on the data of Shen et al. (1998), the triple
point dhcp –fcc – melt is thought by Saxena and Dubrovinsky (2000) to be close to 60 GPa,
possibly at 2,640 K. They propose the following triple points in the phase diagram:
hcp(1)– fcc(g)– b-phase: 40(4) GPa/1,550(100) K
b-phase– fcc(g)– melt: 60(10) GPa/2,600(100) K
The hcp – b-phase boundary shows a very small negative dP/dT. This indicates the
similarity of physical properties, such as molar volume, thermal expansion and bulk
modulus, between hcp and b-phase. But b-phase has a higher entropy and enthalpy than
those of hcp.

1 – g transition. If atoms are regarded as rigid spheres, the packing density for both fcc
and hcp structures is 0.74. Temperature induces 1- to g-phase transformation in iron above
1.5 GPa. For 1-hcp iron, the symmetry is 6m2
while for g-iron the more degenerated 3d
electronic continuum will show a mmm symmetry with lower electronic density.
Mao et al. (1987) reported the 1 – g transition up to , 36 GPa and 1,3758C from
electrical properties measurements. At pressure to 42 GPa (1,560 K), an 1 – g iron
transition has been reported by Dubrovinsky et al. (1998). They have calculated the lattice
parameter of g-Fe using (111) and (200) reflections and observed a volume difference
between 1- and g-phase as 0.3%. However, their X-ray experiments over the pressure
range up to 54 GPa (1,700 K) could find no reflections of dhcp-Fe phase reported earlier by
Yoo et al. (1996). The former could discern only hcp and fcc iron.
Funamori et al. (1996) tracked the 1-hcp to g-fcc phase transformation by XRD in a
multi-anvil press. They found the phase boundary between 1,300 and 1,400 K to be at
, 31.5 GPa. They confirmed a positive Clapeyron slope between 1-hcp and g-fcc and
994 Chapter 14

Figure 14.11. Plot of the variation of c=a for hcp iron as a function in pressure. The c-axis is more compressible
that the a-axis and the linear fit as a function of pressure gives c=a ¼ 1:604ð2Þ 2 4:9ð1:4Þ £ 1025 P; where P is
measured in gigapascals (Mao et al., 1990).

proposed a DV of less than 2% between these two polymorphs. In hexagonal lattice (with
two a priori independent a, c parameters), the ABAB stacking sequence along the c-axis
achieves a maximum density for spherical atoms only when c=a ¼ 8=3 ¼ 1:633: The hcp
form of iron (obtained at RT, 15 GPa) shows a c=a ratio , 1.604 (Poirier and Price, 1999).
That the density of hcp lattice is higher than that of fcc indicates that the iron atoms are not
spherical at high pressure (Andrault et al., 2000). With pressure, the c=a ratio of hcp iron
decreases, but only slightly (Fig. 14.11, read the caption). Yoo et al. (1995) proposed the
occurrence of an intermediate metastable 10 polymorph below 30 GPa, i.e., between the
phase transformation of 1-hcp to g-fcc.

Raman study (15 – 152 GPa). The hcp structure has two atoms per unit cell, which are
located on sites of symmetry D3h : The six normal modes of zero wave vector belong to the
irreducible representations:
A2u þ B1g þ E1u þ E2g:
B1g and E2g are optical phonons polarized along and perpendicular to the optical axis,
respectively. The doubly degenerate E2g mode is Raman active.
Merkel et al. (2000) undertook Raman study of 1-iron in the pressure range
15 –152 GPa. Between 15 and 40 GPa, they observed two Raman bands, the stronger of
which was identified as E2g fundamental predicted by symmetry for the hcp lattice. The
low-frequency band becomes stronger with P, suggesting ordering of the hcp lattice.
The high-frequency weak band is induced by disorder. (Note. Raman spectra were excited
with 488- and 514.5-nm lines of an Arþ-ion laser and measured with a single-grating
ISA.HR-460 spectrometer equipped with a charge-coupled device (CCD) defector. Ruby
fluorescence was used as pressure calibrant (Mao et al., 1986).)
Under pressure, a positive spectral shift of the initial slope dn=dP ¼ 1:0 cm21/GPa
indicates the wide stability range of 1-Fe noted by Dubrovinsky et al. (2000). They opine
Iron and Siderophile Elements: The Earth’s Core 995

TABLE 14.8
g– 1 –l triple point for iron

Authors Triple point (P and T) Structure (P)

Yoo et al. (1995) 50 GPa, 2,500 K dhcp (, 40 GPa)

Shen et al. (1998) 60 GPa, 2,800 (^200) K
Saxena et al. (1993) 65 GPa, 2,700 K dhcp (30–40 GPa)
Boehler (1986) 75 GPa, 2,500 K
Andrault et al. (1997) Orthorhombic

that hcp-Fe is stable to at least 300 GPa and 1,500 K. (Note: Shen et al. (1998) found that
fcc – hcp – liquid tp is at 60 GPa and 3,000 K (see Table 14.8).)
In 1-Fe, the C44 elastic modulus and the E2g Raman mode are properties of the same
phonon branch. Specifically, the E2g mode correlates with a transverse acoustic phonon
and C44 represents the slope of this branch at the centre of the Brillouin zone.

14.2.6.2. 1 – g – l triple point

The location of the 1 – g – liquid tp is found to lie at 2,500 ^ 200 K and
50 ^ 10 GPa (Yoo et al., 1995). Shen et al. (1998a,b) studied the high-pressure melting,
phase transitions and structures of iron up to 84 GPa and 3,500 K with an improved
laser-heated DAC.
Based on their data of the 1 – g phase transition coupled with the observation of
melting by in situ XRD, the 1 – g –l tp was determined. This tp is compared with other
results as below.
Shen et al. (1998) suggested that the in situ appearance of dhcp structure probably
arises from regions of very low temperatures in the sample where incomplete
transformation between 1 and g takes place.

14.2.6.3. Liquid iron: structural change under P

The structure of liquid iron at high pressure remained unknown for a long time
because of experimental difficulties encountered above 1,810 K (melting point of iron),
particularly for the low signal levels above the background Fortuitously, the third-
generation synchrotron facilities have opened up studies using XRD and other methods.
The facilities have helped the discovery that, above the sub-solidus B1 – B2 transition in
molten KCl and KBr, the second-order ion distance decreases with pressure while the
nearest-neighbour distance remains essentially constant (Urakawa et al., 1998).
Near d –g – l tp (2,300 K, 5 GPa) in liquid-iron structural changes occur with
enhanced intermediate range order with increasing P and T. Urakawa et al. (1998) used a
large-volume press (Paris-Edinburg type; Besson et al., 1992) for measurements in
2 –5 GPa and 2,000 –2,300 K range. The intensity data synchrotron X-ray source were
converted to SðQÞ; the structure factor. From SðQÞ; the radial distribution function, gðrÞ;
was calculated using the relation:
ða
1
gðrÞ ¼ 1 þ QðSðQÞ 2 1Þ sinðrQÞdQ ð14-8Þ
2P2 nr 0
996 Chapter 14

where n is the average number density of atoms:

n ¼ rliqFe ðP; TÞN=MFe ðMFe ¼ 55:85 gÞ
rliqFe ðP; TÞ is calculated from ultrasonic measurements of both rliqFe ðP0 ; TÞ and KT ;
the bulk compressibility, KT (2,330 K) ¼ 77 GPa, ›K=›T ¼ 20:0104 GPa/K (Hixon et al.,
1990).
The second- and third-neighbour shells in the gðrÞ shift to shorter distances and
develop new structure with increasing pressure. The result has been interpreted by Sanloup
et al. (2000) in terms of the liquid iron acquiring bcc-like local order which evolves to a
mixture of bcc- and fcc-like local order as pressure and/or temperature are increased.

14.2.7. Elasticity and rheology

To understand the seismological observations — such as the low attenuation of

seismic waves, the low shear-wave velocity (Masters et al., 1981) and the anisotropy of
compressional-wave velocity (Creager, 1992; Tromp, 1993) — a knowledge of the elasticity
and texture of iron (Stixrude and Cohen, 1995) at core pressure is crucial. The change in
rheology of iron at the core’s intense pressure is currently under investigation to identify the
active slip systems which contribute to the deformation of the iron at that condition.
However, computations of the elasticity of iron at the pressures and temperatures at the core
agree well with seismological and free-oscillation results (Steinle-Neumann et al., 2001) .
Shock-wave experiments (Brown and McQueen, 1986) and static compression in the
diamond-anvil cell (Mao et al., 1990; Boehler, 1993; Saxena et al., 1993) offered
information on the thermal properties and sub-solidus phase diagram of iron at high
temperatures and pressures.
The density and bulk modulus of hcp iron have been measured to core pressures by
static (Mao et al., 1990) and dynamic (Duffy et al., 1992) methods. The shear modulus,
single-crystal elasticity tensor, aggregate compressional and shear-wave velocities and the
orientational dependence of these velocities in iron have been determined using radial
XRD (Singh et al., 1998) and ultra-sonic techniques (Li et al., 1996). The shear-wave
velocity in the inner core is lower than the aggregate shear wave velocity of iron,
suggesting the presence of low-velocity components or anelastic effects in the core.
Observation of a strong lattice strain anisotropy in iron samples indicates a large
(, 24%) compressional-wave anisotropy under the isostress assumption (Mao et al., 1998).
The elasticity tensor ðcij Þ of hcp iron as a function of pressure at ambient temperature was
presented earlier in Fig. 14.6c.

14.2.7.1. Experimental
The elasticity and rheology of iron above 220 GPa (Table 14.9) were determined
by Mao et al. (1999) employing the principles and procedure as discussed below
(see Fig. 14.12).
They performed two radial X-ray diffraction (RXD, discussed in Section 4.4.7.1)
experiments with diamond cells. Non-hydrostatic stress was allowed to develop in the
specimen by not adding any pressure medium. The stress state of the specimen compressed
between two anvils is a superposition of the hydrostatic pressure.
Iron and Siderophile Elements: The Earth’s Core 997

TABLE 14.9
Elasticity of hcp Fe at 298 K and high pressures (Mao et al., 1999)

P Density C11 C12 C33 C13 C44 K G VP VS

(GPa) (g/cm3) (GPa) (GPa) (GPa) (GPa) (GPa) (GPa) (GPa) (k/ms) (k/ms)

16.5 9.00a 297 108 6.95 3.47

39 9.67a 500 275 491 284 235 351a 134 7.40 3.73
39c 10.09 747 301 802 297 215 455 224 8.64 4.72
211 12.61b 1,533 846 1,544 835 583 1,071b 396 11.26 5.61
211c 12.80 1,697 809 1,799 757 421 1,085 445 11.45 5.90
210d 12.48 1,554 742 1,749 820 414 1,062d 411d 11.41 5.76
a
Jephcoat et al. (1986); b Mao et al. (1990); c Stixrude and Cohen (1995); d Reussbounds.
Note: The data in the first, second and fourth rows are from theoretical predictions (300 K). The rest are
experimental.

A micro-focus (4 –10-mm diameter) polychromatic X-ray beam, which passes

through the Be gasket in the radial direction, probes the lattice strain of the sample as a
function of the angle (c) to the diamond-cell axis (Singh et al., 1998). At pressures between
16 and 211 GPa, energy dispersive X-ray diffraction (EDXD) patterns containing
diffraction lines (hkl) of hcp iron, gold and tungsten were collected at 108 steps of C
from 0 to 908. The d-spacing varies linearly with cos2 C:

dðhklÞ ¼ dP ðhklÞð1 þ ð1 2 3cos2 CÞQðhklÞÞ

where the intercept dP ðhklÞ denotes the d-spacing under sP and the slope QðhklÞ is the
lattice strain under the uniaxial stress condition (Uchida et al., 1996).
The aggregate compressional-wave speed ðVP Þ and shear-wave speed ðVS Þ of hcp
Fe are calculated from the bulk moduli KFe and GFe. In addition, the aggregate ultra-sonic
VP and VS of hcp Fe are measured directly at 16.5 GPa in a multi-anvil press (Li et al.,
1996). The experiments are carried out at room temperature (298 K). The aggregate VP of
Fe calculated from diamond-cell data (of K and G) are in excellent agreement with low-
pressure ultra-sonic data and compare favourably with high-pressure shock-wave and
inner-core values at high temperatures (dVP =dT ¼ 22:9 £ 1024 km/s/K; Mao et al., 1999)
(see Table 14.9).
The single-crystal VP anisotropies for Fe at 39 and 211 GPa are calculated from the
elastic tensor, Cij , values based on isostress assumption.
The results from RXD (zero-frequency) and multi-anvil (ultra-sonic frequency)
measurements are in good agreement, bracketing a wide frequency range including seismic
waves (Mas et al., 1998) (see Fig. 14.12).
The K=G ratios of hcp Fe at varying pressures have been determined by different
methods as follows:

Pressure Method K=G ratio

20–30 GPa RXD 2.7 ^ 0.7
16.5 GPa Ultrasonic 2.68 ^ 0.1
998 Chapter 14

Figure 14.12. Comparison of results experiments and theories with seismic observations. Filled squares,
ultrasonic experiments with multianvil apparatus; open squares, RXD measurements; solid curves, extrapolation
of X-ray data, based on K=G ¼ 2:7; short-dashed curves, first-principles calculations for 298 K geotherm; long-
dashed curves, shock-wave Hugoniot at high temperature (Mao et al., 1998, q 1998 American Geophysical
Union).

The KFe, GFe, VP and VS, predicted by first-principles theory (e.g., Stixrude and
Cohen, 1995), are higher than the values at low pressures but the difference diminishes at
higher pressures.
The seismic VS of the inner core is lower than the corresponding hcp Fe value. If the
softening of VS represents possible Born –Durand pre-melting effects (Tallon, 1979) or
partial melting, the observed shear values would tightly constrain the inner-core
temperature near melting. The observations may indicate the presence of additional low
VS phases in the inner core.

14.2.7.2. Shear viscosity

The calculated shear viscosity h values for molten iron under Earth’s core
conditions obtained by Wijs et al. (1998) are roughly 10 times those of typical liquid
metals at ambient pressure. Their simulations demonstrate that high-pressure liquid iron
closely resembles the simple system liquids such as the Lennard –Jones and also hard-
sphere liquids. They noted that the value of h at CMB increases by , 25% as T goes from
4,300 to 3,500 K. Assuming an Arrhenius temperature dependence, this means that if T at
the CMB were as low as 3,000 K, h would be 1.8 £ 1022 Pa s. But the overall conclusion
is that the shear viscosity of liquid iron in the Earth’s outer core is ,1.5 £ 1022 Pa s (with
an uncertainty factor of 3). However, the viscosity values deduced from geodetic and
seismic measurements are much higher than this. This may be due to the influence of other
dissipative processes on that time-scale rather than viscosity (Secco, 1995). The influence
Iron and Siderophile Elements: The Earth’s Core 999

of dissolved sulphur on the viscosity, h, of liquid iron has been studied by LeBlanc and
Secco (1996).
Depending on the approximations employed, contributions to the forces such as
Coriolis, Lorentz and viscous forces acting on fluid outer core may be treated as negligible.
This leads to a scenario in which the core is in a state of small-circulation turbulent
convection.

14.2.8. Rigid core: slichter modes of translational motion

A layer about 450 km thick surrounding the solid inner core was called the
“F-layer” by Bullen (1965). The solid inner core is suspended near the centre of the outer
fluid core by gravitational forces.
The inner core’s motion is close to that of a rigid body. The frequency dependence
of the shell Love numbers, although included, has a minor effect on the translational
periods. But the inner core’s rigid translational motions, referred to as Slichter modes,
show three modes: one along the axis of rotation, one prograde in the equatorial plane and
one retrograde in the equatorial plane.
Anomalous splitting of the two equatorial translational modes of oscillation of the
Earth’s solid inner core is used to estimate the effective viscosity just outside its boundary.
Superconducting gravimeter observations give periods of 3.5822 ^ 0.0012 (retrograde)
and 4.0150 ^ 0.0010 (prograde) hours. With the use of Ekman layer theory to estimate
viscous drag forces, an inferred single density of 1.22 £ 1011 Pa s gives calculated periods
of 3.5839 and 4.0167 h for the two modes, close to the observed values. The large effective
viscosity is consistent with a fluid, solid – liquid mixture surrounding the inner core and is
associated with the “compositional convection” that drives the Earth’s geodynamo
(Smylie, 1999).
Inner-core viscosity is estimated from the limit imposed on relative inner-core
rotation by gravitational coupling of the inner core to density anomalies in the shell.
Viscosity of the “solid” inner core is constrained to be between 1016 and 1020 Pa s.

14.2.9. Outer core

Based on geodetic observation at the surface and on modeling of the time-varying

geomagnetic field, it has been found that the tectonic flow of the mantle is approximately
106 slower than that of the outer core (Merrill et al., 1996). This velocity difference may
arise from the large viscosity difference between those of the outer core and the lower
mantle, amounting to a factor of . 1020.
The thermal-boundary layer (TBL) is much thinner for the top of the core (, 1 km)
than the bottom of the mantle (,102 km). TBL thickness is given by the diffusion length
dTBL , ðkL=uÞ1=2; where L and U are the dimension and velocity of the flow and k is the
thermal diffusivity of the material.
Seismological observations constrain the density and bulk modulus of the outer
core to within a few % which, along with the calculated EOS, suggests the material to be
essentially iron. Comparison of density and bulk sound velocity structure allows one to
1000 Chapter 14

address aspects of the dynamics of the outer core. In the outer core, the P-wave velocity is
similar to the bulk sound velocity. VS vanishes in the liquid phase but the bulk sound
velocity remains unaffected in the solid – liquid transition.
However, reports of anomalous velocities or velocity gradients just before the CMB
(Lay and Young, 1990) and just above the inner-core boundary (Song and Helmberger,
1995) may be associated with stratification. These observations may suggest a possible
chemical reaction between the mantle and the outer core.
In the outer-core dynamics, by which the vigorous convection goes from top to
bottom, the temperature gradient is expected to be nearly adiabatic, holding the relation
dT g
¼2 T ð14-9Þ
dr f
where g is the Grüneisen parameter (¼ 1.5 for liquid iron at core condition; Brown and
McQueen, 1986).
The heat conducted ðQÞ along the adiabat is:
dT
Q ¼ 4p2 R2 s ð14-10Þ
dr
For liquid outer core, a thermal conductivity ðsÞ value , 60 W/m/K will give the value of
Q as 8 £ 1012 W.
The existence of a long-lived dynamic magnetic field of the Earth demands the
existence of a dynamo and therefore a large reservoir of a fluid conductor (Merrill et al.,
1996). (Note. Dynamos in the Sun and outer planets are thought to be contained within
layers of metallic hydrogen or other low atomic number fluids.)
The radial density profile of the liquid core, derived from seismology, shows
typically a 10% deficiency compared with the density of pure iron. This deficiency is
explained by the density dilution contributed by H, C, O, S and Si. Fractionation of light
elements into the outer core lowers the gravitational energy. This rearrangement is effected
by convection.

14.2.10. Inner core

The Earth’s solid inner core with a diameter of 2,400 km is surrounded by a liquid
Fe-rich outer core totalling the core diameter of , 7,000 km. The inner core was formed by
the freezing of the liquid outer core as the Earth’s interior started losing its primordial heat.
It is comparable in mass with our Moon but shows super-rotation. The solid inner core
grows slowly with time as the Earth cools. It freezes out almost pure iron from the outer-
core liquid (Masters, 1979) composed of about 90% iron and 10% lighter elements. The
residual liquid becomes enriched in the lighter elements. First-principles theory predicts
that, at high pressure, potassium becomes a transition metal with a siderophile character
(Bukowinski, 1976).
The paleomagnetic data indicate that the inner core was formed at least 4 b.y. ago.
The inner core has an anisotropy of hexagonal symmetry for which P-waves travel 3%
faster along the N –S spin axis than they do in the E – W equatorial plane. It is thought
Iron and Siderophile Elements: The Earth’s Core 1001

that the slow growth of the inner core provides a source of energy to drive the
geodynamo in the fluid outer core, which generates the Earth’s magnetic field. Its energy
drives the convective motions that generate the magnetic field through a self-sustaining
dynamo action (Gubbins et al., 1979) (see the latter under Section 14.2.10.6). Hence, the
process of formation of the inner core provides the primary energy source for the Earth’s
magnetic field.

14.2.10.1. Heat sources

The impact heating during accretion was sufficient to cause a large-scale to
complete melting of the planet (Pierazzo et al., 1997). The descent of liquid iron into the
core released enough potential energy to heat up the planet by several thousand degrees
(Tschauner et al., 1999). At the inner core, temperatures range from 4,000 to 8,000 K and
pressures from 330 to 360 GPa. It consists dominantly of hcp (1) iron crystals. The volume
thermal expansivity of iron falls rapidly as a function of pressure (Chopelas and Boehler,
1989). An effective thermal expansion coefficient for iron lower than , 0.85 £ 1026 K21
(at P ¼ PICB) would require a central core temperature in excess of 104 K to reconcile the
1-iron density with the PREM model.
The inner core grows by incongruent freezing of the overlying liquid. The sinking
of the cooler fluid (þ crystals) releases gravitational energy so there operate two distinct
energy sources: the latent heat of freezing and the release of gravitational energy.
However, the energy sources associated with the inner core depend on the enthalpy of
freezing of the inner core. The release of the latent heat of freezing is thought to be a major
source of energy that drives the production of the geomagnetic field. It is suggested that
some common radioactive heat-producing elements in the Earth may be partially
sequestered in the core. However, the ability of the inner core to incorporate these and also
light elements depends on its crystalline structure. However, radioactive decay is not
considered an important energy source at the core.

14.2.10.2. Rotation
The inner core is rotating faster than the rest of the planet. The travel times of
seismic waves that traverse the Earth’s inner core show a small but systematic variation
over the past three decades. This variation is best explained by a rotation of the inner core
that moves the symmetry axis of its known seismic anisotropy. The inferred rotation rate is
of the order of 18 per year (. 20 km/yr), which is faster than the daily rotation of the mantle
and crust (Song and Richards, 1996).
Using South Sandwich Island events recorded at a network in Alaska, Creager
(1997) first established a map of inner-core anomalies in the region sampled by the rays
and then used it to infer the inner-core rotation rate from a specific path where observations
were available over a period of 30 years. Creager estimated the values of rotation rate of
0.2 – 0.38 per year. Long-term recordings in astronomical observations also manifest the
irregularities of the mantle rotation.
Seismological studies suggest that the inner core rotates faster than the
mantle (Song and Richards, 1996; Su et al., 1996; Creager, 1997). The degree of
1002 Chapter 14

inner-core rotation with respect to the mantle is estimated by different workers of

recent years as:
38 per year Su et al. (1996)
18 per year Song and Richards (1996)
0.2 – 0.38 per year Creager (1997).
This rotation is driven by magnetic coupling between the electrically conductive
inner core and the geomagnetic field (e.g., Kuang and Bloxham, 1997). Maps of the changing
magnetic field at the CMB (Bloxham et al., 1989) which trace the fluid velocity suggest that
westward flow at the surface of the outer core is responsible for the westward drift of the
magnetic field, which first postulated as early as 1683 by Halley. The westward drift is
explained as a result of the outer part of the core rotating less rapidly than the inner part
(Bullard et al., 1950). Again, westward motion at the CMB would imply eastward motion
at depth, which is what Song and Richards (1996) observed for the inner core.

14.2.10.3. Crystalline structure: elastic/seismic behaviour

bcc vs. hcp?. The large range of the estimated temperature (4,000 – 8,000 K) of the inner
core has made the construction of a high-pressure sub-solidus phase diagram of iron at the
core temperature less certain. Throughout the pressure region of the Earth, the energetics
of close-packed phases of iron (bcc, fcc and hcp) were calculated by Stixrude and Cohen
(1995) with elaborate first-principles electronic structure calculations. The bcc structure
was found to be energetically unfavourable and mechanically unstable with respect to a
tetragonal strain at high pressure (P . 150 GPa).
Electronic theory calculations for iron (see Fig. 14.1) presented by Stixrude et al.
(1994) show that the fcc and hcp forms are non-magnetic while the bcc form (and its
tetragonal strain form) shows ferromagnetic behaviour (Stixrude and Cohen, 1995).
A GGA has been developed, which takes into account not only the local charge density but
also includes terms up to sixth order in the charge-density gradient (Perdew et al., 1992).
GGA correctly reveals the ferromagnetic bcc structure as the global ground state (Bagno
et al., 1989) and yields excellent correspondence with experimental observations on the
following:
(1) the bcc to hcp phase transition is predicted to occur at 11 GPa while experimental data
suggest it to be in the 10 –15 GPa range;
(2) for the hcp structure, the c=a ratio (1.58 –1.59) decreases only slightly with pressure
(see Fig. 14.11 and the caption).
The fcc and hcp phases have similar total enthalpies over the entire volume range
but the bcc phase has an enthalpy much higher than either fcc and hcp phases at high
pressures (Fig. 14.13). For the bcc phase to be stable at inner-core pressure, its molar
entropy must exceed that of hcp by , 1:4R (R, gas constant). Because of the constraints
imposed by the estimates of high-pressure entropies, the magnetic, electronic and
vibrational energies at the inner core make the bcc phase unlikely. In addition, bcc phase
has been seen to be unstable in tetragonal strain at high pressure (Stixrude et al., 1994).
The body-centred tetragonal structure relates bcc (c=a ¼ 1) to fcc structure
ðc=a ¼ 2Þ: The change in energy produced by tetragonal strain is related to the
Iron and Siderophile Elements: The Earth’s Core 1003

Figure 14.13. GGA total enthalpy of bcc (solid squares) fcc (open squares) relative to that of hcp (open circles) as
a function of pressure. The enthalpy difference is divided by the gas constant ðRÞ leading to units of temperature.
The differences are compared with the ranges of pressures and temperature estimates (vertical bar) of the inner
core (IC) (from Stixrude and Cohen, 1995b, q 1995 American Geophysical Union).

combination of elastic constants. c11 – c12 is positive for stable crystal structures. As the
tetragonal strain is applied, the total energy of bcc lattice becomes negative ðc11 – c12 , 0Þ
so that the lattice will become mechanically unstable and spontaneously distort the
bcc structure to fcc. At inner-core pressure, the strain energy associated with a small
change in c=a is comparable in magnitude with that of the stable bcc lattice at zero
pressure. At core pressure, if any new phase of iron exists, it is not likely to be the body-
centred tetragonal.
The LAPW method also shows that the close-packed structure of iron (fcc and hcp)
have similar energies throughout the pressure region of the Earth, while the bcc phase
is highly unfavourable energetically at core pressures (Stixrude and Cohen, 1995).
The calculated total energies of bcc, hcp and fcc are shown in Fig. 14.5.
fcc vs. hcp: c44 /c66 ratio. The elastic constants of iron have been predicted by ab initio
and first-principles theory over a range of pressures (densities) that span those of the inner
core (Steile-Neumann et al., 1998).
The P-wave anisotropy of hcp is found to be nearly identical in symmetry and
magnitude to that of the bulk inner core: P-wave anisotropy is 3% at the density of the inner
core and the fast propagation direction is along the symmetry axis.
The anisotropy of fcc is found to be three times greater for the fast direction of
P-wave propagation along the cubic diagonal. The anisotropy of S-waves is much greater
than for P-waves in both hcp and fcc phases by about a factor of 3. The observed
1004 Chapter 14

cylindrical symmetry of the inner core can be explained by either of the hcp and fcc crystal
aggregates, since the bcc phase is not expected to be stable at inner-core conditions.
Fcc crystal aggregate. Considering an aggregate of fcc crystals with their [111]
axes aligned parallel to the Earth’s spin axis, the effective elastic constants of this
cylindrically averaged fcc aggregate is similar to those of an fcc single crystal.
Hcp crystal aggregate. An aggregate of hcp crystals with their c-axis aligned
parallel to the Earth’s spin axis effectively leads to the elastic constant of a large hcp single
crystal. Shear elastic anisotropy expressed by the ratio C44 =C66 (Steile-Neumann et al.,
1998) is near unity for all hcp transition metals other than iron. The latter shows a much
larger ratio, , 2.5. This apparent high value of the ratio C44 =C66 can be explained by the
slip along basal planes (a common deformation mechanism in hcp metals).

14.2.10.4. Anisotropism: axial angle

The inner-core anisotropy can be explained to first order by the intrinsic elastic
anisotropy of iron and the inner core may be composed of a hexagonal phase, either hcp or
a similar structure (e.g., dhcp or slight orthorhombic distortion of hcp). As already
discussed, its formation by the freezing of the overlying outer core is thought to be the
major energy source for driving the fluid motion that produced the geomagnetic field and
its geometry.
The time taken for seismic energy to propagate through the inner core along a N – S
path is observed to be less than that for a path in the equatorial planes. The fast axis (the
axis of anisotropy) is at an angle of 108 to the N – S axis of the Earth. In short, the cause of
the inner-core anisotropy is believed to be the preferred orientation of hexagonal close-
packed iron, the average axis of which is tilted by a few degrees from the N –S axis.

Seismic signatures. Earlier (Fig. 2.14), it was shown that two branches of PKP travel time
delineate the anisotropy: BC traverses the bottom of the outer core and DF samples the top
of the inner core. But the BC – DF travel-time anomalies arise from inner-core anisotropy,
which corresponds very fairly to the hcp model. The BC – DF travel-time residuals,
signifying anisotropy, are reduced in the quasi-eastern hemisphere (40 –1808E) as
compared with the other hemisphere (Tanaka and Hamaguchi, 1997). Small-scale
heterogeneity has also been noted (Creager, 1997).
The seismological data of the inner core present anomalous splitting of free
oscillation frequencies (Tromp, 1993) and anomalous travel-time observations (Creager,
1992; Su and Dziewonski, 1994). The compressional waves traverse the inner core
(PKIKP) 3 – 4% faster along the Earth’s spin axis than in the equatorial plane. By
considering the high-symmetry lattice strains (Mehl et al., 1990), Stixrude and Cohen
(1995) determined the single-crystal elastic-constant tensor, ðCijkl Þ; for each of the iron
polymorphs which are of the densities seismologically observed to occur at the inner core.
They found that both fcc and hcp are substantially anisotropic in compressional (P)- and
shear (S)-wave velocities. They initially modelled a polycrystalline texture in the inner
core. For hcp, an aggregate with all [001] axes aligned with Earth’s spin axis, the effective
elastic constant of such an aggregate became identical to those of an hcp single crystal.
Iron and Siderophile Elements: The Earth’s Core 1005

They approximated the PKIKP travel-time anomalies ðdtÞ due to inner-core anisotropy as:
dtðD; jÞ ¼ 2tðDÞdVP ðjÞ=VPO ð14-11Þ
where D is the angular distance from source to receiver, j is the angle between the
propagation direction and the symmetry axis, Vp is the velocity of the quasi-P waves, t is
the travel time of PKIKP wave in the inner core and the velocity anomaly is given by
dVP ðjÞ ¼ VP ðjÞ 2 VPO , where VPO is the Voigt – Reuss –Hill energy velocity (which is
11.73 km/s for hcp and 11.92 km/s for fcc).
Their forward and inverse hcp models predict that the equatorially polarized
S-waves (Seq) are slow along the spin axis relative to the equatorial plane and that the
meridionally polarized S-wave (Sme) velocities are fastest for j ¼ 458 (see Song, 1997;
Fig. 14.14) The hcp model predicts a large Sme – Seq splitting of the phase, with Sme-waves
arriving as much as 50 s earlier than Seq-waves. In that case, the inner core would be the
source of the largest shear-wave splitting within the Earth.
The velocity perturbations for weak anisotropy are:
dVP ¼ 1 cos2 j 2 s sin2 j cos2 j
dSme ¼ s ða0 =b0 Þ2 sin2 j cos2 j
dSeq ¼ g sin2 j
p p
where a0 ¼ A=r and b0 ¼ L=r are the P-wave velocity in the equatorial plane and the
degenerate S velocity for propagation along the symmetry axis, respectively.

Figure 14.14. Meridional shear velocities Sme of the inner core (solid line) with respect to propagation directions,
which can be derived from a P wave anisotropy model. The dashed line indicates the shear velocity for an
isotropic inner core. The P wave anisotropy model used in this plot is from Song and Richards (1996a), which
suggests a 12% variation in Sme with the maximum at the propagation direction 458 from the anisotropy symmetry
(from Song, 1997).
1006 Chapter 14

Lattice-preferred orientation: solidification texturing. The stress induced by rotation of

the inner core may produce LPO. The non-hydrostatic stress field induced by rotation is
such that the magnitude of the principal stress parallel to the spin axis is everywhere twice
that in the equatorial directions. The strain energy is minimized by aligning with the spin
axis the direction of stiffest elastic response of the constituent crystals. This corresponds to
the direction [001] of maximum P-wave velocity. The hcp texture minimizes the induced
elastic-strain energy and hence is in equilibrium configuration.
Stixrude and Cohen (1995) concluded that the hcp textural model is indistinguish-
able from a single crystal and the inner core behaves as a very large single crystal of hcp
iron. The bulk magnetic susceptibility of the inner core may also be significantly
anisotropic, affecting the geometry of the geomagnetic field (Hollerbach and Jones, 1993).
Directional cooling in the Earth’s core may possibly cause solidification texturing
and the observed anisotropy. Recently, Bergman (1997) measured the elastic anisotropy of
directionally solidified metallic alloys and observed that the anisotropy is due to
solidification texturing, arising from dendrites growing along the crystallographic axis,
aligned along the direction of heat flow.

Plastic deformation. 1-Iron at pressures above 200 GPa reveals ductile deformation
mechanism that may be responsible for the elastic anistropy. Polycrystal plasticity theory
predicts an alignment of the c-axis parallel to the compression direction as a result of basal
slip (primary or secondary in origin).
Wenk et al. (2000) performed an investigation on the plastic deformation and
calculated elastic properties for the polycrystalline 1-iron up to 220 MPa. They used a
polychromatic synchrotron X-ray source to impinge on the sample at an angle of 848 to the
DAC axis. The diffraction patterns were recorded with an energy-dispersive detector in a
symmetrical position ð2u ¼ 128Þ: The cell was rotated in an axis ’ DAC axis so that
different lattice planes are brought into diffracting positions. Systematic differences in
d-spacings of spectra for lattice planes oriented ’ and // to the DAC axis. The shifts in d-
spacings are related to the elastic deformation experienced under stress (Mao et al., 1998).
The intensities were integrated over the peak width in energy and the background was
subtracted. The c-axes were seen aligned parallel to the compression direction.
Elastically hexagonal crystals are divided into two groups on the basis of c=a ratio,
one with high c=a ratio and the other with a low value. The former is elastically more
anisotropic with basal slip being the primary slip system. 1-Iron has a low c=a ratio (1.604
at 49 GPa; Mao et al., 1990). Basal slip is an active deformation mechanism in 1-iron
(Poirier and Price, 1999).
Differentially oriented grains deform on different slip systems and at different rates.
Basal slip becomes dominant as the preferred orientation develops. If deformation in the
Earth’s core occurs during convection and crystals deform by basal and prismatic slip, then
very strong textures develop (Wenk et al., 2000). However, similar deformation may be
caused by magnetic forces (Karato, 1999).

Presence of liquid spheroids. The inner core shows seismological evidence for
anisotropism for P-waves (3 –4% higher along the pole axis than along the equatorial
Iron and Siderophile Elements: The Earth’s Core 1007

plane), low S-wave velocity and high seismic attenuation. These phenomena can be
accounted for by the presence of a volume fraction of 3– 10% liquid (mixture of elements
with little or no iron) in the form of oblate spheroidal inclusions aligned in the equatorial
plane between iron crystals (Singh et al., 2000).
The P-wave velocity jumps from 10.29 to 11.04 km/s at the ICB and then increases
linearly to 11.26 km/s at the centre of the Earth (Kennett et al., 1995). The S-wave velocity
is , 3.65 km/s at the ICB (e.g., Okal and Cansi, 1998).
Singh et al. (2000) considered the inner core to be composed of spheroidal liquid
inclusions with semi-axes a and c. For Cartesian coordinates x; y and z; x2 =a2 þ y2 =a2 þ
z2 =c2 ¼ 1; x; y # a; z # c: The spheroids would have circular cross-sections in the xy
plane. The ratios of the semi-axes a=c vary from 1 (spheres) to 100 (flat discs) to 0.01 (thin
fibres). Random orientations of these spheroids would make the host solid isotropic, while
alignment in one direction would make it anisotropic.
The seismic observations can be explained by considering 3 –10% melt fraction of
spheroidal liquid inclusions with the a=c ratio of 10 –20 (flat disks) aligned with their c-axis
along the polar axis and with a viscosity of the order of 100 kg/m/s. The symmetry axis of the
hcp Fe and c-axes of the inclusions are thus considered to be aligned along the polar axis of
the Earth. To produce a P-wave anisotropy of 3% for fulfilling seismic observations, a
presence of . 14% liquid in a fibre form ða=c , 1Þ aligned to the polar axis would be
required. A melt fraction of , 3% of inclusions ða=c ¼ 100Þ can reduce the S-wave velocity
from 6 km/s of pure hcp Fe to the seismically observed value of 3.65 km/s.
The attenuation of seismic waves P and S shows a maximum peak at a velocity of
, 250 kg/m/s (Fig. 4; Singh et al., 2000) and diminishes completely for lower and higher
viscosities. If the viscosity is high (say, hl ¼ 108 kg/m/s), the composite of crystal –liquid
responds as a purely elastic material on the time scale of the seismic wave and there is no
attenuation. The attenuation is maximum for hl ¼ 100 kg/m/s.
To explain the seismically observed values of Q21 21
P ¼ 0:05 and QS ¼ 0:01; one
requires the following to satisfy both the P-wave anisotropy and S-wave velocity results:

liquid inclusion: , 6–8% 12–16%

h1: 100 kg m21 s21 100 kg m21 s21
a=c: 10 10

Normal mode studies (Widmer et al., 1991) suggest low attenuation (Q , 3,000)
compared with body-wave results (Souriau and Romanowicz, 1996). There is a frequency
dependence for attenuation from liquid inclusions.
The Earth’s rotation could cause alignment of the liquid in the equatorial plane.
The presence of liquid would also increase the invisibility of the inner core and might lead
to convection as well (Morelli et al., 1986).

Isotropic (?) zones. Some observations indicate that the anisotropy may vanish in the
uppermost 150 km of the inner core (Song and Helmberger, 1995). Furthermore, other
observation also indicates that the outermost 50 –100 km of the outer core is isotropic
(Shearer and Masters, 1992). Possibly, the most recently formed outer region may not have
had sufficient time to develop anisotropy through re-crystallization.
1008 Chapter 14

Figure 14.15. Schematic of isotropic UIC and anisotropic LIC structure. The UIC/LIC boundary (solid line)
would give rise to multiple paths for seismic waves travelling nearly NS through the inner core at certain
distances, producing distorted waveforms in long-period seismograms and multiple arrivals in short-period
seismograms. The boundary is speculated to be irregular, which may explain recent reports of large scatter in inner
core travel times (Tanaka and Hamaguchi, 1997) (from Song and Helmburger, 1998).

Although the anisotropy is prevalent throughout the inner core, the upper part of it
has weak or little anisotropy. Song and Helmberger (1998) present evidence for a
transitional structure of a mostly isotropic upper inner core (UIC) surrounding an
anisotropic lower inner core (LIC). The inner-core boundary structure may thus be marked
by the separation of randomly oriented anisotropic iron crystals in the upper part and
aligned preferentially N – S in the lower part.
The schematic isotropic UIC and anisotropic LIC structure are shown in Fig. 14.15.
The UIC/LIC boundary would give rise to multiple paths for seismic waves travelling nearly
N – S through the inner core at certain distances, producing distorted waveforms in long-
period seismograms and multiple arrivals in short-period seismograms. The boundary is
speculated to be irregular, which may explain the recent reports of the large scatter in inner-
core travel times (Tanaka and Hamaguchi, 1997). A change in the inner-core anisotropy may
explain the triplication produced by a sudden increase in velocity along N –S paths.

14.2.10.5. Discontinuities
The model with a P velocity jump of 4.3% at 250 km below the inner-core boundary
(ICB) clearly reproduces the broad (DF) waveforms of the N –S paths. Further tests show
that models with a 3.5% jump at 200 km below the ICB and a 5% jump at 300 km below
the ICB match the waveforms nearly as well (Song and Helmberger, 1998). A direct
evidence for the discontinuity within the inner core is obtained from short-period
Iron and Siderophile Elements: The Earth’s Core 1009

reflections in records from an earthquake in Drake Passage to the German Regional

Seismic Network (GRSN).

14.2.10.6. Geodynamo: convection and av-dynamo

Convection in the core acts as a giant heat engine, called the geodynamo, which is
powered by the release of gravitational energy. Latent heat release and cooling can also
power the geodynamo. In the heat engine analogy, the thermal convection contributes
, 20% of the power to the geodynamo, while the compositional convection contributes
, 80% of that.
Fluid velocities in the core of the order of 10 km/year are sufficiently rapid to
generate a geomagnetic field through the mechanism of the geodynamo. However, the
magnetic field exerts a strong feedback on convection. The convection in the core is linked
to the rate of cooling and the persistence of the magnetic field over most of the Earth’s
history implies continual cooling and convection in the core. Evidence of a magnetic field
during 3.5 Ga suggests that cooling was rapid enough to sustain vigorous thermal
convection. Once the magnetic field exceeds 1023 T, Lorentz forces become important and
the convection becomes large scale. (Note. When the magnetic field is nominally 1023 T,
the Lorentz and Coriolis forces become comparable.)
The dynamo operating in such a field is called a strong-field regime dynamo. The
Earth may revert itself to a weak-field dynamo, at least temporarily. This temporary
transition may possibly be responsible for the disruption in the magnetic field every
30,000 –100,000 years (Langereis et al., 1997). Upwelling along the rotation axis causes
divergence of flow at the CMB and a westward circulation by conservation of angular
momentum. At the ICB, a convergent flow produces an eastward circulation, which is
coupled to the inner core by magnetic stresses. Because the inner core is free to rotate, the
overriding flow sweeps the inner core in an eastward direction (Aurnou et al., 1996).
Planetary rotation causes spiralling convective flows that align with the rotation
axis. The magnetic field is amplified by these helical motions through a mechanism known
as the a-effect. Zonal flows producing large rates of shear across the outer core may
generate the magnetic field through a mechanism called v-effect. A combination of helical
and zonal flows produces the so-called av-dynamo. Similarly, a 2 dynamo relies entirely
on the helical flows. Dynamo action occurs inside the tangent cylinder through a
combination of zonal and helical flows. A combination of helical and zonal flows, i.e., av
dynamos, should produce a dominant dipole field at the CMB. The magnetic fields are
generated primarily by helical flows with relatively little contribution from zonal flows
(e.g., a2 dynamo).
Magnetic waves are expected to have periods of 10– 102 years (e.g., Zarman and
Bloxham, 1997). Some suggest that the generation of a magnetic field is tied in some way
to lateral heterogeneity at the base of the mantle (see Buffett, 2000). Electric currents from
the core flow along conductive pathways through the mantle. Variations in electrical
conductivity may also contribute to the exchange of angular momentum between the core
and the mantle (Holme, 2000).
The inner core is proposed to rotate faster than the mantle (Song and Richards,
1996). The super-rotation may involve electromagnetism (Glatzmaier and Roberts, 1996)
1010 Chapter 14

and viscous (Su et al., 1996) coupling to the outer core. If the core had been immobile, the
geomagnetic field would have decayed on a time scale of , 10,000 years. The energy
supplied within the core to drive the dynamo must be at the rate of at least 3 £ 1010 W
(Merrill et al., 1996). The magnetic energy in the core may be contained in a toroidal field,
which may be larger that the poloidal by a factor Rm , 20 (the magnetic Reynolds
number). The energy required to sustain the geodynamo may be of the order of 4 £ 1012 W,
85% of which is released as heat to the overlying mantle (Stixrude and Brown, 1998).
A finitely conducting solid inner core would stabilize the geodynamo because the
paramagnetic relaxation time of the inner core lags behind external field changes from the
outer core. The outcome is that short-term fluctuations are damped out and the dynamo is
steadied. However, a brief breakdown of the dynamo would not lead to a reversal of the
field because of the stabilizing influence of the inner core on the convection of the outer
core (Gilder and Glen, 1998).
The energy for driving the magnetic dynamo is derived from (a) latent heat of
crystallization of the inner core or (b) from convection due to separation of light elements
during freezing (Buffett, 2002). The Earth’s geodynamo could hardly establish without an
inner core. Since there are distinct pieces of evidence for a geomagnetic field in rocks as
old as 3.5 b.y., the Earth’s inner core must be that old.
If the inner core grew to its present size over the past 4.6 b.y., the heat of
crystallization would provide 0.5 TW (1 TW ¼ 1012 W) of the total core heat flux. The
total core heat flux determined from convective upwellings from the CMB (mantle plumes)
is , 10 TW. This would correspond to a core cooling rate of 120 – 200 K every billion
years. If the inner core formed , 4.6 b.y. ago, then , 4 TW must have been released
through radioactive heating.
Martian magnetic dynamo. The inner core of Mars is thought to have a high sulphur
content and to have formed at lower pressure. It could have very high concentrations of
potassium. Therefore, radioactive decay of 40K could have been the major source of energy
for an early magnetic dynamo in Mars.

14.2.10.7. Geomagnetic-field propagation

The magnetic field reversals, or its manifestation, may possibly be controlled by the
lowermost mantle (Marzocchi and Mulargia, 1992). Iron enrichment of the lowermost
mantle should increase the electrical conductivity of the zone above the CMB. However,
with respect to solid silicates, the conductivity of molten silicates is generally higher by
one to two orders of magnitude (Li et al., 1990). The ULVZ may, therefore, be associated
with high electrical conductivities but the short-period shifts (“jerks”) in the field intensity
may also contrarily suggest the presence of low-conductivity regions near the CMB
(Aurnou et al., 1996).
The core-generated field may experience a screening effect by the heterogeneous
high-conductivity layer within the mantle, with a resultant electromagnetic torque on the
deep mantle. Such a torque is possibly manifested by the decadal fluctuations in the Earth’s
rotational rate (Stewart et al., 1995). Thus, above the CMB in the lowermost mantle, the
temperature field, the compositional variations, partial melting and plume activity bear
potentially profound implications for propagation of the geomagnetic field through the
Iron and Siderophile Elements: The Earth’s Core 1011

mantle, for the manner in which the magnetic field reverses and for the frequency of
reversals (Larson and Olson, 1991). Shifts in plume flux may alter the temperature
distribution in the mantle.

14.2.10.8. Magnetic field, heat flow and plate tectonics

Generation of a magnetic field requires a churning molten iron core that acts as a
geodynamo. The churning is driven by heat flowing out of the core, much as tectonics at
the surface is driven by heat flowing out of the planet.
In the absence of plate tectonics, the heat would not pump out to the surface and
the interior heat would warm up the body as hot as the core, when no heat can move out
of the core. In such a situation, the churning due to heat convection would slow down and
the magnetic field should die out.

14.3. 1-Fe and paramagnetism

The inner core may thus most conveniently be conceived as composed of an

aggregate of preferentially oriented hcp iron (1-Fe) crystals (Clement and Stixrude,
1995). As stated already, such an aggregate could also explain why seismic waves
travelling parallel to the rotational axis appear 1 –4% faster than those travelling
equational paths.
Karato (1993) proposed that the toroidal component of the field could give 1-Fe a
preferred orientation such that its crystallographic c-axis grows parallel to the rotation axis.
This can happen if 1-Fe is paramagnetic with a certain degree of crystalline anisotropy of
magnetic susceptibility (AMS) (DeRango et al., 1991).
Clement and Stixrude (1995) assigned the magnetic susceptibility of 1-Fe to be
1023 –1024 (SI units). Gilder and Glen (1998) opined that, at only 13 GPa, a-Fe
can completely convert to 1 phase and they found that, at 16.9 GPa/2618C, the hcp
1-iron is attracted to a magnet (i.e., either paramagnetic or ferromagnetic) and
show magnetic susceptibilities from 0.15 to 0.001. Magnetization ranges from 1,800 to
15 amp/m.
Recent experiments using Mössbauer-effect measurements detected only internal
magnetization above 0.05 T (Taylor et al., 1991). Band-structure calculations predict 1-Fe
should not order magnetically (Fletcher and Addis, 1974), suggesting that it is
paramagnetic. The estimated susceptibility values have to be valid at high core
temperatures (4,000 –8,0008C) and pressures (330 – 360 GPa).
The P, T boundary (see Fig. 14.17) at which Fe converts from ferromagnetic bcc
phase (a-Fe) to paramagnetic hcp (1-Fe) is hysteretic with near-vertical Clapeyron slopes
of 455, 2283 and 2 4558C per GPa and the corresponding room temperature transition
pressures are 7.4, 13.5 and 9.0 GPa, respectively (Bassett and Weathers, 1990). The width
of the a – 1 transition varies with increasing shear strength of the pressure medium. Bargen
and Boehler (1990) postulated that crystal size is correlated with hysteresis width.
The single crystals transform to high-pressure phases faster than polycrystalline ones
because the former have fewer grain boundaries and defects.
1012 Chapter 14

At the core P – T condition, the paramagnetic inner core can be responsible for a
variety of geomagnetic phenomena.

14.3.1. Hugoniot temperature

Under high-pressure Hugoniot conditions (equivalent to the Earth’s core

condition), the electronic contribution to specific heat is two-thirds as large as the lattice
contribution. Hugoniot temperatures for 1-iron at 200 GPa, determined through integration
of thermodynamic quantities, are found to be within experimental errors of results based on
radiative measurements of the shocked iron interfaces.
Based on Hugoniot measurements, the melting temperature for iron at 243 GPa
is 5,600 ^ 500 K. The Hugoniot temperatures for 1-iron in the pressure range of
0 –290 GPa,as presented in by Bones and Brown (1990), are shown in Table 14.10

14.4. Iron and tungsten: yield strengths

In an experiment by Hemley et al (1997), polycrystalline W and Fe were loaded in a

specially prepared Be gasket mounted at the 10-mm tips of single-bevelled diamonds.
Tungsten remains cubic (bcc) at megabar pressure but transforms to hexagonal (hcp)
structure at 13 GPa (Mao et al., 1990).
The diffraction pattern for the radial X-ray measurement with C ¼ 908 was seen to
be identical to that obtained by axial diffraction (C is the angle between the diffraction
vector and the load axis).

TABLE 14.10
Hugoniot temperatures for 1-iron (Boness and Brown, 1990)

Pressure (GPa) Temperature (K)

0 300
20 345
40 518
60 800
80 1,164
100 1,594
120 2,076
140 2,600
160 3,157
180 3,741
200 4,348
220 4,977
240 5,625

On the basis of measurements of the Grüneisen parameter for liquid iron (Brown, 1989), the adiabatic temperature
difference between the ICB and the CMB is ,1,500 K. This leads to an estimated temperature at the CMB of
3,200 K (Boness and Brown, 1990).
Iron and Siderophile Elements: The Earth’s Core 1013

Hemley et al. (1997) found that the deviatoric stress component t increased with
increasing P, reaching values of , 20 GPa for W and Fe at 200 and 300 GPa,
respectively (Note: t ¼ # sV ¼ 2t; where sV and t are the yield and shear strengths of
the material; for details of t; see Section 5.3.1.1.2.) Their results showed that the yield
strengths of W and Fe exceed 20 GPa at 200 –300 GPa. This indicated a significant
increase relative to ambient condition, where sV (W) ¼ 0.6 GPa (Raffo, 1969) and sV
(Fe) ¼ 0.03 GPa (Spitzig and Leslie, 1971). They also observed that the yield strengths
of W and Fe are higher than the uniaxial stress component. This can also be supported
by theoretical estimates based on scaling to the shear modulus (Kelly and Macmillan,
1986). All these results demonstrate that the strength of these materials enhances
dramatically under ultra-high pressures.

14.5. Fe –Ni alloy

The estimated relative abundance of Ni on Earth is , 2.4 wt% (Anderson, 1977)

and, based on chondritic composition, the estimate of Ni at the core is , 6% (e.g.,
Ringwood, 1977). The presence of Ni in Fe does not appreciably change the volume in
X-ray experiment.
It has been known for a long time that fcc alloys of Fe and Ni with a composition of
, 35% Ni and 65% Fe exhibit almost zero thermal expansion over a broad temperature
range. These are known as Invar alloys. The other variant is “Elinvar” alloy, which shows a
negligible change in elasticity when heated. (These were used to make the springs in
mechanical watches.)
At room temperature, the thermal expansion coefficient, a; of Invar is roughly
constant and usually ranges from 10 £ 1026 to 20 £ 1026 K21. All Invar alloys are
magnetic so that a is only almost zero below the magnetic ordering temperature Tc and
rises to a higher value for temperatures above Tc :
To explain the magnetic behaviour of Invar, the assumption is made of two different
magnetically ordered states (2g model): g1, with magnetic states aligned parallel, having a
larger volume, and g2, an anti-ferromagnetic state with a smaller volume but at a slightly
higher energy. With increase in temperature, the g2 state becomes thermally excited. This
smaller volume state compensates for the vibrational thermal expansion. In effect, the
thermal expansion coefficient becomes low. Above Tc (the magnetic ordering
temperature), the system becomes paramagnetic with magnetic spins randomly oriented
and no further magnetic compensation for the thermal expansion is possible. Thus, a
catches up with the usual high value. With the advent of quantum mechanical calculations
of the electronic and magnetic structure of metallic solids, further insights beyond the
2g-state model have become possible.
In their calculations, van Schilfgaarde et al. (1999) included non-collinear spin
ordering. Thus, spins may be tilted at arbitrary angles to each other if such a
configuration lowers the energy of the overall spin system. The calculations show that,
at large volumes (< low temperature), the ground state is given by parallel
(ferromagnetic) spin alignment. When the volume is reduced (simulating increasing
1014 Chapter 14

temperature), the spins gradually depart from parallel alignment and the spin directions
become increasingly disordered. In magnetic materials, increase in temperature leads to
greater disorder in spin alignment (Note. At Tc ; the disorder is complete and no net
magnetic moment remains.)
For Invar alloys, increasing disorder is volume dependent (viz. volume decreases),
thereby compensating the vibrational thermal expansion. The Invar alloys have
compositions close to structural instabilities. Hence, Fe– Ni alloys with Ni content less
than 32% show a crystal structure which is no longer fcc but bcc.
Mao et al. (1990) made a compression study on iron and an Fe0.8Ni0.2 alloy to above
300 GPa. The a and c unit-cell parameters of 1-Fe (hcp) were determined using XRD,
which indicated that pure Fe remains in the hcp structure to 304 GPa and Fe0.8Ni0.2 is
stable to at least 255 GPa. The pressure – volume data are analyzed with a Birch–
Murnaghan Eulerian finite-strain formalism (BME) modified for a high-pressure
unquenchable phase (Jeanloz, 1981). The pressure –volume curve for pure iron and
Fe0.8Ni0.2 is shown in Fig. 14.16 (see caption; Mao et al., 1990).
A plot of the true variation of c=a with pressure for hcp iron was shown in Fig. 14.11
depicting that the c-axis is more compressible than the a-axis. Above the shock pressure of
200 GPa, a solid – solid transition to g-phase of iron occurs and subsequently moves to a
melt. The density of iron predicted at the pressure of the ICB at 300 K from the recent data
is 13.8 ^ 0.1 mg m23, which is in very good agreement with the extrapolated shock EOS
(see Table 14.11).

Figure 14.16. P – V curve for pure iron and Fe0.8Ni0.2. Crosses are from Jephcoat et al. (1986a) rectangles are
synchrotron X-ray data, and triangles are from Mao et al. (1967). For Fe0.8Ni0.2, diamonds are from Takahashi
et al. (1968) for the composition Fe0.897Ni0.103. Both pure iron and the alloy remain in hexagonal close-
packed structures to the highest pressures. The linear dimensions of rectangles represent P – V error of the data
points with datum at box center. Solid line is a third-order BME. EOS fit to the data; there is no difference on the scale
of the diagram from this EOS formalism and the VFRS method (Mao et al., 1990, q 1990 American Geophysical
Union).
Iron and Siderophile Elements: The Earth’s Core 1015

Figure 14.17. P – T phase diagram of iron. The low-pressure and low-temperature phase of iron has a bcc structure
and is strongly ferromagnetic. The higher-pressure phase has a hcp structure and might be weakly or nearly
antiferromagnetic at low temperatures. Shimizu et al. (2001) show that below 2 K and at pressures above 10 GPa
the hcp phase of iron becomes superconducting. (The superconducting transition temperature, Tc has been
magnified hundred-fold.) The shape of the phase diagram at low temperatures in the border region between the
ferromagnetic bcc and the superconducting hcp phases has still not been established experimentally. The third
structural form of iron shown here is fcc (from Littlewood, 2001).

14.6. Fe –Si alloy

Because of its cosmochemical abundance and its ubiquity in the Earth’s mantle,
silicon has been traditionally proposed as an important alloying element in the Earth’s
outer core (Birch, 1952) Zhang and Guyot (1999) carried out in situ synchrotron XRD
study on iron and iron –silicon alloy (Fe0.91Si0.09) at simultaneously high pressure
and temperature. They observed that substitutional silicon has very little effect on the
thermoelastic parameters of iron-rich alloys in the Fe –Si system. The effect of silicon
content on the room-temperature bulk modulus of iron-rich Fe – Si alloys is less than
0.3 GPa for 1 wt% silicon. Also, the substitution of silicon in iron is not expected to change
the thermoelastic properties of such alloys. The similarity of thermoelastic parameters of

TABLE 14.11
EOS parameters for 1-Fe and 1-Fe0.8Ni0.2

Composition EOSa V02 (cm3/mol) K02 (GPa) K 002 rb (mg/m3)

Fe BMEc 6.73 (1) 164.8 (3.6) 5.33 (9) 13.8(1)

Fe0.8Ni0.2 BMEd 6.737 (5) 171.8 (2.2) 4.95 (9) 14.1(1)
Fe, shock BMEe 6.724 (8) 172.6 (2.7) 5.14 (7) 13.8
a
EOS formalism (Birch expansion truncated at third order term of energy strain).
b
Density evaluated at P ¼ PICB ¼ 330 GPa.
c
EOS parameters determined from fit of the 1-phase data from Jephcoat et al. (1996a,b) and Mao et al. (1967).
d
EOS parameters determined from fit to data of Mao et al. (1990) and Takahashi et al. (1968).
e
Fit to reduced isotherm from Brown and McQueen (1982).
1016 Chapter 14

pure iron and iron-rich Fe– Si alloys is of significant geophysical importance. This relation
holds true to the Earth’s outer core. If this relation persists, the relative density contrast
between iron and iron-rich Fe –Si alloys at outer-core condition could be close to that
measured at ambient conditions, i.e., 0.6% for 1 wt% silicon. This, indeed, corresponds to
a much earlier conclusion by Balchan and Cowan (1966), who, on the basis of shock-
compression experiments, opined that the 10% density deficit in the outer core could be
interpreted by the presence of 14– 20 wt% silicon in iron.
In the Fe – Si binary system, the presence of a closed fcc loop in the Fe-rich region
has been known. In such a system, silicon appears to be a ferrite (bcc phase) stabilizer.
A knowledge of the bcc-stabilizing character of Si and the maximum solubility of silicon
in fcc phase change with pressure is of great importance in realistic modelling of the
thermodynamics of silicon incorporation in ferrous metallic phases at the Earth’s core.
In the Fe –Si system, the observations by Zhang and Guyot (1999) indicate that the
ferrite (bcc phase)-stabilizing behaviour of silicon persists at high pressures and that the
maximum solubility of silicon in the fcc phase increases with increasing pressure.
The transformation from the bcc phase to the fcc phase was observed in Fe0.91Si0.09
at 6.0, 7.4 and 8.9 GPa at varying temperatures and the temperatures measured at the onset
of the transformations were 3008C higher than those in iron at similar pressures. The
transformation rate in Fe0.91Si0.09 was extremely sluggish compared with that of iron.

14.7. Fe –H system

The affinity of hydrogen for iron increases significantly with pressure This behaviour
makes the iron –water reaction and model reaction in the primordial Earth (e.g., Sugimoto
and Fukai, 1992). The Fe – H system has been investigated by many workers for its structural
properties (e.g., Badding et al., 1991) and high P – T melting (e.g., Okuchi, 1998).
Fe –H crystallizes in double-hexagonal close-packed (dhcp) structure near 3 GPa
and beyond (Badding et al., 1991). There is a 17% expansion of the unit cell in the dhcp
relative to that of hcp 1-Fe phase.
The surface tension of molten iron hydride as a function of T (at 5 GPa) is much
greater than that of silicates. Hence, gravitational separation of molten iron occurs when
the degree of melting is very high. It is seen that in the iron –enstatite – water system, iron
reacts with water to form iron hydride above 2.8 GPa and 5508C (Yagi and Hishinuma,
1995). The melting temperature of FeH3 is found to have a minimum at 3 – 4 GPa, some
6,0008C below that of pure iron.
Iron can react with water through two pressure-dependent reactions:
MgSiO3 þ Fe þ H2 O ! ðMg; FeÞ2 SiO4 þ H2
and
3Fe þ H2 O ! 2FeH þ FeO
According to Okuchi (1997), in a hydrous magma ocean . 95% of H2O should react with
Fe to form FeH. Presence of hydrogen as a light element may account for 60% of the
Iron and Siderophile Elements: The Earth’s Core 1017

density deficit. Okuchi also opined that 95% of the H2O is consumed at temperatures above
the dry solidus (,1,8008C). At lower T, the silicate (i.e., (Mg, Fe)2SiO4) crystallizes and all
H is consumed to form FeH. Thus, the formation of FeH helps the drying out of the silicate
mantle in the primordial Earth. Increasing P, T further stabilizes FeH.

14.8. Sulphur in the core

It is suggested that up to 10% S may be present in the inner core and FeS in liquid
form may get trapped there Brown et al. (1984) determined the outer core to be well
matched with 10 wt% sulphur. Employing first-principles calculations, Sherman (1995)
observed that the inner core is more likely to contain sulphur than oxygen. Boehler (1992)
suggested that the Fe– O – S system has a melt temperature near 3,300 K at the CMB. Later
(1996) working on Fe – FeS system he opined an enhanced solid solution at high pressure.
Amongst Fe, S and O, sulphur has the largest radius at low pressure. Iron is
intermediate in size and oxygen is significantly smaller. Above 100 GPa, sulphur assumes
a radius identical to that of iron. Oxygen experiences a much smaller variation in radius
with pressure and remains some 20% smaller than iron or sulphur throughout most of the
core pressure. The reason for this may be stated as in the following.
Sulphur (with [Ne] 3s23p4 structure) has both a spherical closed-shell core and
unoccupied 3d electronic states whose energy levels are only slightly higher than the 3p
states. Oxygen ([He] 2s22p4) lacks a closed-shell configuration with p electrons. As a
result, the valence electrons are more tightly bound and energy levels for unoccupied
electronic states are at a substantially higher energy. An intrinsically non-spherical
electron distribution may inhibit formation of close-packed phases. Calculations for each
orbital quantum number of the partial density of states as a function of increasing
compression reveal an electronic transition for sulphur. With increasing compression,
sulphur d bands are filled at the expense of the p and s bands. But in oxygen no such filling
of inner orbitals is possible in the pressure regime up to 400 GPa (Boness and Brown,
1990). Thus, its EOS is stiffer and it maintains p-band character under high compression.
At ambient condition, sulphur is molecular and forms crystalline rings with S8 species.
S below 90 GPa behaves as a semiconductor but at 93 GPa S becomes metallic and
superconducting with Tc ¼ 10:1 K.
First-principles calculations predict that S compressed under pressure from 280 to
540 GPa favours the simple cubic structure in which the average phonon frequency kvl
increases, leading to a smaller l and smaller Tc : Ultimately, S transforms to a bcc phase.

14.8.1. Oxygen and sulphur solution in iron

Both the atomic radius (smaller than iron) and electronic structure of oxygen make
it incompatible with iron in close-packed metallic alloys. Hence, a solid solution of oxygen
in iron under pressure is not possible although oxygen and iron can form compounds at
ambient condition.
1018 Chapter 14

However, sulphur has a spherical closed-shell electronic configuration with an

atomic volume that is compatible with iron. Its s, p and d electronic states hybridize under
pressure and thus sulphur forms a binary metallic alloy with iron, which has a solid-
solution chemistry.
Sulphur and oxygen have differing high-pressure behaviour. Sulphur experiences a
marked s –p state ! d state electronic transition with increasing compression. This leads
to an atomic volume compatible with ideal mixing in close-packed iron-dominated
structures. In the iron-sulphide system, sulphur and iron have overlapping bands that
further enhance solid-solution behaviour. In contrast, oxygen, with no closed-shell
p-electrons, does not become a d-band metal at high pressure. Its smaller atomic volume
excludes solid-solution behaviour with iron.

14.8.1.1. S, Se and Te
S, Se and Te under pressure undergo transitions to extended structures (non-
molecular) and become metallic. At low pressure, they crystallize in base-centred
orthorhombic (bco) structures and are metallic and superconducting. On further
compression, they transform to b-Po structure.
In selenium, superconductivity is observed from 15 to 25 GPa with Tc changing
from 4 to 6 K and, above 150 GPa, with Tc of 8 K (Gregoryanz et al., 2002).

14.9. Iron sulphides

Ferrous iron is coordinated in a number of minerals by an octahedral array of sulphur

atoms. The monosulphide, troilite (FeS), is hexagonal (average d(Fe –S) < 2.50 Å),
whereas pyrrhotite (Fe12xS) exists in both hexagonal and monoclinic forms. In cubic
disulphide pyrite (FeS2), the FeS6 octahedron is highly symmetric (d(Fe –S) < 2.26 Å)
although the kS – Fe– Sl is 868 instead of 908. The tetragonal marcasite (FeS2) has an
octahedral geometry with c-axis elongated by shortening of the edges, shared between
adjacent FeS6 polyhedra.
Many aspects of the electronic structure of ferrous iron in octahedral sites are better
understood by employing a FeS102 6 cluster model. In the cluster approach, the relative
energies of different electron configurations such as different spin states can be made
amenable for calculation.
Cosmochemical and geophysical arguments (e.g., Boehler, 1992) suggest that
sulphur may be the lighter element in the Fe-rich cores such as the Earth and Mars.
Because the Earth consists of a liquid outer core and a solid inner core, the melting
relations in the system Fe –FeS at high pressure are used to estimate the core temperature.
Many models of core T and core formation processes were based on extrapolation
of melting data in the system Fe – FeS obtained at relatively low P(, 10 GPa)
(e.g., Usselman, 1975).
Hence, a knowledge of the properties of iron sulphide under high pressure and
temperature is key to the understanding of the core of terrestrial planets. In situ XRD
measurements revealed that FeS, a possible core material for the terrestrial planets
Iron and Siderophile Elements: The Earth’s Core 1019

(such as Earth and Mars), transforms under high pressure and temperature to a
hexagonal NiAs superstructure with axial ratio ðc=aÞ close to an ideal close-packing
value of 1.63.
In situ synchrotron XRD measurements of FeS using an externally heated diamond-
anvil cell with the sample loaded in a hydrostatic Ne pressure medium (Fei et al., 1992)
indicate that troilite (FeS) with a ð3a; 2cÞ unit cell transforms to a MnP-type structure at
about 3.4 GPa (King and Prewitt, 1982). Above 600 K, troilite transforms to a simple
NiAs-type structure.

14.9.1. FeS: five polymorphs

Under ambient pressure, FeS is an anti-ferromagnetic insulator (TN ¼ 598 K) and

has a NiAs-related (troilite) structure. FeS undergoes two structural phase transitions at
ambient temperature, from the NiAs-related to a MnP-related structure at 3.5 GPa and then
to a monoclinic phase at 6.5 GPa. The latter transition causes shortening in the c-parameter
from 5.70 to 5.54 Å (Fei et al., 1995).
Fei et al. (1995) and Fei and Prewitt (1996), using high-pressure experiments, found
five polymorphs of FeS (Fig. 14.18). Three are NiAs-type hexagonal structures, which can
be described with ð3a; 2cÞ; ð2a; cÞ and ða; cÞ unit cells. There are two other polymorphs of
lower symmetry. All the high P – T phases of FeS are non-quenchable and the transitions
are reversible. The stoichiometric troilite (FeS I), having a NiAs-type structure with a
ð3a; 2cÞ unit cell, transforms at , 3.4 GPa to a MnP-type structure (FeS-II) (Fig. 14.19). At
pressures above 6.7 GPa, FeS III phase forms. Upon heating at pressures below 4 GPa,
troilite transforms to a hexagonal phase with a ð2a; cÞ unit cell (FeS IV) (Fig. 14.20) and
then to a simple NiAs-type structure with ða; cÞ unit cell (FeS V) (Fei et al., 1995). The
experimentally determined phase diagram of FeS is shown in Fig. 14.18. The phase
boundary of the FeS III –FeS IV transition was determined to be

P ¼ 11:25 þ 0:485 T ðP in GPa; T in KelvinÞ

The energy dispersive XRD spectra of FeS compounds at varying temperatures and
pressure are shown in Fig. 14.21. A discussion on the phases of FeS III, IV and V follows.

14.9.1.1. FeS III, monoclinic

A high-pressure FeS phase (FeS III) forms at pressures above 6.7 GPa at room
temperature (King et al., 1978). The high-pressure FeS III is monoclinic with space group
P21 =n: The stability field of FeS III was summarized by Fei et al. (1995), who (in 1998)
obtained XRD data for the high-pressure phase FeS III at 10.95 GPa and 3,000 K by using
both energy-dispersive and image plate techniques (Fig. 14.22a,b). They indexed the
image plate diffraction data on a monoclinic unit cell and obtained cell parameters:


a ¼ 8:025ð4Þ A


b ¼ 5:614ð2Þ A
1020 Chapter 14

Figure 14.18. P – T phase diagram of FeS. The boundaries among phases I – V are determined by X-ray diffraction
with resistance-heated diamond cells. Phase IV, which has B8 structure, is pertinent to the P – T conditions of the
Martian core as indicated (Fei et al., 1995).


c ¼ 6:414ð4Þ A and
b ¼ 93:018
These values are similar to those obtained by Kusaba et al. (1997), who proposed P2 or
P21 =m as possible space groups for the monoclinic phase FeS III.
The FeS III phase transforms to a hexagonal phase, akin in structure to FeS IV but
with an ideal close-packing value of 1.63. At about 6 GPa, an abrupt shortening in the
c-axis is observed, arising from a shortening of the Fe –Fe distance. Further shortening
would lead to metallization of FeS. The c-axis shortening may possibly accompany a spin-
pairing transition. X-ray spectra of FeS through the transition are shown in Fig. 14.24
(Rueff et al., 1999). The shorter Fe –Fe distances in FeS (III) are associated with the change
in spin state. Recent electrical conductivity measurements have confirmed the insulating
state of FeS (III) (Takele and Hearne, 1999).

14.9.1.2. FeS IV, hexagonal (2a,c)

Upon heating, the monoclinic phase FeS III transforms to FeS IV, a phase with
NiAs-type hexagonal structure with a ð2a; cÞ unit cell (Fei et al., 1995). FeS IV is seen to be
Iron and Siderophile Elements: The Earth’s Core 1021

pffiffiffiffiffi
Figure 14.19. NiAs-type substructure lattice parameters a and c and the c=a ratio of FeS I ð 3a; 2cÞ FeS IV
ð2a; cÞ and FeS V ða; cÞ as function of temperature at a pressure of about 3 GPa (Fei and Prewitt, 1996).

stable over a high and wide P – T range. At about 6 GPa, an abrupt shortening of the c-axis
with 4% density change is observed but with no structural change. This change may be due
to an electronic/magnetic transition (Fei et al., 1995). The determined cell parameters at
5.2 GPa are:


2a ¼ 6:7175 A


c ¼ 5:6542 A:

57
Fe Mössbauer spectra taken between 3.0 and 7.7 GPa (Fig. 14.23a – d) also showed
a dramatic change. This transition pressure at room temperature occurs at the
electronic-transition boundary dividing the FeS IV from the FeS III phase. Experiments
by Fei and Campbell (1995) indicate that eutectic composition has a sharp change at
, 7 GPa, coinciding with electronic-transition pressure. Above this pressure, sulphur
becomes soluble in pure iron.
1022 Chapter 14

Figure 14.20. NiAs-type substructure lattice parameters a and c and the c=a ratio of FeS IV as a function of
pressure at 500 K (solid squares) and 600 K (open squares) (Fei And Prewitt, 1996).

14.9.1.3. FeS V: hexagonal (a,c)

The hexagonal FeS IV phase with a ð2a; cÞ unit cell transforms under higher
temperature to a simple NiAs-type hexagonal structure with ða; cÞ unit cell, assigned as
FeS V (Fei et al., 1995). The transition is marked by the extinction of reflections in XRD
related to the ð2a; cÞ superlattice, indexed as 101, 112, 211, 301 and 103 in the FeS IV
phase (space group P63 =mmc). The refined lattice parameters for FeS V at 3.4 GPa and
650 K are:
a ¼ 3:4386ð8Þ A 

c ¼ 5:6760ð8Þ A
In the energy-dispersive XRD spectra of FeS under pressure (Fig. 14.21), the
disappearance of the 211 peak of FeS I and the emergence of the 301, 103 and 311
Iron and Siderophile Elements: The Earth’s Core 1023

Figure 14.21. Energy-dispersive X-ray diffraction spectra of FeS I, FeS IV, and FeS V, collected at temperatures
of 380, 392 and 667
pffiffiffi K respectively, at a pressure of about 3 GPa FeS I, and FeS IV were identified as NiAs-type
structure with ð 3a; 2cÞ; ð2a; cÞ; and ð2a; cÞ unit cell, respectively. The hkl indices are indicated for each
diffraction peak. Gold was used as internal calibrant (Fei and Prewitt, 1996).

peaks of FeS IV mark the transition from FeS I to FeS IV with increasing temperature. All
the super-lattice diffraction peaks, such as 301, 103 and 311, eventually disappear and,
above , 640 K, a simple NiAs structure of FeS V is formed. A plot of c=a ratios also
clearly demonstrates the discontinuity between FeS I ð3a; 2cÞ; FeS IV ð2a; cÞ and FeS V
ða; cÞ (Fei and Prewitt, 1996) as a function of T (K) and P (GPa) (Figs. 14.18 and 14.19).
The high-pressure transformation in FeS observed by various studies is presented in
Fig. 14.23 (Hemley et al., 1998) (read the caption).

14.9.1.4. Spin state of ferrous iron

At ambient condition, in troilite and pyrrhotite, the d6 electrons occur in a high-spin
quintet state. In pyrite, however, Fe2þ is in a low-spin singlet state. Both troilite and
pyrrhotite under pressure show a change in their Mössbauer spectrum. The multi-peak
magnetic hyperfine spectrum, produced by the anti-ferromagnetic ordering of the high-
spin Fe2þ, collapses under pressure to a single peak with a reduced isomer shift.
This change reverses when pressure is released.
Fe2þ occurs as a substitutional impurity in MnS2, which contains S42 2 anion but has
a Mn – S distance of 2.59 Å, which offers a much larger metal site than in FeS2. Hence,
MnS is high spin at atmospheric pressure but converts to low-spin state at high pressure.
The Fe – S distance (not S22 vs S22 distance) apparently determines the spin state of
Fe in sulphides. Thus, Fe2þ in MnS2 or FeS2 will spin pair near a characteristic (average)
2þ

Fe –S distance. In the FeS102

6 cluster, with an average Fe –S distance of 2.50 Å, the radii of
1024 Chapter 14

Figure 14.22. (a) Powder XRD pattern and Rietveld refinement of FeS(III) (Fei et al., 1998). (b) Structure of
FeS(II) and FeS(III) (Source: Hemley et al., 2000b, q The Mineralogical Society).

overlapping (by about 20%) Fe and S ions can be taken as roughly equal in size. In this
cluster, the maximum azimuthal quantum numbers for expansion functions are l ¼ 3 for
the Fe spheres and l ¼ 1 for S (i.e., d functions are excluded). A Watson sphere of uniform
positive charge totalling 10 electronic charges would pass through the S nuclei, thus
stabilizing the entire cluster.
The MO diagrams for the quintet state, the singlet state and a transition state with
orbital occupations half-way between those of the quintet and singlet states are compared
at fixed R (Fe – S) ¼ 2.26 Å (Tossell, 1977). In each case, the orbitals are labelled
according to the irreducible representations of the assumed octahedral point group.
Iron and Siderophile Elements: The Earth’s Core 1025

Figure 14.23. High-pressure transformation in FeS. (a) Mössbauer spectra of phases I, II, and III showing the
collapse of the quadrupole splitting in the (room-temperature) high-pressure phase. (b) K0b satellite as a function of
pressure. The solid line is a guide to the eye; the dashed line shows the zero intensity level. (d) Near infrared
reflectivity as a function of pressure showing that the transition is not accompanied by a major-change in
reflectivity as expected for a transformation to a highly conducting metal. The pressures are listed in gigapascals
on the left (Source: Hemley et al., 1998, q 1998 Mineralogical Society of America).
1026 Chapter 14

The orbital energies are given in eV with respect to an assumed constant average energy
for the S 3p lone-pair orbitals of 1t1g ; 3t1u and 1t2u symmetries.

14.9.1.5. R (Fe –S) change and spin splitting

In FeS, which is a high-spin system with a long average Fe – S distance (2.50 Å),
there are three occupied Fe 3d-type crystal-field orbitals. These extend from that of the S
3p non-bonding orbitals to about 3 eV above the non-bonding orbital. Again, due to the
longer Fe –S distance in FeS, the separation of S 3p non-bonding orbital from Fe – S 3p
bonding orbital will be smaller and the non-binding energy shoulder will be resolved
corresponding to the bonding orbital set.
In the optical spectra of FeS, an intense absorption occurs at an energy of less than
1.9 eV. The calculated separation ðRÞ of 2t2g # and 3eg # orbitals in perfectly octahedral
FeS102
6 is (R ¼ ) 2.50 Å (< 1.4eV). The energy calculated for the lowest S 3p non-bonding
Fe, e.g., the crystal-field transition ð1t1g # !3eg #Þ is about 4.5– 5.0 eV, which is
substantially lower than that in FeS2.
In the FeS102
6 system, a reduction of the FeS distance in the quintet state causes
(i) an increase in the 2t2g – 3eg splitting, (ii) a destabilization of the crystal-field orbital
weighted average energy (baricentre), (iii) a stabilization of the Fe– S 3p bonding orbitals
(2a1g ; 2t1u and 1t2u ) and (iv) a stabilization of the S 3s-type orbitals. These effects are
qualitatively similar to those observed in oxides.
The destabilization of the crystal-field orbital baricentre increases by R(Fe –S)21,
while the spin-averaged 2t2g – eg splitting increases at a rate approximately proportional to
R25 : The difference in orbital energies between quintet and singlet states is also dramatic.
In FeS, the spins pair at quite low pressures (Vaughan and Tossell, 1973), while the
spin pairing in FeO has not been observed at pressures as high as 30 GPa. The ease with
which FeS can be spin-paired is found to be primarily a result of its high compressibility
and its large molar-volume change upon spin-pairing.
For FeO6, (R(Fe –O) ¼ 2.17 Å) the 2t2g spin splitting is 3.31 eV. Thus, near their
inter-nuclear distances, the spin –spin and the crystal-field splitting are similar for FeO
and FeS at their equilibrium distances. (Note. The SCF-Xa calculation substantially
overestimates the crystal-field (i.e., t2g – eg ) splittings in FeO102
6 and slightly overestimates
this splitting in FeS.)
Under pressure, the spin splitting first increases slightly, then levels off and begins
to decrease slightly. The 2t2g " !3eg " splitting increases at slightly less than the R25 rate.
The predicted (Tossell, 1976) Fe– O distance at which the internal energies of the high-
spin and low-spin forms will be equal is about 1.91 Å (which is about 0.21 Å less than the
Fe –O equilibrium distance). The MO studies by Tossell (1977) suggest spin pairing in
FeS at about R ¼ 2:28 Å (which is 0.21 Å less than the average experimental Fe– S
distance in troilite). The Fe– S distance at which the internal energy of the singlet state
drops below that of the quintet state is predicted to be 2.28 Å.

HS ! LS transition: XES studies. Rueff et al. (1999) have reported the pressure-induced
HS ! LS transition in FeS using new HP synchrotron X-ray emission spectroscopy
Iron and Siderophile Elements: The Earth’s Core 1027

techniques. The transition is evidenced by the disappearance of the low-energy satellite in

the Fe Kb emission spectrum of FeS.
Moreover, the phase transition is reversible and closely related to the structural
phase transition from MnP-like phase to monoclinic phase.
The spin state of Fe in FeS is monitored by high-resolution measurement of the
Fe Kb emission line. The Fe in HS state is characterized in the emission spectrum by a
main peak with energy of 7,058 eV and a satellite peak about 12 eV lower in energy
(Fig. 14.24). The transition at 6.5 GPa is also accompanied by the disappearance of the
magnetic moment-induced hyperfine splitting in the Mössbauer spectra. Under pressure
beyond 6.9 GPa, the satellite line vanishes, indicating a collapse of the Fe 3d moment to a
low-spin state. Under ambient condition, Fe in FeS is considered to be divalent with HS
3
electron configuration ðt2g " e2g " t2g
1
Þ (Koyabashi et al., 1997). Mössbauer spectra of FeS
indicate divalent Fe and the isomer shift of þ 0.76 mm/s is consistent with Fe2þ in HS
configuration. This is also confirmed by the large effective moment (estimated to be
5:521mB by magnetic susceptibility measurements (Horwood et al., 1976). In the LS state
3 3
ðt2g " t2g #Þ; the two final states of the Kb emission line, 3p "# 3d; are degenerate because of
the absence of a 3d magnetic moment and thus the XES spectrum reduces to a single peak
(Rueff et al., 1999). This is consistent with the disappearance of the hyperfine field above
6.5 GPa (cf. Kobayashi et al., 1997). The XES spectra of HS FeS (room P) and low-spin
(under P ¼ 11:5 GPa) states, along with room P reference spectra of two iron compounds
with Fe2þ, e.g., FeO(HS) and FeS2(LS), have been shown by Rueff and Kao (1999).
The spin pairing on the Fe 3d shell may result in diminution of the Fe radius and
drives the transition. Metallization is a competing process which can cause pressure-
induced reduction of satellite amplitude. The magnetic moments of HS FeS and metallic

Figure 14.24. XES spectra of FeS in the high-spin (room pressure) and low-spin (11.5 GPa) states, along with
room pressure reference spectra of two iron compounds with þ2 oxidation states for iron, namely, FeO (HS) and
FeS2 (LS). The difference between the low-spin spectra of FeS and FeS2 is within the measured error
(Rueff et al., 1999).
1028 Chapter 14

FeS are 5.5 and 2.2mB, respectively. However, the sequence of transitions (e.g., HS – LS
metallization) and Mott transition in such materials are of current theoretical interest
(Cohen et al., 1997). Pressure-induced measurements of the anion band-width is
important for determining the origin of metal – insulator transitions in such systems (Chen
et al., 1993).

14.9.2. FeS: Martian CMB and core

On the basis of cosmochemical constraints and the thermal history of the planet, the
iron-dominant Martian core may contain as much as 34% sulphur by weight (Longhi et al.,
1992). This corresponds to stoichiometric FeS. The depth of the CMB was assumed to lie
between 1,370- and 1,990-km depths (e.g., Ohtani and Kamaya, 1992) on the basis of the
density data of FeS III phase (Mao et al., 1981). But the study on FeS polymorphs at high
pressure and high temperatures by Fei et al. (1995) showed that the stable polymorph at
Martian core pressures (, 25 GPa) and temperatures (.1,600 K) is a hexagonal NiAs
superstructure (FeS IV).
For a model mantle composition of Mars (Morgan and Anders, 1979) with
13.9 wt% sulphur in the core, the Martian CMB is located at a depth of 2,000 km
(, 24 GPa), which is deeper than the silicate perovskite stability field. Mars may have only
a thin layer of lower mantle (, 200 km) but the density discontinuities are similar to those
of the Earth (Fei et al., 1995).

14.9.3. Fe –FeS system: eutectic points

Melting relations in the Fe – FeS system have been used to constrain the
temperatures of the interior of planets in many planetary models (Boehler, 1992). The
finding of the high-pressure – high-temperature transformation of FeS III to hexagonal FeS
IV now helps our better understanding of the melting relations in Fe – FeS and in FeS
systems and the solution behaviour in systems such as Fe – S– O under high P – T: Changes
in properties of FeS under pressure will contribute to the melting behaviour in binary (e.g.,
Fe –S) or ternary (e.g., Fe– S –O) systems. These changes will indeed significantly
influence the process of incorporation of lighter elements (e.g., S, O, H, etc.) into the iron-
dominant cores of terrestrial planets (Fei and Mao, 1994).
Fei et al. (1997) reported a Fe– S compound formed at P . 14 GPa, which changes
the melting relations in the Fe – FeS system. Fe and FeS form a binary system at ambient P
with a eutectic melting point at 9888C and 31% S (Hansen and Anderko, 1958). With
increasing P, the eutectic composition becomes more Fe-rich, whereas eutectic T remains
nearly constant to at least 6 GPa. At P . 6 GPa, Usselman (1978) reported that the
eutectic T rose with a slope , 34 K/GPa to at least 10 GPa. Pressure, however, showed
little effect on the eutectic composition between 7 and 14 GPa.

14.9.3.1. Fe3S2, Fe3S, Fe2S

The melting relations and sub-solidus mineralogy in the system Fe – FeS at high
pressure are fundamentally important for understanding the thermal and physical state of
Iron and Siderophile Elements: The Earth’s Core 1029

S-bearing iron core. Fe –FeS form a binary eutectic system up to 14 GPa, beyond which an
intermediate sulphide compound Fe3S2 forms until 18 GPa. Fe3 S2 melts incongruently
(Fei et al., 1997). At pressures greater than 21 GPa (T ¼ 950– 1,4008C), two more
compounds, Fe2S and Fe3S, form (Fei et al., 2000).
Powder XRD data revealed that Fe3S has a tetragonal cell, isostructural with Fe3P (I
4). The cell dimensions are:

a ¼ 9:144ð2Þ A

c ¼ 4:509ð2Þ A
and the density of this iron-rich sulphide is 7.0339 cm23. The RT bulk moduli is:
K0 ¼ 150 ^ 2; when K 00 ¼ 4:0
In many Fe –Ni meteorites, FeS and [(Fe, Ni)3P] phases have been found. Because FeS,
Fe3P and Ni3P are isostructural, P may be easily incorporated into Fe3S at high pressures.
This is how P gets incorporated into the core. Solid solutions in the Fe3S – Fe3P – Ni3P
system may help unravel the complex nature of the core chemistry.

14.9.4. FeS2

14.9.4.1. FeS2, pyrite

Pyrite (FeS2, S.G., Th6 ðPa3Þ) static compressed up to 40 GPa shows no phase
transition (e.g., Drickamer et al., 1986) while shock compression up to 320 GPa showed no
phase transition (Ahrens and Jeanloz, 1987). However, subsequent measurements and
calculations by Merkel et al. (2002) showed the pressure derivatives of the elastic moduli:
dC11 =dP ¼ 5:76ð^0:15Þ
dC12 =dP ¼ 1:41ð^0:11Þ
dC44 =dP ¼ 1:92ð^0:06Þ
were obtained from diffraction data, assuming the previously reported zero-pressure
ultrasonic data:
C11 ¼ 382 GPa
C12 ¼ 31 GPa
C44 ¼ 109 GPa
The third-order Birch– Murnaghan EOS parameters for pyrite are found (Markel et al., 2002):
K0T ¼ 133:5ð^5:2Þ GPa
K 00T ¼ 5:73ð^0:58Þ
The elastic moduli of pyrite up to 50 GPa seem to change linearly with pressure. For studies
on elastic moduli and strength of materials of cubic symmetry under very high pressure,
1030 Chapter 14

the radial diffraction method (Singh et al., 1998a,b) seems to offer better accuracy than other
methods.
In FeS2, the effect of a Fe –S distance reduction by about 0.1 Å upon the 2t2g orbital
energy is to raise it by about 0.3 eV. A reduction in R (S– S) by 0.1 Å results in SCF-Xa
(for S12
2 ) for destabilization of 2su (with respect to 1pg ). Therefore, the change in the
relative energy will depend mainly on the relative compressibilities of the S –S with
respect to the Fe –S bonds (Tossell, 1977). In FeS2, the force constant for S –S bond
compression in S12 2 and S222 has been estimated as 3.3 mdyn/Å, which is near to
3.68 mdyn/Å observed in diatomic FeS.
The X-ray photo-electron (XPS) and UV photo-electron spectra (UPS) of FeS2
show an intense narrow peak with a binding energy of about 1 eV and a broad two-peak
feature extending in about 3 –8 eV. The narrow lowest binding-energy peak should be
assigned to the 2t2g crystal-field orbital, which contains six electrons in the singlet ground
state of FeS2. The broad features from 3 to , 8 eV arise from orbitals of predominantly S
3p character while the features at 14 and 17 eV correspond to S 3s-type orbitals.

14.9.4.2. Pyrrhotite: magnetic transition

Pyrrhotite is one of the very few minerals exhibiting ferrimagnetism or
ferromagnetism In natural pyrhotite, as many as four six-line fits in the MS spectra have
been reported (Vaughan and Ridout, 1970). At 1.6 GPa, pyrrhotite is reported to show a
single paramagnetic line (Vaughan and Tossell, 1973). In pyrrhotite (as in troilite),
magnetic transition occurs before any structural change.

14.10. Fe3S2

At P . 14 GPa, Fei et al (1997) synthsized a Fe– S compound, Fe3S2 (27:9^

0.3% S). The compound may be non-stoichiometric because the samples synthesized at
18 GPa show a lower sulphur content (27.1 ^ 0.2%). They synthesized Fe3S2 phase at
15 GPa and 8808C using stoichiometric composition with a slight excess of metallic Fe
as the starting material with only 7.0% S. In XRD, the line seen at 2.0277 Å (22.7058) is
due to the presence of a-Fe while the others are due to troilite in the hexagonal cell.
The formation of Fe3S2 changes the melting relations from a simple binary eutectic
system to an intermediate compound that melted incongruently. In a quenching experiment,
the observed relative amounts of FeS and Fe were evidence against the argument for a
breakdown of Fe3S2 to 2FeS þ Fe.
Because iron sulphide, such as FeS, typically undergoes a phase transition at high P
and T, it is presumed that the Fe3S2 phase is not quenchable. However, theoretical
calculations showed that it could form at higher P (Sherman, 1995). It has been possible to
identify the presence of Fe3S2 in meteorite bodies, whose pressure at the centre should be
greater than 14 GPa, and T is less than the eutectic temperature (, 9008C).
The melting relations in the Fe– FeS system at the core pressure of the Earth,
i.e., 135– 360 GPa, may be different from what was experimentally observed (viz. Fei et al.,
1997).
Iron and Siderophile Elements: The Earth’s Core 1031

14.11. Mn –S system

14.11.1. a-MnS

The compression data of a-MnS at room temperature obtained by experiments are

plotted in Fig. 14.25 (McCammon, 1991). The data were fit to a third-order Birch–
Murnaghan EOS. Taylor expansion of energy in terms of the Eulerian finite strain
(Birch, 1952, 1978) is:

P ¼ 3f ð1 þ 2f Þ5=2 K0 ð1 2 3f =2ð4 2 K 00 Þ

where the Eulerian finite strain is given by f ¼ 1=2ððv=v0 Þ2=3 2 1Þ:

A least square fit of the MnS data yields the values K0 ¼ 72 ^ 2 GPa and
K 00 ¼ 4:3 ^ 1:3 (McCammon, 1991).

14.11.2. MnS2

Substitutional iron occurs in MnS2, which is isomorphous with FeS2. The lattice
parameter of MnS2 is 6.10 Å while that of FeS2 (pyrite) is 5.504 Å. It is not surprising,
therefore, that, although Fe(II)S is low spin at all pressures, Fe(II) as a dilute substitutional
impurity in MnS2 is in high-spin state at atmospheric pressure.
Nevertheless, it is reasonable to conceive of iron atoms in MnS2 to experience
a large negative pressure (expanded lattice) compared with its situation in FeS2.

Figure 14.25. Compression data of a-MnS. The data were fit to a third order Birch-Murnaghan equation of state
with parameter K0 ¼ 72ð2Þ GPa and K 00 ¼ 4:3ð13Þ (McCammon, 1991, q 1991 Springer-Verlag).
1032 Chapter 14

Figure 14.26. Conversion to low spin versus pressure MnS2(57Fe) (Source: Drickamer and Frank, 1973).

The majority of pressure measurements have involved 2% 57Fe in MnS2 (Bargeron et al.,
Inorg. Chem. 10, 1,338, 1971). The IS (0.84 mm/s relative to bcc iron) and QS (1.50 mm/s)
differ considerably from the typical high-spin Fe(II) values observed in ionic compounds
but these compare closely with the values obtained for other high-spin ferrous sulphides.
There was usually a trace (, 10% or less) of material with IS , 0.35 mm/s and QS
, 0.60 mm/s, which compare closely with the values for FeS2 (IS ¼ 0.30 mm/s and
QS ¼ 0.60 mm/s) (see also McCammon et al., 1984).
No significant change was noted till 4.0 GPa. At high pressures, the relative
amounts of low-spin Fe(II) increased; at 6.5 GPa, the conversion was . 50%. At 13.8 GPa,
the conversion was complete. Figure 14.26 plots conversion vs. pressure. The process is
reversible but involves considerable hysteresis. This is not surprising since the bonding and
the volume associated with the two spin states are different. Similar transition was
observed for 57Fe as an impurity in MnSe2 and MnTe2. These are examples of electronic
transition from a paramagnetic to a diamagnetic state. The transition is brought about by
the increasing field of ligands with decreasing Fe(II) – ligand distance. This spin change is
of significance in mantle geochemistry because pressures in the mantle reach 150 GPa at
the core boundary. It appears that Fe2þ ions would be all at low-spin state before the core
of the Earth is reached.

14.11.3. (Fe,Mg)S and (Fe,Mn)S

Fe2þ ions in two different hosts, such as NaCl structure (B1) and NiAs (B8)
structures at high pressure, have been studied by XRD and the by Mössbauer spectroscopy
of a quenched sample at 1 atm.
The spectra of (Fe,Mg)S – (Fe,Mn)S (B1 structure) solid solutions show quadrupole
doublets at RT. At 4.2 K, the spectra split into eight lines. The line-widths of the magnetic
Iron and Siderophile Elements: The Earth’s Core 1033

spectra are broad, consistent with a multi-axial spin arrangement but Mg-rich (Fe,Mg)S
(B1) shows only quadrupole doublets.
FeS (B1) has a molar volume 7.2% larger than the B8 phase at RT; B1 phase is more
ionic than B8 (McCammon et al., 1984). In the case of FeS, there is a 1.8% decrease of
kFe– Sl bond length in the transition from FeS (B1) to FeS (B8), which accounts for the
larger DV. A high-spin ! low-spin transition of Fe2þ causes a substantial reduction of
inter-atomic distance; this might accompany a B1 ! B8 transition. (Note. Most first-row
transition-metal oxides and sulphides that crystallize in the B1 structure are Mott insulators
(see Wilson, 1972).)

14.12. Pressure behaviour of FeS vs. FeO

FeO bond is quite rigid, with the bond length decreasing by about 0.035 Å per
0.1 GPa applied pressure. The sulphide force constant is equal to about two-thirds of the
oxide force constant. This difference in behaviour of FeO and FeS must be almost entirely
a result of differences in the pressure required to reduce the Fe –ligand (L) distance by
about 0.20 Å. Both the crystal-field and bonding orbital energies are consistently smaller in
FeS102
6 than FeO102
6 .
A possible difference in magnitude of the change in molar volume DVm between the
high-spin and low-spin state of FeO and FeS may explain the difference in their behaviour
under pressure. The experimental difference in equilibrium distances at atmospheric
pressure is of the order of 0.2 Å. Thus, the PDVm term in the free energy will be more
negative for sulphide than for oxide at any given pressure, favouring the conversion of the
sulphide to the low-spin state. The change in equilibrium distance calculated for the
quintet ! singlet transition in FeF326 is much smaller, , 0.03 Å.

14.13. Fe(Ni) – Cu – S compounds

Sulphides of Fe(Ni) – Cu –S solid solution are common accessory phases in upper-

mantle peridotites from both xenolithic and massif occurrences. From upper-mantle
peridotites, a big range of sulphide compositions have been reported, namely pyrite,
pyrrhotite, pentlandite, chalcopyrite, troilite, millerite and cubanite (Fig. 14.27)
(McDonough and Rudnick, 1998). These phases, in general, reflect exsolution from
primary mono-sulphide solid solution (MSS). Low-temperature alteration of these may
produce additional phases such as millerite, mackinawite, etc.

14.13.1. Nickel sulphides

Ni3S2, heazelwoodite, structure consists of a network of Ni3 triangles inter-

connected by Ni– Ni bonds. S atoms are in hexagonal layers, stacked one above the other in
the direction of the unit cell. The low-pressure Ni3S2 I form is rhombohadral in structure,
which under high pressure transforms to Ni3S2 III forms with Ni coordinated by five S
1034 Chapter 14

Figure 14.27. The Fe–Ni– S ternary diagram for sulfides from spinel peridotites (after Lorand, 1989). Squares
and circles represent measured composition which formed by exsolution from monosulfide solid solution (MSS),
here shown as re-integrated compositions (stars) (Source: McDonough and Rudnick, 1998, q 1998 Mineralogical
Society of America).

atoms in a square pyramid. Pressure induces a charge disproportionation of the nickel

atoms:

3Nið4=3Þþ ¼ Ni2þ þ 2Niþ

which arises from the reorganization of the sulphide array (Prewitt et al., 2002).

14.13.2. Cubanite, CuFe2S3

Cubanite (CuFe2S3) is a mixed valent compound with monovalent Cu ions and

Fe2þ –Fe3þ ion pairs Under pressure above 3.3 GPa, cubanite is reported to transform from
an orthorhombic structure to a derivative of NiAs structure, with a volume decrease of 29%
by the change of coordination of Fe from 4- to 6-fold.
High-pressure metallization and the electromagnetic behaviour of a natural sample
of hexagonal cubanite (CuFe2S3) have been investigated by Rozenberg et al. (1997). The
results of energy-dispersive XRD for the transformation of cubanite from orthorhombic to
hexagonal form are presented in Fig. 14.28.
An insulator – metal transition occurs in the range 3.4 – 5.8 GPa, coinciding with a
structural transition from orthorhombic to hexagonal (NiAs, B8) symmetry. This
transformation also corresponds to a magnetic transition from a magnetically ordered
phase at low P to a non-magnetic or para-magnetic phase at high P. The data may be
attributed to a reduction in the Fe– S –Fe super-exchange angle formed by edge-sharing
octahedra in the NiAs phases. On decompression, the non-magnetic or para-magnetic
phase is recovered, indicating a substantial hysteresis associated with the transition.
The pressure evolution of both the 57Fe Mössbauer hyperfine interaction parameters
and resistance behaviour is consistent with the transition from mixed-valence character in
Iron and Siderophile Elements: The Earth’s Core 1035

Figure 14.28. Energy-dispersive X-ray diffraction spectrum for (a) orthorhombic and (b) hexagonal cubanite. The
arrows indicate peaks from gasket diffraction (Rozenberg et al., 1997, q 1997 Springer-Verlag).

the low-P orthorhombic structure to extended electron delocalization in the hexagonal

phase at high pressure. MeCammon (1995) studied the effect of pressure on the isomer
shift (Dd) of orthorhombic cubanite relative to the zero-pressure value (Fig. 14.29).

14.14. Phosphates

14.14.1. Berlinite, AlPO4: memory glass (?)

The crystallographic structural characteristics of berlinite have already been

discussed in Section 3.5.2 (see Fig. 3.13). Berlinite (AlPO4) exists in the a-quartz
structure. Since 1990, it has been postulated to have the property of a “memory glass”, such
that its amorphous phase under pressure reverts to crystallinity with the earlier
orientations. In addition, Brillouin scattering results suggest that the high-pressure
amorphous phase (”p-glass”; Sharma and Sikka, 1996) is anisotropic (Polian et al., 1993).
For further discussion see Section 3.5.2.
Structurally, as in quartz, the oxygen atoms in a-AlPO4 also have a tendency to
approach the bcc lattice. With increasing pressure, the spiralling AlO4 and PO4 tetrahedra
distort and rotate into empty spaces in the crystal, resulting in higher Al –O coordination.
1036 Chapter 14

Figure 14.29. Effect of pressure on the isomer shift of orthorhombic cubanite relative to the zero-pressure value.
Observed ranges for ionic and covalent compounds from Drickamer et al. (1969) are included for comparison
(McCammon, 1995).

On releasing the pressure, the oxygens around Al atoms relax back to the PO4 tetrahedra,
which in turn displaces Al atoms to their initial positions.
Both Al and P continue to be four-coordinated to the neighboring oxygens up to a
pressure of , 30 GPa. Beyond 30 GPa, a discontinuous change in Al– O coordination is
noted (Tse and Klug, 1992). Around 30 GPa, Al –O coordination increases to , 5.2 and
beyond this pressure it gradually increases to 5.8 at 6.5 GPa (Garg and Sharma, 2000).
However, Tse and Klug (1992) found it only increased to 4.6 by 80 GPa.
The computed diffraction patterns from the equilibrated atomic positions in
berlinite have been indexed at various pressure (Garg and Sharma, 2000).
Around 15 GPa, the oxygen sublattice becomes disordered and the intensities of
the Bragg diffraction peaks are reduced. MD calculation showed that a-AlPO4 undergoes
a first-order phase transformation at , 30 GPa to a disordered structure. However, even
beyond 30 GPa, the calculated diffraction pattern continues to show sharp peaks.
Systematic reduction in intensity occurs at higher pressure but, even beyond 45 GPa, the

and (1014)
(1012)
peaks remain while others vanish. Thus, translational order may
persist well beyond the “pressure of amorphization”. The results of Garg and Sharma
(2000) at , 15 GPa also indicated that the structure is not amorphous but holds oxygen
sublattice disorder. Earlier Raman measurements by Gillet et al. (1995) and the XRD
results of Sun et al. (1994) also indicated this.
Calculations suggest this state continues to , 30 GPa but even some XRD peaks
continue to persist up to , 60 GPa. The calculated variation of V=V0 with pressure for
berlinite does not show any first-order phase transformation until 30 GPa. As a function
of volume, the calculated total energy per molecular unit of a berlinite is shown in
Fig. 14.30.
Iron and Siderophile Elements: The Earth’s Core 1037

Figure 14.30. Calculated total energy per molecular unit for a-berlinite as a function of volume. The energy is
referenced to the equilibrium energy. The volume is per molecular unit. The curve is a Murnaghan fit to the
calculated points (Source: Christie and Chelikowsky, 1998).

The compression of the c- and a-axis is found to be anisotropic up to , 12 GPa,

beyond which the c=a ratio reduces to form a plateau. This loss of anisotropy could be due
to the distortion of the tetrahedral network. An analysis of the atomic translational order
showed that the oxygen atoms are disordered, while there is no substantial disordering of
Al and P atoms. This disordering of oxygen atoms supports the suggestion of Gillet et al.
(1995) that, around 12 GPa, berlinite transforms to a disordered crystalline phase, although
the coordination of Al – O and P– O still remains four at this pressure range.
Simulations employing instantaneous compression suggest that, beyond 12 GPa,
the CmCm phase is more stable than the a-phase. A CmCm phase has been opined to be
stable at very high pressure because this phase is of lower energy and can be generated
through non-hydrostatic stress or shock loading. However, this phase transforms to a four-
coordinated disordered phase at ambient conditions and can only be stabilized on
compression beyond 20 GPa (Garg and Sharma, 2000).
Berlinite single crystals were shocked to peak pressures of 12 and 24 GPa by
Cordier et al. (1994) by a 6.5-m-long two-stage gun. In the sample shocked at 12 GPa, the
prominent shock-induced defects are dislocations and basal glide appears to be the only
glide system activated. In contrast, the sample shocked at 24 GPa exhibits no dislocations.
Conversion to partially amorphous phase occurs in the form of thin amorphous
lamellae parallel to the {101n} planes ðn ¼ 0; 2; 3; 4Þ: This microstructure is very similar
to the one observed in experimentally shocked quartz.

14.14.2. Farringtonite – Mg3 (PO4)2-II

Brunet (1995) reports the curve of the farringtonite to Mg3(PO4)2-II polymorphic

transformation to be tightly bracketed in an internally heated pressure vessel (IHPV) used
at Clermont-Ferrand, France. In the temperature range 565 –8258C and for pressures
1038 Chapter 14

between 6.1 and 8.3 kbars, the equilibrium curve has been described as:
PðbarsÞ ¼ 2772 þ 8:13 TðKÞ ^ 200 bars
The small value of the Clapeyron slope is comparable with that of the quartz –coesite
transformation and makes the farringtonite – Mg3(PO4)2-II transformation a good
candidate for the calibration of piston-cylinder assemblies at low P (, 1 GPa).
The farringtonite to Mg3(PO4)2-II transformation is unique for calibrating piston-
cylinder assemblies at low pressure (, 10 kbar). Furthermore, this curve fulfils the
requirements (e.g., low Clapeyron slope and fast kinetics of transformation) for an accurate
pressure calibration.
Finally, the IHPV data allowed Brunet and his co-workers to extract the thermo-
chemical properties such as DfH 0298 ¼ , 3,745.8 kJ/mol, and S 0298 ¼ 180.7 J/mol/K
for Mg3(PO4)2-II, the high-pressure form with an olivine-related structure (Brunet, 1995).

14.14.3. Apatite (Ca5(PO4)3, (F,Cl,OH)) – monazite topotaxy: REE

Under VHP conditions, apatites serve as the major sinks for REE and, during
decompression, monazite lamellae are likely to exsolve out from the apatite host Apatite
from apatite-bearing garnet peridotite, with high REE concentrations and obtained from
Dabie Shan (also Kokcherav) Massif, shows a topotaxial relationship with lower LREE
and higher MREE and HREE compared with those present in the host apatite.
For extremely REE-rich lithologies such as those of the Dabie Shan, discrete
monazite grains occur in the matrix in addition to the monazite lamellae in apatite.

14.14.4. Bearthite (Ca2Al(PO4)2OH)

This very high-pressure hydrous phosphate phase is reported from Dora Maira
Massif as a matrix phase as well as inclusions in both metapelites and pyrope – phengite –
quartzite (Chopin et al., 1993) and corundum and Mg-staurolite (XMg ¼ 0.85 –0.95) in
metapelitic rocks (Chopin et al., 1991).
 Section E

Transport Properties at
Deep Depths & Related Condensed
Matter Phenomena
This page is intentionally left blank
 1041

Chapter 15
Transport Properties in Deep Depths and Related Condensed-Matter
Phenomena

15.1. Transport properties under pressure

15.1.1. Introduction

Notwithstanding the unique development in high-pressure studies, probing the

electronic transport properties of materials at megabar pressures has been a major
challenge. It is difficult to run electric current through samples of a few micron
dimensions in the pressure cell and to measure it accurately. Moreover, the necessary
complementary magnetic measurements are equally difficult because of the small
sample sizes.
The measurement of transport properties such as electrical conductivity,
thermal conductivity and rheology poses technical difficulties under high-pressure
and high-temperature conditions. Techniques have been developed for studying the
magnetic variations with pressures, such as magnetic moment changes, paramagnetic –
diamagnetic transitions and superconductivity. The pressure and absolute temperature
dependence of transport properties are accurately obtained from the measurements of
Grüneisen parameter gTh, bulk modulus KT and thermal expansivity. The pressure
dependence of thermal conductivity in Earth models are presented in Fig. 15.1
(Hofmeister, 1999) (read the caption).
In insulators, the energy gap between the valence and conduction bands are large to
inhibit electron hopping and conduction. In semiconductors, the gap is so narrow that a
slight energy will boost the valence electrons into the mobile conduction band. Under
pressure, when the originally empty conduction band is forced to overlap the full valence
band, an insulator may become a conductor (see for example, Fig. 1.6). Similarly, when an
energy band is pushed above or below the Fermi energy of a metal, discontinuous changes
in the topology of the Fermi surface may occur.
It has long been known that small quantities of hydrogen (100-ppm level)
influence the electrical conductivity of minerals at low-pressure conditions (e.g., Griggs,
1967). For this reason, numerous dielectrical and rheological studies of H-bearing minerals
have been carried out (e.g., Karato, 1995; Sweeney, 1997; Chen et al., 1998). Such studies
revealed important relationships involving the defect concentration and the nucleation and
migration of dislocations. These control the transport properties, often induced.
1042 Chapter 15

Figure 15.1. Thermal conductivity versus pressure. Long dashed line, radiative contribution multiplied by 10 for a
dry solidus, whole mantle convection, and the presence of Fe2þ. All other lines are total thermal conductivity. The
short-dashed lines for the pyrolite model in the upper mantle (UM) and transition zone (TZ) were calculated for wet
solidi and decreasing water content with depth. The following upper-mantle and transition-zone curves were
calculated using a dry solidus in the LVZ: Light solid line pyrolite model. Dotted line, piclogite model. Heavy solid
line k for a chondritic mineralogy of for the upper mantle this is not shown for the upper mantle but lies between the
curves for piclogite and pyrolite. For the lower mantle only a chondritic composition (80 % Pv, 20 % Mw) was
considered. Heavy dot-dashed line lower mantle assuming whole mantle convection and a dry solidus. Solid line with
dots, lower mantle using the adiabat of which is probably the lowest temperature possible. Light dot-dashed line, lower
mantle allowing the maximum possible temperature and layered convection. The discontinuities in k and 410 km
and 670 km are attributable to the high conductivity of ringwoodite (from Hofmeister, 1999, q 1999 AAAS).

Transition-metal oxides exhibit a wide range of exotic structural, magnetic and

electronic behaviour, which cannot be explained by the usual one-electron band theory.
Generally, copper-oxide superconductors cover their peculiar transport properties to a
dynamic-phase segregation. Non-adiabatic polarons containing just five or six copper
centres in under-doped oxides condense into large “stripe” domains in the more heavily
doped, superconducting compositions. It is now known that such dynamic phase
segregation does, indeed, occur in real systems and this property is opening a new
frontier in solid-state science.

15.1.2. Electrical conductivity

A knowledge of the conductance of minerals is necessary to understand phenomena

such as the origin of the Earth’s dynamo. Electrical conductivity can be electronic or ionic
and the former can be intrinsic or extrinsic. By ionic conductivity, the electric current can
be carried by ions. This conductivity is remarkably seen at higher temperatures with the
Transport Properties in Deep Depths and Related Condensed-Matter Phenomena 1043

occurrence of pre-melting or sublattice melting. For the gap between valence and
conduction bands, the electronic conductivity can be thermally activated at finite
temperature. As the electrons move through this, the lattice experiences distortion,
designated as small polaron (e.g., Kittel, 1996). Electrical conductivity is sensitive to Fe
content, f O2, volatile content and grain boundary. Under pressure as the grain – grain
contact increases, the grain-boundary conduction decreases.
Development of electrical experiments in a DAC is also interesting for comparison
of static pressure data with data obtained in shock-wave experiments (Weir et al., 1996). In
spite of these objectives, electrical experiments in a DAC were limited to pressures
, 70 GPa and conductivity measurements seemed too difficult to perform at high pressures
because of the small sample size.
However, conductivity measurement techniques are well developed for large-
volume pressure cells and diamond cells (e.g., Kavner et al., 1995). For ultra-high
pressures, a new method, that uses a lithographic layout of electrical leads covered by
epitaxial CVD diamond overgrowth, provides “designer diamonds” for micron-size
conductivity measurements.
High-pressure metal –insulator transition also shows conductivity changes in the
optical range and thus graphite transforms to an insulator and becomes transparent
(e.g., Utsumi and Yagi, 1991).
Electrical measurements in a diamond-anvil cell are challenging. Few experiments
have been carried out and in those the pressure was limited to 1 Mbar. A knowledge of the
electrical properties of minerals at extreme conditions will serve as complementary to
DAC experiments and should resolve many of the controversies brought about by
conflicting data sets. Eremets and his co-workers studied the optical and electrical
properties of sulphur at pressures up to 1.5 Mbar, during which sulphur was metallized and
became superconductive. In their investigation to pressures up to 220 GPa bar, Eremets
et al. (1998) obtained metallization of CsI and found superconductivity in it. The high-
pressure experiments were performed in combination with millikelvin temperatures and
high magnetic fields. Oxygen was also seen to be transformed to superconductivity.
Studies on secular variations of the Earth’s magnetic field have indicated that there
is a steep increase in electrical conductivity in the 400 –1,000 km region, corresponding to
the transition zone phase change. Within the mantle, the lower mantle is estimated to have
electrical conductivities four to five orders of magnitude higher than those of the upper
mantle.
The electrical conductivity, s; related to charge transfer (CT) at a given pressure is
related to absolute temperature as:
X
s¼ soi expð2Ea =kTÞ; ð15-1Þ
i

where Ea is the energy of activation, k is the Boltzmann constant, and soi is the electrical
conductivity arising from charge-transfer conduction mechanism (e.g., O ! M CT and
Fe2þ ! Fe3þ IVCT).
The electrical conductivities of many silicate minerals are still unknown. In general,
silicates are insulators and their s (573 K) is , 1028 (V cm)21. The conduction
1044 Chapter 15

mechanism is due to hopping transfer Fe2þ ! Fe3þ and is intimately related to lattice
defects including vacancies. High conductivities have also been reported from Fe-rich
amphiboles with mixed valencies (Fe2þ and Fe3þ).
Electrical conductivities of minerals are seen to increase with iron content, oxygen
fugacity, temperature and also with pressure (e.g., Shankland and Peyronneau, 1993).
Pressure- and temperature-induced opacities and red-shifts of O ! M CT absorption edges
into the visible region spectra are observed for Fe-bearing olivines, silicate spinels,
periclase, magnesio-wüstite, perovskite, etc. This observation indicates that, in the lower
mantle, an extrinsic conduction mechanism of the type O ! Fe CT is important. The same
mechanism, which causes electronic transitions and inhibits radiative heat transfer within
the deep Earth (Shankaland et al., 1979), would enhance its electrical conductivity.
In perovskite structure lattice defects, permitting Fe3þ ions to exist, allow O ! Fe
and Fe2þ 3þ
A ! FeB CT to occur in the face-sharing polyhedra which enclose these two
valence states of iron. The opacity of the hydrous “phase D” ((Mg, Fe)SiH2O4) also
indicates that extensive electron delocalization exists in its structure. Trace amounts of
water present in polycrystalline olivine can account for large increases in its conductivity.

15.1.2.1. Electrical conductivity and activation energy

The relationship between electrical conductivity (s), temperature and activation
energy, Ea, may be expressed as (see equation (15-1)):
 
AðnÞ Ea
s¼ exp 2 ; ð15-2Þ
Tn kT
where Ea is the activation energy and AðnÞ is a constant depending on n. The values for n
are determined as (Appel, 1968):
n¼1 for adiabatic small polarons
n ¼ 3=2 for non-adiabatic small polarons
n¼2 for large polarons:
For adiabatic small polarons, the activation energy Ea is connected to the binding energy
Eb and the exchange integral J through the relation (Emin, 1973) as
Eb
Ea ¼ 2J ð15-3Þ
2
The activation energy, Ea, is related to pressure and volume as
Ea ¼ DE0 þ PDV; ð15-4Þ
where
Eb R
DE0 ¼ 2 J0 and DV ¼ 2h 0 J0 ; ð15-5Þ
2 3K
where DE0 is the activation energy at atmospheric pressure and DV is the activation
volume.
Transport Properties in Deep Depths and Related Condensed-Matter Phenomena 1045

The binding energy Eb may vary with pressure through the static and optical
dielectric constants. The dependence of these constants on pressure has been studied for
ionic crystals (e.g., Mayburg, 1950) at pressures varying between 0 and 0.8 GPa.

15.1.2.2. Techniques employed

For a study of the electrical conductivity of (Mg, Fe)O up to 30 GPa, a four-lead
technique at very high pressure was first used by Mao and Bell (1981).
Direct-resistance techniques for determining the pressure dependence of super-
conductivity have been developed at low temperature (Akahama et al., 1993). This
technique has been used to study the pressure dependence of superconductivity in high-
temperature superconductors (Gao et al., 1994). Techniques have also been developed for
studying the magnetic variations with pressures such as magnetic-moment changes,
paramagnetic – diamagnetic transitions and superconductivity.
For studies on changes in electrical conductivity under pressure, magnetic methods
have been used such as in superconductivity measurements. The use of inductive
techniques to detect superconductivity through the Meissner effect is the most reliable one.
These experiments employ superconducting flux transformers, multiple modulation
techniques and SQUIDS.
In an electrical impedance measuring device of Poe and Rubie (1998), an MgO
sleeve insulates the sample from a low-voltage graphite furnace capable of heating the
sample to 1,2008C. Interference from the AC furnace on the impedance measurements is
only observed near 50 Hz. The impedance can be measured over a frequency range of
1025 –107 Hz, making it possible to distinguish and quantify contributions from both grain
interior and grain boundary conduction mechanisms.

15.1.2.3. Conductivity of minerals

Parkhomento (1982) enlisted the electrical resistivity of minerals and rocks at high
temperatures and pressures. An enhanced concentration of Fe is linked with higher values
of s and apparently Fe is involved in charge transport processes.
Pressure-induced electrical conductivity has been seen to occur in olivine (forsterite
to fayalite) (Mao, 1993; Mashimo et al., 1980), spinel (g-Fe2SiO4) and various magnesio-
wüstite at 3 GPa. In the lower mantle, conductivity is seen to be six orders of magnitude
higher than forsteritic olivine in the upper mantle; this also explains the observed rapid
increases in conductivity in the transition zone. Under pressure, the activation energies of
olivine and magnesio-wüstite seem to decrease (Mashimo et al., 1980).
The conduction mechanism in the olivine Mg2SiO4 – Fe2SiO4 series has been
thoroughly investigated at temperatures around 573 K (e.g., Robert and Tyburczy, 1993;
Wanamaker and Duba, 1993). Close to the end-member Fe2SiO4 (fayalite), oxidizing
conditions cause the co-existence of Fe2þ and Fe3þ. The conduction mechanism is
apparently related to the electron hopping transfer Fe2þ ! Fe3þ, which is related to
the vacancies and other lattice defects causing as high a value for s (623 K) as
,1023 (V cm)21.
Electrical conductivity of an assemblage of (Mg, Fe) SiO3 –perovskite and phase D
(with 4 wt% H2O) shows an increase of about three orders of magnitude w.r.t. an
1046 Chapter 15

anhydrous assemblage at pressures of 48 – 57 GPa. Contamination by water greatly affects

the conductivity measurement (Li and Jeanloz, 1991).
In amphiboles containing mixed valences of Fe2þ, Fe3þ in the structure, high
conductivities are reported as shown below.

Amphibole Conductivity (s, V21 cm21) Author(s)

at 573 K
Riebeckite 5 £ 1025 Parkhomenko (1982)
Arfvedsonite 5 £ 1022 Schmidtbauer et al. (1996)
Ilvaite 2 £ 1023 Schmidtbauer and Amthauer (1998)

In ilvaite, there could be band- or a hopping-type charge transport.

The magnetic field on the Earth’s surface is controlled by the electrical conductivity
of the mantle phases. Therefore, a study of the electrical conductivity of silicate perovskite
at high P and T is of great interest.

Viscosity and conductivity. The rate of hydrogen diffusion to the extent of 100– 1,000
hydrogen ions per 106 Si atoms (6 – 60 ppm water) may elevate the electrical conductivity
of the mantle by one to three orders of magnitude (Karato, 1990). Such a H-induced
conductivity increase may explain the observed enhancement of conductivity near the top
of the asthenosphere (, 90– 150 km) without invoking partial melt at these depths (Karato,
1990). Again, to match magnetotelluric sounding results, the high conductivity pathway
may be ascribed to hydrogen diffusion (Farber et al., 1994).
The viscosity of olivine-bearing rock is more than two orders of magnitude less
than that of dry olivine (Hirth and Kohlstedt, 1996). A drastic lowering of viscosity of
silicate rock by the presence of water has important geodynamic consequences. A strong
partitioning of water into melts, migrating upwards, leaves behind a largely dehydrated
residual solid. The dehydration-induced strengthening may be the reason for the
narrowness of the upwelling zone beneath ridges. Dehydrated melt is sufficiently viscous
for keeping lateral corner-flow gradients capable of focusing melt into a narrow ridge axis
(Hirth and Kohlsted, 1996).
The drop in yield strength of hydrated olivine is almost a factor of two larger for
a-phase than the b- and g-phases (Chen et al., 1998). The manner, rather than the amount,
in which hydrogen is incorporated controls the degree of weakening of such phases.

Electrical conductivity of LM and D00 zone. The results of Xu et al. (1998) show that
electrical conductivity reaches a value of about 1 S m21 at the top of the lower mantle,
which is in agreement with geomagnetic determinations (Petersons and Constable, 1996).
They also found a conductivity jump of two orders of magnitude at the 410-km
discontinuity and a minor conductivity increase at 660-km depth, where ringwoodite
disproportionates to perovskite þ magnesiowüstite. The effect of pressure on conductivity
is seen to be weak and the activation volume should be small. The negative activation
volume of 0.26 cm3 mol21 for the perovskite – magnesiowüstite assemblage is consistent
with the hopping conduction mechanism (Poirier et al., 1996).
Transport Properties in Deep Depths and Related Condensed-Matter Phenomena 1047

The conductivity of (Mg, Fe)O is significantly higher than that of (Mg, Fe)SiO3
perovskite and the former very likely contributes most to the conductivity of the lower
mantle (Wood and Nell, 1991). Indeed, the electrical conductivity of the Mg end-member
of (Mg, Fe)SiO3 at high P and T (determined by using the laser-heated diamond-cell
technique; Li and Jeanloz, 1990) is seen to be several orders of magnitude lower than
geophysical observations for the lower mantle (Li and Jeanloz, 1987).
However, measurement of the electrical conductivity of Fe-bearing perovskite
becomes very pertinent for understanding the electromagnetic processes such as the
transmission of geomagnetic signals from the Earth’s interior (Burns, 1993). Analysis of
transient and secular variations of the geomagnetic field gives a value of 1 S m21 at a depth
of 1,000 km. An extrapolation of the results of Poirier and Peyronneau (1992) to lower-
mantle pressure also gives similar values. To explain the high conductivity values they
surmised electron hopping between Fe2þ and Fe3þ and reported the activation energies of
conduction to be in the range 0.21 –0.35 eV. The values decrease with increasing Fe
content. It may be noted here that activation energy amounting to 0.12 eV has been
reported for the electron exchange in Zn-bearing spinels (Lotgering and Van Diepen,
1977). Also notable is the similarity between the activation energies of conduction and
electron exchange. The oxidation state is of prime importance in controlling both
the conduction mechanism and the electrical conductivity. In perovskite structure, the
Fe2þ –Fe3þ pairs most likely form clusters within the three-dimensional polyhedral
framework. This co-operative electron exchange is likely to be higher in perovskite than
that in spinel or ilvaite.

15.2. Electron/hole transfer and magnetic behaviour

Investigations on the electronic and magnetic properties of strongly correlated 3d-

electron systems under high pressure have been found to be useful. Application of pressure
and the consequent enhancement of the transfer interaction (t) affects the electronic
properties, especially near the insulator – metal phase boundary (Kuster et al., 1989;
Jin et al., 1994).
Pressure increases (i) the itineracy of the carriers influencing the magnetic
behaviour, (ii) the Néel temperature, TN, in a band filling ðn ¼ 1 2 xÞ manner, suggesting a
crossover behaviour from localized to itinerant magnetism with increasing x, (iii) the Curie
temperature, Tc, which enlarges the ferromagnetic phase and (iv) the electron (or hole)
transfer interaction or the one-electron band-width. The important parameter of strongly
correlated 3d-electron systems is the nominal hole concentration x (or the degree of band
filling n ¼ 1 2 x).
The magnetic as well as the electronic properties are governed by the one-electron
band-width (W) of the eg band and the on-site exchange interaction (Hund coupling, J )
between the itinerant eg electron and the local t2g spin (Furukawa, 1994). Increase of W
(one-electron band-width) in ferromagnets (itinerant) generally suppresses Tc.
1048 Chapter 15

Pressure is shown to have substantial effects on the electrical conduction process

and on Tc, as seen in a number of RE-Mn perovskites, e.g., La12xCaxMnO3, discussed in
Section 15.4.1.

15.2.1. Polarons: small and large

A strong electron –phonon coupling may localize carriers with the formation of a
“self-trapped” state called a polaron. The polaron formation usually involves a single
electron coupled to a deformable medium. In “low-mobility semiconductors”, the
interaction of electrons with the lattice (with large effective mass) is strong and the band
formalism becomes inadequate. Only through thermal activation do the charges (electrons
or holes) move from one site to another.
When an extra charge sits on a cationic site (e.g., a M3þ ion in a M2þO oxide)
for a period greater than the typical lattice vibration period, the lattice deforms through
electrostatic interaction. The potential resulting from the lattice deformation acts to trap
the carrier. Thus, in ionic crystals, when a charge induces deformation of the
surrounding lattice, it gets self-trapped. If the interaction with the lattice is strong, this
“self-trapping” will form a “small polaron” and, if the interaction is weaker and extends
over a large distance, it will form a “large polaron”. Put another way, if the deformation
extends over distances smaller than or of the same order as the lattice parameter, the
polaron is a small polaron; if the deformation extends over larger distances, beyond a
lattice parameter, it is called a large polaron.
At low temperature, a small polaron moves in a band but, when the temperature
is higher than half the Debye temperature, the small polaron moves essentially by a
thermally activated diffusive process, i.e., by hopping. If the small polaron readily adjusts
to a modification of the ionic positions, the hopping is adiabatic. Otherwise, it is non-
adiabatic.
Hopping is controlled by the exchange integral, J, which corresponds to the
interaction energy between the starting and arrival sites in the lattice potential. The
exchange integral depends on the interaction potential between the charge carrier and
the lattice and on the spread of the wave functions at both sites. The exchange integral is
expected to be smaller than the binding energy (Goddat et al., 1999).
Polaron hopping is commonly referred to as charge transfer (CT). In magnetite,
Fe3O4, for instance, holes hop from Fe3þ ions to Fe2þ ions (e.g., Morris and Williams,
1997). The mobility is due to thermally activated jumps of the hole from Fe3þ to
Fe2þ ions.
The concentration of the charge carriers (electrons or holes) is related to the
concentration of the vacancies. The conductivity changes as diffusion varies with
temperature, following the Arrhenius law:

 
DE0 þ PDV
s ¼ s0 exp 2 ; ð15-6Þ
kT
Transport Properties in Deep Depths and Related Condensed-Matter Phenomena 1049

where s0 is a pre-exponential factor (depending on the oxygen fugacity), DE0 is the

activation energy at 1 atm (, n £ 1021 eV), DV is the activation volume, k is Boltzmann’s
constant and other symbols carry their usual meaning.
For adiabatic small polarons, the pre-exponential term (s0) is proportional to a 2
(Austin and Mott, 1969) where a is the lattice parameter, which varies with pressure as
 
2 P
a2 < a20 1 2 ; ð15-7Þ
3 K

where a0 is the lattice parameter at 1 atm and K is the bulk modulus.

In cases where the pressure dependence of the activation volume has been
determined, the activation-volume change, DV, is small and negative. For a small
activation energy and a small negative activation volume, small polaron hopping occurs
which results in conduction.
It has to be noted here that, where conductivity decreases with pressure
increase, the density of vacancies decreases and thereby the conduction by diffusion of
ions is eliminated. Conduction by diffusion of ions has larger activation energies.
However, conductivity also depends on oxygen fugacity and thermopower (e.g., Hirsch
et al., 1993).

15.2.2. Ferroelectricity: regimes and local well potential

Ferroelectric materials are characterized by a switchable macroscopic polarization.

Technologically, most important ferroelectrics are oxides with perovskite structure. The
balance between long-range Coulomb forces (which favour the ferroelectric state) and
short-range repulsions (which favour the non-polar cubic structure) are delicate enough to
respond to pressure. Ferroelectric perovskites provide well-known examples of displacive
transitions. Ferroelectric distortions involve small displacements of the cations relative to
the anions, leading to a net dipole moment per unit volume or polarization.
Ferroelectrics are ionic insulators with long-range forces. Real ferroelectrics have
long-range dipolar forces, not just local nearest-neighbour coupling. (Note: Short-range
forces favour the symmetric state.)
Ferroelectricity is a bulk property, arising out of intrinsic multiple well potentials
that underlie atomic motions. The zero potential surface has multiple minima. In
ferroelectrics, the atoms experience some well potentials around each, leading to two
regimes with intermediates in between. These two regimes are determined as stated below:
(i) Order –disorder regime — when each atom sees an average multiple well potential;
and
(ii) Displacive regime — when multiple wells disappear and each atom sees only one well
in the field of other atoms.
Much above the ferroelectric transition temperature, most perovskite ferroelectrics
seem to occur in the displacive regime. But the transition itself occurs in the order –
disorder regime, as evidenced from infrared studies (Fontana et al., 1984).
1050 Chapter 15

In a model of this behaviour, the total energy can be represented as

X X
E¼ ðAx2i þ Bx2j Þ þ C xi xj ðxi ¼ instantaneous position of each ionÞ:
i ij

The first sum is equivalent to the local potential in the field of other ions while the
second, summing over the neighbouring cells, gives the coupling effect that leads to a
phase transition. Commonly, in the order – disorder regime, phase transition A is p 0 and
the local potential is a double-well potential; in the displacive regime, A is q 0 and the
local potential has a single well.
So far, no appropriate microscopic coefficients for any real ferroelectric have been
ascertained. Even for the frozen phonon, total energy calculations give only the total
energy as a function of displacements and do not allow an easy separation into “local” and
“coupling” parts.

15.3. Insulator to superconductivity

Pressure would enhance high-temperature superconductivity. Superconductivity

may arise from electron pairing induced by coupling with lattice vibrations. Because
pressure generally enhances the frequencies of lattice vibrations, pressure studies helped
revealing the HT superconductivity (e.g., in cuprates Y2BaCu3O7 with Tc ¼ 90 K).
Pressure can enhance superconductivity mainly by an increase in average phonon
frequencies with applied pressure. Very high critical temperatures have been measured in
Hg-cuprates, with the highest Tc materials having a value of 135 K. The Tc increases to
164 K at 30 GPa (the highest Tc on record; Gao et al., 1994).
The high-Tc materials have structures based on perovskites, which also manifest
properties such as ferroelectricity and fast-ion transport. These interesting physical
properties bear important implications for the physical properties of the deep Earth.
The “Wunderkinds” of the material world — high T superconductors — can be
formed from layered ceramic insulators spiked with small amounts of other elements.
Post-modern investigations (Science, 282, p. 2067, 1998) show that these doped
insulators bear electrical signatures reminiscent of a metal. In these, the electronic states
are organized in bands with the valence band containing lower energy electrons with
restricted movement and the conduction band containing higher energy mobile electrons.
These two bands overlap in metals, wherein the valence-band electrons hop easily into
the conduction bands and they can whiz around and conduct electricity through
the material.
d-Wave symmetry: In a study of a flat crystal of Ca2CuO2Cl2 using an angle-
resolved photoemission spectroscopy, the surface was bombarded with X-rays at precisely
controlled energies and the emitted electrons are measured for their energies and directions
of travel. The energies of the electrons vary depending on the direction they are travelling.
Superconducting electrons travelling in pairs can move only within the planes of
copper and oxygen atoms and only along the two axes of the crystal (not along the 458
diagonals). This gives the wave function a cloverleaf pattern, known as d-wave symmetry.
Transport Properties in Deep Depths and Related Condensed-Matter Phenomena 1051

Shen et al. (1998) found this as “a new kind of insulator that has a sign it could be metal”.
In this, perhaps the electrons are pairing up briefly and then separating again.

15.3.1. Superconductivity and magnetism: “co-habitation”

In the superconducting state, the pairs of spins are aligned anti-parallel but a large
enough magnetic field can align the spins and thus destroy the superconducting state.
When a magnetic field is applied to a common superconductor (e.g., type II), individual
field lines run through the sample and around each line there is a small region where
superconductivity vanishes. In such an ambient condition, normal metallic regions co-exist
with superconducting regions. Under certain conditions, magnetic moments can possibly
co-exist with superconductivity; such a co-existence might be possible in hybrid
compound contributing conduction electrons and the other magnetic moments.
This spatial dichotomy between the atoms contributing magnetism and the atoms
contributing superconducting electrons is analogous to Na and Cl existing in distinct but
periodic regions of salt crystals.
Copper atoms play such a dual role in high-Tc cuprates (Ramirez, 1999). In
lanthanum cuprates, an insulator, the copper – copper interaction leads to anti-
ferromagnetism. When this is doped with 2% electron vacancies (holes) per Cu atom,
e.g., by substituting divalent Sr for trivalent lanthanum or by introducing oxygen into
interstitial sites of the crystal, anti-ferromagnetism is destroyed (Kastner et al., 1998). For a
higher concentration of holes, superconductivity occurs. Thus, the same copper atoms are
responsible for both superconductivity and magnetism. Neutron-scattering experiments
alone (probing both the energy and wavelength dependence of magnetic excitations) have
revealed that anti-ferromagnetism actually co-exists with superconductivity and both the
phenomena originate from the same atom (e.g., copper).
However, an incommensurate magnetic behaviour is known to be betrayed by
doped cuprates (Dai et al., 1998), resulting from the frustration of the magnetic moments in
trying to achieve a low-energy configuration within the constraints imposed by the crystal
lattice. The Tc value increases with increasing incommensurability in the layered cuprates,
which manifests competition between magnetism and superconductivity.
Like the cuprates, heavy-fermion systems exhibit delicate interplay of magnetism
and superconductivity (although with Tc , 1 K). Heavy-fermion phenomena are
associated with the same atomic species such as uranium in uranium intermetallics
(Aeppli et al., 1989).
Conventional wisdom suggests that superconductivity and ferromagnetism are
incompatible properties. Shimizu et al. (2001) showed the possible presence of
superconductivity in iron at high pressures but at low temperatures. Superconductivity
may occur in conventional non-magnetic metals when they are cooled to low enough
temperatures for their conduction electrons to pair up and flow without resistance.
However, tiny magnetic impurities can destroy superconductivity in conventional
superconductors by breaking up the electron pairs.
In addition to showing zero electrical resistance, superconductors expel magnetic
fields from their interior — the so-called Meissner effect (see Section 15.3.2).
1052 Chapter 15

This dramatic effect allows a permanent magnet to be levitated above a superconductor.

Contrary to the conventional wisdom, superconductivity and ferromagnetism can mix
when the magnetization is small (Pfleiderer et al., 2001).
Ferromagnetic order in iron is destroyed at pressures above 10 GPa and below
2 K. Shimizu et al. (2001) even observed the Meissner effect confirming that iron reaches a
true superconducting state at high pressures. Some, however, opine that the hcp phase of
iron is weakly anti-ferromagnetic with no net magnetization.
The origin of electron pairing in all new superconductors needs to be resolved. In
conventional superconductors of elemental metals, elastic vibrations (phonons) in the
crystal lattice mediate the pairing between electrons.
In the Earth, the liquid outer core generates geomagnetic field but the solid inner
core can damp out the fast fluctuations and stabilize the geodynamo (as happens in an
electromagnet), inhibiting field reversals (,105-yr scale). Since the solid inner core is too
hot to be superconducting, the dynamo can be stabilized by large electrical conductivity
and by strong paramagnetic fluctuations (Gilder and Glen, 1998). If the magnetic
fluctuations (either anti-ferromagnetic or paramagnetic) are large, the superconductivity
noted by Shimizu et al. (2001) might be explained like the unconventional superconductors
such as heavy-fermion compounds and the copper-oxide high-temperature superconduc-
tors. Such an exotic mechanism requiring strong magnetic fluctuations may go well with
the model of a strongly paramagnetic core. Probing these two types of mechanism for
superconductivity — conventional phonon-mediated and exotic magnetically mediated —
would provide better clues for the Earth’s geomagnetism.

15.3.1.1. “Magnetic glue” and failed spin: ghost magnetism

In conventional superconductivity, electrons form superconducting Cooper pairs
through an attraction mediated by lattice vibrations.
Some heavy-fermion superconductors, which are metallic materials, have
conduction electrons (or, collectively, quasi-particles) acquiring very large effective
masses (n100 £ mass of an electron). The large masses arise from the interaction of the
conduction electrons with localized (generally f-shell) magnetic moments in the materials.
The interaction may favour anti-parallel alignment of the local and conduction-electron
spins. The conduction electrons may screen the local magnetic moment and acquire mass
from the local moments they screen.
Magnetic interactions tend to break anti-parallel Cooper spin pairs. With the first
discovery of superconductivity in such a material as CeCu2Si2, the question arose if anti-
parallel spin pairing can actually be mediated by a magnetic interaction. Mathur et al.
(1998) have inferred that a magnetic pairing mechanism operates in such materials. In the
experiments of Mathur et al. (1998), a magnetically ordered state of cerium 4f-local
moment competes with a heavy-fermion state that has all these moments compensated by
conduction electrons. Pressure increases the conduction-electron — 4f-moment coupling
in cerium materials, favouring the heavy-fermion state. However, in between low-
temperature magnetic order and heavy-fermion behaviour, superconductivity emerges (see
Fisk and Pines, 1998). Mathur et al. argue that superconductivity arises here from
magnetically mediated pairing, which they call “magnetic glue”.
Transport Properties in Deep Depths and Related Condensed-Matter Phenomena 1053

One can also think of magnetism and superconductivity as different manifestation

of the same basic interactions. Spin-density waves, generated by moving charge carriers,
bind these particles into pairs. The quasi-particles participate in a nascent spin-density
wave (the “ghost” of magnetism). The signature of magnetic interaction between quasi-
particles is evidenced from an electrical resistivity relationship with temperature. Similar
interactions have been postulated in high-temperature cuprate superconductors. These
systems show the presence of “failed” spin or charge-density waves, either of which might
bring about superconductivity. Thus, a superconducting phase may reflect the ghost of
magnetism. This new physical phenomenon, whose offspring are magnetic and
superconducting states, seen for example in cuprates, has also been theoretically
approached (e.g., Zhang, 1997).

15.3.2. Novel physics: paramagnetic Meissner effect

Some novel physics has emerged from the studies under pressure. The most
common examples of the outcome of novel physics are the discoveries of superfluidity and
superconductivity.
Physical properties of fundamental and technological interest, such as supercon-
ductivity, can also be favourably tuned or enhanced by pressure. For example, the
mercury-based oxides HgBa2Can21CunO2nþ2þd, when n ¼ 1; 2, 3, show superconduc-
tivity on a set of temperatures (Tc) of 94, 128 and 135 K, respectively. Their Tcs were found
to increase with pressure (Chu et al., 1993). When studied in a diamond cell with four-lead
electrical conductance technique (maximum pressure ¼ 45 GPa), record high Tcs of 164,
154 and 118 K were reached.
A superconductor placed in a magnetic field and cooled down through the transition
temperature expels magnetic flux. This phenomenon, known as Meissner effect, is the most
essential property of superconductors (Gelm et al., 1998). Strangely, a well-known high-
temperature superconductor, Bi2Sr2CaCu2O8 in highly disordered granular form, is seen to
show a paramagnetic response to a small external field (Braunisch et al., 1992). This
unusual behaviour, known as the paramagnetic Meissner effect (PME) or the Wohlleben
effect, arises from an oscillating function of the magnetic field (due to flux quantization)
and is related to the surface superconductivity. The PME is a property of these
superconductors in granular form with grains of “mesoscopic” size. However, some
superconducting samples may attract magnetic fields manifesting the paramagnetic
Meissner effect. Gelm et al. (1998) concluded that PME is only related to the surface
superconductivity.

15.3.2.1. Mesoscopic magnetism: frustration and superconducting loop

When a metal becomes superconducting, the magnetic properties change
remarkably as the magnetic fields are expelled from the material’s interior. The magnetic
field is diminished in a superconductor, which behaves as diamagnetic (Tinkham, 1996).
However, in semiconductors of type II, magnetic fields may penetrate in vortices of
circulating supercurrent (the resistance-free current carried by paired electrons).
1054 Chapter 15

An IBM research group prepared miniature loops (, 50 mm across) of various high-

temperature superconductors to frustrate the phase of the superconducting state. A
spontaneous current flows, generating a magnetic flux. Loops were designed to frustrate
the phase of the quantum-mechanical wave function that describes the superconducting
electron pairs. This only works in superconductors with an unconventional electron-
pairing symmetry (the so-called d-wave pairing in contrast to the s-wave pairing of
standard superconductors such as Nb, Pb and Al).
Kirtley et al. (1998) found that the frustration led to spontaneous supercurrents
running without dissipation around the loop. The magnetic moment of such a
superconducting loop should indeed act like a paramagnet as it aligns parallel to the
external field in a similar manner to a magnetic needle. This observed frustration
phenomenon is crucial in proving the unconventional symmetry of the superconducting
state in high-temperature superconductors (Gelm et al., 1998).

Spin fluctuations. Fluctuations in spins associated with the incomplete outer electronic
shells (such as 3d9 in copper ions) give rise to magnetic excitations of the material and
might mediate the electron pairing that leads to superconductivity.
Neutron-scattering data reveal clear differences between the spin fluctuations for
the following two major types of high Tc materials in which the building blocks are CuO2
layers and bilayers, respectively:
(i) La22xSrxCuO4 (e.g., Mason et al., 1992)
(ii) YBa2Cu3O72x (e.g., Tranquada et al., 1992).
In their studies, Mook et al. (1998) used two-dimensional neutron scattering. One of
these reveals that the low-frequency magnetic excitations of YBa2Cu3O6.6 are virtually
identical to those of similarly doped La22xSrxCuO4. The magnetic fluctuations at low
frequencies are beginning to achieve the same universality across widely different
materials with different values of Tc, which characterizes many of their macroscopic
properties. As for the bulk characteristics, the controlling factor seems only to be the hole
density (Mook et al., 1998).

15.3.3. Double exchange in magnetic transition

Strong correlations between the electrical resistivity and ferromagnetism in

these materials were qualitatively explained by Zener (1951) using a double-exchange
(DE) model. Because of strong Coulomb repulsion, no d-orbital can be occupied by more
than one electron and, because of a large Hund’s rule coupling, all electron spins on a given
Mn are ferromagnetically coupled. It thus emerges that three of the (4 2 x) d electrons fill
up the t2g levels to form an electrically inert core spin, Sc, of magnitude 3/2. The remaining
(1 2 x) electron goes into a linear combination of eg orbitals and may move through the
crystal, subject to the constraint that, when it is on site I, its spin must be parallel to Sic.
This means that the amplitude for a carrier to hop from site i to site j is modulated by
the overlap between its spin parallel to Sic and having its spin parallel to S jc.
So ferromagnetic order maximizes the hopping while anti-ferromagnetic order minimizes
it. This DE connects the magnetic correlations and transport (Kubo and Ohata, 1972).
Transport Properties in Deep Depths and Related Condensed-Matter Phenomena 1055

Resistivity changes by DE on passing through the paramagnetic to ferromagnetic

transition. In the high-temperature phase, the spins are disordered and the electrons are
scattered while, in the low-temperature phase, the spins are ordered and the electrons are
not scattered.
Manganese oxides (like copper-oxide superconductors) show a dynamic phase
segregation at a transition from localized to itinerant electronic behaviour. Phase
segregation is normally accomplished by atomic diffusion, which can occur only at
higher temperatures. But, at a transition from localized to itinerant electronic behaviour,
phase segregation may be accomplished at lower temperatures by co-operative atomic
displacements, which create domains of shorter bond lengths containing delocalized
electrons separated by regions where the atoms contain localized electrons.
When the insulator LaMnO3 is substituted by Sr 2þ for La3þ (i.e., hole doping),
itinerant eg holes are produced, coupling Mn4þ ions. The eg holes hybridize with O(2p).
A DE Mn3þ interaction mediates ferromagnetic interaction, which, below the Tc,
shows metallic conduction. Near Tc, an applied magnetic field aligns the local spins and
the electron transfer increases, causing the consequent drop in resistivity.

15.3.4. “Skutterudites” and chalcogenides: “holey” and “unholey” semiconductors

Recent results on materials called skutterudite and cage-like materials called

clathrates dispelled the “30-year drought in new thermoelectric (TE) materials” (Sales
et al., 1997). With the voids or cages in these materials filled with “rattler atoms”, the
materials exhibit very low thermal conductivity (k , 1.5 W m21 K) (Nolas et al., 1996,
1998). The vibration amplitudes of the individual atoms have potential for scattering
different phonon (quantized lattice vibrations) modes, which determine thermal
conductivity. These phases show “pressure tuning” when subjected to high pressure
(P , 2 GPa) with increase in their power factors. This class of materials has been termed
as “holey semiconductors” (see Tritt, 1999).
In TE materials, heat is not transported by the lattice but by the electrical charge
carriers — electrons or holes. An application of electrical current through a TE material
cools one end and transports heat to the other end of the device. A good TE material has
to have the electronic properties of a crystalline material and the thermal properties
of a glass.
The potential of a material for TE applications is determined by the material’s
dimensionless figure:

ZT ¼ ða2 s=kÞT;

where a is the Seebeck coefficient or thermopower (a , DV/DT ), s is the electrical

conductivity and k is the total thermal conductivity ( ¼ k1 þ kE, where k1 and kE are the
lattice and electronic contributions, respectively).
Slack’s concept of “holey” semiconductors (see Slack and Ross, 1985) for new TE
materials centres around minimizing the lattice thermal conductivity of a material
by inserting loosely bound atoms or “rattlers” into voids or holes in the structure.
1056 Chapter 15

These rattlers will move or bounce around inside the voids and thus scatter phonons,
effectively reducing the lattice thermal conductivity.
The “unholey semiconductors” showing TE properties include naturally occurring
two-dimensional structure such as chalcogenides (e.g., CsBi4Te6) and the penta-telluride
materials (HfTe5 and ZrTe5). The chalcogenide materials can be viewed as “filled Bi2Te3”
materials. Both these groups show promising electronic properties and may hopefully lead
to low temperature (T , 220 K) TE materials. Quasicrystals are also representative of the
unholey group of materials. They exhibit glass-like thermal behaviour (Tritt, 1999).

15.4. RE-Mn perovskite

Mn-oxide perovskites have been discussed in Section 10.8. In this Section,

electronic and magnetic behaviour of manganese-bearing rare-earth perovskites would be
taken up with examples of such compounds as La12xAxMnO3 and Pr12xCaxMnO3. In this
perovskite structure, Mn ion is in a locally cubic environment and the crystal field splits the
five d-orbitals into a t2g triplet and an eg doublet, separated by , 2 –4 eV. The degree of
filling n ( ¼ 1 2 x) of the conduction eg band is an important parameter that governs the
electronic nature in the perovskite manganese oxide.
A phase diagram of La12xCaxMnO3 in x – T plane is presented by Millis (1998).
When x ¼ 2; the material is insulating at all temperatures. At high temperature, it
is paramagnetic (with no long-range magnetic ordering) but below ,140 K it becomes a
(0, 0, p) anti-ferromagnet (the magnetic moment direction alternates from plane to plane in
the direction ’ to c-axis) and finally to metallic.
At x ¼ xM1 ; the ground state changes from the insulating one. For xM1 , x , 0:5;
the ground state is a ferromagnetic metal. For x . 0.5, the ground state again becomes
insulating and anti-ferromagnetic and, in addition, is “charge-ordered” (Millis, 1998).
These compounds show magnetoresistance (see Section 15.4.1.2). The ferromag-
netic transition is a metal – insulator transition and the resistivity depends on the magnetic
field near this transition. There is a mechanism which localizes the carriers at T . Tc and
“turns off” the localization as T is lowered through Tc.
The ferromagnetic and the charge-ordered phases have a large strain mismatch.
External perturbations, such as magnetic field, light, X-ray or stress, change the relative
proportions of the ferromagnetic and charge-ordered phases. Recently, Fiebig et al. (1998)
have shown that well-conducting paths can be created simply by shining light on insulating
Pr0.7Ca0.3MnO3.

15.4.1. La12xCaxMnO3

When larger Sr or Ba substitutes for Ca, larger values of Tc are observed. This is
because Ca is more covalent than Sr and hence competes with Mn – O – Mn interaction.
Such covalency would make dTc=dP less positive. Mn –O –Mn bond angle is an important
factor in determining Tc. The angle varies systematically with A-ion size. As the angle
decreases from 1808, the magnetic exchange and electronic hopping are seen to reduce
(Hawang et al., 1995).
Transport Properties in Deep Depths and Related Condensed-Matter Phenomena 1057

In La12xCaxMnO3, the values of dTc=dP are larger than in conventional

ferromagnets and eight times larger than in Ca12xSrxMnO3.
Billinge et al. (1996) used atomic-pair distribution function (PDF) analysis of
neutron powder diffraction data to study the local structure of La12xCaxMnO3 for x ¼
0:12; 0.21 and 0.25 as a function of temperature. They observed a local structural response
to the MI transition, which is evident as abrupt broadening of certain PDF peaks. The two
PDF peaks which are most strongly affected involve the nearest-neighbour Mn –O and
O – O correlations. The bonds make up the MnO6 octahedra.
They (1996) directly observed the formation of small lattice polarons in the local
structures of La12xCaxMnO3 using the pair distribution function analysis of neutron-
powder diffraction data. The polaronic distortion of the lattice is well modelled as a
uniform collapse of the MnO6 octahedron associated with the Mn4þ site. The oxygen
displacements of the distortion are of the order of d ¼ 0.12 Å when it is assumed that a
hole localizes itself on one in four of the Mn sites. The polaron formation is closely
associated with the MI transition.
The charge-ordered stripes in La0.33Ca0.67MnO3 are shown in Fig. 15.2 (see the
caption), showing the periodic ordering of Mn3þ and Mn4þ ions. High-resolution electron
microscopy reveals pairing of stripes with a period three times the fundamental lattice
constant.

15.4.1.1. Resistivity and magnetism under pressure

The influence of pressure on the resistively, r, qualitatively resembles the influence
of the magnetic field on r at zero pressure (Kuster et al., 1989). Therefore, under pressure,
the decrease of r not only mimics the effect of magnetic field but also destroys the
anomalously large magnetoresistance (MR) of the system. Thus, a reduction in volume
reduces the unusual physical state which is responsible for the anomalously large
magnetoresistance. Hence, the changes in volume (i.e., pressure) not only play an
important role in electrical transport but in magnet transport as well.

Figure 15.2. Charge-ordered stripes seen in La0.33Ca0.67MnO3 (adapted from Mori et al., 1998). Charge ordering
is a feature of transition-metal oxides, such as the magnetoresistive compounds studied by Uehara et al. (1999).
The high-resolution image made by electron microscopy shows pairing of stripes with a period of three times the
fundamental lattice constant, B. This is interpreted as a periodic ordering of Mn3þ and Mn4þ ions, with the
occupied d-orbitals (from Littlewood, 1999).
1058 Chapter 15

Pressure induces rapid movement of p toward higher temperatures, primarily

because of the sensitivity of resistivity, r, to pressure in the neighbourhood of Tc.
The resistivity above Tc can be considered as arising from the propagation of magnetic
polarons (Kuster et al., 1989), which move via thermally activated hopping.
The transfer integral for electrical conduction between neighbouring Mn sites is
approximated as
teff ¼ t0 cos{ðuÞ=2}:
teff is maximized when the angle (u) between the two neighbouring spins is zero (i.e.,
ferromagnetic alignment). A dramatic decrease of r with decreasing temperature near Tc
(i.e., as H goes to zero) at zero pressure occurs as a result of broadening of the
bandwidth, which is proportional to teff, resulting in an increase of charge-carrier
mobility.
Pressure is shown to have substantial effects on the electrical conduction
process and Tc in La12xCaxMnO3 perovskite. The electrical resistivity is strongly
reduced on application of pressure and dTc=dP is a strong function of x, indicating
that pressure increases charge-carrier mobility and total charge-carrier concentration,
which in turn enhance the ferromagnetic double-exchange interaction and Tc.
Perovskite-based La12x (Ca, Sr or Ba)xMnO3 compounds show magnetic exchange
(i.e., positive super-exchange interaction), believed to originate within the Mn3þ –O–
Mn4þ complex. Electrical resistance measurements showed activated behaviour above
the ferromagnetic transition Tc, followed by the onset of a transition to metallic
behaviour at Tc.
In the La12xCaxMnO3 magnetic lattice, numerous anti-ferromagnetic lattice (in the
ranges 0 # x , 0.2 and 0.45 , x , 1) and ferromagnetism (0.2 , x , 0.45) have been
reported. The DE model of Zener (1951) may in part explain the large MR effect, although
some have suggested an additional role of magnetic polarons (Kuster et al., 1989).
Polaronic effects associated with Jahn – Teller splitting of the Mn “d” states could also be
important.
The magnetisation density near and below Tc is related to the bulk magnetization M
through the phenomenological relation r=a ¼ exp{ 2 MðH; TÞ=M0 }; where M0 is a
constant. In interpreting the measurements using this relation, evidence is seen of magnetic
polarons evolving from small polarons above Tc to large polarons below Tc.
Similarly, pressure on La0.79Ca0.21MnO3 is seen to reduce strongly the resistance r
near room temperature at a rate of , 45%=GPa (2 80.4 mV cm GPa21) and it moves Tc to
a higher temperatures. Tc is observed to be strongly pressure dependent with a slope near
zero pressure of dTc=dP ¼ þ 36.7 K GPa21 (Fig. 15.3).
Pressure on La0.67Ca0.33MnO3 decreases the electrical resistivity near RT at the rate
of 2 49%=GPa (2 16.5 mV cm GPa21) and moves the Tc to higher temperatures. A sharp
drop in electrical resistivity coincides with the magnetic transition in the entire pressure
range.
In La0.60Ca0.40MnO3, pressure drives Tc to higher temperature at the rate of
dTc/dP ¼ þ 16.0 K GPa21. Tc nearly structures at P < 1.7 GPa. The electrical resistivity
near room temperature is reduced with P at a rate of 2 55%/GPa (2 16.5 mV cm GPa21),
Transport Properties in Deep Depths and Related Condensed-Matter Phenomena 1059

Figure 15.3. Electrical resistivity versus temperature at three hydrostatic pressures for La0.79Ca0.21MnO3. In the
inset the pressure dependence of Tc (left abscissa) and the pressure dependence of the energy Eg (right abscissa)
are plotted versus pressure (Neumeir et al., 1995).

which is the same as with La0.67Ca0.33. Possibly, a magnetic transition from the
ferromagnetic state to a canted ferromagnetic state would decrease the electrical
conduction within the DE description. Pressure on the Pb-doped single crystal of La(Pb)
MnO3 reduces the MR effect.

15.4.1.2. CMR: RE12xAxMnO3

The family of manganese perovskites manifests colossal (or giant) magnetoresis-
tance (Jin et al., 1994), in which significant variation of electrical resistance with magnetic
field is observed. This is crucial for devices such as magnetic data storage, etc. Magnetites,
other spinels and pyrochlores also display this behaviour.
Manganese oxide perovskites exhibit intrinsic GMR, which originates from an
applied magnetic field affecting spin-charge-coupled magneto transport phenomena. How,
in an applied magnetic field, the electrical resistance of GMR material undergoes a large
change can be illustrated with the case of LaMnO3 as an example. The essential physics of
the “colossal magnetoresistance” of RE12xAxMnO3 is the interplay between a strong
electron-photon coupling and the DE effect of spin alignment on electronic kinetic energy.
The colossal magnetoresistance is observed for temperature near Tc ðxÞ:
The electronically active orbitals are the d-orbitals and, as already stated, the mean
number of “d” electrons per Mn is 4 2 x. The cubic anisotropy and Hund’s rule coupling
are so large that three electrons go into tightly bound t2g core states and make up an
electrically inter-core spin (Sc) of magnitude 3/2; the remaining (1 2 x) electrons go into a
band of width , 2.5 eV, mostly made up of the outer-shell eg orbitals. The eg electrons are
aligned to the core states by a Hund’s rule coupling, JH, which is believed to be large. The
large JH means that the hopping of an outer-shell electron between two Mn sites is effected
1060 Chapter 15

Figure 15.4. Crystal and electronic structure of managanese perovskite. (a) Oxygen motions in Jahn –Teller
distortion indicated by arrows in a cubic lattice with Mn–O bonds stretching along z, and compression along x – y
plane. (b) The change in energy levels of Mn ion with distortion. The central portion is for undisturbed lattice. A
Jahn–Teller distortion of the surrounding O6 octahedron may occur (as shown in a), causing a split of the eg
doublet by an energy of EJT. The left side of the figure shows that if eg level is unoccupied, a ‘breathing’ distortion
may occur, which lowers the energy of the unoccupied eg doublet by an amount EB relative to the energy of the
ideal structure (Millis, 1998, q 1998 Nature Publishing Group).

by the relative alignment of the core spins, which are maximal when the core spins are
parallel and minimal when they are anti-parallel.
Also, electron hopping promotes ferromagnetic order. The DE is significant in the
regime 0.2 , x , 0.5. However, the DE phenomenon cannot alone account for the large
resistivity at T . Tc and for the sharp drop in resistivity just below Tc. To explain this, a
strong electron – phonon coupling due in part to a Jahn – Teller splitting in the Mn eg states
is invoked (Fig. 15.4).
The cubic –tetragonal phase transition observed for 0 , x , 0.2 is known to be due
to a frozen Jahn – Teller distortion with long-range order at the wave vector (p, p, p).
Millis et al. (1996) propose that, for x . 0.2 and T . Tc ðxÞ; the strong electron-phonon
coupling localizes the conduction-band electrons as polarons but the polaron effect is
“turned off” as T decreases through Tc, permitting the formation of a metallic state. Thus,
the metal –insulator (MI) transition is a consequence of the large-to-small polaron
transition, induced by the reduction of effective hopping integral temperatures near Tc.
However, some workers argue that random hopping in the paramagnetic phase, due to the
DE mechanism, is sufficient to localize electrons and induce a MI transition at Tc.
In a perovskite of composition La12xAxMnO3, ferromagnetic clusters of manganese
atoms (, 12 Å across) occur within the paramagnetic matrix. On application of a magnetic
field, the number of clusters decreases as their size grows. The CMR is developed through
the confinement of the charge carriers to finite ferromagnetic domains in the temperature
interval Tc , T , 1.8Tc (where Tc ¼ 260 K).
In La12xAxMnO3, the transition from localized to itinerant (delocalized)
behaviour of the conduction electrons occurs in the presence of localized electron
configurations having a total localized spin S ¼ 3/2. So the phase containing mobile charge
carriers is ferromagnetic, whereas the phase with no delocalized charge carriers remains
Transport Properties in Deep Depths and Related Condensed-Matter Phenomena 1061

paramagnetic in the range Tc , T , 1.8Tc (e.g., in La0.67Ca0.33MnO3). Application of a

magnetic field above Tc stabilizes the ferromagnetic phase relative to the paramagnetic
phase because of dynamic-phase segregation. This in turn increases the volume fraction of
the more conductive ferromagnetic phase to give a colossal magnetoresistance.
Small-angle neutron scattering (SANS) can detect a magnetic heterogeneity
involving larger ferromagnetic clusters in a paramagnetic background. The magnetic
susceptibility data of La0.67Ca0.33MnO3 obtained by De Teresa et al. (1997) indicate that at
temperatures . 1.8Tc there is a presence of super-paramagnetic clusters consisting of
ferromagnetically coupled Mn(III) þ Mn(IV) pairs (see, for example, Fig. 15.2). These
Zener pairs (Zener, 1951) are charge carriers and segregate below 1.8Tc to larger
ferromagnetic regions within a paramagnetic matrix. The magnetic field makes the
ferromagnetic regions grow until they overlap. Below Tc, the conduction electrons become
itinerant and develop a long-range magnetic order and low resistivity. Possibly, in the
range Tc , T , 1.8Tc, a trapping out of Zener-pair charge carriers occurs in larger
ferromagnetic clusters.

15.4.2. Pressure on polarons, activation energy and charge carrier mobility

The charge carrier and the associated lattice deformation (or cloud of phonons)
constitute a polaron (see “Electrical conductivity of LM and D00 zone” in Section 15.1.2.3).
The polaron can be characterized by its binding energy. The bandwidth of a polaron is
related by

JB , expð2aRÞ;

where a is a constant and R is the distance between two neighbouring polarons. Pressure
decreases R and has a strong positive effect on JB (due to its exponential dependence on R),
thereby increasing the charge-carrier mobility and decreasing the resistivity, r: The
pressure-induced increase of charge-carrier mobility would increase Tc and be partly
responsible for enhancing the DE ferromagnetic coupling strength, which increases Tc.
DE depends directly on the charge-carrier mobility and the number of total charge carriers,
which usually increases with pressure.
An application of pressure stabilizes the low-temperature phase and leads to an
increase in the transition temperature. At high pressure, an increase in overlap between
adjacent orbits occurs, which means a greater role for electronic kinetic energy for
determining the stable structure for the system. As pressure increases the transition
temperature, it decreases the magnitude of MR peak. The increase in transition
temperature can also be interpreted as a consequence of increased electron hopping
amplitudes as a function of pressure.
An increase in pressure would accentuate the stability of the low-temperature
ordered phase. In contrast, the decrease in the bond angle (e.g., in Mn – O –Mn) upon
introduction of chemical pressure decreases the relevant overlap integrals and
consequently decreases the Tc (Hwang et al., 1995).
1062 Chapter 15

External pressure also lowers the activation energy. A decrease in activation energy
leads to a decreased high-temperature resistivity and a lower peak for magnetoresistance.
The decrease in activation energy can be explained as due to a reduced local relaxation
around the more quickly hopping electrons.
In the ground state, each electron has its spin parallel to the localized spin. In the
first excited state, the electron is transferred from one site to another and the energy
becomes higher than the ground state by 2SJ. In the second excited state, the spin of
any of the electrons is reversed and the energy becomes higher than the ground state
by ð2S þ 1ÞJ:
Hund’s rule dictates that the site energy of a charge carrier aligned with the
underlying ionic spin is lower that that of an unaligned carrier. In the high-temperature
spin-disordered phase, the set of spin-up carriers predominantly occupy the spin-up ions,
while the set of spin-down electrons is mostly restricted to the spin-down ions. Upon spin
ordering, the entire set of charge carriers may occupy all of the ionic sites. The
delocalization energy is such that there is overlap between the wave functions of the
adjacent orbitals.
In the case in which each electron has its spin parallel to the localized spin, as well
as N equals Ne ; the electrons cannot move because of the strong Hund coupling. In the case
when N , Ne ; electrons can move to ðN 2 Ne Þ unoccupied sites.
When N 2 Ne p Ne ; the unoccupied sites can be visualized as moving among the
sites with spin S þ ð1=2Þ: In such a situation, by removing an electron from an occupied
site, a “hole” is produced but holes and electrons cannot simultaneously occupy the
same site. Since the total spin of an ion with a hole is S, the hole is regarded as coupling
anti-parallel with the localized spin of size S þ ð1=2Þ:
Under perturbation, the degeneracies are removed. The perturbation (the transfer
terms) has matrix elements between states which belong to the same eigenvalues (the
secular part) as well as the states which belong to different eigenvalues (non-secular part).

15.4.2.1. Electron –lattice coupling

Electron– lattice coupling has an important role in optical activity and explains the
strong pressure dependence of physical properties (Neumeier et al., 1995; Khazeni et al.,
1996; Morimoto et al., 1996). The lattice distortions can generate three modes:
(i) a “breathing” distortion of the O6 octahedron around a given cation, e.g., Mn,
which couples to the changes in eg charge density and (ii) two linearly independent,
even-parity uniaxial volume-preserving distortions (Jahn– Teller modes), which couple to
preferential occupancy of one eg orbital over the other.
One such distortion, indicated by arrows on Fig. 15.4a, involves a stretching of the
Mn –O bonds along z and compression in the x – y plane and is favoured by occupancy of
the 3z2 2 y2 . eg orbital. The corresponding change in the energy levels is shown
in Fig. 15.4b.

Tolerance factor. In RE12xAxMnO3, different choices for RE and A, having different

ionic radii, will produce different internal stresses acting on the Mn –O – Mn bonds. Under
compression, this bond is buckled and the Mn –Mn hopping decreases. This effect of static
Transport Properties in Deep Depths and Related Condensed-Matter Phenomena 1063

crystal structure on electron hopping is called the “tolerance factor” arising from electron –
lattice coupling.
The other effect is the dynamical electron – phonon coupling. In this, the deviations
of atoms from their ideal crystallographic positions would simultaneously bring forth
change in the electronic configuration from the average value.
The ratio of the lattice energy (Elatt) gained from the electron – phonon coupling in
the absence of hybridization and the “bare” electron kinetic energy, teff, is denoted by the
dimensionless quantity lð< Elatt =teff Þ:
In manganite (such as RE12xAxMnO3), l can be changed over a wide range by
varying magnetic field and temperature (affecting spin correlation). The tolerance factor
and carrier concentration can be changed by changing RE and A. As temperature is
decreased through Tc ðxÞ; the growing ferromagnetic order increases teff and l decreases to
cause metallic behaviour to manifest. However, to explain the insulating behaviour, the
mechanism of electron – lattice coupling may be invoked.

15.4.3. LaMnO3 perovskite

The nominal Mn3þ ion in the parent compound LaMnO3 (x ¼ 0) has an electron
3 1
configuration of t2g eg : Among the four 3d electrons, t2g electrons are localized on the Mn
site and give rise to a local spin (S ¼ 3/2) while the eg state, hybridized with the O2p state,
is either itinerant or localized as the case may be.
There is a strong exchange interaction (Hund coupling J) between eg electron and
localized t2g spins. In this sense, the system can be viewed as a Kondo lattice system with
ferromagnetic coupling J $ 0 (Kubo and Ohata, 1972). With hole doping, an I– M
transition takes place close to a critical concentration, xc < 0.17, and the itinerant eg
carriers ferromagnetically align the local spins by way of the so-called DE interaction
(Zener, 1951).
LaMnO3 is an insulator with one eg electron per Mn. The Jahn – Teller coupling
causes an “anti-ferrodistortive” deviation from the basic perovskite structure, resulting
in a decrease in some Mn – O bond lengths and an increase in others. This occurs in a
“(p, p, 0)” pattern which alternates from one Mn site to the other Mn site throughout
the crystal (Ellemaans et al., 1971).
Bragg diffraction experiments show that the changes in bond lengths are long-range
ordered and are about 10% of the mean Mn – O distance. This may occur due to a very
strong Jahn – Teller coupling in La12xCaxMnO3 (where x ¼ 0). When x ¼ 1, viz. in
CaMnO3, no distortion occurs at all because the eg band is empty but large-amplitude local
distortions may exist even when no coherent part is visible in the conventional Bragg
scattering. As temperature is lowered through Tc the distortion drops to an undistorted
material.

15.4.3.1. M– I cohabitation
Interpenetrating structures of metal and insulator are nowadays designed
for technological purposes. But nature can deliver such structures for free, if one can
1064 Chapter 15

look for them. Mn-perovskites of LaMnO3 type show remarkable electronic and
magnetic properties. Many of these are ferromagnetic, similar to iron, but on heating
they lose their magnetic order and this is accompanied by a huge increase in electrical
resistance.
The M – I transition in such perovskites as LaMnO3 can be manipulated.
Substituting La with divalent cations such as Sr2þ or Ca2þ (or their ilk) will vary the
valence states of manganese to Mn3þ and Mn4þ in changing proportions. Again, replacing
Mn2þ with a smaller trivalent rare-earth atom (such as Pr3þ) will change the lattice
constant, producing a kind of chemically induced negative pressure. Changing the Mn
valence adds potentially mobile carriers whereas smaller cations tend to induce insulating
behaviour.
When the nominal Mn valence is intermediate between 3þ and 4 þ , at least three
different states of the solid can be seen: (i) a ferromagnetic metal, (ii) a charge-ordered
insulator and (iii) a poorly conducting paramagnet at higher temperatures.
As discussed earlier (see “Viscosity and conductivity” Section 15.1.2.3), a local
lattice distortion produces a potential minimum which can trap the electron in a given Mn
orbital. In manganites, self-trapping occurs for a high density of electrons but the self-
trapping has to compete with the tendency for delocalization arising from electron
hybridization.
In the ferromagnetic state, the electrons are mobile and the charge density on
each Mn site is identical. The mobile electrons promote ferromagnetism by aligning the
core spins on the Mn ions in the DE process. Alternatively, the charge segregates
preferentially between different sites with nominally Mn3þ or Mn4þ valence, producing
a charge-ordered state (Mori et al., 1998) which is also an insulator (Fig. 15.2
(from Littlewood, 1999)).
Such charge-ordered structures are not only seen in manganese oxides, but also in
cuprates (Tranquanda et al., 1995) and nickelates (Chen et al., 1993). Charge stripes are a
common phenomenon of transition-metal oxides. At high enough temperature, this ordered
array can “melt” and become dynamic while retaining charge fluctuations. This liquid-like
paramagnetic state is sensitive to magnetic fields and this is when CMR becomes most
pronounced (Millis, 1998).
These metallic materials are characterized by the co-existence of metallic
behaviour for one electron spin and insulating behaviour for the other. This implies that
the Fermi energy (EF) falls in a gap. Thus, the electronic DOS is completely spin-polarized
at the Fermi level and the conduction is dominated by these metallic single-spin
charge carriers.

15.5. La12xSrxMnO3

In this compound, with increasing x, the room-temperature structure changes from

orthorhombic (Pbnm; Z ¼ 4) to rhombohedral (R 3c̄; Z ¼ 2) at a critical concentration
xs , 0.17 (Asamitsu et al., 1995).
Transport Properties in Deep Depths and Related Condensed-Matter Phenomena 1065

Figure 15.5. (Top) Pressure dependence of resistivity for the La12xSrxMnO3 (x ¼ 0.175) crystal. Filled triangles
stand for the temperatures determined by susceptibility measurements. Open triangle is due to a structural
transition (Bottom) Magnetoresistance for the x ¼ 0.175 crystal with current parallel to the magnetic field
(Morimoto et al., 1995).

The upper panel of Fig. 15.5 shows a prototypal example of the pressure
dependence of resistivity (r) for the La12xSrxMnO3 (x ¼ 0.175) crystal. The
pressure-induced change in the r –T exhibited no hysteresis in repeated pressure
cycles. The r – T curve gradually increases with decreasing temperature and then
sharply drops around the Curie temperature, indicating a transition from the
paramagnetic non-metal to ferromagnetic metal. Application of pressure enlarges the
ferromagnetic metal phase through the pressure-enhanced transfer interaction (t) of
the holes generated by doping. The pressure-enhanced t value is also responsible for
the reduction of r in the paramagnetic phase ðT $ Tc Þ: The magnetoresistance effect
for the x ¼ 0.175 crystal is shown in the lower panel of Fig. 15.5 with a current
parallel to the field ðIkhÞ:
Morimoto et al. (1996) investigated the pressure effect on the magnetic and
electronic properties of prototypal DE ferromagnets, such as La12xSrx MnO3, by
systematically varying the nominal hole concentration, xð0:15 # x # 0:5Þ: The relative
change of the Curie temperature, Tc, against pressure of La12xSrxMnO3 crystals with
varying x is shown in Fig. 15.6.
Park et al. (1998) have reported spin-resolved photo-emission measurements
of a ferromagnetic manganese perovskite, La0.7Sr0.3MnO3, which directly manifests
the half-metallic nature well below the Curie temperature. For the majority spin, the
1066 Chapter 15

Figure 15.6. Relative change of the Curie temperature Tc against pressure for La12xSrxMnO3 crystals with
various x values. The solid lines are the results of least-square fitting (from Morimoto et al., 1995).

photo-emission spectrum clearly shows a metallic Fermi cut-off whereas, for the minority
spin, it shows an insulating gap with disappearance of spectral height at , 0.6 eV
(binding energy).

15.6. Pr-manganates

The stability of one phase of RE-manganate perovskite over another is a matter of

chemistry and can be tuned. Uehara et al. (1999) have studied a series of compounds
where the La site is partially substituted by the smaller Pr, but the nominal Mn valence is
fixed.
The Pr-rich compounds have stable charged-ordered phases and the La-rich
compounds are ferromagnetic, so one would expect to see a sudden insulator – metal
transition as the Pr/La ratio is varied. Instead, Tomioka et al. (1996) find a broad range over
which the two phases co-exist. Electron microscopy shows that both metal and insulator
are present (co-habit) in the same crystal, segregated on length scales of hundreds of
nanometres to micrometres. Other workers (Kimura et al., 1999) have imaged similar
structures. As the La/Pr ratio is varied in the (La, Pr, Ca) MnO3 system, only the proportion
of metal to insulator changes.
In this case, CMR is produced by the co-existence of two phases, not by the
homogeneous transformation of one to the other as observed with changing temperature in
conventional manganites. Although the phases co-exist over wide variations in
composition, it is known that the charge-ordered phase can be transformed to the
Transport Properties in Deep Depths and Related Condensed-Matter Phenomena 1067

ferromagnetic state by modest magnetic fields and even by substitution of 16O with
18
O. This implies that the ground-state energies of the two phases remain very close,
despite large compositional change. The co-existence of metal and insulator thus becomes
a real possibility (discussed in Section 10.8).

15.6.1. Pr12xCaxMnO3

In this family of perovskites, the end-members having x ¼ 0 and x ¼ 1 are

insulating or anti-ferromagnetic with manganese ions in the Mn3þ or Mn4þ valence states,
respectively. For intermediate x, the average Mn valence is a non-integer.
In the compound Pr0.7Ca0.3MnO3, a transition from the insulating anti-
ferromagnetic state to the metallic ferromagnetic state can be driven by illumination
with X-ray at low temperatures (, 40 K) (Fiebig et al., 1998). This transition is
accompanied by significant changes in the lattice structure and can be reversed by thermal
cycling. This effect, undoubtedly a manifestation of the strong electron – lattice
interactions, is believed to be responsible for the magnetoresistive properties of these
materials and provides insight into the physical mechanisms of persistent photo-
conductivity. This may also find applications in X-ray detection and X-ray lithographic
patterning of ferromagnetic nanostructures.

15.6.2. (La12yPry)12xAxMnO3: short- and long-range order

In the (La12yPry)12xAxMnO3 system, the short- and long-range order may be

significantly changed by chemical and physical means (see the caption to Fig. 15.7;
Mathur, 1999). By replacing La with larger Sr atoms, rhombohedral lattice distortion
is effected to cause chemical strain. By varying x in the La/Ca ratio to x ¼ 0.15, an
unusual layered structure is attained, where every other microscopic layer contains no
charge carrier. By applying a magnetic field, B; in the range 0.1 Tesla , B , 1 Tesla, a
mixture of long-range ferromagnetic and anti-ferromagnetic order is produced. Thus, in a
continuous ferromagnetic crystal, an artificial magnetic twist can be generated
(see Fig. 15.7).

15.7. Fe31 in perovskite and conductivity

As discussed in Section 10.2.6, perovskite in the lower mantle contains substantial

amounts of Fe3þ (even in equilibrium with metallic Fe; McCammon et al., 1992) causing
charge-transfer bands contributing to optical absorption throughout the mantle. Over the
entire IR and visible range, perovskite shows strong absorption because of the combination
of crystal-field and charge-transfer bands. The intervalence charge-transfer bands reduce
the thermal conductivity and causes a strong increase in electrical conductivity by electron
hopping (polaron hopping). This implies that, at high temperatures where radiative
transport dominates, perovskite is an extremely poor thermal conductor. All these
contribute to the electrical semiconductivity of the lower mantle.
1068 Chapter 15

Figure 15.7. Long-range and short-range order may be significantly varied in the (La12yPry)12xAxMnO3 system
by chemical and physical means. (a) Chemical strain is introduced by replacing La with larger Sr atoms,
producing a rhombohedral lattice distortion. (b) Varying the ratio of La: Ca, x, leads to an unusual layered
structure when x ¼ 0.15, with every other microscopic layer containing no charge carriers. (c) Physical control is
demonstrated by applying a magnetic field B in the range 0.1 Tesla , B , 1 Tesla to produce a mixture of long-
range ferromagnetic and antiferromagnetic order. (d) A thin film device can be used to impose an artificial
magnetic twist in a continuous ferromagnetic crystal (from Mathur, 1999, q Nature Publishing Group).

The CFSE of Fe2þ in perovskite (CFSE ¼ 4350 cm21) is almost the same as in
magnesiowüstite (CFSE ¼ 4320 cm21; Burns, 1993). Therefore, to explain the partition-
ing of iron in perovskite, the electron delocalization (Fe2þ , Fe3þ) between these two
phases needs to be considered.

15.8. Al2O3 content and conductivity

Electrical conductivity measurements of natural orthopyroxene show that an

addition of small amounts of Al2O3 increases electrical conductivity significantly and that
conductivity is relatively independent of oxygen fugacity (Huebner et al., 1979). It is,
therefore, likely that the electrical conductivity of (Mg, Fe, Al)SiO3 perovskite would also
be significantly higher than for the Al-free phase. Nevertheless, the conductivity
mechanism is believed to involve Fe3þ as well (Poirier and Peyronneau, 1992).
Transport Properties in Deep Depths and Related Condensed-Matter Phenomena 1069

TABLE 15.1
Electrical conductivity and the amount of Fe3þ in Al-bearing and Al-free perovskite (Pv) at 25 GPa and ,1,6008C
(Xu et al., 1998)

Al-bearing Pv Al-free Pv Al-bearing Reference

Pv/Al-free Pv

(S/m) 0.97 ,0.28 ,3.5 Xu et al. (1998)

Fe3þ=SFe 42 ^ 5% 12 ^ 3% 3.5 Xu et al. (1998)
Fe3þ=SFe 50 ^ 5% 16 ^ 3% ,3.1 McCammon (1997)

The lower mantle consists predominantly of (Mg, Fe)SiO3 perovskite co-existing

with about 20% (Mg, Fe)O by volume. The partitioning of Fe and Mg between
perovskite and magnesio-wüstite is strongly coupled to Al2O3 concentration (Wood and
Rubie, 1996). Again, Fe3þ=SFe concentration (determined by Mössbauer spectroscopy)
in perovskite is greatly enhanced when aluminium is present even in small amounts
(McCammon, 1997).
Assuming that the electrical conduction is dependent on the electron hopping
between Fe2þ and Fe3þ ions (when they are present), a measurement of electrical
conductivity in synthetic perovskite at lower-mantle condition can also be related to the
aluminium substitution in perovskite. Measurements of (Mg, Fe)SiO3 at 25 GPa and
1,400 –1,6008C showed that a content of 2.89 wt% Al2O3 increases the electrical
conductivity by 3.5 times the conductivity of Al-free (Mg0.915Fe0.085)SiO3 perovskite
(Xu et al., 1998). The conduction mechanism is apparently by polarons (e.g., electron
hopping, Fe2þ ! Fe3þ þ ē) because the former is found to have , 3.5 times the amount of
Fe3þ/SFe than the Al-free sample (Table 15.1). Thus, the electrical conductivity of
perovskite is sensitive to the amount of Fe3þ coupled with Al2O3 content and so the effect
of oxygen fugacity on Al-bearing perovskite is small. The effect of OH2 on the
conductivity of silicate perovskite is still unknown.
In peridotite mantle, the perovskite phase should contain 4– 5% Al2O3 by weight
(Ringwood, 1975). A conductivity depth profile for the lower mantle can be constructed
using the Al-bearing perovskite parameters and the following relation:

s ¼ s0 e2DH=kT ;
where s0 is a pre-exponential factor, DH is the activation enthalpy, k is the Boltzmann
constant and T is the temperature.

15.9. Conductive TiO2, SiO2, FeO, Fe2O3 and Fe3O4

The electrical resistivity of rutile (TiO2) has been found to drop from , 105 to
21
10 V at about 2.2 Mbar. This resistance change may be associated with the Mott
transition (insulator ! metal transition) or be related to the crystal transition. The cubic
phase of TiO2 may be responsible for this resistance change. SiO2 also at higher pressure
displays a similar change from 105 to 1021 V accompanying a hysteresis loop on release of
1070 Chapter 15

pressure. Kawai and Nishiyama (1974) attributed this to the Mott transition or to the
breakdown of SiO2 into oxygen and silicon in the metallic state.
Fe0.95O was found by Kawai and Nishiyama (1974) to become conductive at
high P, which supports the hypothesis of an FeO outer core, proposed earlier by
Dubrovskiy and Pan’Kov (1972). However, the large change in density which is
characteristic of phase transformation of rutile –fluorite – cubic structure and the presence
of perovskite-mixed oxides are not corroborated by the data of seismic velocities of the
lower mantle.
The electrical resistivity of Fe2O3 measured in DAC is shown in Fig. 15.8 with a
comparison of shock-wave measurements of Kondo et al. (1998). Fe2O3 remains a
semiconductor in the high-pressure phase and its resistivity of , 1024 V m remains two
orders of magnitude higher than is typical of metals. However, at above 48 GPa, the
electrical resistivity of a-Fe2O3 shows a value similar to that obtained for magnetite

Figure 15.8. The electrical resistivity of Fe2O3 measured in the diamond-anvil cell (room temperature) compared
with a summary (dashed line) of shock-wave measurements of Kondo et al. (1980). The experimental data were
collected upon compression (solid symbols) and decompression (open symbols) for two samples, as marked by the
circles and triangles. Above 50 GPa the shock-wave measurements only yielded an upper limit for the resistivity
(indicate by the arrow), but the same compression pressure is found by both static and shock wave techniques
(Knittle and Jeanloz, 1986b; Kondo et al., 1998).
Transport Properties in Deep Depths and Related Condensed-Matter Phenomena 1071

(Fe2þFe3þ2 O4) in which Fe occurs in two valence states. Thus, the conduction mechanism
in Fe2O3 changes at 48 GPa (Fig. 15.8) from conduction of electrons in the iron d-bands
(Morin, 1954) at low pressure to an electron hopping mechanism between iron sites of
different valence at high pressure akin to the mechanism in magnetite (Knittle and
Jeanloz, 1986b).
Electrical resistance study at ultrahigh static pressure two anomalous drops of
resistance are observed. The resistance remains constant at , 107 V but at 0.4 – 0.5 Mbar it
begins to drop before tending to level out at , 104 V. From there, it slides down to a few
ohms. The first decrease corresponds to a phase transition while the second decrease may
indicate a transition to the conductive state (Endo and Ito, 1982).

15.10. Thermal conductivity, k

Heat conductivity is a complex function of T and varies proportionally to the third

power of temperature. Near zero temperature, the heat conductivity approaches zero. In
metals, heat is transferred by electrons and magnetic quasi-particles such as magnons. Both
electrons and magnons collide with phonons.
Heat is conducted through phonons colliding with each other and possibly with
defects and grain boundaries. Because rising pressure (or lowering temperature) raises
vibrational frequencies and because densification increases the chances of collision, the
thermal conductivity k increases as P increases (or as T decreases).
Hofmeister (1999) has developed a model for thermal conductivity, k; based on
phonon lifetimes obtained from infrared reflectivity, which replicates the experimental
data (for salts, silicates and oxides) at ambient conditions. The thermal conductivities with
depths in the Earth are presented in Fig. 15.1 (read the caption). For the lower mantle, only
the chondritic composition is considered. The discontinuities in k at 410 and 670 km are
attributed to the high conductivity of ringwoodite.
An understanding of heat transport within planetary bodies requires knowledge of
thermal conductivity (k) and thermal diffusivity ðD ¼ k=ðrCP Þ; where r is density and CP
is heat capacity at constant pressure) as functions of pressure (P) and absolute temperature
(T). The thermal conductivity under pressure within the Earth is presented in Table 15.2.

TABLE 15.2
Thermal conductivities from lower mantle to the core

Parameters Depths (km)

2,700 2,900
Lower mantle 00
D zone Outer core
21 21
K (W m k ) 6.3 6.3–4.8 4.3
Conductivity Q (TW) 0.2 5.4 4.4
Total Q (TW) 5.7 5.7 5.7
1072 Chapter 15

Thermal conductivity, k; generally increases with increasing pressure but several

exceptions are known. The volume dependence is conveniently characterized by the
derivative:

g ¼ ðd ln k=d ln rÞT : ð15-8Þ

For ordered systems, g-values fall in the range 6 – 12 (e.g., alkali halides). For plastic
phases, which show translational symmetry but orientational disorder, g-values are in the
range 4 – 6. For glasses and water, g ¼ 2. For the simple Debye theory, the value of k is
normally 2 and positive.
However, there are several instances where k is negative, e.g., for ordinary ice 1 h.
This happens in cases when the crystalline structure involves tetrahedral bonding.
The phenomenon may be qualitatively explained in terms of negative mode Grüneisen
parameters for transverse acoustic phonons.
Increased density may also result in dramatically decreasing k in certain phase
transitions, e.g., when KBr transforms at 1.8 GPa from NaCl to CsCl structure.

15.10.1. k at mantle depths

For mantle conditions, the thermal conductivity (k) has been approximated by
Hofmeister (1999) as
 ðT 
a P 0
kðP; TÞ ¼ kð298Þð298=TÞ exp 4q =K 0 2 ð4gTh þ 1=3Þ aðuÞdu
298
 
0 0 K 0 2
£ ð1 þ K 0 P=K0 Þ exp 4gTh þ ð1=3Þ=K 0 2 4q0 =ðK 0 Þ þ f ðTÞ; ð15-9Þ

where K 00 ¼ dKT =dP is assumed to be constant, the temperature and pressure dependence
of KT are assumed to be independent and q ¼ dg=dP: For mantle substance, the thermal
Grüneisen parameter, gTh ¼ 1– 1.4, K00 ¼ 4– 5 and q < 0. Within the uncertainty of these
parameters, the above equation is simplified to
 ð 
kðP; TÞ ¼ kð298Þð298=TÞa exp 2 4g þ 1=3 aðuÞdu ð1 þ K 00 P=K0 Þ þ f ðTÞ ð15-10Þ

The temperature dependence of thermal conductivity for dense silicate minerals is nearly
equal and the ambient k values of those silicates are also similar. High temperature
results from inefficient conduction in the deep lithosphere. The new conductive geotherm
attains 1,440 K at 70 km where it intersects the solidus of pyrolite with 8.3 wt% H2O
(Inoue and Sawamoto, 1992).
In the lithosphere, thermal conductivity decreases to 2.7 W m21 K21 at 64 km and it
may be as low as 2 W m21 K21 if the thermal boundary layer extends to 81 km (see also
Fig. 15.1). At 1,675 K and 81 km, the geotherm intersects the dry solidus of peridotite
(Takahashi, 1986a).
Transport Properties in Deep Depths and Related Condensed-Matter Phenomena 1073

The temperature in the upper mantle rises adiabatically with depth at the rate of
0.33 K km21 (Turcotte and Schubert, 1982). For the case of whole-mantle convection, the
lower mantle follows the same adiabat. Extrapolation to the CMB gives a temperature of
2,830 K. An adiabat through the garnet – ilmenite – perovskite triple point attains
3,050 ^ 100 K at the CMB (Hofmeister, 1999). Garnet shows relatively low thermal
conductivity. On the whole, the total conductivity in the upper mantle is roughly
independent of mantle mineralogy but, in the transition zone, ringwoodite possesses a high
thermal conductivity.
Condition for steady-state flow along the Z direction in an isotropic medium is
described by the Fourier law:
Q ¼ 2kðdT=dZÞ; ð15-11Þ

where Q is the heat flux (Carslaw and Jaeger, 1959). However, during compression
experiments, deformation produces uncertainties in the geometrical parameters employed
in solving the equation and also cracking reduces the thermal contacts.
A large polycrystalline sample of MgSiO3 was measured between 160 and 340 K
for its thermal diffusivity (D ¼ rkCp) by Osako and Ito (1991) and the fit resulted in the
relation:
1=D ¼ A þ BT; ð15-12Þ
4 22 3 2
where A ¼ (2 6.2 ^ 1.4) £ 10 m s and B ¼ (2.15 ^ 0.05) £ 10 m s K. They
obtained k ¼ 5.1 W m21 K21 under ambient conditions. The value is similar to other
silicates. They extrapolated to lower-mantle conditions and obtained values
k ¼ 3.0 W m21 K21 at 670 km (1,900 K) and 12 W m21 K21 at the top of the D00 layer
(2,500 K). The thermal conductivity was found to be high enough to prevent the D00 layer at
the base of the mantle from being a thermal boundary layer.

15.10.1.1. Radiative and lattice contribution

At 1,200 K, for (001) faces of olivine crystals (with 90% Mg) k ranges from 3.0 to
5.6 W m21 K21 (Koyabashi, 1974). Some geophysicists (Labrosse et al., 1997) believe
that the core has a conductive layer outside but the mantle also seems to manifest a thermal
conductivity parameter that can be separated into a radiative contribution, krad, which is
the flow of energy by radiation (as in a black body) and a lattice contribution, klat, which is
the energy flow through the minerals in the mantle. Across the mantle, the radiative
contribution is small, reaching 0.67 –0.82 W m21 K21 (10 –15% of the total k) at the core,
depending on the geotherm. Hence, the lattice contribution dominates yielding ktot ¼ 5.8–
6.7 W m21 K21 at the top of the D00 . (Note: The radiative component is pressure
independent because the absorption line-width is pressure independent.)
The lattice contribution to the thermal conductivity is
 ðT 
g
klat ¼ k298 ð298=TÞ exp 2 ð4gTh þ 1=3Þ aðuÞdu
298

with a ¼ 0.33 for silicates (or 0.9 for MgO).

1074 Chapter 15

If Fe2þ is present, the smaller pressure-independent radiative contribution is

krad 0:0175 2 0:0001037 T þ ð2:245 T 2 =107 Þ 2 ð3:407 T 3 =1011 Þ in W m21 K21 Þ:

The lithosphere geotherm is steep. Consequently, the mantle geotherm is hot if the
low-velocity zone is anhydrous but cold if the LVZ is hydrated (Hofmeister, 1999).
Hofmeister (1999) has shown that krad contributes substantially to deep-Earth
thermal conductivity and this improves the existing ideas of klat. In equations for krad,
Hofmeister includes the connection of photon and phonon life times reflected in infrared
peak widths. She has determined the values of k and the P and T dependence of klat of
insulators. At the base of the mantle, she obtains a thermal conductivity of
6.3 W m21 K21, which is higher than the value of 4.2 W m21 K21 obtained by
Keiffer (1976) for the deep mantle (see Fig. 15.1). The pressure dependence of thermal
conductivity and ambient values are presented in Table 15.3 (Hofmeister, 1999).

15.10.2. k at D00 zone

The calculation (Hofmeister, 1999) suggests that there is a convective flux of

0.3 TW in the D00 region near the mantle. Thus, D00 is considered as a rapidly flowing
region of low viscosity at the base of the mantle. The convective power in flowing D00 can
be regarded as the source of power to form the plumes. Seismic evidence for partial melt in
D00 under the Pacific Ocean lends credence to this (Vidale and Hedlin, 1998).
The sk value at the D00 – core boundary is 4.8 W m21 K21. A mechanism needs to
be found by which the conduction k of D00 near the core be made equal to the total power of
the core, i.e., 5.7 TW. Three reasons can be stipulated by which the composition of D00 may
be modified to increase the value of k (Anderson, 1999):
(i) Small chunks of core may be interspersed in D00 (mainly of mantle material) causing k
to increase.
(ii) Periclase (MgO) and stishovite (SiO2), the breakdown products of MgSiO3
perovskite, have high k values.
(iii) The ancient subducted oceanic crust may rest at D00 as its graveyard (Kendall and
Silver, 1996), causing an increase in k values.
Near the D00 boundary, the mantle must now carry away the total power of 5.7 TW
and conduction is limited to , 0.2 TW.
Plumes carry the heat from the core to the base of the lithosphere. Such plumes
are inferred from hot spots on Earth’s surface (e.g., Hawaii, Yellowstone and Iceland).
A measurement of thermal buoyancy above a hot spot gives a measure of the heat flux
and the power involved. Sleep (1990) measured the global buoyancy flux of 37 hot spots
and calculated an average heat flux of 4 £ 1023 W m22 or a mantle convective power of
2 TW. It should be noted, however, that estimates of buoyancy flux of the same hot spot
may differ by as much as 40% (Sleep, 1990).
In “mid-Cretaceous superplume episode” (Larson, 1995), a vast amount of heat was
transported to the Earth’s surface 120 my ago. The rate at which new hot material was
delivered then was twice that of today and persisted for 40 my.
 Transport Properties in Deep Depths and Related Condensed-Matter Phenomena
TABLE 15.3
Pressure dependence of thermal conductivity and ambient values. The last digit is uncertain by ^1 unless indicated. Assumed values for K0 are denoted by ; (from
Hofmeister, 1999)

Structure Composition Thermal properties Calculated Measured k (ambient) Reference

gTha KTa K 0a d[ln(k)/dP d[ln(k)/dP (W m21 K21)
(GPa21) (GPa21)

Olivine Fo ¼ Mg2SiO4 1.25 127.9 4.0 0.0417 0.18b 5.2c Fujisawa et al. (1968)
Olivine Fo90Fa10 1.28 128.1 4.6 0.0426 0.04 4.7 Chai et al. (1996)
0.046 5.0c Katsura (1995)
0.06 Beck et al. (1978)
0.165b Scharmli (1982)
Olivine Fa ¼ Fe2SiO4 1.45 135.1 ;4 0.0454 – 3.16 Horai (1971)
Garnet Mg3Al2Si4O12 1.43 171.5 3.8 0.0353 – 3.2 (some Fe) Horai (1971)
Garnet Mg1.5Fe1Ca0.5 ,1.4 169.5 4.09 ,0.035 – 3.3 Chai et al. (1996)
Opx MgSiO3 0.96 107 5–10 0.039 0.07 4.5d Schloessin and Dvorak (1972)
Cpx CaMgSi2O6 0.09 112 4.7 0.035 – 4.9 (average) Horai (1971)
b Mg2SiO4 1.0e 172 4.8 0.025 – –
g Mg2SiO4 1.25 183 5.2 0.029 – 7.7 (calc.) Hofmeister et al. (1999)
Garnet MgSiO3 1.33 154 ;5 0.0369 – –
Ilmenite MgSiO3 1.70f 210 ;5 0.0339 – –
Pv MgSiO3 1.2–1.7g 261 ;5 0.025(2) – 4.7h
Stishovite SiO2 1.2i 306 2.8 0.026 0.090j 1.7j Yutatake and Shimada (1978)
8.6 Osako and Kobayashi (1979)
Coesite SiO2 0.41 113 8.4 0.017 0.01–0.04 1.4– 1.5 Beck et al. (1978)
0.039j 8.0j Yutaatake and Shimada (1978)
Quartz SiO2 0.667 37.5 6.4 0.080 0.017 7.7c Beck et al. (1978)
0.17j Yutatake and Shimada (1978)
0.17j Kieffer et al. (1976)
0.25 Horai and Sasaki (1989)

1075
 1076
TABLE 15.3 (continued)
Structure Composition Thermal properties Calculated Measured k (ambient) Reference
a a 21 21
gTh KT K 0a d[ln(k)/dP d[ln(k)/dP (W m K )
(GPa21) (GPa21)

Rocksalt MgO 1.54 160 4.3 0.041 0.02 McPherson and Schloessin (1982)
0.04 Katsura (1997)
0.05 55.2 Anderson and Backstron (1986)
0.068j 41j Yutatake and Shimada (1978)
Spinel MgAl2O4 1.4 195.2 4.9 0.030 – 9.5 Horai (1971)
Corundum Al2O3 1.27 255.4 4.3 0.021 – 18 Horai (1971)
Rocksalt NaCl 1.58 23.8 4.45 0.280 0.32(4)k 6.6c Ross et al. (1984)
Oxysalt NaClO3 1.85 24.6 ;4 0.206 0.22 1.0 Franson and Ross (1983)
a
From Sumino and Anderson (1984) unless noted.
b
These studies are discounted because the results are not consistent with other studies, particularly with PTGS data (Chai et al., 1996) or with the average for NaCl.
c
The 1 atm value is from Horai (1971) rather than the referenced study at pressure.
d
For Opx with 10% Fe, k values are 4.2 (Fijisawa et al., 1968), 4.4 (Koyabasci, 1974) or 3.46 W m21 K21 (Chai and Brown, 1996).
e
From Cynn and Hofmeister (1994).
f
Used thermal expansion from Ashida et al. (1988).
g
The range exists because thermal expansion values (Ross and Hazen, 1989) depend on whether volume above or below room-temperature experiments is used.
h
Calculated from Osala and Ito’s (1991) diffusivity. The ambient value resembles those of other dense Mg silicates.
i
From Hofmeister (1996).
j
Because substantial cracking was observed, the results are unlikely to be accurate.
k
Average of six experiments, not including (Fujisawa et al., 1968) from Ross et al. (1984) compilation.

Chapter 15
Transport Properties in Deep Depths and Related Condensed-Matter Phenomena 1077

15.10.3. k and convective power of core

The thermal conductivity, k, of the iron core is considered to be ,10 times the
rocky mantle.
Most authors have proposed that the convective power of the core is negligibly
small (Buffet et al., 1996). Some suggest that the convective power of the core is
0.2 TW (Stacey and Loper, 1983), while the conductive power of D00 is considered to be
,4.4 TW. Using 30% of the core conductive power, the total power leaving the core is
5.7 TW (4.4 TW conduction plus 1.3 TW convection). The thermal conductivity and
ambient values in the depths of the Earth have been shown in Table 15.2. For 40K as heat
source, see Section 1.22.

15.10.4. k under shock pressure

Thermal conductivity, k, is related to the measured diffusivity, D ¼ kr/CP ; where r

and CP are, respectively, the density and specific heat of the shocked state. The temperature
dependence of conductivity may be sensitive to defect concentration. Defects in the
shocked state inhibit photon transport. For this reason, k is much smaller for amorphous
solids than for crystalline solids.
The value of k for iron at core conditions needs to be determined experimentally,
which will require new shock-wave measurements. A confirmation by experiment of the
values of k calculated by Hofmeister (1999) for mantle rocks at lower mantle would be
valuable.
Tang (1994) used a model for thermal conductivity with theoretical dependence on
T and r0 originally developed by Libfried and Schlomann (see Hofmeister, 1999).
The equation is
 5=3   
k r r T0
¼ exp 3g0 1 2 0 : ð15-13Þ
k0 r0 r T

15.11. Ferroelectric transitions

In perovskite-type ferroelectrics, the dielectric polarization is related to the

polarizability of oxygen octahedra, either relative to the small six-coordinated interstitial
cation or relative to the larger 8 –12-coordinated central cation by rotation of the O6
octahedra.
Some of the orthorhombic modifications of perovskite structure satisfy the
requirements for distortion of the cubic structure so that dielectric and dielastic
polarization becomes operative.
The high dielectric constant values (typically, 103 – 104) associated with ferro-
electric transitions could well explain the screening and attenuation of electromagnetic
radiation passing through the mantle. Moreover, ferroelectric structures in the lower
mantle would influence core – mantle coupling.
1078 Chapter 15

There is a steep rise in electrical conductivity by four to six orders of magnitude in

the transition zone and below. This is too large for a change in activation energy with
pressure and temperature for an extrinsic condition mechanism. A major change in
conduction mechanism is therefore needed. Since the dielectric constant is related to the
refractive index n (as k ¼ n 2), the radiative heat conduction is expected to be high in
the ferroelectric phase near its Curie temperature.
In materials with different dielectric and elastic properties, the ferroelectric
transitions are manifested. Also, because the inherent coupling between dielectric
polarization and spontaneous strain (“dielastic polarization”) finds expression in the
piezoelectric and electrostrictive coefficients, any ferroelectric structures present in the
Earth’s interior would exhibit elastic anomalies and affect elastic-wave propagation.

15.11.1. Ferroelectric phenomena in large planets

In large Jovian planets, the structural transformations involving ferroelectric

transitions become significant. In these planets, the dielectric polarization may occur as a
result of three different mechanisms: (i) proton motion in hydrogen-bounded structures and
hydrogen bonds exhibiting double-well potentials, (ii) order –disorder transitions of non-
spherical radicals such as (NH4)þ, glycine, C(NH2)3þ and NO3, and (iii) transitions in
numerous methylammonium – metal alums.
In all mechanisms, however, it is seen that the stability regions are strictly bounded
by the separations between the charged units or the displaced ions. Above these bounds,
some of the structures are anti-ferroelectric (Cochran, 1967).
Ferroelectric phenomena are expected to be more prominent in Jovian planets. The
bursts of radio-emission from Jupiter with peaks at 17.6– 22.2 MHz of less than 1 –2 s
duration are reminiscent of giant Jovian lightning bolts, which are possibly generated from
large ferroelectric depolarization pulses (by stresses and release mechanisms), associated
with tidal strain accumulations due to Io and others. Shock or impact depolarization of
ferroelectric samples may produce electrical bursts or pulses of microsecond duration.
Again, depolarization by decompression can also cause ferroelectric – anti-ferroelectric
transitions (Gonnard et al., 1972).
If the electrical bursts on Jupiter are caused by ferroelectric phenomena
involving highly efficient converters from mechanical to electrical energy, then the heat
sources can also be deciphered. On the other hand, if the bursts could be correlated with the
motion of the inner satellites, the radio bursts may be caused or triggered by the conversion
of tidal energy.

15.12. Non-elastic transport properties

In the dynamics of the Earth’s interior, the depth –variation of non-elastic transport
properties plays an important role. The non-elastic properties show a P – T dependence
following the Arrhenius-type relation, where the rate of a given process is proportional to
Transport Properties in Deep Depths and Related Condensed-Matter Phenomena 1079

the expression 2 (E p þ PV p)/RT, where E p is activation energy, V p is the activation

volume, P is pressure, T is temperature and R is gas constant.

15.12.1. Power law: fractal distribution

In physical sciences, power laws occur in several situations (Poirier and Duba,
1997):
(1) They may be derived from physical principles, such as gravitational force F / r 22 or
radiated energy E / T 4 (Stefan’s law).
(2) They may express a geometrical relation, e.g., a quantity depending on volume is
proportional to a length to the third power.
(3) They may reveal the existence of a self-similar and scale-invariant ( fractal)
distribution, for which the cumulative number of objects N, larger than a given size L,
is N / L2D ; where D is fractal dimension.
(4) Finally, power laws may be purely empirical. A common usage of a log –log plot is
when a linear plot fails to yield a straight line. In such cases, one measures the slope n
for the line and finds a power: Y / X n :
Diffusion coefficients, electrical conductivity and creep of minerals under constant
stress are usually reported to depend on oxygen fugacity through a power law. Likewise,
power laws govern the stress dependence of creep rate at constant temperature under
various applied stresses. However, in most cases, the exponents of theoretical models and
those obtained from experimental data do not agree.
If a thermally activated observable Y (diffusion coefficient, electrical conductivity,
creep rate, etc.), in addition to depending on temperature, also depends on another variable
X (oxygen fugacity, stress, etc.), it is usually assumed to obey a constitutive relation in
the form:
 
Q
Y / X n exp 2 : ð15-14Þ
RT

where Q is an apparent activation energy derived from the slope of an Arrhenius plot and is
the power-law exponent derived from the slope of a log Y versus X plot (for a better
discussion, see Poirier and Duba, 1997).
Electrical conductivity in silicates is a thermally-activated process, as expressed by
equation (15-14). Wanamaker and Duba (1993) showed that the activation energy for
conduction of olivine is dependent on oxygen fugacity. The conductivity versus f O2 shows a
slope of 1/5.5 with a little error. Also, the activation energy for conduction varies by a factor
of 2 as a function of f O2 across the olivine stability field. Lower activation energies are
associated with higher f O2. However, a part of the power-law dependence of f O2 should in
fact belong to the activation energy and Poirier et al. (1996) found a similar dependence as

Q ¼ Q0 2 A log f O2 with A ¼ 20:020 ^ 0:003 eV at:21

Increased temperatures within D00 can possibly drive the deformation mechanism of the
deep mantle from a super-plastic regime (Karato et al., 1995) into a regime of power-law
1080 Chapter 15

creep (see Section 15.14.2.2). Deformation maps of olivine and other silicates have revealed
such a transition (Frost and Ashby, 1982). A deformation mechanism of this kind could
produce an apparent onset of anisotropy within D00 , which could again be a simple
manifestation of either the hotter temperature in this region or the flow regime producing
stretching of melt inclusions.

15.12.2. Diffusion: self and co-operative

Diffusion of the average density of particles and of an individual particle is referred

to as co-operative diffusion and self diffusion, respectively. Diffusion can be detected
in dilute systems by observations under a microscope, specifically by treating them
with a photochromatic dye. In this method, the quantitative measurements are made by
“forced Rayleigh” scattering experiments, in which the diffracted intensity of the light
source determines DS, the self-diffusion. Self-diffusion can also be detected via incoherent
neutron scattering. Co-operative diffusion gives rise to density changes and can be probed
by inelastic light scattering that measures the density correlation function related via the
fluctuation dissipation.
In planetary interiors, diffusion is rate-limiting in many kinetic processes.

15.13. Defects, dislocations and deformation

15.13.1. Defects

The transport processes such as creep and electrical conduction are controlled
by point defects (Wanamaker, 1994). Defects enable plastic deformation. In
crystalline lattice vacancies, interstitial atoms and impurity atoms contribute to
deformation. For a plastic (or rather, ductile) deformation, microscopic defects must
be present. The kinetics of defect creation and migration are thermally activated with
exponential temperature dependence. The macroscopic strain rate depends on
temperature.
The total energy of a material increases with the presence of defects. A tendency for
the elimination of defects constantly competes with all processes that generate defects and,
above 0 K, a dynamic equilibrium exists between the spontaneous creation and
annihilation of defects.
Defect densities can be reduced through recovery, polygonalization and division of
crystals into sub-grains (Poirier, 1985, Chapter 6). Grain growth reduces total energy
through a reduction in total energy by decreasing the grain-boundary area per unit volume.
However, highly deformed materials may develop aggregates of grains, which are strain-
free but smaller in size and crystallized at T $ Tm=2. Re-crystallization reduces the stored
strain energy. For dynamic (under external stress) crystallization, the higher the external
stress, the smaller the re-crystallized grain size.
A strong partitioning of dissolved H2O-related defect species from olivine to melt
may cause hardening effects.
Transport Properties in Deep Depths and Related Condensed-Matter Phenomena 1081

15.13.2. Dislocations

Dislocations are linear defects and account for shear strengths. Strain due to atomic
displacement is high around the dislocations, which can contribute to phase transitions,
chemical reactions, etc.
Each of the line defects or dislocations has an associated crystallographic slip
vector, designated as the Burgers vector, b, which denotes the slip (relative atomic
movement) as the dislocation moves through. The simplest form is edge dislocation, where
an extra layer of atoms in a crystal is involved. An edge dislocation is normal to its b, while
a screw dislocation is parallel to it. A screw dislocation transforms successive planes of
atoms into the surface of a helix. The strain energy of a dislocation (lbl2) is much greater
near an edge dislocation than a screw dislocation. The trace (usually planar) of the
dislocation on the surface is called the glide surface.
Thus, a combination of glide plane and b marks the dislocation slip system as
denoted below:
Screw dislocation: when the dislocation line is parallel to b.
Edge dislocation: when the line is perpendicular to b.
Mixed dislocation: when the screw and edge components occur together.
Dislocation-induced deformation generally involves both glide and climb
components. Dislocations following power-law creep is called dislocation creep. The
rate of dislocation creep in some cases may be limited by diffusion, i.e., by climb activity.
The activation energies of non-elastic properties in olivine are well known (Karato,
1989). The pressure dependence of dislocation creep in olivine up to 2.5 GPa has been
determined (see Borch and Green, 1987). The effect of pressure on creep comes both from
the activation volume, V p, and from the change in water fugacity (Karato, 1989). Borch
and Green (1987) observed a very high activation volume for creep (V p ¼ 27 cm3 mol21),
which would preclude any convective motion in the deep upper mantle. They suggested
that the activation volume will be significantly reduced under high pressures.
To unravel the ductile flow in the mantle at depth, the effect of pressure on the
generation and velocity of dislocations needs to be investigated (Ando et al., 1997).
The component of thermo-remanent magnetization (TRM) that resides in stressed
regions surrounding dislocations has the highest stability because the highly strained
regions behave much as single-domain particles. (Note: The magnetic quiet belt (Green,
1976) in Africa may be caused by the upwelling of the Curie point isotherm.)

15.13.3. Dislocation recovery

Dislocation creep is often seen to be controlled by dislocation recovery. Hence,

dislocation recovery experiments may help to determine the rate of deformation due to
dislocation creep (Bai and Kohlstedt, 1992). Karato et al. (1993) conducted dislocation
recovery experiments in olivine up to 10 GPa and 1,5008C (corresponding approximately
to the depth of 300 km in the upper mantle) using a multi-anvil apparatus. Micro-structural
observations indicate that dislocation recovery occurs through the climb of edge
dislocations and hence is controlled by diffusion. They observed that, during annealing,
1082 Chapter 15

a decrease in the density of homogeneously distributed dislocations and an organization of

dislocations into sub-boundaries occur. After annealing, the dislocations are often curved
out of their glide plane and dislocation climb becomes significant during static recovery.
From SEM pictures, the initial and final dislocation densities (ri and rf,
respectively) in olivine were measured and the rate constant, A, was calculated through:
A ¼ ð1=rf 2 1=ri Þ=t; ð15-15Þ
where t is the annealing time.
The kinetics of dislocation recovery depend on the slip systems and on the type of
dislocation. Karato and Ogawa (1982) analysed the kinetics of recovery of homogeneously
distributed dislocations assuming second-order kinetics, namely:

dr=dt ¼ 2Ar2 ;
where r is the dislocation density and A is the rate constant, which depends on T and P as
 
A ¼ k0 exp 2 ðEp þ PV p Þ=RT ;

where E p is the activation energy and V p is the activation volume. The dislocation
recovery is controlled by the mobility of dislocations. Thus, the rate constant, A,
is proportional to dislocation mobility (Karato and Ogawa, 1982). Using the activa-
tion energy, E p ¼ 290 kJ mol21 (at T ¼ 1,5008C) (Yan, 1992), the mobility of
dislocations (i.e., the diffusion coefficient of the slowest diffusion species of olivine) is
seen to be essentially constant with depth in the deeper portion of the upper mantle for
V p , 6 cm3 mol21 (Karato et al., 1993). The results of an investigation by Karato and
his co-workers were employed in determining the value of V p as 6.1 ^ 0.2 cm3 mol21
(Fig. 15.9, read the caption; from Karato et al., 1993).
H2O also increases the rate of dislocation recovery. Hydrogen-bearing olivine
shows a weakening of strength due to an increase in the rates of dislocation climb.
Hydrogen within the dislocation core may (i) enhance the ability of the dislocation to glide
or climb or (ii) reduce the number of kinks and jogs associated with dislocation
(e.g., Mackwell et al., 1982). The flow law, the dependence of strain rate or stress as a
function of temperature, changes with water content in olivine.

15.13.4. Deformation

Minerals deform by dislocation glide and climb. Some deformation mechanisms

depend on the local electronic structure around the large dislocation core (e.g., Poirier and
Price, 1999). In the deformation of an aggregate of low-symmetry minerals, where some
orientations deform more easily than others to maintain stress equilibrium, heterogeneous
deformation occurs at intra-crystalline scale.
With increasing temperature and decreasing deviatoric stress, the deformation
mechanism may change from dislocation glide to power-law creep. Deformation
mechanism may also be influenced by chemical composition.
Transport Properties in Deep Depths and Related Condensed-Matter Phenomena 1083

Figure 15.9. The pressure dependence of dislocation recovery kinetics. The rate constant k in the recovery kinetic
equation, dr/dt ¼ 2 kr 2, is plotted as a function of pressure at T ¼ 1,5008C. The room pressure data obtained at
the same oxygen fugacity buffer (Ni–NiO) and oxide activity as the high pressure runs are used to estimate the
activation volume. The errors are due primarily to the heterogeneity of dislocation density. A linear regression was
performed to determine the activation volume V p, assuming that it is independent of pressure. The result
yields V p ¼ 6.1 ^ 0.2 cm3 mol21. Earlier results by Karato and Ogawa (1982) are also shown for comparison.
(from Karato et al., 1993, q 1993 American Geophysical Union).

15.13.4.1. Olivine deformation

Kohlstedt et al. (1976) determined a relation between dislocation density and
applied differential stress for olivine single crystals in steady-state deformation
experiments. The number of dislocations (dislocation density) generated on the limited
number of slip systems, which are a function of the resolved shear stress (t) on those,
greatly depends on the crystallographic orientation with respect to the stress axis (Wang
et al., 1988). The sine function of the resolved shear stress on the slip plane,
t ¼ 12 ðs1 2 s3 Þ sin 2u; where u denotes the angle between s1 direction and slip plane
normal, makes the orientation effect less significant. Therefore, orientation effect on
dislocation density is usually small. (Note: Dislocation piezometer established by
deformation experiments under relatively low confining pressure is available.)
In general, multiplication and recovery of dislocations govern the creep of olivine,
calcite and quartz as well as metals within the dislocation creep regime. Perhaps the
absolute value of dislocation density corresponding to the applied differential stress also
depends on the pressure.
1084 Chapter 15

15.13.4.2. Deformation: single crystal to polycrystalline mass

In the deformation process, several planer defects involving stacking faults, twins
and grain boundaries may be operative.
Plastic deformation of polycrystalline materials is governed by an interplay
between applied stress and the deformation within individual grains through dislocation
movement. Dislocations in the crystals leads to deformation. Pressure causes the
deforming grains to rotate and anisotropism develops in the elasticity or electrical
conductivity of the polycrystalline body. Orientational changes in a single grain is
commonly studied with electron (transmission or scanning) microscopy.
The synchrotron-based approach of Margulies et al. (2001) allows the rotation path
of the deformed material to be studied from the undeformed material. A focussed high-
energy X-ray beam can monitor the orientation changes of lattice reflections from simple
grains within a deforming polycrystal. The width of the reflections corresponds to the
spread of orientations due to deformation in a grain. The polycrystal plasticity theory can
predict the rotation path and the deformation behaviour of a polycrystalline mass.
After transient creep, olivine single crystals change to polycrystal due to a dynamic
recrystallization at , 50% strains and creep is approximately in a steady state. Activation
energy for steady-state creep of polycrystalline olivine is 540 ^ 40 kJ mol21, suggesting a
good agreement with previous creep experiments of polycrystals (500 – 590 kJ mol21).

15.14. Diffusion, creep and viscoplastic deformation

Many kinetic processes in Earth and planetary interiors such as solid-state creep
(leading to mantle convection), an elastic deformation (contributing to seismic
attenuation), phase transformations and chemical mass transfer reactions are rate-limited
by the process of diffusion.
Seismic anisotropy in the lower mantle is explained, as diffusion creep, to be the
dominant deformation mechanism. While above the CMB the anisotropy (patchy) may
arise from dislocation creep.

15.14.1. Diffusion

During volume diffusion, the volume fraction of reacted components, c, is given by

dc D
c ¼ 2; ð15-16Þ
dt r

where r is the typical grain radius and D is the diffusion coefficient (Ahrens and Schubert,
1975). c is obtained by integration over time.
Temperature, pressure, crystallographic orientation, composition and oxygen
fugacity were found to have significant effects on diffusivity. The diffusion may be
more sensitive to changes in the lattice environment of a cation when the structure is in a
compressed state. Therefore, diffusion rates of large cations such as Ca and Fe (e.g., in
garnet) show a stronger dependence on composition than small cations (e.g., Mg) at a fixed
Transport Properties in Deep Depths and Related Condensed-Matter Phenomena 1085

pressure (Chakraborty and Ganguli, 1998). The cation diffusion during mantle convection
even controls the sharpness of mantle discontinuities and the discontinues are likely
to be diffusion-controlled in such regions (Bina and Kumazawa, 1993; Solomatov and
Stevenson, 1994).
The data on volume diffusion in silicates show that Si and O diffuse much more
slowly than divalent and trivalent cations. Thus, the Si – O groups provide a static
framework through which the cations diffuse (Freer, 1981). If the volume diffusion is too
slow, the reaction rate will be limited by grain-boundary diffusion.

15.14.1.1. Mg, Fe diffusion: olivine, pyroxene and garnet

The diffusion rates of Mg2þ and Fe2þ in garnet and olivine (e.g., Morioka and
Nagasawa, 1991) have been experimentally determined. Of these, the diffusion rate of
Mg2þ in garnet (DMg,Gt) is the lowest. The slowest diffusion rate is likely to limit the
reaction rate. Therefore, the diffusion coefficient, DMg,Gt, for this process may be taken as
the upper bound on D in equation (15-16).
Diffusion rates, when not determined experimentally, may be estimated from
observed compositional gradients or from measurements at a single temperature. The
minimum value of these estimates may be taken as the lower bound of the diffusion
coefficient. Such a minimum for the diffusion of Al3þ in orthopyroxene was estimated
from an analysis of compositional gradients in natural assemblages (Smith and Barron,
1991). The estimated rate depends on assumptions of cooling rate and pressure. Cation
diffusion in olivine and garnet is generally faster than in pyroxene.
The diffusion of Ca2þ in garnet has been estimated by numerical simulation only at
1,2008C (Chakraborty and Ganguli, 1992) and is found to be faster than that of Al3þ in
orthopyroxene at that temperature. The mobilities of Ca2þ and Fe2þ are held to be
comparable in garnet and pyroxene. These considerations reinforce the view that DAl,Opx
gives a reasonable lower bound on the diffusion coefficient D in equation (15-16). The two
bounds on D are

D # DMg;Gt ¼ 2:8 £ 1028 exp½2ð270 kJ þ 3:2 £ 1026 PÞ=ðRTÞ ð15-17Þ

D $ DAl;Opx ¼ 1:1 £ 1025 exp½2400 kJ=ðRTÞ ð15-18Þ

where P is in Pa.
The diffusion rates in plagioclase are generally less than in pyroxene (Smith and
Brown, 1988) and inter-diffusion of CaAl – NaSi in plagioclase is several orders of
magnitude slower than the lower bound on D (Grave et al., 1984).
Experiments on Fe diffusion in dunite show that, under conditions of chemical and
mechanical equilibrium, CO2 exists in isolated pores and has no effect on diffusional
transport (Watson, 1991).

15.14.1.2. Grain-boundary diffusion

During metamorphism, grain-boundary diffusion apparently governs nucleation
and growth mechanisms but, in the absence of an inter-granular fluid, grain-boundary
1086 Chapter 15

diffusion may be a less efficient process (Ahrens and Schubert, 1975). However, grain-
boundary diffusion coefficients can be several orders of magnitude greater than those for
volume diffusion although this may not be strictly so in cases where mass transport by
grain-boundary diffusion is confined to a thin layer.
In the lower mantle, the grain-boundary effects may play an important role. It might
entail grain-boundary diffusion on the coating of the perovskite and ferro-periclase grains
with another phase such as a hydrous mineral. Transport along the grain boundary could
very well be the controlling mechanism for diffusion (Ita and Cohen, 1998). Assuming that
diffusion in perovskite is extrinsically controlled and has the same vacancy concentration
as periclase, estimates by Wright and Price (1993) of the migration energy in perovskite
indicate that its effective diffusion coefficient will be six orders of magnitude smaller than
that of periclase (Ita and Cohen, 1998).

15.14.2. Creep

The type of rheology of linear creep can be expressed with a stress exponent n ¼ 1
and the grain-size exponent m ¼ 2 or 3. Diffusion-controlled creep has a linear-creep
constitutive relation. At high temperature, diffusion through the crystalline lattice
dominates and the grain-size exponent m ¼ 2 (Nabarro – Herring creep). At lower
temperatures, grain-boundary diffusion becomes more important and m becomes equal to 3
(Cobble creep). Viscoplastic deformation (creep) of crystalline materials under constant
stress involves the motion of a large number of interacting dislocations. Analytical
methods and “dislocation dynamics” simulations have proved to be very effective in the
study of dislocation patterning and have led to macroscopic constitutive laws for plastic
deformation.
Diffusion creep. Diffusion creep is grain-size sensitive, involves diffusion of all
species along all available paths. It does not lead to significant crystal preferred orientation.
The rate equation is
e_ ¼ ð13:3sVDeff =RTd 2 Þ
where d is the grain size, V the molar volume and Deff a weighted mean of the bulk
diffusivities of atoms (e.g., Mg and O),
Dislocation creep. Dislocation creep is a power-low function of shear stress and
leads to strong preferred crystal orientation at large strains.
The rate law is
e_ ¼ Asn exp½2ðE þ PVÞ=RT;
with n ¼ 4:0; E ¼ 327 kJ/mol and A ¼ 2:4 £ 10224 Pa23 s21 for ferri-periclase at low
pressure (Stretton et al., 2001).

15.14.2.1. Creep rate

The creep rate at steady state is given by an Arrhenius equation:
 
1 ¼ Aðs=mÞn exp 2ðEp þ PV p Þ=RT ;
Transport Properties in Deep Depths and Related Condensed-Matter Phenomena 1087

where 1 is the creep rate, A is a rate constant, s the differential stress, m the shear modulus,
n the stress component, E p the activation energy, V p the activation volume, R the gas
constant and T the absolute temperature. This equation also shows that the creep rate
exponentially decreases with increasing pressure in spite of the constant differential stress
condition (e.g., Karato, 1989).
Based upon Orowan’s equation,
1 ¼ Obv;
˛
the creep rate ð1Þ
 is determined by these parameters, namely dislocation density (O ˛ ),
Burgers vector (b) and dislocation velocity (v). Much of the exponential change of the
creep rate with pressure would be due to changes of dislocation density and dislocation
velocity because the length of the Burgers vector is expected to change little.
The activation energy for parabolic creep is 670 ^ 20 kJ mol21 and stress
dependence is expressed as a power law of s 2.5. In the cell-wall climb model, activation
energy is approximated to be the sum of oxygen diffusion and jog-formation energy. Jog-
formation energy is thus calculated to be , 167 kJ mol21 olivine (Toriumi et al., 1984).

15.14.2.2. Dislocation (power-law) creep ! diffusion creep

Miguel et al. (2001) have reported acoustic emission measurements on stressed ice
single crystals, the results of which indicate that dislocations move in a scale-free
intermittent fashion. They find that dislocations generate a slowly evolving configuration
which co-exists with rapid collective rearrangements.
Dislocation creep is often referred to as “power-law creep” (see Section 15.12.1).
The power-law creep and the diffusion-controlled climb of edge dislocations control the
creep rate (see Poirier, 1985). The density of dislocations in equilibrium in their mutual
stress field varies as s 2 and, if the climb velocity of dislocations is assumed to vary
linearly with stress, the creep rate should vary as s 2 (Poirier, 1985). It is natural that the
stress dependence of climb-controlled dislocation creep should follow a power law with
exponent 3 (Poirier and Duba, 1997). The activation energy, E*, is stress independent.
Power-law dependence of a thermally-activated physical property is an Arrhenius
process.
There is the possibility of a transition from dislocation creep to diffusion creep in
lithospheric shear zones to deep in the upper mantle. The deformation mechanism in the
upper mantle may change from dislocation creep in the shallower upper mantle to diffusion
creep in the deep upper mantle (Karato and Wu, 1993). Diffusion creep is, however, likely
to play an important role under some upper-mantle conditions (Karato, 1992).
Measurements of the effect of pressure on diffusion or dislocation (diffusion-controlled)
recovery are more tractable than measurements of creep properties under high pressure.
Estimates of pressure effects on diffusion have a direct application to the estimation of
viscosity due to diffusion creep. Also, dislocation creep is sometimes controlled by
dislocation recovery. Karato et al. (1986) showed that, under dry (water-free) conditions,
volume-diffusion creep (Nabarro – Herring creep) (see Section 15.14.2) dominates over
grain-boundary diffusion creep. Dislocation creep results in the preferred orientation of
minerals and a resultant seismic anisotropy but diffusion creep does not (Karato, 1988).
1088 Chapter 15

Thus, the seismic anisotropy in the deep upper mantle will be much smaller than that in the
shallow upper mantle if in such a transition deformation mechanism occurs (Karato et al.,
1993). However, this transition may be responsible for the Lehmann discontinuity at
200 –300 km.
In olivine climb, (recovery)-controlled creep operates in the (010) [100] slip system
under some thermodynamic conditions (Bai and Kohlstedt, 1992). Creep due to the softest
slip system (010) [100] might be responsible for the low strain creep such as that involved
in post-glacial rebound. Water increases the dislocation mobility.
Edge or screw dislocations (Yan, 1992) occur through climb and involve the
diffusion of atoms (Karato and Ogawa, 1982). The diffusion of the slowest diffusion
species is the rate-controlling process for dislocation climb. The atom paths in change of
shape by Nabarro – Herring creep and dislocation climb are shown in Fig. 15.10. In olivine,
the species may be oxygen (Karato and Ogawa, 1982) or silicon (Houlier et al., 1990). The
depth variation of climb mobility of edge dislocations in the (010) [100] slip system in
olivine in the Earth for several values of activation volume V p is shown in Fig. 15.11
(read the caption). The activation volume obtained is small (V p ¼ 6 ^ 1 cm3 mol21).
This is possibly because of an interstitial mechanism of diffusion for the relevant transport-
limiting species (oxygen or silicon). In general, ionic diffusion in olivine lattice has
small activation volumes and hence V p would hardly change significantly with depth in
the deep upper mantle. However, Borch and Green (1987) suggested that pressure does
have an effect.
Since the volume change of a crystal associated with the formation of an interstitial
atom is significantly smaller compared with that associated with the formation of a
vacancy site, it is expected that diffusion by an interstitial mechanism has a significantly
smaller activation volume than diffusion through a vacancy mechanism.
Thus, the dominant mechanism of deformation may change from dislocation creep
in the shallow upper mantle to diffusion creep in the deep upper mantle.

Figure 15.10. Atom paths (dashed lines) envisaged in changes of shape (a) by Nabarro–Herring creep and (b) by
dislocation climb (Nabarro creep) (Karato et al., 1993, q 1993 American Geophysical Union).
Transport Properties in Deep Depths and Related Condensed-Matter Phenomena 1089

Figure 15.11. The depth variation of climb mobility of edge dislocations in the (010)[100] slip system in olivine
in the Earth for several values of activation volume V p. Dislocation mobility is normalized to the value at T ¼ 1.
The depth variation results both from the effects of pressure and temperature. In this calculation, a temperature
distribution for a typical old sub-oceanic mantle is assumed (e.g., Mercier and Carter, 1975). Activation volume
V p is assumed to be independent of pressure (Karato et al., 1993, q 1993 American Geophysical Union).

15.14.2.3. Dislocation creep and spinel deformation

High-pressure, high-temperature, large-strain shear deformation experiments were
performed by Karato et al. (1998) on g-spinel (ringwoodite) with different grain sizes.
The synthesized samples were first pressurized to 16 GPa (in 5 h). Then the
temperature (T) was raised (in , 10 min) to 1,600 K (T/Tm ¼ 0.61, where Tm is the melting
temperature , 2,600 K; Ohtani and Kumazawa, 1981) and to 1,400 K (T/Tm ¼ 0.54).
Under these conditions, (Mg, Fe)2SiO4 olivine (Fe/(Fe þ Mg) ¼ 0.4) transforms to spinel
structure (Katsura and Ito, 1989).
The sample strain was measured from the rotation of a strain marker in each
experiment. Most of the deformation was simple shear and homogeneous. In a similar
study, Karato and Rubie (1997) found that the mode of deformation was predominantly
stress relaxation at differential stresses of 1 –3 GPa having average strain rates of
1025 –1024 s21.
In small grains, Karato et al. (1998) found numerous [110] stacking faults,
characteristic of spinel structure (Vaughan and Kohlstedt, 1981). In large grains, a high
density of dislocations (, 1,013 –1,014 m22) is observed. Also, four-grain junctions are
seen, which are expected to occur during grain-switching events involving grain-boundary
sliding, characteristic of “superplasticity” (Ashby and Verrall, 1973). These results show
that direct experimental studies of plastic deformation of deep-mantle materials can now
become possible. The evidence suggests a grain-size-sensitive creep mechanism (diffusion
creep or “superplasticity”) to have operated on the sample. However, in grains larger than
, 3 mm, the deformation is largely by dislocation creep.
1090 Chapter 15

Most of the dislocations are straight along k110l while others are straight along
k100l. In nearly stoichiometric oxide and germanate spinels, these dislocations have
Burger vectors b ¼ (1/2)k110l{111} slip system. The germanate spinels are good
analogues of ringwoodite for dislocation creep. The creep strength of a spinel-rich layer
is significantly higher than that of an olivine-rich layer if both of them are to deform by
dislocation creep (Dupas-Bruzek et al., 1999).

15.14.2.4. Slip systems

Isostructural structures may show different slip systems, e.g., slip systems of
magnesite are entirely different from calcite and halite has a different slip system from
periclase or galena. In halite at low temperature, the {110}k1Ī0l slip mode is soft and the
{100}k011l and {111}k110l slip modes are much harder. With the hard slip system, the
{100} and {111} modes are 10 times harder than the {110} soft mode (Merkel et al., 2002).
At high temperature, many slip systems are active, including {111}k110l. Polycrystal
plasticity simulations suggest that {110}k1Ī0l is the only significantly active slip system. In
alkali halides, there is a high plastic anisotropy at room temperature which decreases with
increasing temperature.

Yield strength (sy) and slip system. The maximum uniaxial stress supported by a material
is determined by its yield strength, i.e., t # sy ; where sy is the material yield strength. The
yield strength of a polycrystalline material depends on the possible slip systems and their
critical resolved shear stress (CRESS) and also on its texture. Increase in yield strength
may be caused by “hardening”, resulting from an increase in the density of dislocations
(Poirier, 1985). Yield stress depends on the stress history of the polycrystalline samples
and is caused by an increase in the dislocation density.

15.14.2.5. Creep, diffusion rate and conduction

In the deep interior, the solid-state creep, accounting for the mantle convection, and
electrical conduction (serving as a probe for the structure and composition of the Earth) are
physical processes that depend on diffusion. The diffusion rates also determine the rates of
chemical processes such as the mixing of mantle heterogeneities, mineral transformations
and element partitioning during partial melting.

H diffusion and conductivity. The rate of hydrogen diffusion in olivine indicates that
100 –1,000 hydrogen ions per 106 Si atoms (6 –60 ppm water) may elevate the electrical
conductivity of the mantle by one to three orders of magnitude (Karato, 1990). Such H-
induced conductivity increase may explain the observed enhancement of conductivity
near the top of the asthenosphere (, 90 –150 km) without invoking partial melt at these
depths (Karato, 1990). Again, to match magnetotelluric sounding results, the high
conductivity pathway may be ascribed to hydrogen diffusion (Farber and Williams,
1996).
Hydrogen-bearing olivine shows weakening of strength due to an increase in the
rates of dislocation climb. Hydrogen within the dislocation core may (i) enhance the ability
Transport Properties in Deep Depths and Related Condensed-Matter Phenomena 1091

of the dislocation to glide or climb or (ii) reduce the number of kinks and jogs associated
with dislocation (e.g., Mackwell et al., 1982).

15.14.3. Diffusivity and viscosity: Stokes – Einstein diffusivity

As discussed in Section 11.3.1, the diffusion – viscosity relation via the Stokes–
Einstein (see equation (11-1)) is

Dh ¼ kB T=2pa: ð15-19Þ

This equation from the theory of Brownian motion gives an exact relation between the
diffusion coefficient D of a Brownian particle of diameter a and the viscosity h of the
surrounding liquid; kB is the Boltzmann constant. Simulation studies on the hard-sphere
liquid demonstrate that the accuracy of the Stokes –Einstein relation holds good even at
high pressures where the free volume becomes small.
Applications of Stokes – Einstein diffusivity (extrinsic) in the estimation of
viscosities at elevated pressures have been documented by Poe and Rubie (1998).
The direct calculation of shear viscosity h from the first-principles simulations is
theoretically possible but this has not yet been achieved for any liquid because of the very
long simulations that are needed to obtain useful statistical accuracy.
The shear viscosity, h; is connected with the self-diffusion coefficient, D; which can
be calculated via the Einstein relation:

kDrðtÞ2 l ! B þ 6Dt; ð15-20Þ

where B is a constant. The mean square distance, kDrðtÞ2 l; travelled by any atom in finite t
becomes linear in long times.

15.14.3.1. Silicate melts: O, Si diffusion

There is a negative pressure dependence of the viscosity of silicate melts and an
inverse relationship is likely to exist between diffusivities and viscosity. Simulation studies
using molecular dynamics (Angell et al., 1982), etc., have predicted a maximum in the
diffusivities of several components including Si.
It is seen that, at high viscosities and sufficiently low temperature, the transport of
oxygen within the melt can be decoupled from the constraints of viscous flow (see the
review by Chakraborty, 1995). Again, the estimation of melt viscosity from oxygen
diffusivity carries restrictions which distinguish it from silicon. Such observations can be
described in terms of extrinsic and intrinsic diffusivities in silicate melt (see discussions by
Chakraborty (1995) and Dingwell (1998)).
Si diffusivity data can be used under all P – T – X conditions for estimation of
viscosity. Oxygen diffusivities may be used at high temperatures and low viscosities; and
other component diffusivities should be used only where extrinsic behaviour is confidently
known.
1092 Chapter 15

15.14.4. Intrinsic and extrinsic regimes

A transition from an extrinsic to an intrinsic diffusion mechanism is marked by a

kink in the Arrhenius plot (change in activation energy) of diffusion coefficient. An
absence of the kink would suggest that the measurements are in a single regime. The low-
temperature side of the measured data is the extrinsic regime. In the extrinsic regime, the
diffusivities (vacancy concentrations) at a given major element composition are functions
of some component activities in addition to P and T.
If the vacancies predominantly responsible for the transport process are thermally
generated, the process is termed intrinsic and the vacancy concentrations are P, T
dependent. In the intrinsic regime, the activation energy is made up of two parts: an energy
of formation of the thermally generated vacancies and an energy of migration of these
vacancies to the neighbouring sites. In the extrinsic regime, the activation energy consists
only of the migration energy term and is hence smaller than in the intrinsic regime.
The intrinsic diffusivities are not related to the viscous flow mechanism and cannot
be used to obtain viscosity data. Intrinsic diffusivities are always higher than extrinsic
diffusivities and exhibit lower activation energies. Their pressure dependences often have
opposite signs.
The minimum defect formation energies obtained from model calculations are
, 800 kJ mol21 (Ottonello et al., 1990). A very conservative estimate of intrinsic vacancy
formation energy of , 550 kJ yields the concentration of vacancies from the relation:

Xv ¼ e2DHv =2RT : ð15-21Þ

At 1,3008C, this yields XV of , 2 £ 10210 and, at 1,0008C, XV is even lower, viz.

, 1 £ 10212 mol fraction.
Intrinsic diffusion behaviour in silicates is unlikely ever to be observed unless (i)
intrinsic vacancy formation energies are unusually low (a few kJ), (ii) temperatures are
high (. 1,3008C) and (iii) the crystals are absolutely pure (Chakraborty et al., 1994). The
absolute values of diffusivities may depend on concentrations of impurities, at least for
cases where diffusion is by a simple vacancy mechanism.
Defects can be formed thermally (instrinsic regime) or by the presence of impurities
(extrinsic regime). Extrinsic vacancies on Mg sites are introduced by the substitution of
two trivalent (M3þ) cations for two Mg atoms. Vacancies on O sites are formed by the
replacement of two Mg atoms with two monovalent (M1þ) cations. In the intrinsic regime,
the defect concentration is
 
DGf
Ci ¼ exp 2 ; ð15-22Þ
WKb T

where DGf is the energy of formation of the vacancy pair and W is the solubility factor for
polyatomic materials.
Ionic conductivity measurements indicate that Mg diffusion is controlled by
impurities (Sempolinski and Kingery, 1980) whereas O diffusion is intrinsic in nature in a
nominally pure sample (Yang and Flynn, 1994). At ambient pressure, movement of O is
Transport Properties in Deep Depths and Related Condensed-Matter Phenomena 1093

accommodated by grain-boundary diffusion and is faster than Mg bulk diffusion.

Therefore, Deff ¼ DMg.
The intrinsic Mg defect concentration is always much less than the extrinsic
concentration. Thus, Mg diffusion in the lower mantle will be in the extrinsic regime.

15.15. Cation (Mg, Fe and Ni) diffusion in olivine: a ! b –g transition

A knowledge of cation-diffusion rates in olivine would help a better understanding

of the creep in the Earth’s upper mantle and the closure of ion-exchange geothermometers
involving olivine. Compositional zoning profiles in olivine may present clues to the
cooling rates and thermal histories of terrestrial rocks and meteorites.
Mg tracer diffusion coefficients in crystals of forsterite (Fo100) and San Carlos
olivine (Fo92) between 1,000 and 1,3008C were determined by Chakraborty et al. (1994).
Mg tracer diffusion in a nominally pure forsterite is found to be anisotropic
(D11c . D11a . D11b) and is a function of f O2. This diffusion mechanism of Mg
in forsterite differs from that in Fe-bearing olivine (over the same range of f O2).
Pressure dependence of Mg tracer diffusivity in forsterite measured to 10 GPa in a
multi-anvil apparatus yields an activation volume of , 1 –3.5 cm3 mol21 (Chakraborty
et al., 1994). The values of activation volume for inter-diffusion in olivine group of
minerals are:

Activation volume Inter-diffusion Authors

(cm3 mol21)

,2 Fe– Mg Bertran-Alvarez et al. (1992)

,2 Fe– Mg Chakraborty et al. (1994)
5.5 Fe– Mg Misener et al. (1974)
21 Fe diffusion in Fe2SiO4 Takeda (1990)

If it is assumed that Fe always occupies its preferred M-site in the olivine structure,
then the f O2 dependence of diffusivity (as well as vacancy concentration), controlled by
Fe2þ –Fe3þ equilibrium, would be proportional to approximately f O1/6 2 for a vacancy
mechanism. At a relatively reducing condition, the concentration of vacancies created by
Fe2þ –Fe3þ equilibria is overwhelmed by other vacancies, which may be intrinsic defects
or vacancies related to the presence of other impurities so that the diffusivity becomes
independent of f O2.
EPR studies of forsterite showed that Fe3þ may occur in the Si site, accompanied by
3þ
Fe in the M-site for charge balance (e.g., Nakamura and Schmalzried, 1983). However,
the defect involving Fe would also be an interstitial site.
Diffusion coefficient for high-P mantle phases (e.g., wadsleyite, ringwoodite
and silicate perovskites) are poorly known because of experimental difficulties at the
P – T range at which these phases are stable. Nevertheless, a few experiments were
performed at high pressure for both tracer diffusion and chemical diffusion (e.g.,
Chakraborty and Rubie, 1996). In olivine, the Mg –Fe inter-diffusion coefficient has been
1094 Chapter 15

determined in pressures up to 7 GPa and the activation volume has been determined as
V p ¼ 2.2 cm3 mol21 (Bertran-Avarez et al., 1992).
Farber et al. (1994), from a study on b- and g-phases of (Mg, Ni)2SiO4,
concluded that Mg –Ni inter-diffusion is three orders of magnitude higher in the
a-olivine structure. However, this result on Ni-bearing orthosilicate may not be
adequately good for modelling the diffusion behaviour in Fe-bearing mantle
compositions.
Chakraborty et al. (1999) determined the rates of cation (Mg, Fe and Ni) diffusion in
a- and b-phases of (Mg, Fe)2SiO4 at 9– 15 GPa (and 1,100 – 1,4008C) for compositions
that are relevant to the Earth’s mantle (Xmg ¼ Mg/(Mg þ Fe) ¼ 0.8– 1.0). The f O2 during
the experiments was in the range 1028 –1029 bar (corresponding about 15 GPa). Knowing
the composition and volume of the a-phase, the free energy of the system is minimized to
obtain the equilibrium oxygen fugacity (for details, see Dohmen et al., 1998). For the b-
phase at 15 GPa, they obtained diffusion coefficients of 7(^1) £ 10215 m2 s21 at 1,2008C
and 3(^ 1) £ 10215 m2 s21 at 1,1008C. These results gave an estimate for the activation
energy of 145 kJ mol21. This is much lower than what was obtained for the a-phase of
226 kJ mol21 (Chakraborty, 1997). Diffusion coefficients change by about an order of
magnitude in the range Xmg ¼ 0.86 – 0.98 at 12 GPa and 1,4008C. The general trend is that,
at all pressures, the diffusion rates increase as the composition becomes richer in the
Fe2SiO4 component.
In wadsleyite (b phase), diffusion is seen to be one or two orders of magnitude
faster than in olivine, depending on the temperature. Nickel diffusion is much faster in
wadsleyite than in olivine. However, the diffusion depends strongly on the composition
(Xmg) of the latter phase.
The rate processes that are controlled by the diffusion of cations such as Mg, Fe
and Ni should be at least one order of magnitude faster in the transition zone (410 –
660-km depth and T ¼ 1,500 – 1,6008C) than at shallower depths in the upper mantle
(Deng et al., 1998).

15.15.1. a , b , g transitions and upper-mantle rheology

The transitions between modified spinel (b-) and spinel (g-) polymorphs of
(Mg, Fe)2SiO4 have been studied between 15 and 20 GPa (800 – 9508C) by Rubie and
Brearley (1994). This rheological change associated with this phase change may trigger the
deep-focus Earthquakes.
The b ! g transformation involves grain-boundary nucleation and interface-
controlled growth, while the g ! b transformation occurs by a shear mechanism. Such
phase boundaries are believed to coincide with the seismic discontinuities located at depths
of 410 km, possibly 520 and 670 km.
Localized shear may initiate shear instabilities for the phase transformations in
subducted slabs. In the cold subduction zones, transfer motions occur under near-
equilibrium conditions at temperatures of 1,500 –1,6008C.
Olivine ! spinel transition can occur as a martensitic-like mechanism involving
the synchro-shear mechanism. Martensitic or shear transformations are most likely to
Transport Properties in Deep Depths and Related Condensed-Matter Phenomena 1095

cause a reduction in strength. In addition to the change in crystal structure, the migration of
partial dislocation also results in plastic deformation. The driving force for this migration
is the change in Gibbs free energy associated with the phase transformation.
For b ! g transformation, the mechanism involves the migration of 1/2
[1̄01](010) partial dislocation whereas g ! b transformation occurs by the migration
of 1/4[112](110) partial dislocations (Madon and Poirier, 1983). It has been shown
that g-Mg2SiO4 transforms to b by shear mechanism at 15 GPa and 9008C (Brearley
et al., 1992).
With increasing depth in the interior of subducting slabs, the expected sequence of
the stability fields of the Mg1.8Fe0.2SiO4 polymorph is: a ! a þ g ! a þ b ! b !
b þ g ! g (Akoagi et al., 1989). In the third stage, transformation g ! b takes place.
In (Mg, Fe)2SiO4, a ! b transition transformational faulting occurs and deep-focus
earthquakes by this mechanism are as likely as from a ! g transformation since this
involves relatively large volume change and high latent-heat production (Kirby et al.,
1991). Again, g ! b transition, by shear during transformation faulting, could enhance the
transformation plasticity and the faulting process.
In the transition zone in the Earth’s mantle, the g ! b transformation will occur
primarily in rising plumes. If material fluctuates periodically across the b– g phase
boundary as a result of instability of mantle flow, the transformation between these two
phases may occur. If shear mechanism operates during g ! b phase transformation,
the changing rheology of the rising plume may cause mechanical decoupling across the
b –g phase boundary.

15.15.2. Chemical diffusion in the slab and transition zone

In the transition zone, a high rate of chemical diffusion will rapidly mix the
chemical heterogeneities generated by the subducted oceanic lithosphere. Chemical
diffusion acts on the scale of metres in the transition zone, as opposed to centimetres in the
shallow upper mantle.
By mantle convective mixing, the subducted oceanic crust is thinned out from its
initial thickness of 6 km until it is thin enough for diffusion to destroy its distinctive
chemical signature. In the transition zone, the chemical diffusivity is three orders of
magnitude higher and, as a result, the mixing rate becomes faster than in the upper mantle.
Heterogeneities of less than about a metre thickness in the transition zone will be rapidly
destroyed by diffusive homogenization.
A high concentration of water has been reported in the transition zone (Nolet
and Zeilhuis, 1994) associated with subduction. High chemical diffusivity of the b-phase
and spinel (Farber et al., 1994) provides a mechanism for homogenization of volatiles
from the slab into the transition zone. The higher chemical diffusivity could also affect
the rheology of transition-zone material. Mantle-rock flow by thermally activated creep
and high diffusivity of incompatible elements and volatiles would reduce the viscosity but
the transition zone is known to have less viscosity than the shallow upper mantle.
However, viscosity would tend to increase with the presence of excess garnet
(Karato, 1989).
1096 Chapter 15

15.15.2.1. Homogenization rate in mantle

Homogenization of mantle heterogeneities (chemical mixing) and mineral
transformations involving a Mg – Fe exchange will therefore occur considerably faster in
the transition zone than at depths of , 410 km (Chakraborty et al., 1999).
Chakraborty et al. (1999) argued that, in subducting slabs where temperatures are
relatively low, the enhancement in homogenization rates should be even greater because of
the relatively low activation energy for diffusion in wadsleyite. For example, in a coarse-
grained mantle at 1,4008C, where volume diffusion dominates, the diffusive-mixing time
scale is given by

x2 =D , t; ð15-23Þ

where D is diffusion coefficient, x is distance and t is time.

For homogenization of a heterogeneous volume of , 1 m in radius, the time
required is , 50 my in the a-part and , 1 my only in the transformed b-part. In colder
mantle (1108C), the difference in time for homogenization is more drastic. In a mantle
bearing b-phase, the volume will be homogenized in 10 my, whereas in the a-phase-
bearing region it will take up to 2 by.
The data of Chakraborty et al. (1999) imply that the 410-km discontinuity should be
thinner in regions of upwelling than in regions of downwelling.

15.15.2.2. Diffusion in lower mantle: MgO (Ita and Cohen, 1998)

In the lower mantle, diffusion is thought to be the dominant deformation
mechanism (Li et al., 1996). Indirect evidence for a diffusion-dominated deformation
process in the lower mantle is obtained from the analysis of post-glacial rebound
(Wu, 1996), convection simulations (van der Berg and Yuen, 1996) and studies of the
analogue materials (Li et al., 1996). If the deformation process in MgO (periclase) under
lower-mantle conditions is bulk diffusion, then the viscosity, h; will be Newtonian
(i.e., strain rate depends linearly on stress) and can be calculated using the Nabarro–
Herring expression:

d 2 Kb T
hm ¼ ; ð15-24Þ
Deff Vm

where d is the characteristic grain size, Vm is the molecular volume and Deff ¼ 2DMg
D0 =ðDMg þ D0 Þ is the effective diffusion coefficient for mass transport in MgO (Poirier,
1985). In the mantle, the major impurities affecting Mg vacancy concentrations in MgO
are Al3þ, Cr3þ and Na1þ. Fe3þ is supposed to be strongly partitioned into the
perovskite phase (McCammon, 1997). These impurities should induce vacancy
concentrations of , 0.001 –0.1% for Mg (e.g., Wood and Rubie, 1996; Bertka and
Fei, 1997).
In this ionic compound, MgO, the dominant point defect is the pair vacancy
(Yang and Flynn, 1994) with Mg and O sites vacant. The self-diffusion coefficient
Transport Properties in Deep Depths and Related Condensed-Matter Phenomena 1097

Di (i ¼ Mg or O), is given by
 
Zm 2 DGm;j
Di ¼ Zf l nci exp 2 ; ð15-25Þ
6 kT

where Zf is the number of crystallographically equivalent ways of forming a vacancy type,

Zm is the number of equivalent diffusion paths, l is the jump distance, n is the attempt
frequency, c is the defect concentration, DGm is the energy of migration, k is Boltzmann’s
constant and T is temperature (Tilley, 1987). Because MgO has a rock-salt structure,
l 2 ¼ a 2/2, where a is the cubic-cell parameter. For isolated Mg and O (Schottky) defects,
W ¼ 2, Zf ¼ 1 and Zm ¼ 12. Highly correlated Mg and O defects (bound pair) require
W ¼ 1, Zf ¼ 6 and Zm ¼ 8 (Ita and Cohen, 1997).
Determination of the absolute diffusion coefficient requires knowledge of
the energetics of defect formation and migration, the cell parameter and the attempt
frequency, n:

15.15.3. Rheology of lower mantle: strain rate

Little is known about deformation of perovskite under lower-mantle conditions.

For detailed modelling of convection within the Earth, an estimation of the rheological
properties of silicate perovskite at high pressures and temperatures becomes important.
The plasticity of single crystals of MgSiO3 perovskite at 300 K is seen to be higher
than those of olivine and enstatite (Karato et al., 1990). Below the depth of 700 km, the
absence of seismicity may arise from the superplasticity associated with the fine-grain
texture of perovskite and magnesiowüstite assemblages in the lower mantle (Ito and Sato,
1991). Meade and Jeanloz (1990) speculated that the lower mantle might be more ductile
than the transition zone. It is possible that the increased T within D00 could drive the
deformation mechanism of the deep mantle from a super-plastic regime (Karato et al.,
1995) into a regime of power-law creep.
In the lower mantle under high temperatures and low strain rates, the relationship of
stress versus the viscosity or strain rate becomes important. The strain rate under lower-
mantle conditions is unknown for perovskite. Beauchesne and Poirier (1990) studied the
high-temperature creeps in a series of perovskites and drew conclusions about perovskite
deformation at high temperatures. They reached a relation,

1 ¼ Asn expð2Ep =RTÞ; ð15-26Þ

where 1 is the strain rate, s is the shear stress, R is the gas constant, E p is the activation
energy, T is the temperature and h is the stress component. They found the following
values in a series of perovskites, ranging as

h : 1 – 3:7; As : 1262ðs in PaÞ; Ep : 116 – 469 kJ mol21 :

1098 Chapter 15

Although strain rates fluctuate slightly during creep, the empirical creep law holds as
(Karato et al., 1982):

1 ¼ 3:4 £ 104 s3:5 expð2643; 000=RTÞ:

The activation energy for steady-state creep of polycrystalline olivine is larger than that of
oxygen and silicon self-diffusion in olivine (Karato et al., 1982).
These data may imply that there is a drop in strength (or increase in strain rate) of
many orders of magnitude between room and mantle temperatures.
Plastic flow by lattice processes such as dislocation glide occur at higher pressures
and stresses (Ranalli, 1995). Thermally activated viscous creep takes over at greater
mantle depths at higher temperatures. The creep accommodated by dislocation climb
results in a non-linear stress (s) versus strain rate ð1Þ
 relationship as

1 ¼ AD sn expð2H=RTÞ; ð15-27Þ

where n < 3; H < 500 kJ mol21 and AD < 2.0 £ 103 MPa23 s21 for hydrated olivine (in
the mantle rocks).
In addition, strain weakening may occur such as is manifested in shear zones in the
form of dynamic recrystallization, reaction with volatiles and heat production through
viscous dissipation (Jin et al., 1998).

15.16. Transformational plasticity: partial dislocation, martensitic or synchro-shear

mechanism

In the rheological changes, transformation plasticity attains significance, as is seen

in metals and ceramics. This often exhibits a reduction in mechanical strength on phase
transition. Under low differential stress, unusually high strains can develop, especially
when the temperature is changed cyclically back and forth across the phase transition.
During mineralogical phase transformations in the Earth’s interior, transformation
plasticity may attain importance. (Note: Plastic deformation in solids results from the
creation and motion of dislocation. A dislocation is a lattice mismatch along a slip plane
and is characterized by long-range stress and strain fields.)
Theoretical studies have shown that transformational plasticity in such phase
boundaries as observed at seismic discontinuities (e.g., 410 km, possibly 520 and 660 km)
would cause mechanical decoupling and favour layered convection (Christensen and
Yuen, 1985). A different mechanism for transformation plasticity will presumably operate,
depending on the mechanism of phase transition (e.g., Rubie, 1990). Martensitic or shear
transformations are particularly likely to result in a reduction in strength. Such
transformation can occur by the migration on certain slip planes of partial dislocations,
each of which changes the stacking sequence and leaves behind a layer of the product
phase. In addition to the changing of the crystal structure, the migration of the partial
dislocations also results in plastic deformation.
Transport Properties in Deep Depths and Related Condensed-Matter Phenomena 1099

The driving force for the migration of the partial dislocations and the resulting
deformation is the change in Gibbs free energy associated with the phase transformation.
Thus while the phase transformation is in progress, the material undergoing transformation
would deform at much smaller externally applied stress than is required at normal
condition. Indeed, a martensitic-like mechanism is involved when the olivine –spinel
transformation occurs. However, shear mechanism dominates during the spinel to
b-olivine transformation (Rubie and Brearley, 1994). The transformation may occur by
topotactic coherent nucleation of g on b followed by interfacial controlled growth, which
involves diffusion of atoms across the interface boundary. The transformation of a- to
g-phase by the synchro-shear mechanism can only be achieved by glide on one particular
slip system.

15.17. Oxygen fugacity in the Earth’s dynamics

15.17.1. Solid-state diffusion: fO2

In ionic crystals at high temperature, solid-state diffusion of ionic species generally

occurs by a vacancy mechanism.
The self-diffusion coefficient of the cation on its sublattice is equal to the product of
the diffusion coefficient of the cationic vacancies by their atomic fraction. When a crystal
is non-stoichiometric and is with or without impurities with variable oxidation states, the
cation vacancies are controlled by the degree of non-stoichiometry (or fraction of oxidized
species). The self-diffusion coefficient of the cations then depends on oxygen fugacity f O2
(see Poirier and Duba, 1997).
The diffusion coefficient of cations depends on f O2 through the power law derived
from the quasi-chemical equations. (Note: The activation energy for migration does not
depend on f O2.) The f O2 dependence of diffusion coefficient leads to a power-law
dependence with exponents ranging from 1/2 to 1/6.
In non-stoichiometric wustite (Fe12xO), where Fe2þ can be partly oxidized to Fe3þ,
1=6
the atomic fraction of Fe vacancies depends on f O2 through a power law: lVFe l / f O2 :

15.17.1.1. Olivine/iron buffer

Most of the transport properties, e.g., electrical conductivity, ionic diffusion, etc.,
largely depend on the oxygen fugacity. In an experiment, the f O2 is controlled by inserting
a metal/metal oxide buffer (e.g., Ni/NiO). The buffer imposes its f O2 to the assembly.
However, it is assumed that the kinetics of chemical exchanges between phases is rapid
enough for the equilibrium to take place.
In studies of Fe-bearing silicates such as olivine, pyroxenes or garnets, the Fe
capsule is often kept in contact with the sample to control the f O2 (e.g., Béjina et al., 1997).
In such cases, the silicate/Fe-metal couple serves as an f O2 buffer. Raterron et al. (1998),
through an experiment on San Carlos olivine crystals, found the olivine/Fe-metal buffer
does efficiently control the f O2 in high-pressure experiments. They estimated the annealing
f O2, theoretically controlled by the olivine/Fe-metal equilibrium, to be 1– 2 log units
1100 Chapter 15

above the f O2 of the iron/wüstite buffer. This buffering technique can be used for other Fe-
bearing silicates such as pyroxene and garnets.
The fugacity of free oxygen in the mantle barely exceeds 1027 bar. Several redox
equilibria could buffer the oxygen fugacity in the mantle. The oxygen fugacity becomes
dependent on the bulk chemical composition (the bulk ratio Fe2O3/FeO), bulk mineralogy
and the chemical potentials of the ferric iron-bearing components or of CO2.

15.17.1.2. Ferric iron and redox zone: crustal recycling

Ferric iron, which in the peridotitic upper mantle is concentrated in the minor
phases: spinel, garnet and pyroxene (the major phase olivine does not incorporate ferric
iron), finds a vastly increased reservoir in the transition zone. The bulk composition (and
ferric iron content) of the transition zone is the same as that of the fertile peridotitic mantle
material, the source of basaltic melt. With increase in pressure, fertile peridotite undergoes
a series of phase changes. The transitions of plagioclase to spinel at , 25-km depth and of
spinel to garnet at , 190 km are well known. A further increase in pressure will gradually
push the ferric iron and aluminium from pyroxene into the garnet phase until, at around
9 GPa (200-km depth), the silicate end-members of pyroxene become soluble in aluminous
garnet.
In the overlying silicate mantle, due to the partial melting at low oxygen fugacity,
the volatiles present will be mostly the reduced species, methane and hydrogen with minor
amounts of water and virtually no carbon dioxide. Methane does not react with mantle
silicates nor is it soluble in the melt. Therefore, methane- and H2-rich fluids infiltrate the
overlying more oxidized silicate mantle. They will gradually become enriched in water and
carbon dioxide until the H2O and CO2 activities are high enough for partial melting of the
mantle to occur. The low-velocity zone may be explained by this redox zone within the
peridotitic mantle.
Oxidized crust recycled into the convecting mantle may cause long-term oxidation
of the mantle (Arculus, 1985). Some basalts may owe their oxidized nature to a crustally-
derived component that re-emerges in the course of large-scale mantle stirring and
the oxygen isotopes in basalts may be used to quantify the crustal input (Woodhead
et al., 1993).

15.17.1.3. Core –mantle partitioning

Metal – silicate partition coefficients depend strongly on oxygen fugacity, which is
calculated on the basis of calibration of the iron –wüstite buffer (along with the
concentrations of FeO in the silicate and of Fe in the metal). The oxygen fugacity relative
to the IW buffer can be calculated as

aFeO
D IW ¼ 2:0 log ; ð15-28Þ
aFe

where aFeO and aFe are, respectively, the relative activities of FeO in the silicate liquid and
of Fe in the metal. The activities of Fe in the metal and of FeO in the silicate liquid can be
estimated as their mole fractions.
Transport Properties in Deep Depths and Related Condensed-Matter Phenomena 1101

It was found by Hillgren et al. (1994) that the Ni and Co metal – silicate partition
coefficients decrease with increasing temperature and pressure. The partition coefficients
for both V and Ni increase with increasing temperature and pressure (see Section 1.26.1.1).
Simple mass balance between the core and the mantle of the Earth may be used to
calculate the metal –silicate partition coefficients that are necessary to produce the
observed abundance of these elements in the mantle. The mean core – mantle partition
coefficients for Ni, Co and W need to be 40 ^ 5 (Hillgren et al., 1994). The partition
coefficient of W increases with increasing temperature and pressure.
The metal – silicate partition coefficients of siderophile elements do not decrease
uniformly with increasing temperature. However, more complex processes, rather than
simple metal – silicate equilibrium, must have operated to lead to the siderophile element
pattern in the mantle.
This page is intentionally left blank
 1103

References

Abbott, D.H. and Hoffman, S.E. (1984) Archean plate tectonics revisited. 1. Heat flow, spreading rate, and the age
of subducting oceanic lithosphere and their effects on the origin and evolution of continents. Tectonics,
3(4), 429–448.
Abe, Y. (1993) Thermal evolution and chemical differentiation of the terrestrial magma ocean. Geophys. Monogr.
Am. Geophys. Union 74/IUGG, 14, 41–54.
Abe, Y., Ohtani, E., Okuchi, T., Righter, K. and Drake, M.J. (2000) In: Canup, R.M. and Righter, K. (eds.), Origin
of Earth and Moon. Tucson: University of Arizona Press, pp. 413–433.
Abers, G.A. (2000) Hydrated subducted crust at 100–250 km depth. Earth Planet. Sci. Lett., 176, 323–330.
Abramson, E.H., Brown, J.M., Slutsky, L.J. and Zaug, J. (1997) The elastic constants of San Carlos olivine to
17 GPa. J. Geol. Phys. Res., 102, 12253–12264.
Abu-Eid, R.M. (1974) Absorption spectra of transition metal-bearing at high pressure. In: Strens, R.G.J. (ed.),
Physics Chemistry Minerals and Rocks. New York: Wiley, pp. 641–675.
Abu-Eid, R.M. and Burns, R.G. (1974) The effect of pressure on the degree of covalency of the cation–oxygen
bond in minerals. Am. Mineral., 61, 391–397.
Acuna, M.H. et al. (1999) Global distribution of crustal magnetization discovered by the Mars Global Surveyor
MAG/ER experiment. Science, 284, 790 –793.
Adachi, H. and Takano, M. (1991) J. Solid State Chem., 93, 556.
Adam, L.H. and Williamson, E.D. (1923) On the compressibility of minerals and rocks at high pressures.
J. Franklin Inst., 195, 475 –529.
Adams, D.M. (1978) Inorganic Solids: An Introduction to Concepts in Solid State Structural Chemistry.
New York: Wiley.
Adams, L.H. and Gibson, R.E. (1929) The elastic properties of certain basic rocks and of their constituent
minerals. Proc. Natl. Acad. Sci., 15, 713 –724.
Aeppll, G. (1989) Phys. Rev. Lett., 63, 676–679.
Agee, C. (1993a) High-pressure melting of carbonaceous chondrite. J. Geophys. Res., 98, 5419–5426.
Agee, C.B. (1993b) Petrology of the mantle transition zone. Annu. Rev. Earth Planet. Sci., 21, 19– 42.
Agee, C.B. (1998) Phase transformations and seismic structure in the upper mantle and transition zone. Rev.
Mineral., 37, 165 –203.
Agee, C.B. and Walker, D. (1990) Aluminum partitioning between olivine and ultrabasic silicate liquid to 6 GPa.
Contrib. Mineral. Petrol., 105, 243 –254.
Agnon, A. and Bukowinski, M.S.T. (1990) dS at high pressure and d ln Vs/dVp in the lower mantle. Geophys. Res.
Lett., 17, 1149–1152.
Agrinier, P., Javoy, M., Smith, D.C. and Pineau, F. (1985) Carbon and oxygen isotopes in eclogites, amphibolites,
veins and marbles from the Western Gneiss Region, Norway. Chem. Geol. (Isotope Geosci. Sec.), 52,
145–162.
Ahmed-Zaid, I. and Madon, M. (1991) A high pressure-form of Al2SiO5 as a possible host of aluminium in the
lower mantle. Nature, 353, 426–428.
Ahmed-Zaid, I. and Madon, M. (1995) Electron microscopy of high-pressure phases synthesised from natural
garnets in a diamond anvil cell: implications for the mineralogy of the lower mantle. Earth Planet. Sci.
Lett., 129, 233–247.
Ahrens, T.J. (1980) Dynamic compression of Earth materials. Science, 207, 1035–1042.
Ahrens, T. and Graham, E. (1972) A shock-induced phase change in iron-silicate garnet. Earth Planet. Sci. Lett.,
14, 87 –90.
1104 References

Ahrens, T.J., Anderson, D.L. and Ringwood, A.E. (1969) Equation of state and crystal structure of high-pressure
phases of shocked silicates and oxides. Rev. Geophys., 7, 667–707.
Ahrens, T.J., Lyzenga, G.A., and Mitchell, A.C. (1982) Temperatures induced by shock-waves in minerals.
Applications to Geophysics, High Pressure Research in Geophysics. Akimoto, S. and Manghnani, M.H.
(eds.), Center for Academic Publishing, Tokya, 579–594.
Ahrens, T.J., Holland, K.G. and Chen, G.Q. (1998) Phase diagram of iron, Revised core temperatures. Geophys.
Res. Lett., 29, 54-1–54-4.
Aikawa, N., Kumazawa, M. and Tokanami, M. (1985) Temperature dependence of intersite distribution of
Mg and Fe in olivine and the associated change of lattice parameters. Phys. Chem. Miner., 12,
1–8.
Aines, R.D. and Rossman, G.R. (1985a) The water content of mantle garnets. Geology, 12, 720– 723.
Akahama, Y., Kobayashi, M., Kawamura, H. and Endo, S. (1993) Rev. Sci. Instrum., 64, 1979–1983.
Akaogi, M. and Akimoto, S. (1977) Pyroxene-garnet solid solution equilibria in the system Mg4Si4O12 –
Mg3Al2Si3O12 and Fe4Si4O12 and Fe4Si4O12 –Fe3Al2Si3O12 at high pressures and temperature. Phys.
Earth Planet. Interiors, 15, 90 –106.
Akaogi, M. and Akimoto, S. (1979) High pressure phase equilibria in a garnet lherzolite, with special reference to
Mg2þ –Fe2þ partitioning among constituent minerals. Phys. Earth Planet. Interiors, 19, 31–51.
Akaogi, M. and Akimoto, S. (1980) High pressure stability of a dense hydrous magnesium silicate Mg23Si8O42H6
and some geophysical implications. J. Geophys. Res., 85, 6944–6948.
Akaogi, M. and Ito, E. (1993a) Heat capacity of MgSiO3 perovskite. Geophys. Res. Lett., 20, 105–108.
Akaogi, M., Ross, N.L., McMillan, P. and Navrotsky, A. (1984) The Mg2SiO4 polymorphs (olivine, modified
spinel and spinel) — thermodynamic properties from oxide-melt solution calorimetry, phase relations,
and models of lattice vibrations. Am. Mineral., 69, 499–512.
Akaogi, M., Ito, E. and Navrotsky, A. (1989) Olivine-modified spinel-spinel transitions in the system Mg2SiO4 –
Fe2SiO4: calorimetric measurements, thermochemical calculation, and geophysical application.
J. Geophys. Res., 94, 15671–15685.
Akaogi, M., Ito, E. and Navrotsky, A. (1992) J. Geophys. Res., 94, 15671.
Akaogi, M., Yusa, H., Shuraishi, K. and Suzuki, T. (1995) Thermodynamic properties of a-quartz, coesite and
stishovite and equilibrium phase relations at high pressures and high temperatures. J. Geophys. Res.,
100(B11), 22337–22347.
Akimoto, S. (1972) The system MgO– FeO–SiO2 at high pressure and temperature: phase equilibria and elastic
properties. Tectonophysics, 13, 161–187.
Akimoto, S. and Akaogi, M. (1980) The system Mg2SiO4 –MgO–H2O at high pressure and temperatures —
possible hydrous magnesian silicates in the mantle transition zone. Phys. Earth Planet. Interiors, 23,
268–275.
Akimoto, S., Komada, E. and Kushiro, I. (1967) Effect of pressure on the melting of olivine and spinel polymorphs
of Fe2SiO4. J. Geophys. Res., 7, 679–686.
Akimoto, S., Katsura, T., Syono, Y., et al., (1965) Polymorphic transition of pyroxenes FeTiO3 and CoSiO3 at
high pressures and temperatures. J. Geophys. Res., 70, 5269–5278.
Akimoto, S., Suzuki, T., Yagi, T. and Shimomura, O. (1987) Phase diagram of iron determined by high pressure/
temperature X-ray diffraction using synchrotron radiation. In: Manghnani, M.H. and Soyono, Y. (eds.),
High Pressure Research in Mineral Physics. Tokyo/Washington, DC: Terra Scientific/American
Geophysical Union, pp. 149–154.
Aksandrov, K.S. (1976) The sequences of srtuctural phase transitions in perovskites. Ferroelectrics, 14, 801–805.
Aldridge, L.P., Bancroft, G.M., Fleet, M.E. and Herzberg, C.T. (1978) Omphacite studies, II Mössbauer spectra of
C2=c and P2=n omphacites. Am. Mineral., 63, 1107–1115.
Aleksandrov, I.V., Goncharov, A.F. and Stishov, S.M. (1986) Direct determination of the Gruneisen parameter of
the LTO(G) mode of diamond at high pressures. JETP Lett., 44, 611–614.
Alexander, C.M.O’D., Boss, A.P. and Carlson, R.W. (2001) The early evolution of the inner solar system: a
meteoritic perspective. Science, 293, 64 –68.
Alidibirov, M. and Dingwell, D.B. (1996) Magma fragmentation by rapid decompression. Nature, 380,
146–148.
References 1105

Allan, D.R. and Angel, R.J. (1997) A high-pressure structural study of microcline (KAlSi3O8) to 7 GPa. Eur. J.
Mineral., 9, 263 –275.
Allan, D.R. and Nelmes, R.J. (1992) The crystal structure of potassium titanyl phosphate (KTP) in its high-
pressure phase. J. Phys. Condens. Matter, 4, L395–L398.
Allègre, C.J. (1997) Earth Planet. Sci. Lett., 150, 1 –6.
Allègre, C.J. and Turcotte, D.L. (1985) Geodynamic mixing in the measosphere boundary layer and the ring of
oceanic islands. Geophys. Res. Lett., 12, 207–210.
Allègre, C.J. and Turcotte, D.L. (1986) Implications of a two-component marble-cake mantle. Nature, 323,
123–127.
Allegre, C.J., Boctor, N.J., Strixrude, L. (1982) Chemical geophysics. Tectonophysics, 81, 109–132.
Allègre, C.J., Gerad, M. and Gopel, C. (1995a) Geochim. Cosmochim. Acta, 59, 1445–1456.
Allègre, C.J., Moreira, M. and Staudacher, T. (1995b) Geophys. Res. Lett., 22, 2325–2328.
Allègre, C.J., Poirier, J.P., Humler, E. and Hofmann, A.W. (1995c) The chemical composition of the Earth. Earth
Planet. Sci. Lett., 134, 515–526.
Allègre, C.J., Schiano, P. and Lewin, P. (1995d) Earth Planet. Sci. Lett., 129, 1– 12.
Allen, F.M. and Buseck, P.R. (1988) XRD FTIR, and TEM studies of optically anisotropic grossular garnets. Am.
Mineral., 73, 568 –584.
Allsopp, H.L. and Roddick, J.C. (1984) In: Erlank, A.J. (ed.),, Special Publication, 143. Geological Society of
South Africa, pp. 267– 271.
Amelin, Y. et al. (1999) Nature, 399, 252–255.
Amthauer, G., Annersten, H. and Hafner, S.S. (1976) The Mössbauer spectrum of 57Fe in silicate garnets.
Z. Kristallogr., 143, 14–55.
Ancilotto, F., Chiarotti, G.L., Scandolo, S., Tosatti, E., Hubbard, W.B. et al. (1997) Science, 275, 1288.
Anders, E. and Ebihara, M. (1982) Solar system abundances of the elements. Geochim. Cosmochim. Acta, 46,
2363– 2380.
Anders, E. and Grevesse, N. (1989) Abundances of the elements: meteoritic and solar. Geochim. Cosmochim.
Acta, 53, 197– 214.
Anderson, P.W. (1950) Phys. Rev., 79, 350.
Anderson, D.L. (1967a) A seismic equation of state. Geophys. J. R. Astron. Soc., 13, 9 –30.
Anderson, O.L. (1967b) Equation for thermal expansivity in planetary interiors. J. Geophys. Res., 72,
3661– 3668.
Anderson, D.L. (1970) Petrology of the mantle. Mineral. Soc. Am., 3, 85–93, Special paper.
Anderson, D.L. (1979) The upper mantle transition region: eclogite? Geophys. Res. Lett., 6, 433–436.
Anderson, D.L. (1980) Bulk attenuation in the Earth and viscosity of the core. Nature, 285, 207.
Anderson, D.L. (1982) The chemical composition and evolution of the mantle. In: Akimoto, S. and Manghnani,
M.H. (eds.), Advances in Earth and Planetary Sciences, High Pressure Research in Geophysics. MA: D.
Reidel, pp. 301–318.
Anderson, P.W. (1986) Science, 235, 1196–1198.
Anderson, O.L. (1988) A ferroelectric transition in the lower mantle? EOS Trans. Am. Geophys. Union, 69, 1451,
Abstracts.
Anderson, D.L. (1989) Theory of the Earth. Boston: Blackwell Scientific Publication, p. 366.
Anderson, O.L. (1990) J. Geophys. Res., 95, 21697.
Anderson, O.L. (1993) The phase diagram of iron and the temperature of the inner core. J. Geomagn. Geolectr.,
45, 1235–1248.
Anderson, O.L. (1995) Equation of State of Solids for Geophysics and Ceramic Science. Oxford: Oxford
University Press.
Anderson, M.H., Ensher, J.R., Matthews, M.R., Wieman, C.E. and Cornell, E.A. (1995) Science, 269, 198–201.
Anderson, O.L. (1998) Themoelastic properties of MgSiO3 perovskite using the Debye approach. Am. Mineral.,
83, 23 –35.
Anderson, O.L. (1999) A thermal balancing act. Science, 283, 1652.
Anderson, W.W. and Ahrens, T. (1994) An equation of state for liquid iron and implications for the Earth’s core.
J. Geophys. Res., 99, 4273–4284.
1106 References

Anderson, D.L. and Anderson, O.L. (1970) The bulk modulus–volume relationship for oxides. J. Geophys. Res.,
75, 3494–3500.
Anderson, S. and Backstron, G. (1986) Rev. Sci. Instrum., 57, 1633.
Anderson, D.L. and Bass, J.D. (1986) Transition region of the Earth’s upper mantle. Nature, 320, 321–328.
Anderson, O.L. and Isaak, D.G. (2000) Calculated melting curves for phases of iron. Am. Mineral., 85, 376–385.
Anderson, O.L. and Nafe, J.E. (1965) J. Geophys. Res., 70, 3951–3957.
Anderson, O.L., Isaak, D.G. and Yamamoto, S. (1989) Anharmonicity and the equation of state for gold. J. Appl.
Phys., 65, 1534–1543.
Anderson, O.L., Chopelas, A. and Boehler, R. (1990) Thermal expansivity versus pressure at constant
temperature: a re-examination. Geophys. Res. Lett., 17, 685–688.
Anderson, O.L., Isaak, D. and Oda, H. (1992) High-temperature elastic constant data on minerals relevant to
geophysics. Rev. Geophys, 30, 57–89.
Anderson, M.H., Ensher, J.R., Matthews, M.R., Wieman, C.E. and Cornell, E.A. (1995) Science, 269,
198–201.
Andrault, D. (2001) Evaluation of (Mg,Fe) Partitioning between silicate perovskite and magnesiowustite up to
120 GPa and 2300 K. J. Geophys. Res., 106, 2079–2087.
Andrault, D. and Poirier, J.P. (1991) Evolution of the distortion of perovskites under pressure: an EXAFS study of
BaZrO3-SrZrO3 and CaGeO3. Phys. Chem. Miner., 18, 91–105.
Andrault, D., Bouhifd, M.A., Itie, J.P. and Richet, P. (1995) Compression and amorphization of (Mg,Fe)2SiO4
olivines: an X-ray diffraction study up to 70 GPa. Phys. Chem. Miner., 22(2), 99–107.
Andrault, D., Fiquet, G., Kunz, M., Visocekas, F. and Hausermann, D. (1997) The orthorhombic structure of iron:
an in situ study at high-temperature and high-pressure. Science, 278, 831–834.
Andrault, D., Fiquet, G., Guyot, F. and Hanfland, M. (1998a) Pressure-induced Landau-type transition in
stishovite. Science, 282, 720–724.
Andrault, D., Fiquet, G., Kunz, M., Visocekas, F. and Hausermann, D. (1998b) The orthorhombic structure of
iron: an in situ high-T/high-P structure solution and refinement. Science, 278, 831–834.
Andrault, D., Fiquet, G., Charpin, T. and Bihan, T.L. (2000) Structure analysis and stability field of b-iron at
high P and T. Am. Mineral., 85, 364 –371.
Angel, R.J. (1988) High pressure structure of anorthite. Am. Mineral., 73, 1114–1119.
Angel, R.J. (1994) Feldspars at high pressure. In: Parsons, I. (ed.), Feldspars and Their Reactions. Dordrecht:
Kluwer, pp. 271– 312.
Angel, R.J. (1996) New phenomena in minerals at high pressures. Phase Transition, 59, 105–119.
Angel, R.J. (1997) Mineralogy of the Earth’s mantle: remodeled. Nature, 385, 490–491.
Angel, R.J. and Hugh-Jones, D.A. (1994) Equations of state and thermodynamic properties of enstatite pyroxenes.
J. Geophys. Res., 99(B10), 19777–19783.
Angel, R.J. and Ross, N.L. (1997) Equations of state of mantle minerals from high-pressure diffraction. Phys.
Chem. Earth, 22(1– 2), 119–123.
Angel, R.J., Gasparik, T., Ross, N.L., Finger, L.W., Prewitt, C.T. and Hazen, R.M. (1988a) A silica-rich pyroxene
phase with six-coordinated silicon. Nature, 335, 156–158.
Angel, R.J., Hazen, R.M., McCormick, T.C., Prewitt, C.T. and Smyth, J.R. (1988b) Comparative compressibility
of end-member feldspars. Phys. Chem. Miner., 15, 313–318.
Angel, R.J., Finger, L.W., Hazen, R.M., Kanzaki, M., Weidner, D.J., Liebermann, R.C. and Veblen, D.R. (1989a)
Structure and twinning of single-crystal MgSiO3 garnet synthesized at 17 GPa and 18008C. Am.
Mineral., 74, 509 –512.
Angel, R.J., Redfern, S.A.T. and Ross, N.L. (1989b) Spontaneous strain in the I1̄ 2 P1̄ transition in pure anorthite.
Phys. Chem. Miner., 16, 539 –544.
Angel, R.J., Carpenter, M.A. and Finger, L.W. (1990) Structural variation associated with compositional variation
and order-disorder behaviour in anorthite-rich feldspars. Am. Mineral., 75, 150–162.
Angel, R.J., Chopelas, A. and Ross, N.L. (1992) Stability of high-density clinoenstatite at upper-mantle pressures.
Nature, 358, 322–324.
Angell, C.A., Cheeseman, P.A. and Tamaddon, S. (1982) Pressure enhancement of ion mobilities in liquid
silicates from computer simulation studies to 800 kilobars. Science, 218, 885–888.
References 1107

Aoki, K. (1984) Petrology of materials derived from the upper mantle. In: Sunagawa, I. (ed.), Materials Science of
the Earth’s Interior. Tokyo: Terra Scientific, pp. 415–444.
Aoki, K., Fujino, K. and Akaogi, M. (1976) Titanochondrodite and titanoclinohumite derived from the upper
mantle in the Buell Park Kimberlite, Arizona, USA. Contrib. Mineral. Petrol., 56, 243–253.
Aoki, K., Yamawaki, H., Sakashita, M., Gotoh, Y. and Takemura, K. (1994) Science, 265, 356–358.
Aoki, K., Yamawaki, H., Sakashita, M. and Fujihisa, H. (1996) Infrared absorption study of the hydrogen-bond
symmetrization in ice to 110 GPa. Phys. Rev. B, 54, 15673–15677.
Appel, J. (1968) Polaron. In: Seitz, F. and Turnbull, D. (eds.), Solid State Physics, Vol. 21. New York: Academic
Press, pp. 193–391.
Arco, D., Sandrone, G., Dovesi, R., Apra, E. and Saunders, V.R. (1994) A quantum-mechanical study of
the relative stability under pressure of MgSiO3-ilmenite, MgSiO3-perovskite, and MgO-periclase þ
SiO2-stishovite assemblage. Phys. Chem. Miner., 21, 285–293.
Arculus, R.J. (1994) Aspects of magma genesis in arcs. Lithos, 33, 189–208.
Arlt, T. and Armbruster, T. (1997) The temperature dependent P21 =c – C2=c phase transition in the
clinopyroxene, kanoite MnMg(Si2O6): a single crystal X-ray and optical study. Eur. J. Mineral.,
953–964.
Arlt, T., Angel, R.J., Miletich, R., Armbruster, T. and Peters, T. (1998) High-pressure P21 =c – C2=c phase
transition in clinopyroxenes: influence of cation size and electronic structure. Am. Mineral., 83,
1176– 1181.
Armbruster, T. and Larger, G.A. (1989) Oxygen disorder and the hydrogen position in garnet–hydrogarnet solid
solutions. Eur. J. Mineral., 1, 363–369.
Armbruster, T., Burgi, H.B., Kunz, M., Gnos, E., Bronnimann, S. and Lienert, C. (1990) Variation of
displacement parameters in structure refinements of low albite. Am. Mineral., 75, 135– 140.
Armstrong, R.L. (1968) Rev. Geophys., 6, 175–199.
Arora, A.K. and Sakuntala, T. (1992) J. Phys. Condens. Matter, 4, 8697.
Artioli, G., Fumagalli, P. and Poli, S. (1999) The crystal structure of Mg8 (Mg2Al2) Al8Si12(O,OH)56 pumpellyite
and its relevance in ultramafic systems at high pressure. Am. Mineral., 84, 1906–1914.
Asamitsu, A., Moritomo, Y., Tomioka, Y., Arima, T. and Tokura, Y. (1995) Nature, 373, 407.
Ashby, M.F. and Verrall, R.A. (1973) Diffusion accommodated flow and superplasticity. Acta Metall., 21,
149–163.
Ashcroft, N.W. (1990) Pairing instabilities in dense hydrogen. Phys. Rev. B, 41, 10963–10971.
Ashcroft, N.W. (1995) Phys. World, 8, 43–47.
Ashida, T., Kume, S., Ito, E. and Navrotsky, A. (1988) MgSiO3 ilmenite: heat capacity, thermal expansivity, and
enthalpy of transformation. Phys. Chem. Miner., 16, 239–245.
Asimow, P.D. and Hirschmann, M.M. (1995) The effect of pressure-induced solid-solid phase
transitions on decompression melting of the mantle. Geochim. Cosmochim. Acta, 59(21),
4489– 4506.
Askarpour, V., Manghnani, M.H., Fassbender, S. and Yoneds, A. (1993) Elasticity of single-crystal MgAl2O4
spinel up to 1273 K by Brillouin spectroscopy. Phys. Chem. Miner., 19, 511–519.
Auld, B.A. (1973) Acoustic Waves and Fields in Solids, Vol. 1. New York: Wiley.
Aurnou, J.M., Buttles, J.L., Neumann, G.A. and Olson, P.L. (1996) Electromagnetic core– mantle coupling and
paleomagnetic reversal paths. Geophys. Res. Lett., 23, 2705–2708.
Aust, R.B. and Drickamer, H.G. (1963) Science, 140, 817.
Avrami, A. (1941) J. Chem. Phys., 9, 177.
Ayers, J.C. and Eggler, D.H. (1995) Partitioning of elements between silicate melt and H2O–NaCl fluids at 1.5
and 2.0 GPa pressure: implications for mantle metasomatism. Geochim. Cosmochim. Acta, 59(20),
4237– 4246.
Aziz, M.J., Circone, S. and Agree, C.B. (1997) Vanishing atomic migration barrier in SiO2. Nature, 390,
596–599.
Babuska, V., Fiala, J., Kumazawa, M. and Ohne, I. (1978) Elastic properties of garnet solid-solution series. Phys.
Earth Planet. Interiors, 16, 157 –176.
Bada, J.L. (1991) Amino acid cosmogeochemistry. Phil. Trans. R. Soc. London B, 1991, 349–358.
1108 References

Badding, J.V., Hemley, R.J. and Mao, H.K. (1991) High pressure chemistry of hydrogen in metals — in situ study
of iron hydride. Science, 253, 421 –424.
Badding, J.V., Mao, H.K. and Hemley, R.J. (1992) High pressure crystal structure and equation of state of iron
hydrite: implications for the Earth’s core. In: Syono, Y. and Manghnani, M.H. (eds.), High Pressure
Research: Applications to Earth and Planetary Science. Tokyo/Washington, DC: Terra Scientific/AGU,
pp. 363 –371.
Badro, J., Varel, B. and McMillan, P. (1995) Phys. Rev., 51, 11262.
Badro, J., Barrat, J.L. and Gillet, P. (1996) Nuemrical simulation of a-quartz under nonhydrostatic compression:
memory glass and five-coordinated crystalline phases. Phys. Rev. Lett., 76, 772–775.
Badro, J., Teter, D.M., Downs, R.T., Gillet, P., Hemley, R.J. and Barrat, J.L. (1997) Theoretical study of a five-
coordinated silica polymorph. Am. Phys. Soc., 56(10), 5797–5806.
Badro, J., Viktor, V., Struzhkin, J.S., Russell, J.H., Ho-Kwang, M., Struzhkin, V.V., Shu, J., Hemley, R.J. and
Mao, H.K. et al. (1999) Magnetism in FeO at megabar pressures from X-ray emission spectroscopy.
Phys. Rev. Lett., 83, 4101–4104.
Badro, J., Fiquet, G. et al. (2002) Nature of high-pressure transition in Fe2O3 hematite. Phys. Rev. Lett., 89(20),
205504-1-4.
Badro, J., Fiquet, G. et al. (2003) Iron partitioning in Earth’s mantle: toward a deep lower mantle discontinuity.
Science, 300, 789 –791.
Baer, A.J. (1981) Geotherms, evolution of the lithosphere and plate tectonics. Tectonophysics, 72, 203–227.
Bai, Q. and Kholsedt, D.L. (1992a) Substantial hydrogen solubility in olivine and implications for water storage in
the mantle. Nature, 375, 672 –674.
Bai, Q. and Kholsedt, D.L. (1992b) High temperature creep of olivine single crystals, III. Mechanical results for
unbuffered samples and creep mechanism. Philos. Mag., 66, 1149–1181.
Bai, Q. and Kohlstedt, D.L. (1992c) High-temperature creep of olivine single crystals, II. Dislocation structures.
Tectonophysics, 206, 1–29.
Bai, Q. and Kholsedt, D.L. (1993) Effects of chemical environment on the solubility and incorporation mechanism
for hydrogen in olivine. Phys. Chem. Miner., 19, 460–471.
Bakker, H.J., Hunsche, S. and Kurz, H. (1993) Phys. Rev. B, 48, 9331.
Balchan, A.S. and Cowan, G.R. (1966) Shock compression of two iron-silicon alloys to 2.7 megabars. J. Geophys.
Res., 71, 3577–3588.
Ball, P. (2001) Table top astrophysics. Nature, 411, 628–630.
Bancroft, G.M., Osborne, M.D. and Fleet, M.E. (1983) Next-nearest-neighbour effects in the Mössbauer spectra of
Cr-spinels: an application of partial quadrupole splittings. Solid State Commun., 76, 623–625.
Banghman et al. (1998) Nature, 279, 1522–1524.
Bania, T.M., Rood, R.T. and Balser, D.S. (2002) Nature, 415, 54–57.
Baranowski, B. (1992) A hypothesis concerning the low temperature phase transition in solid hydrogen and
deuterium at about 150 GPa. Pol. J. Chem., 66, 1637–1640.
Barbier, J. (1995) Structure refinement of Na2(Mg,Fe)6[(Ge,Fe)6O18]O2, a new aenigmatite-analog. Z. Kristal.,
210, 19 –23.
Barker, F. and Arth, J.G. (1976) Generation of trondhjemitic– tonalitic liquids and Archean bimodal trondhjemite-
basalt suites. Geology, 4, 596 –600.
Basaltic Volcanism Study Project (BVSP) (1981) Basaltic Volcanism on the Terrestrial Planets. New York:
Pergamon Press.
Bass, J.D. (1984) Elasticity of single crystal SmAlO3, GdAlO3 and ScAlO3 perovskites. Phys. Earth Planet.
Interiors, 36, 145 –156.
Bass, J.D. (1987) Mineral and melt physics. Rev. Geophys., 25, 1265–1276.
Bass, J.D. (1989) Elasticity of grossular and spessartine garnets by Brillouin spectroscopy. J. Geophys. Res., 94,
7621– 7628.
Bass, J.D. and Kanzaki, M. (1990) Elasticity of majorite-pyrope solid solution. Geophys. Res. Lett., 17,
1989– 1992.
Bass, J.D. and Weidner, D.J. (1984) Elasticity of single-crystal orthoferrosilite. J. Geophys. Res., 89, 4359– 4371.
Bassett, W. and Weathers, M. (1990) J. Geophys. Res., 95, 21709.
References 1109

Bassett, W.A. and Weathers, M.S. (1993) High pressure science and technology. Proceedings of the AIRAPT and
APS Conference, Colorado Springs, CO, 651.
Bassett, W.A., Shen, A.H., Bucknum, M. and Chou, I.M. (1993) Rev. Sci. Instrum., 64, 2340.
Bassett, W.A. et al. (1996) The hydrothermal diamond anvil cell (HDAC) and its application. In: Dyar, M.D.,
McCammon, N.C. and Schaefs, M.W. (eds.), Mineral Spectroscopy; A Tribute to Roger, G. Burns, Vol.
5. The Geochemical Society, pp. 261–272.
Bassett, W.A., Reichmann, H.J., Angel, R.J., Spetzler, H. and Smyth, J.R. (2000) New diamond anvil cells for
gigahertz ultrasonic interferometry and X-ray diffraction. Am. Mineral., 85, 283– 287.
Basso, R., Giusta, A.D. and Zefiro, L. (1983) Crystal structure refinement of plazolite: a highly hydrated natural
hydrogrossular. N. Jb. Miner. Mh., 251–258.
Basson, J.M., Hamel, G., Grima, T., Nelmes et al. (1992) A large volume pressure cell for high temperatures. High
Press. Res., 8, 625–630.
Bauer, M. and Klee, W.E. (1993) The monoclinic hexagonal phase transition in chloraoatite. Eur. J. Mineral., 5,
307–316.
Baughman, R.H. and Galvao, D.S. (1993) Nature, 365, 735.
Baughman, R.H., Stafstrom, S., Cui, C. and Dantas, S.O. (1998) Materials with negative compressibilities in one
or more dimensions. Science, 279, 1522–1533.
Baur, W.H. (1978) Crystal structure refinement of lawsonite. Am. Mineral., 63, 311–315.
Beatti, P. (1994) Sytematics and energetics of trace-element partitioning between olivine and silicate melts:
implications for the nature of mineral/melt partitioning. Chem. Geol., 117, 57–71.
Beauchesne, S. and Poirier, J.P. (1990) In search of a systematics for the viscosity of perovskites: creep of
potassium tantalate and niobate. Phys. Earth Planet. Interiors, 61, 182–198.
Beck, A.E., Darba, D.M. and Schloessin, H.H. (1978) Phys. Earth Planet. Int., 17, 35.
Becker, L. and Bunch, T.E. (1997) Meteorit. Planet. Sci., 32, 479–487.
Becker, L., Bunch, T.E. and Allamandola, L.J. (1999) Higher fullerenes in the Allende meteorite. Nature, 400,
227–228.
Béjina, F., Raterron, P., Zhang, J., Jaoul, O. and Liebermann, R.C. (1997) Activation volume of silicon diffusion
in San Carlos olivine. Geophys. Res. Lett., 24, 2597–2600.
Belayachi, A., Nogues, M., Dormann, J.L. and Taibi, M. (1996) Eur. J. Solid State Inorg. Chem., 33, 1039.
Bell, D.R. (1992) Water in mantle minerals. Nature, 357, 646.
Bell, P.M. and Mao, H.K. (1975) Preliminary evidence of disproportionation of ferrous iron in silicates at high
pressures and temperatures. Carnegie Inst. Washington Yearbook, 74, 557–559.
Bell, D.R. and Rossman, G.R. (1992a) The distribution of hydrogen in garnets from the subcontinental mantle of
southern Africa. Contrib. Mineral. Petrol., 111, 161 –178.
Bell, D.R. and Rossman, G.R. (1992b) Water in Earth’s mantle: the role of nominally anhydrous minerals.
Science, 255, 1391–1397.
Bell, P.M., Yagi, T. and Mao, H.K. (1979) Iron magnesium distribution coefficients between spinel
[(Mg,Fe)2SiO4], magnesiowüstite [(Mg,Fe)O], and perovskite [(Mg,Fe) SiO3]. Carnegie Institute
Washington Yearbook, 78, 618 –621.
Bellissent-Funel, M.C. (1998) Is there a liquid– liquid phase transition in supercooled water? Europhys. Lett., 42,
161–166.
Belonoshko, A.B. and Dubrovinsky, L.S. (1997) A new high-pressure silica phase obtained by molecular
dynamics — reply. Am. Mineral., 82, 1043.
Belonoshko, A.B., Dubrovinski, L.S. and Dubrovinski, N.A. (1996a) A new high-pressure silica phase obtained
by molecular dynamics. Am. Mineral., 81, 785–788.
Belonoshko, A.B., Dubrovinsky, L.S., Dubrovinsky, N.A. and Saxena, S.K. (1996b) Phase diagram and properties
of silica: lattice and molecular dynamics study. J. Petrol., 4, 519– 535.
Benioff, H. (1949) Seismic evidence for the fault origin of oceanic deeps. Geol. Soc. Am. Bull., 60, 1837– 1856.
Benoit, M., Bernasconi, M., Focher, P. and Parrinello, M. (1996) Phys. Rev. Lett., 76, 2934.
Benoit, M., Marx, D. and Parrinello, M. (1998) Nature, 392, 258.
Benz, W. and Cameron, A.G.W. (1990a) In: Newson, H.E. and Jones, J.H. (eds.), Origin of the Earth. Oxford:
Oxford University Press, pp. 61– 67.
1110 References

Benz, W. and Cameron, A.G.W. (1990b) Terrestrial effects of the giant impact. In: Newson, H.E. and Jones, J.H.
(eds.), Origin of the Earth. Oxford: Oxford University Press, pp. 61– 67.
Berg, G.W. (1986) Evidence for carbonate in the mantle. Nature, 324, 50–51.
Bernard, S., Chiarotti, G.L., Scandolo, S. and Tosatti, E. (1967) Phys. Rev. Lett., 81, 2092.
Berlinsky, A.J. (1975) Phys. Rev. B, 12, 1482.
Berman, R.G. (1988) Internally-consistent thermodynamic data for minerals in the system Na2K2O–CaO–MgO–
FeO–Fe2O3 – Al2O3 –SiO2 –TiO2 –H2O–CO2. J. Petrol., 29, 445–522.
Bernal, J.D. (1936) Geophys. Discuss. Observatory, 59, 268.
Bernal, J.D. and Fowler, R.H. (1933) J. Chem. Phys., 1, 515–548.
Bernasconi, M., Parrinello, M., Chiarotti, G.L., Focher, P. and Tosatti, E. (1995) Anisotropic a-C:H form
compression of polyacetylene. Phys. Rev. Lett., 76, 2081–2084.
Bernasconi, M., Silvestrelli, P.L. and Parrinello, M. (1998) Phys. Rev. Lett., 81, 1253.
Bersuker, I.B. and Polinger, V.Z. (1989) Vibronic Interactions in Molecules and Crystals. New York: Springer.
Bertie, J.E., Labbe, H.J. and Whalley, E. (1968) Infrared spectrum of ice VI in the range 4000–50 cm21. J. Chem.
Phys., 49, 2141–2144.
Bertka, C. and Fei, Y. (1996) Constraints on the mineralogy of an iron-rich Martian mantle from high-pressure
experiment. Planet. Space Sci., 44(11), 1269–1276.
Bertka, C. and Fei, Y. (1997) Mineralogy of the Martian interior upto core– mantle boundary. J. Geophys. Res.,
102(B3), 5251–5264.
Bertka, C.M. and Fei, Y. (1998a) Implications of Mars pathfinder data for the accretion history of the terrestrial
planets. Science, 281, 1838–1840.
Bertka, C.M. and Fei, Y. (1998b) Density profile of an SNC model Martian interior and the moment-of-inertia
factor of Mars. Earth Planet. Sci. Lett., 157, 79–88.
Bertran-Alvarez, Y., Jaoul, O. and Liebermann, R.C. (1992) Fe –Mg interdiffusion in single crystal olivine at very
high pressure and controlled oxygen fugacity: technology advances and initial data at 7 GPa. Phys. Earth
Planet. Interiors, 70, 102 –118.
Bertrand, P. and Mercier, J.C. (1985) The mutual solubility of coexisting ortho and clinopyroxene: towards an
absolute geothermometer for the natural system. Earth Planet. Sci. Lett., 76, 109– 122.
Bertrand, P., Sotin, C., Mercier, J.C. and Takahashi, E. (1986) From the simplest chemical system to the natural
one: Garnet peridotite barometry. Contrib. min. Petrol., 93, 168– 178.
Besson, J.M. et al. (1992) High Press. Res., 8, 625.
Beyeler, H.U. (1976) Cationic short-range order in the hollandite K1.54Mg0.77T7.23O16: evidence for the
importance of ion–ion interactions in superionic conductors. Phys. Rev. Lett., 37, 1557–1560.
Biellmann, C., Gillet, P., Guyot, F., Peyronneau, J. and Reynard, B. (1993a) Experimental evidence for carbonate
stability in the Earth’s lower mantle. Earth Planet. Sci. Lett., 118, 31– 41.
Biellmann, C., Gillet, P., Guyot, F., Peyronneau, J. and Reynard, B. (1993b) High-pressure stability of carbonates:
quenching of calcite II, high-pressure polymorph of CaCO3. Eur. J. Mineral., 5, 503–510.
Billinge, S.J.L., DiFrancesco, R.G., Kwei, G.H., Neumeier, J.J. and Thompson, J.D. (1996) Direct observation of
lattice polaron formation in the local structure of La12xCaxMnO3. Phys. Rev. Lett., 77, 4.
Bina, C.R. (1995) Confidence limits for silicate perovskite equations of state. Phys. Chem. Miner., 22, 375–382.
Bina, C.R. (1998a) Lower mantle mineralogy and the geophysical perspective. In: Hemley, R.J. (ed.), Ultrahigh
Pressure Mineralogy, Reviews in Mineralogy, Mineralogical Society of America, Vol. 37, pp. 205–239.
Bina, C.R. (1998b) Olivine emerges from isolation. Nature, 392, 650– 652.
Bina, C.R. (2001) Mantle cookbook calibration. Nature, 411, 536–537.
Bina, C.R. and Kumazawa, M. (1993) Phys. Earth Planet. Interiors, 76, 329.
Bingeli, N. and Chelikowsky, J.R. (1991) Structural transformation of quartz at high pressures. Nature, 353,
344–346.
Bingeli, N. and Chelikowsky, J.R. (1992) Phys. Rev. Lett., 69, 2220.
Binggeli, N., Chelikowsky, J.R. and Wentzcovitch, R.M. (1994) Simulating the amorphization of a-quartz under
pressure. Phys. Rev. B, 49, 9336.
Birch, F. (1947) Finite elastic strain of cubic crystal. Phys. Res., 71, 809–824.
Birch, F. (1952) Elasticity and constitution of the Earth’s interior. J. Geophys. Res., 57, 227– 286.
References 1111

Birch, F. (1960) The velocity of compressional waves in rocks to 10 kbar. J. Geophys. Res., 65, 1083–1102.
Birch, F. (1961) Composition of the Earth’s mantle. Geoph. J. R. Astr. Soc., 4, 295–311.
Birch, F. (1978) Finite strain isotherm and velocities for single-crystal and polycrystalline NaCl at high pressures
and 300 K. J. Geophys. Res., 83, 1257– 1268.
Birnie, D.P. III. (1991) Determination of the lithium Frenkel energy in lighium tantalate. J. Appl. Phys., 69,
2485– 2488.
Bismayer, U., Schmahl, W., Schmidt, C. and Groat, L.A. (1992) Linear birefringence and X-ray
diffraction studies of the structural phase transition in titanite, CaTiSiO5. Phys. Chem. Miner., 19,
260–266.
Biswas, R., Martin, R.M., Needs, R.J. and Nielsen, O.H. (1987) Phys. Rev. B, 35, 9559.
Bloxham, J., Gubbins, D. and Jackson, A. (1989) Philos. Trans. R. Soc. A, 329, 415–502.
Bochu, B., Chenavas, J., Joubert, J.C. and Marezio, M. (1974) High-pressure synthesis and crystal structure of a
new series of perovskite-like compounds CMn7O12 (C ¼ Na, Ca, Cd, Sr, La, Nd). J. Solid State. Chem.,
11, 88 –93.
Bocquet, A.E., Mizokawa, T., Saitoh, T., Namatame, H. and Fujimori, A. (1992) Phys. Rev., 46, 3771.
Boctor et al. (1999) Ann. Report Geophys. Lab. Washington 1988–1999.
Boehler, R. (1986) The phase diagram of iron to 430 kbar. Geophys. Res. Lett., 13, 1153–1156.
Boehler, R. (1992a) Melting of the Fe–FeO and Fe–FeS systems at high pressure: constraints on core
temperature. Earth Planet. Sci. Lett., 111, 217.
Boehler, R. (1992b) Measurement of the adiabats of quartz, forsterite, and magnesium oxide at high pressures and
high temperatures, and adiabatic gradients in the mantle of the Earth. Phys. Earth Planet. Interiors, 29,
105–107.
Boehler, R. (1993) Temperatures in the Earth’s core from melting - point measurements of Iron at hgih static
pressures. Nature, 363, 534– 536.
Boehler, R. (1996a) Melting temperature of the Earth’s mantle and core Earth’s thermal structure. Annu. Rev.
Earth Planet. Sci., 15–40.
Boehler, R. (1996b) Philos. Trans. R. Soc. London A, 354, 1265– 1278.
Boehler, R. (2000) High-pressure experiments and the phase diagram of low mantle and core materials. Rev.
Geophys., 38(2), 221–245.
Boehler, R. and Chopelas, A. (1992) Phase transitions in a 500 km 3000 K gas apparatus. In: Soyono, Y. and
Manghnani, M.H. (eds.), High Pressure Research in Minerals Physics: Applications to Earth and
Planetary Science. Tokyo/Washington: Terra Scientific/AGU, pp. 55–62.
Boehler, R. and Ramakrishnan, J. (1980) Experimental results on the pressure dependence of the Gruneisen
parameter: a review. J. Geophys. Res., 85, 6996–7002.
Boehler, R. and Ross, M. (1997) Melting curve of aluminium in a diamond cell to 0.8 Mbar: implications for iron.
Earth Planet. Sci. Lett., 153, 223 –227.
Boehler, R., Von Bargen, N. and Chopelas, A. (1990) Melting, thermal expansion and phase transitions of iron at
high pressure. J. Geophys. Res., 95, 21731– 21736.
Boeriogoates, J., Artman, J.I. and Woodfield, B.F. (1990) Heat capacity studies of phase transitions in
langbeinites, 2K2Mg2(SO4)3. Phys. Chem. Miner., 17, 173–178.
Boettger, J.C. and Trickey, S.B. (1985) Phys. Rev., 32, 3391– 3398.
Bogard, D.D. and Johnson, P. (1983) Martian gases in an Antarctic meteorite. Science, 221, 651–654.
Bogard, D.D., Nyquist, L.E. and Johnson, P. (1984) Noble gas contents of Shergottites and implications for the
Martian origin of SNC meteorities. Geochim. Cosmochim. Acta, 48, 1723–1739.
Bolhuis, P.G., Frenkel, D., Mau, S. and Huse, D.A. (1997) Entropy difference between crystal phases. Nature,
388, 235 –236.
Bonaccorsi, E., Merlino, S. and Pasero, M. (1989) Crystal structure of the meteoritic mineral krinovite,
NaMg2CrSi3O10. Z. Kristall., 187, 133–138.
Bonczar, L.J. and Graham, E.K. (1977) The pressure and temperature dependence of the elastic constants of
pyrope garnet. J. Geophys. Res., 82, 2529–2534.
Bones, D.A. and Brown, J.M. (1990) The electronic band structures of iron, sulfur, and oxygen at high pressure
and the Earth’s core. J. Geophys. Res., 95, 21721– 21730.
1112 References

Boness, D.A., Brown, J.M. and McMahan, A.K. (1986) The electronic thermodynamics of iron under Earth core
condition. Phys. Earth Planet. Int., 42, 227– 240.
Borch, R.X. and Green, H.W. II. (1987) Dependence of creep in olivine on homologous temperature and its
implications for flow in the mantle. Nature, 330, 345–348.
Born, M. (1939) Thermodynamics of crystal and melting. J. Chem. Phys., 7, 560–591.
Born, M. and Huang, K. (1982) Dynamical Theory of Crystal Lattices. London: Oxford University Press.
Bose, K. and Ganguly, J. (1995a) Experimental and theoretical studies of the stabilities of talc, antigorite and
phase A at high pressures with applications to subduction processes. Earth Planet. Sci. Lett., 136(3–4),
109–121.
Bose, K. and Ganguly, J. (1995b) Quartz–coesite transition revisted: reversed experimental determination at
500–12008C and retrieved thermodynamic properties. Am. Mineral., 80, 231–238.
Bose, K. and Navrotsky, A. (1998) J. Geophys. Res., 103, 9713.
Bostock, M.G. et al. (2002) Nature, 417, 536.
Bottinga, Y. and Richet, P. (1995) Silicate melts: the anomalous pressure dependence of the viscosity. Geochem.
Cosmochim. Acta, 59(13), 2725–2731.
Bouhifd, M.A., Itie, J.P. and Richet, P. (1995) Compression and amorphization of (Mg, Fe)2SiO4 olivines: X-ray
diffraction study up to 70 GPa. Phys. Chem. Miner., 22, 99–107.
Boyd, F.R. and Nixon, P.H. (1975) Origins of the ultramafic nodules from some kimberlites of northern Lesotho
and the Monastery, South Africa. Phys. Chem. Earth, 9, 431–455.
Boyer, L.L., Cohen, R.E., Krakauer, H. and Smith, W.A. (1990) First principles calculations for ferroelectrics — a
vision. Ferroelectrics, 111, 1–7.
Bozhilov, K.N., Green, H.W. and Dobrzhinetskaya, L. (1999) Clino-enstatite in Alpe Arami peridotite: additional
evidence of very high pressure. Science, 284, 128–132.
Bradon, A.D., Walker, R.J., Morgan, J.W., Norman, M.D. and Prichard, H.M. (1998) Science, 280,
1570– 1573.
Bramwell, S.T. (1999) Ferroelectric ice. Nature, 397, 212–213.
Brannon, P.J. et al. (1983) J. Appl. Phys., 54, 6374.
Braunisch, W., Knauf, N., Kataev, S., Neuhausen, A., Grütz, A., Kock, B., Roden, D., Khomskii, D., and
Wohlleben, W. (1992) Paramagnetic Meissner effect in Bi high-temperature superconductors. Phys. Rev.
Lett., 68, 1908–1911.
Brearly, M. (1990) Ferric silicate melts in the system Na2O–Fe2O3 – SiO2 at high pressure. J. Geo. Res., 95,
15703–15716.
Brearley, A.J., Rubie, D.C. and Ito, E. (1992) Mechanisms of the transformations between the a, b and g
polymorphs of Mg2SiO4 at 15 GPa. Phys. Chem. Miner., 18, 343–358.
Brett, R. (1976) The crust status of speculations on the composition of the core of the Earth. Rev. Geophys. Space
Phys., 14, 375–383.
Breuer, D. and Spohn, T. (1995) Possible flush instability in mantle convection at the Archean–Proterozoic
transition. Nature, 378, 608 –610.
Brewster, M.Q. (1992) Thermal Radiative Transfer and Properties. New York: Wiley, pp. 229– 230, 374–382,
385–386, and 424– 426.
Brey, G., Doroshev, A. and Kogarko, L. (1991) The join pyrope–knorringite: experimental constraints for a new
geothermobarometer for coexisting garnet and spinel. Proceedings of the Fifth International Kimberlite
Conference, 5, 26–28.
Bridgman, P.W. (1923) The compressibility of thirty metals as a function of pressure and temperature. Proc. Am.
Acad. Arts Sci., 58, 165– 242.
Bridgman, P.W. (1925) Linear compressibility of fourteen natural crystals. Am. J. Sci., 210, 483–498.
Bridgman, P.W. (1928) The linear compressibility of thirteen natural crystals. Am. J. Sci., 15, 287– 296.
Bridgman, P.W. (1948) Rough compressions of 177 substances to 40,000 kg/cm2. Proc. Am. Acad. Arts Sci., 76,
71–87.
Bridgman, P.W. (1949) The Physics of High Pressure. London: Bell.
Brigatti, M.F., Contini, S., Capedri, S. and Poppi, L. (1993) Crystal chemistry and cation ordering in
pseudobrookite and armalcolite from Spanish lamproites. Eur. J. Mineral., 5, 73– 84.
References 1113

Brindley, G.W. (1984) Order –disorder in clay mineral structures. In: Brindley, G.W. and Brown, G. (eds.),
Crystal Structures of Clay Minerals and Their X-ray Identification, Mineralogical Society Monograph,
Vol. 5, pp. 125 –129.
Brodholt, J. and Wood, B.J. (1993) Simulations of the structure and thermodynamic properties of water at high
pressures and temperatures. J. Geophys. Res., 98, 519–536.
Bromberg, S.E. and Chan, I.Y. (1992) Rev. Sci. Instrum., 63, 3670–3673.
Brougham, D.F., Caciuffo, R. and Horsewill, A.J. (1999) Coordinated proton tunnelling in a cyclic network of four
hydrogen bonds in the solid state. Nature, 397, 241.
Brown, J.M. (1989) Thermodynamics of liquid iron at high pressure and the iron core. EOS Trans. Am. Geophys.
Union, 70, 1417.
Brown, I.D. (1995) Anion-anion repulsion, coordination number, and the asymmetric of hydrogen bonds. Can. J.
Phys., 73, 676–682.
Brown, M.J. (2002) Annu. Rev. Earth Planet. Sci., 30, 307 –345.
Brown, J.M. and McQueen, R.G. (1982) The equation of state for iron and the Earth’s core. In: Akimoto, S. and
Manghnani, M.H. (eds.), High Pressure Research in Geophysics. Tokyo: Center for Academic
Publications, pp. 611–623.
Brown, J.M. and McQueen, R.G. (1986) Phase transitions, Grüneisen parameter, and elasticity for shocked iron
between 77 GPa and 400 GPa. J. Geophys. Res., 91, 7485–7494.
Brown, N.E. and Navrotsky, A. (1989) Structural, thermodynamic, and kinetic aspects of disordering in the
pseudobrookite-type compound karrooite, MgTi2O5. Am. Mineral., 74, 902– 912.
Brown, J.M. and Shankland, T. (1981) Thermodynamic properties in the earth determined from seismic profiles.
Geophys. J.R. Astron. Soc., 66, 579–596.
Brown, I.D. and Shannon, R.D. (1973) Empirical bond-strength –bond length curves for oxides. Acta Crystallogr.
A, 29, 266–282.
Brown, J.M., Ahrens, T.J. and Shampine, D.L. (1984) Hugoniot data for pyrrhotite and the Earth’s core.
J. Geophys. Res., 89, 6041–6048.
Brown, N.E., Ross, C.R. and Webb, S.L. (1994) Atomic displacements in the normal incommensurate phase
transition in Co-Åkermanite (Ca2CoSi2O7). Phys. Chem. Miner., 21, 469–480.
Brunet, F. and Vielzeuf, D. (1996) The farringtonite/Mg(PO4)2-II transformation: a new curve for pressure
calibration in piston-cylinder apparatus. Eur. J. Mineral., 8(2), 349–354.
Brunet, F. et al. (1995) A new curve for calibrating piston-cylinder assemblies at low pressure, Annual Report.
Bayerisches Forschumgsinstitut für experimentelle Geochemic and Geophysik, Universität Bayreuth,
pp. 146 –147.
Bruzek, C.D., Sharp, T.G., Rubie, D.C. and Durham, W.B. (1998) Mechanisms of transformation and deformation
in Mg1.8Fe0.2SiO4 olivine and wadsleyite under non-hydrostatic stress. Phys. Earth Planet. Interiors,
108, 33 –48.
Buffett, B.A. (2000) Earth’s core and geodynamo. Science, 288, 2007–2012.
Buffett, B.A. and Wenk, H.R. (2001) Texturing of Earth’s inner core by maxwell stresses. Nature, 413, 60 –63.
Buffett, B.A., Huppert, H.E., Lister, J.R. and Woods, A.W. (1996) J. Geophys. Res, 101, 7989.
Bukowinski, M.S.T. (1976) On the electronic structure of iron at core pressures. Phys. Earth Planet. Inter., 13,
57–66.
Bukowinski, M.S.T. (1977) A theoretical equation of state for the inner core. Phys. Earth Planet. Interiors, 14,
333–344.
Bukowinsky, M.S.T. (1980) Effect of pressure on bonding in MgO. J. Geophys. Res., 85(B1), 285– 292.
Bukowinski, M.S.T. and Aidun, A. (1985) First principles equations of state of MgO and CaO. Geophys. Res.
Lett., 12(8), 536 –539.
Bukownski, M.S.T. (1994) Quantum Geophysics. Am. Rev. Earth Planet. Sci., 22, 167–205.
Bukowinski, M.S.T. (1999) Phonons rewrite structural recipes. Nature, 398, 372–373.
Bukownski, M.S.T. and John, A. (1985) First principles versus spherical ion models of the B1 and B2 phases of
NaCl. J. Geophys. Res., 90, 1794–1800.
Bukowinski, M.S.T. and Wolf, G.H. (1986) Equation of state and stability of fluorite-structured SiO2. J. Geophys.
Res., 91(B5), 4704–4710.
1114 References

Bukowinski, M.S.T. and Wolf, G.H. (1988) Equation of state and possible critical phase transitions in MgSiO3
perovskite at lower-mantle conditions. In: Ghosh, S., Coey, J.M.D. and Salje, E. (eds.), Advances in
Physical Geochemistry, Structural and Magnetic Phase Transitions, Vol. 7. New York: Springer,
pp. 91 –112.
Bullard, E.C., Freeman, C., Gellman, H. and Nixon, J. (1950) Phil. Trans. R. Soc. A, 243, 61–92.
Bullen, K.E. (1965) An Introduction to the Theory of Seismology, 3rd edn. New York: Cambridge University
Press, pp. 220–242.
Bundy, F.P. (1965) Pressure temperature phase diagram of iron to 200 kbar 9008C. J. Appl. Phys., 36, 616–620.
Bundy, F.P. and Kasper, J.S. (1967) J. Chem. Phys., 46, 3437.
Bundy, F.P., Bassett, W.A., Weathers, M.A., Hemley, R.J., Mao, H.K. and Goncharov, A.F. (1996) The pressure-
temperature phase and transformation diagram of carbon: updated through 1994. Carbon, 34, 141–153.
Burger, D.M. (1951) In: Smoluchowski, R., Meyer, J.E. and West, W.A. (eds.), Phase Transformation in Solids.
New York: Wiley.
Burnham, C.W. (1975) Thermodynamics of melting in experimental silicate-volatiles systems. Geochim.
Cosmochim. Acta, 39, 1077–1084.
Burnham, C.W. (1994) Development of the Burnham model for predicting of H2O solubility in magmas.
Mineralogic. Soc. Am. Rev. Mineral., 30, 123–130.
Burnley, P.C. and Navrotsky, A. (1997) Synthesis of high-pressure hydrous magnesium silicates: observations and
analysis. Am. Mineral., 81, 317 –326.
Burns, R.G. (1970) Mineralogical Applications of Crystal Field Theory. Cambridge: Cambridge University Press.
Burns, R.G. (1975) On the occurrence and stability of divalent chromium in olivines included in diamonds.
Contrib. Mineral. Petrol., 51, 213 –221.
Burns, R.G. (1976) Partitioning of transition metal ions in mineral structures of the mantle. In: Strens,
R.G.J. (ed.), Physics and Chemistry of Minerals and Rocks. New York: Wiley, pp. 555–572.
Burns, R.G. (1985) Electronic spectra of minerals. In: Berry, J. and Vaughan, D.J. (eds.), Chemical bonding and
Spectroscopy in Mineral Chemistry. London: Chapman & Hall, pp. 63–101.
Burns, R.G. (1987) Polyhedral bulk moduli from high-pressure crystal field spectra. In: Manghnani, M.H. and
Syono, Y. (eds.), High-Pressure Research in Mineral Physics. Tokyo/Washington, DC: Terra Scientific/
AGU, pp. 360–369.
Burns, R.G. (1989) Spectral mineralogy of terrestrial planets: scanning their surfaces remotely. In: Vaughan, D.J.
(ed.), Spectroscopic Studies of Minerals: Principles, Applications, and Advances, Mineral. Mag., 53, pp.
135–151.
Burns, R.G. (1993,1994) Mineralogical Applications of Crystal Field Theory, 2nd edn. Cambridge, UK:
Cambridge University Press, p. 551.
Burns, R.G. and Burns, V.M. (1984) Crystal chemistry of meteoritic hibonite. Proceedings of the 15th Lunar
Planetary Science Conference. J. Geophys. Res., 89, C313–C321.
Burns, R.G. and Sung, C.M. (1978) The effect of crystal field stabilization on the olivine ! spinel transition in the
system Mg2SiO4 –Fe2SiO4. Phys. Chem. Miner., 2, 349–364.
Burton, B.P. and Cohen, R.E. (1994) Theoretical study of cation ordering in the system Pb(Sc1/2Ta1/2)O3.
Ferroelectrics, 151, 331 –336.
Burton, B.P. and Cohen, R.E. (1995) Nonempirical calculation of the Pb(Sc0.5Ta0.5)O3 –PbTiO3 quasibinary
phase diagram. Phys. Rev. B, 52, 792–797.
Callaway, J. (1974) Quantum Theory of the Solid State. New York: Academic Press.
Cameron, A.G.W. and Benz, W. (1991) The origin of the moon and the single impact hypothesis, IV. Icarus, 92,
204–216.
Canil, D. (1994) Stability of clinopyroxene at pressure – temperature conditions of the transition region. Phys.
Earth Planet. Interiors, 86, 25 –34.
Canil, D. and Wei, K. (1992) Constraints on the origin of mantle-derived low Ca-garnets. Contrib. Mineral
Petrol.
Canil, D. and Scarfe, C.M. (1990) Phase relations in peridotite þ CO2 systems at 12 GPa: implications for
origin of kimberlite and carbonate stability in the Earth’s upper mantle. J. Geophys. Res., 95,
15805–15816.
References 1115

Canil, D., O’Nell, H. and Ross, C.R. II. (1990) A preliminary look at phase relations in the system Fe3O4 –
gFe2SiO4 at 7 GPa. Terra Abstr., 3, 65.
Cannillo, E., Mazzi, F., Fang, J.H., Robinson, P.D. and Ohya, Y. (1971) The crystal structure of aenigmatite. Am.
Mineral., 56, 427 –446.
Capobianco, C.J. and Amelin, A.A. (1994) Metal-silicate partitioning of nickel and cobalt: the influence of
temperature and oxygen fugacity. Geochim. Coscmochim. Acta, 58, 125–140.
Capobianco, C. and Carpenter, M.A. (1989) Thermally induced changes in kalsilite, (KAlSiO4). Am. Mineral., 74,
797–811.
Capobianco, C.J., Jones, J.H. and Drake, M.J. (1993) J. Geophys. Res., 98, 5433.
Carbonin, S., Russo, U. and Della Giusta, A. (1996) Cation distribution in some natural spinels from X-ray
diffraction and Mössbauer spectroscopy. Mineral. Mag., 60, 355– 368.
Carlisle, D.B. and Braman, D.R. (1991) Nature, 352, 708–709.
Carlson, R.W. (1998) A conduit to the core. Nature, 394, 11.
Carlson, R.W. (2002) Osmium remembers. Science, 296, 475–476.
Carlson, W.D. and Rosenfeld, J.L. (1981) Optical determination of topotactic aragonite-calcite growth kinetics:
metamorphic implications. J. Geol., 89, 615–638.
Carmichael, I.S.E., Turner, F.J. and Verhoogen, J. (1974) Igneous Petrology. New York: McGraw-Hill.
Carpenter, M.A. (1992) Thermodynamics of phase transitions in minerals: a macroscopic approach. In: Price, G.D.
and Ross, N.I. (eds.), The Stability of Minerals. Boston, MA: Chapman & Hall, pp. 172–216.
Carpenter, M.A., Angel, R.J. and Finger, L.W. (1990) Calibration of Al/Si order variations in anorthite. Contrib.
Mineral. Petrol., 104, 471–480.
Carpenter, M.A., Hemley, R.J. and Mao, H.K. (2000) High pressure elasticity of stishovite and the P42 =mnm ,
Pnnm phase transition. J. Geophys. Res., 105(B5), 10807–10816.
Carr, R. and Perrinello, M. (1985) Unified approach for molecular dynamics and density functional theory. Phys.
Rev. Lett., 55, 2471–2474.
Carroll, M., Hollway, J.R. (ed.) (1994) Volatiles in Magmas, Rev. Mineral., Vol. 30. Washington, DC:
Mineralogical Society of America.
Carslaw, H.S. and Jaeger, J.C. (1959) Conduction of Heat in Solids. Oxford: Clarendon Press.
Carswell, D.A. and Gibb, F.G.F. (1980) Geothermometry of garnet lherzolite nodules with special reference to
those from the kimberlite of northern Lesotho. Contrib. Mineral. Petrol., 74, 403–416.
Carter, J.L. (1970) Mineralogy and chemistry of the Earth upper mantle based on the partial fusion-partial
crystallization model. Geo. Soc. Am. Bull., 81, 2021– 2034.
Carter, N.L. and Ave Lallemant, H.G. (1970) High temperature flow in dunite and periclotite. Gel. Soc. Am. Bull.,
81, 2181–2202.
Casetx, J. and Madon, M. (1990) Test of the vibrational modelling for lambda-type transitions: application to the
alpha-beta phase transition. Phys. Chem. Miner., 22, 1– 10.
Castex, J. and Madon, M. (1995) Test of vibrational modelling for the g-type transitions: application to the a–b
quartz transition. Phys. Chem. Miner., 22(1), 1–10.
Catlows, E.J. and Sorensen, S.S. (2003) Phengite-based chronology of K- and Ba-rich fluid flow in two
paleosubduction zones. Science, 299, 92–94.
Catti, M., Ferraris, G., Hull, S. and Pavese, A. (1995) Static compression and H disorder in brucite, Mg(OH)2, to
11 GPa: a powder neutron diffraction study. Phys. Chem. Miner., 22, 200–206.
Catti, M., Fava, F.F., Zicovich, C. and Dovesi, R. (1999) High pressure decomposition of MCr2O4 spinels
(M ¼ Mg, Mn, Zn) by ab initio methods. Phys. Chem. Miner., 26, 389–395.
Causà, M., Dovesi, R., Pisani, C. and Roetti, C. (1986) Electronic structure and stability of different crystals
phases of magnesium oxide. Phys. Rev. B: Condens. Matter, 33, 1308–1316.
Cavazzoni, C., Chiarotti, G.L., Scandolo, S., Tosatti, E., Bernasconi, M. and Parrinello, M. (1999) Superionic and
metallic states of water and ammonia at giant planet conditions. Science, 283, 44–46.
Cellai, D., Carpenter, M.A. and Heaney, P.J. (1992) Phase transitions and microstructures in natural kaliophilite.
Eur. J. Mineral., 4, 1209–1220.
Cellai, D., Carpenter, M.A., Kirkpatrick, R.J., Salje, E.K.H. and Zhang, M. (1995) Thermally-induced phase
transitions in tridymite: an infrared spectroscopy study. Phys. Chem. Miner., 22, 50–60.
1116 References

Chabrier, G. and Baraffe, I. (1997) Astron. Astrophys., 327, 1039.

Chai, M. and Brown, J.M. (1996) Geophys. Res. Lett., 23, 3539.
Chai, M. and Brown, J.M. (1997) The elastic constants of an aluminous orthopyroxene to 12.5 GPa. J. Geophys.
Res., 102(B7), 14779–14785.
Chai, M., Brown, J.M. and Stutsky, L.J. (1996) Phys. Chem. Miner., 23, 470.
Chakraborty, S. (1995) Diffusion in silicate melts. Rev. Mineral., 32, 411– 503.
Chakraborty, S. (1997) Rates and mechanisms of Fe-Mg interdiffusion in olivine at 980-13008C. Phys. Rev. Res.,
102(B6), 12317–12331.
Chakraborty, S. and Ganguly, J. (1992) Cation diffusion in aluminosilicate garnets: Experimental determination in
spessartine-almandine diffusion couples, evaluation of effective binary diffusion coefficients, and
applications. Contrib. Mineral. Petrol., 111, 74–86.
Chakraborty, S. and Rubie, D.C. (1996) Mg tracer diffusion in aluminosilicate garnets at 750-85008C, 1atm. and
13008C, 8.5 GPa. Contrib. Mineral. Petrol., 122, 406–414.
Chakraborty, S., Farver, J.R., Yund, R.A. and Rubie, D.C. (1994) Mg tracer diffusion in synthetic forsterite and
San Carlos olivine as a function of P, T and fO2. Phys. Chem. Miner., 21, 489–500.
Chakraborty, S., Knoche, R., Schulze, H., Rubie, D.C., Dobson, D., Ross, N.L. and Angel, R.J. (1999)
Enhancement of cation diffusion rates across the 410-kilometer discontinuity in Earth’s mantle. Science,
283, 362 –365.
Chambers, J.E. and Wetherill, G.W. (1998) Icarus, 136, 304.
Chamorro-Perez, E., Gillet, P., Jambon, A., Badro, J. and McMillan, P. (1998) Nature, 393, 352–355.
Chang, L.L.Y. (1965) Subsolidus phase relations in the systems BaCO3-SrCO3, SrCO3 –CaCO3, and BaCO3 –
CaCO3. J. Geol., 73, 346– 368.
Chang, Z.P. and Barsch, G.R. (1969) Pressure dependence of the elastic constants of single-crystalline magnesium
oxide. J. Geophys. Res., 74, 3291–3294.
Chang, Z.P. and Barsch, G.R. (1973) Pressure dependence of single-crystal elastic constsnts and anharmonic
properties of spinel. J. Geophys. Res., 78, 2418–2433.
Chang, K.J. and Cohen, M.L. (1984) High pressure behaviour of MgO: structural and electronic properties. Phys.
Rev. B: Condens. Matter, 30, 4774– 4781.
Chaplot, S.L. and Sikka, S.K. (1992) In: Singh, A.K. (ed.), Recent Trends in High Pressure Research. New Delhi:
IBH, 259 pp.
Chaplot, S.I. and Sikka, S.K. (1993) Phys. Rev. Lett., 71, 2674.
Charbonnel, C. (2002) A baryometer is back. Nature, 415, 27–29.
Chatillon-Colinet, C., Newton, R.C., Perkins, D. and Kleppa, O.J. (1983) Thermochemistry of (Fe2þ, Mg)SiO3
orthopyroxene. Geochim. Cosmochim. Acta, 47, 1597–1603.
Chatterjee, N.D. and Schreyer, W. (1972) The reaction enstatite þ sillimantle ¼ sapphirine þ quartz in the
system MgO–Al2O3 – SiO2. Contrib. Mineral. Petrol., 36, 49 –62.
Chau, P.-L. and Hardwick, A.J. (1998) A new order parameter for tetrahedral configurations. Mol. Phys., 93,
511–518.
Chau, R., Mitchell, A.C., Minich, R.B. and Nellis, W.J. (1999) Electrical conductivity of water at high pressure
and temperatures, International Conference on High Pressure Science and Technology, July 25 –30,
Honolulu, HI, P. 28, Abstr. AIRAPT-17.
Chelikowsky, J.R., King, H.E., Troullier, N., Martians, J.L. and Glinnemann, J. (1990a) Structural properties of a-
quartz near the amorphous transition. Phys. Rev. Lett., 65, 3309–3312.
Chelikowsky, J.R., King, H.E. and Glinnemann, J. (1990b) Interatomic potentials and the structural properties of
silicon dioxide under pressure. Phys. Rev., 41, 10866–10869.
Chelikowsky, J.R., Troullier, N., Martians, J.L. and King, H.E. (1991) Pressure dependence of the structural
properties of a-quartz near the amorphous transition. Phys. Rev., 44, 489–497.
Chen, W.-P. and Brundzinski, M.R. (2001) Science, 292, 2475.
Chen, R. and Kirsh, Y. (1981) Analysis of Thermally Stimulated Processes. Oxford: Pergamon Press.
Chen, J.H. and Wasserberg, G.J. (1996) Live 107Pd in the early solar system and implications for planetary
evolution. In: Bass, A. and Hart, S. (eds.), Earth Processes: Reading the Isotopic Code, AGU
Monograph, 95, pp. 1–20, Washington, DC.
References 1117

Chen, F.R., Davis, J.G. and Fripiat, J.J. (1992) J. Catal., 133, 263.
Chen, C.H., Cheong, S.W., and Cooper, A.S. (1993) Charge modulations in La2-xSrxNiO4þy: Ordering of
polarons. Phys. Rev. Lett., 71, 2461–2464.
Chen, G., Spetzler, H.A., Getting, I.C. and Yoneda, A. (1996a) Selected elastic moduli and their temperature
derivatives for olivine and garnet with different Mg/(Mg þ Fe) contents-results from GHZ ultrasonic
interferometry. Geophys. Res. Lett., 23, 5– 8.
Chen, M., Sharp, T.G., El Goresy, A. et al. (1996b) The majorite pyrope þ magnesiowustite assemblage:
constraints on the history of shock veins in chondrites. Science, 271, 1570–1573.
Chen, G., Libermann, R.C. and Weidner, D.J. (1998) Elasticity of single-crystal MgO to 8 gigapascals and 1600
Kelvin. Science, 280, 1913–1916.
Chen, M., Shu, J., Xie, X. and Mao, H.K. (2003) Natural CaTi2O4-structured FeCr2O4 polymorph in the
Shizhou meteorite and its significance in mantle mineralogy. Geochim. Cosmochim. Acta, 67(20),
3937– 3942.
Chenavas, J., Joubert, J.C., Marezio, M. and Bochu, B. (1975) The synthesis and crystal structure of
CaCu3Mn4O12: a new ferromagnetic perovskite-like compound. J. Solid State Chem., 14, 25–32.
Chester, R. (1990) Marine Geochemistry. London: Unwin Hyman.
Chiari, G. and Ferraris, G. (1982) The water molecule in crystalline hydrates studied by neutron diffraction. Acta
Crystallogr., 38, 2331–2341.
Chopelas, A. (1990) Thermal expansion, heat capacity, and entropy of MgO at mantle pressures. Phys. Chem.
Miner., 17, 142–148.
Chopelas, A. (1991a) Single crystal Raman spectra of forsterite, fayalite, and monticellite. Am. Mineral., 76,
1101– 1109.
Chopelas, A. (1991b) Thermal properties of b-Mg2SiO4 (modified spinel) at mantle pressures derived from
vibrational spectroscopy: implications for the mantle at 400 km. J. Geophys. Res., 96, 11817–11830.
Chopelas, A. (1996a) The fluorescence sideband method for obtaining acoustic velocities at high compressions:
application to MgO and MgAl2O3. Phys. Chem. Miner., 23, 25–37.
Chopelas, A. (1996b) Thermal expansivity of lower mantle phases MgO and MgSiO3 perovskite at high pressure
derived from vibrational spectroscopy. Phys. Earth Planet. Interiors, 98, 3–16.
Chopelas, A. (1999) Estimates of mantle relevant Clapeyron slopes in the MgSiO3 system from high pressure
spectroscopy data. Am. Mineral., 84, 233–244.
Chopelas, A. (2000) Thermal expansivity of mantle relevant magnesium silicates derived from vibrational
spectroscopy at high pressure. Am. Mineral., 85, 270 –278.
Chopelas, A. and Boehler, R. (1989) Thermal expansion measurements at very high pressure, systematics, and a
case for a chemically homogeneous mantle. Geophys. Res. Lett., 16, 1347–1350.
Chopelas, A. and Boehler, R. (1992a) Raman spectroscopy of high pressure MgSiO3 phases synthesized in a CO2
laser heated diamond anvil cell. In: Syono, Y. and Manghnani, M. (eds.), High Pressure Research:
Application to Earth and Planetary Science. Tokyo: Terra Scientific, pp. 101–108.
Chopelas, A. and Boehler, R. (1992b) Thermal expansivity in the lower mantle. Geophys. Res. Lett., 19,
1983– 1986.
Chopelas, A. and Hoffmeister, A.M. (1991) Vibrational spectrsocopy of aluminate spinels at 1 atm and of
MgAl2O4 to over 200 kbar. Phys. Chem. Miner., 18, 279–293.
Chopelas, A., Boehler, R. and Ko, T. (1994) Thermodynamics and behaviour of g-Mg2SiO4 at high pressure:
implications for Mg2SiO4 phase equilibrium. Phys. Chem. Miner., 21(6), 351–359.
Chopelas, A., Reichmann, H.J. and Zhing, L. (1996) Sound velocities of five minerals to mantle pressures
determined by the side band fluorescence method. Geochem. Soc. Spl Publ., 5, 229–242.
Chopin, C., Henry, C. and Michard, A. (1991) Geology and petrology of the coesite-bearing terrain, Dora Maira
massif, Western Alps. Eur. J. Mineral., 3, 263–291.
Chopin, C., Brunet, F., Gebert, W., Medenbach, O. and Tillmanns, E. (1993) Berthite, Ca2Al[PO4]2(OH), a new
mineral from high-pressure terranes of the western Alps. Schweiz. Mineral. Petrogr. Mitt., 73, 1–9.
Chou, I.M. (1987) Oxygen buffer and hydrogen sensor techniques at elevated pressures and temperatures.
In: Ulmer, G.C. and Barnes, H.L. (eds.), Hydrothermal Experimental Techniques. New York: Wiley,
pp. 61 –99.
1118 References

Chou, I.M., Blank, J.G., Goncharov, A.F., Mao, H. and Hemley, R.J. (1998) In situ observations of a high-pressure
phase of H2O ice. Science, 281, 810–813.
Christensen, N.I. (1972) Elastic properties of polycrystalline magnesium, iron, and manganese carbonates to 10
kilobars. J. Geophys. Res., 77, 369 –372.
Christensen, N.I. (1984) The magnitude, symmetry and origin of upper mantle anisotropy based on faberic
analyses of ultramafic tectonites. Geophys. J. R. Astron. Soc., 76, 89–111.
Christensen, U.R. (1995) Annu. Rev. Earth Planet. Sci., 23, 65 –87.
Christensen, U.R. and Hoffman, A.W. (1994) Segregation of subducted oceanic crust in the convecting mantle.
J. Geophys. Res., 89, 4389–4402.
Christensen, U.R. and Yuen, D.A. (1985) The interaction of subducting lithorpheric slab with a chemical or phase
boundary. J. Geophys. Res., 90, 10291–10300.
Christie, D.M. and Chelikowsky, J.R. (1998) Structural properties of a-Berlinite (AlPO4). Phys. Chem. Miner.,
25, 222 –226.
Christie, D.M., Carmichael, I.S.E. and Langmier, C.H. (1986) Oxidation states of mid-ocean ridge basalt glasses.
Earth Planet. Sci. Lett., 79, 397 –411.
Christy, A.G. (1995) Isosymmetric structural phase transitions: phenomenology and examples. Acta Crystallogr.
B, 51, 7753–7757.
Christy, A. and Angel, R.J. (1995) A model for the origin of the doubling phase transition in clinopyroxene and
body-centered anorthite. Phys. Chem. Miner., 22, 129 –135.
Chu, C.W. et al. (1993) Nature, 365, 323.
Chudinovskikh, L. and Boehler, R. (2001) Nature, 411, 574–577.
Chyba, C.F. and Sagan, C. (1992) Endogenous production, exogenous delivery and impact-shock synthesis of
organic molecules: an inventory for the origins of life. Nature, 335, 125–132.
Ciesla, F.J., Lauretta, D.S., Cohen, B.A. and Hood, L.L. (2003) A nebular origin for chondritic fine-grained
phyllosilicates. Science, 299, 549– 552.
Clark, S.P. (1957a) Absorption spectra of some silicates in the visible and near infrared. Am. Mineral., 42,
732–742.
Clark, S.P. Jr. (1957b) Trans. Am. Geophys. Union, 38, 931.
Clark, S.P. and Ringwood, A.E. (1964) Density distribution and constitution of the mantle. Rev. Geophys., 2,
35–42.
Class, C., Goldstein, S. and Galer, S. (1993) Young formation age of a mantle plume source. Nature, 362,
715–721.
Clayton, R.N. and Mayeda, T.K. (2001) Oxygen isotope signatures of hydration reactions in solar system
minerals, 11th Annual V.M. Goldsmidt Conference, Abstract No. 3648. LPI Contribution No. 1088.
Houston: Lunar and Planetary Institute.
Cleary, J.R. and Haddon, R.A.W. (1972) Nature, 240, 549.
Clement, L. Stixrude. (1995) Earth Planet. Sci. Lett., 130, 75.
Cohen, R.E. (1987a) Calculation of elasticity and high pressure instabilities in corundum and stishovite with the
potential induced breathing model. Geophys. Res. Lett., 14, 37–40.
Cohen, R.E. (1987b) Elasticity and equation of state of MgSiO3-perovskite. Geophys. Res. Lett., 14, 1053– 1056.
Cohen, R.E. (1992a) In: Syono, Y. and Manghnani, M.H. (eds.), High-Pressure Research: Application to Earth
and Planetary Sciences. Tokyo/Washington, DC: Terra Scientific/American Geophysical Union,
425 pp.
Cohen, R.E. (1992b) Nature, 358, 136.
Cohen, R.E. (1993a) Origin of ferroelectricity in perovskites: the principal problems from a theoretical
perspective. Ferroelectrics, 150, 1–12.
Cohen, R.E. (1993b) Science, 261, 307– 308.
Cohen, R.E. (1994) First-principles theory of crystalline SiO2. In: Heaney, P.J., Prewitt, C.T. and Gibbs, G.V.
(eds.), Silica: Physica Behavior, Geochemistry and Materials Applications, Reviews in Mineralogy,
Mineralogical Society of America, 29, pp. 369– 402.
Cohen, R.E. (1996) Periodic slab LAPW computations for ferroelectric BaTiO3. J. Phys. Chem. Solid, 57(10),
1393– 1396.
References 1119

Cohen, R.E. (1998a) First-principles investigations of solid iron at high pressure and implications for the Earth’s
inne core. Am. Geophys. Union, 101, 159 –171.
Cohen, R.E. (1998b) Thermal conductivity of MgO at high pressures. High Press. Sci. Tech., 7, 160–162.
Cohen, R.E. (1999) Theory of ferroelectrics: a vision for the next decade and beyond. Carnegie Inst. Washington
Yearbook, 98.
Cohen, R.E. (2000) Theory of ferroelectrics: a vision for the next decade and beyond. J. Phys. Chem. Solids, 61,
139–146.
Cohen, R.E. and Gong, Z. (1994) In: Schmidt, S.C. et al. (eds.), High-Pressure Science and Technology– 1993.
New York: American Institute of Physics, 379 pp.
Cohen, R.E. and Gong, Z. (1994) Melting and melt structure of MgO at high pressures. Phys. Rev. B, 50,
12301–12311.
Cohen, A.J. and Gordon, R.G. (1976) Modified electron-gas stability, elastic properties and high-pressure
behaviour MgO crystals. Phys. Rev. B, Solid State, 14, 4593–4605.
Cohen, R.E. and Krakauer, H. (1990) Lattice dynamics and origin of ferroelectricity in BaTiaO3: linearized-
augmented-plane-wave total-energy calculations. Phys. Rev. B, 42, 6416–6423.
Cohen, R.E., Boyer, L.L., Mehl, M.J., Pickett, W.E. and Krekener, H. (1989) Electronic structures
and total energy calculations of oxides perovskites and superconductors. In: Navtrosky, A. and
Weidner, D.J. (eds.), Perovskite: A Structure of Great Interest to Geophysics and Materials
Science, Geophysics Monograph, Vol. 45. Washington: American Geophysical Union,
pp. 55 –65.
Cohen, R.E., Stixrude, L. and Papaconstantopoulos, D.A. (1994) In: Schmidt, S.C., Shaner, J.W., Samara, G.A.
and Ross, M. (eds.), High Pressure Science and Technology — 1993. New York: American Institute of
Physics, pp. 891 –894.
Cohen, R.E., Mazin, I.I. and Isaak, D.E. (1997) Magnetic collapse in transition metal oxides at high pressure:
implications for the earth. Science, 275, 654 –657.
Cohen, R.E., Stixrude, L. and Wasserman, E. (1997a) Tight-binding computations of elastic anisotropy of Fe, Xe,
and Si under compression. Phys. Rev. B 56(14), 8575–8589.
Cohen, R.E., Fei, Y., Downs, R., Mazin, I.I. and Isaak, D.G. (1998) Magnetic collapse and the behavior of
transition metal oxides: FeO at high pressure. Mater. Res. Soc. Symp. Proc., 499, 27 –37.
Cohen, R.E., Gulseren, O. and Hemley, R.J. (2000) Accuracy of equation-of-state formulations. Am. Mineral., 85,
338–344.
Coleman, R.G. and Wang, V. (1995) Ultrahigh Pressure Metamorphism. Cambridge: Cambridge University
Press.
Collins, G.W., Silva, L.B.D., Celliers, P., Gold, D.M., Foord, M.E., Wallace, R.J., Ng, A., Weber, S.V., Budil,
K.S. and Cauble, R. (1998) Measurements of the equation of state of deuterium at the fluid insulator–
metal transition. Science, 281, 1178.
Committee on Fracture Characterization and Fluid Flow (1996) Rock Fractures and Fluid Flow: Contemporary
Understanding and Applications. Washington, DC: National Academic Press.
Comodi, P. and Zanazzi, P.F. (1995) High pressure structural study of muscovite. Phys. Chem. Miner., 21(3),
170–177.
Comodi, P. and Zanazzi, P.F. (1996) Effects of temperatures and pressure on the structure of lawsonite. Am.
Mineral., 81, 833 –841.
Comodi, P. and Zanazzi, P.F. (1997a) The pressure behaviour of clinozoisite and zoisite: an X-ray diffraction
study. Am. Mineral., 82, 61 –68.
Comodi, P. and Zanazzi, P.F. (1997b) Pressure dependence of structural parameters of paragonite. Phys. Chem.
Miner., 24, 274–280.
Comodi, P., Zanazzi, P.F., Poli, S. and Schmidt, M.W. (1997) High pressure behaviour of kyanite: compressibility
and structural deformations. Am. Mineral., 82, 452 –459.
Conard, P.G. (1998) The stability of almandine at high pessures and temperatures. Geophys. Monogr., 101,
393–399.
Copley, J. (2001) The story of O. Nature, 410, 862–864.
Cord, B. and Courths, R. (1985) Surf. Sci., 152/153, 1141.
1120 References

Cordier, P., Gratz, A.J., Doukhan, J.C. and Nellis, W.J. (1994) Microstructures of AIPO4 subjected to high shock
pressures. Phys. Chem. Miner., 21(3), 133 –139.
Cornish, S.L., Claussen, N.R. et al. (2000) Phys. Rev. Lett., 85, 1795–1798.
Cottet, F., Hallouin, M., Romain, J.P., Fabbro, R., Faral, B. and Pepin, H. (1986) In: Gupta, Y.M. (ed.), Shock
Waves in Condensed Matter. New York: Plenum Press.
Cowley, J.M. (1981) Diffraction Physics, 2nd ed. Amsterdam: North-Holland, 430 pp.
Cox, P.A. (1992) Transition Metal Oxides — An Introduction to their Elestronic Structure and Properties.
Oxford: Oxford University Press.
Craig, H. and Lupton, J. (1976) Earth Planet. Sci. Lett., 31, 369.
CRC (1983) Handbook of Chemistry and Physics, 64th edn. Boca Raton, FL: CRC Press.
Creager, K.C. (1996) Science, 278, 1284.
Creager, K.C. (1997) Anisotropy of the inner core from differential travel times of the phases PKP and PKIKP.
Nature, 356, 309–314.
Crichton, W.A. and Gasparik, T. (1999) Equation of state of magnesium silictes anhydrous B and superhydrous B.
Physics and Chemistry of Minerals, 26(7), 570 –575.
Cristensen, U.R. and Yuen, D.A. (1984) The interaction of a subducting lithospheric slab with a chemical or phase
bounding. J. Geophys. Res., 89, 4389–4402.
Cronin, J.R. and Chang, S. (1998) In: Greenberg, J.M. and Pironello, V.T. (eds.), The Chemistry of Life’s Origin.
Dordrecht, Netherlands: Kulwer, 209 pp.
Cronin, J.R. and Pizzarello, S. (1997) Enantiomeric excesses in meteoritic amino acids. Science, 275,
951–955.
Cronin, J.R., Pizzarello, S. and Cruikshank, D.P. (1988) Organic matter in carbonaceous chondrites, planetary
satellites, asteroids, and comets. In: Kerridge, J.F. and Matthews, M.S. (eds.), Meteorites and the Early
Solar System. Tucson, AZ: University of Arizona Press, pp. 819–857.
Curl, R.F. (1993) Phil. Trans. R. Soc. Lond., 343, 19– 32.
Cuzzi, J.N. et al. (2001) J. Astrophys., 546, 496.
Cygan, R.T., Boslough, M.B. and Kirkpatrick, R.J. (1992) NMR Spectroscopy of experimentally shocked quartz
and plagioclase feldspar powders. Proceedings of the 22nd Lunar Planetary Science Conference, 22,
127–136.
Cynn, H. and Hofmeister, A.M. (1994) High Pressure IR spectra of lattice modes and O–H vibrations in
Fe-bearing wadsleyite. J. Geophys. Res., 99, 17717–17727.
Cynn, H., Isaak, D.G., Cohen, R.E., Nicol, M.F. and Anderson, D.L. (1990) A high-pressure phase transition of
corundum predicted by the potential induced breathing model. Am. Mineral., 75, 439–442.
Cynn, H., Hofmeister, A.M., Burnley, P.C. and Navrotsky, A. (1996) Thermodynamic properties and hydrogen
speciation from vibrational spectra of dense hydrogen speciation form vibrational spectra of dense
hydrous magnesium silicates. Phys. Chem. Miner., 23, 361–376.
Dachillie, F., Zeto, R.J. and Roy, R. (1963) Science, 140, 991.
Dai, P., Mook, H.A. and Dogan, F. (1998) Phys. Rev. Lett., 80, 1738–1741.
Dal, Z.R., Bradley, J.P., Joswiak, D.J., Brownlee, D.E., Hill, H.G.M. and Genge, M.J. (2002) Possible in situ
formation of meteoritic nanodiamonds in the early solar syastem. Nature, 418, 157–158.
Dandekar, D.P. (1968) Variation of the elastic constants of calcite with temperature. J. Appl. Phys., 39,
3694– 3699.
Daniel, I., Gillet, P. and Ghosh, S. (1995) A new high-pressure transition in anorthite (CaAl2Si2O8) revealed by
Raman spectroscopy. Am. Mineral., 80(5–6), 645 –648.
Daniel, I., Gillet, P., McMillan, P.F. and Wolf, G.H. (1997) High pressure behaviour of anorthite compression and
amorphization. J. Geophys. Res., 102, 74– 86.
Daniel, L., Fequet, G. et al. (1999) P – V – T equation of state of lawsonite. Phys. Chem. Miner., 26,
406–414.
Daniels, W.B. (1993) High-pressure techniques. Encycl. Appl. Phys., 7, 497–508.
Daniels, L.R.M., Gurney, J.J. and Harte, B. (1996) A crustal mineral in a mantle diamond. Nature, 379, 153–156.
D’Arco, P.H., Sandrone, G., Dovesi, R., Orlando, R. and Saunders, V.R. (1993) A quantum mechanical study of
MgSiO3 perovskite. Phys. Chem. Miner., 20, 407–414.
References 1121

D’Arco, Ph., Sandrone, G., Dovesi, R., Aprà, E. and Saunders, V.R. (1994) A quantum-mechanical study of
the relative stability under pressure of MgSiO3-ilmenite, MgSiO3-perovskite and MgO-periclase þ
SiO2-stishovite assemblage. Phys. Chem. Miner., 21, 285–293.
Da Silva, C.R.S., Karki, B.B., Stixrude, L. and Wentzcovitch, R.M. (1999) Ab inito study of elastic behaviour of
MgSiO3 ilmenite at pressure. Geophys. Res. Lett., 26, 943–946.
Datchi, F., LeToullec, R., Loubeyre, P., Goncharov, A.F., Somayazulu, M., Hemley, R.J. and Mao, H.K. (1996)
Synchrotron infrared spectroscopy of Ar(H2)2 to 220 GPa. Bull. Am. Phys. Soc., 41, 564.
Davies, G.F. and Dziewonski, A.M. (1975) Homogeneity and constitution of the Earth’s lower mantle and outer
core. Phys. Earth Planet. Interiors, 10, 336 –343.
Davies, R.E. and Koch, R.H. (1991) Philos. Trans. R. Soc., 334B, 391–403.
Davis, H.J. and Stevenson, D.J. (1992) Physical model of source region of subduction zone volcanics. J. Geophys.
Res., 97, 2037–2070.
Dawson, J.B. (1980) Kimberlites and Their Xenoliths. Berlin: Springer, 252 pp.
Deb, S.K., Rekha, M.A., Roy, A.P., Vijayakumar, V., Meenakshi, S. and Godwal, B.K. (1993). Raman-
scattering study of high-pressure phase transition and morphization of resorcinol. Phys. Rev. B, 47,
11491–11494.
Debenedetti, P.G. (1998) One substance, two liquids? Nature, 392, 127–129.
Decker, D.L. (1971) High-pressure equation of state for NaCl, KCl, and CsCl. J. Appl. Phys., 42, 3239–3244.
Decker, D.L. and Zhou, Y.X. (1989) Dielectric and polarization measurements on BaTiO3 at high pressures to the
tricritical point. Phys. Rev. B, 39, 2432–2438.
de Gennes, P.G. (1960) Phys. Rev., 118, 141–154.
de Lingny, D. and Richet, P. (1996) High-temperature heat capacity and thermal expansion of SrTiO3 and SrZrO3
perovskites. Phys. Rev. B, 53, 3013.
Della Giusta, A., Carbonin, S. and Ottonello, G. (1996) Temperature dependent disordering in a natural Mg –Al–
Fe2þ –Fe3þ –spinel. Mineral. Mag., 60, 603– 616.
Dell’Angelo, L.N. (1991) EOS, 71, 458.
Deloule, E. and Robert, F. (1995) Interstellar water in meteorites? Geochim. Cosmochim. Acta, 59, 4695– 4706.
Deloule, E., Albarede, F. and Sheppard, S.M.F. (1991) Hydrogen isotope heterogeneities in the mantle from ion
probe analysis of amphiboles from ultramafic rocks. Earth Planet. Sci. Lett., 105, 543– 553.
Deng, J. et al. (1998) Science, 282, 1689–1692.
Dengel, O., Eckener, V., Plitz, H. and Riehl, N. (1964) Phys. Lett., 9, 291–292.
Depmeier, W. (1988) Aluminate sodalites a family with strained structures and ferroic phase transitions. Phys.
Chem. Miner., 15, 419– 426.
Depmeier, W. (1992) Remarks on symmetric occurring in the sodalite family. Zeit. Kristall., 199, 75 –89.
Dera, P., Prewitt, C.T., Boctor, N.Z. and Hemley, R.J. (2002) Characterization of a high- pressure phase of silica
from the Martian meteorite Shergotty. Am. Mineral., 87, 1018–1023.
Des Marias, D.J., Strauss, H., Summons, R.E. and Haves, J.M. (1992) Nature, 359, 605–609.
De Teresa, J.M. et al. (1997) Nature, 386, 256– 259.
De Wijs, G.A., Kresse, G., Vocadlo, L., Dobson, D., Alfe, D., Gillan, M.J. and Price, G.D. (1998) The viscosity of
liquid iron at the physical conditions of the Earth’s core. Nature, 392, 805.
Dexter, D.L. (1957) Phys. Rev., 108, 707.
di Brozolo, F.R., Bunch, T.E., Fleming, R.H. and Macklin, J. (1994) Fullerenes in a impact crater on the LDEF
spacecraft. Nature, 369, 37–40.
Dickey, J.S. (1970) Partial fusion products in an alpine type peridotites, Serrania di Ronda and other examples.
Mineral. Soc. Am, 3, 33–49, Special paper.
Dickey, J.S. Jr. (1975) A hypothesis of origin for podiform chromite deposits. Geochem. Cosmochim. Acta, 39,
1061– 1074.
Dickinson, J.E., Scarfe, C.M. and McMillan, P. (1990) J. Geophys. Res. B, 95, 15675.
Dingwell, D.B. (1998) The glass transition in hydrous granitic melts. Phys. Earth Planet. Interiors, 107, 1–8.
Dingwell, D.B. and Webb, S.L. (1989) Phys. Chem. Miner., 16, 508–516.
Dingwell, D.B. and Webb, S.L. (1990) Relaxation in silicate melts. Eur. J. Mineral., 2, 427– 449.
1122 References

Dingwell, D.B., Hess, K.U. and Romano, C. (1998) Viscosity data for hydrons peraluminous granitic melt:
comparison with the metaluminous model. Am. Mineral., 83, 236–239.
Dobrzhinetskaya, L., Green, H.W. and Wang, S. (1996) Alpe Arami: a peridotite massif from depths of more than
300 kilometers. Science, 271, 1841–1845.
Dobson, D.P., Meredith, P.C. and Boon, S.A. (2002) Simulation of subduction zone seismicity by dehydration of
serpentine. Science, 298, 1407– 1408.
Dohmen, R., Chakraborty, S., Palme, H. and Rammensee, W. (1998) Am. Mineral., 83, 970.
Dolino, G. and Vallade, M. (1994) Lattice dynamical behaviour of anhydrous silica. Rev. Mineral., 29, 403–431.
Dollase, W.A. and Gustafson, W.I. (1982) 57Fe Mössbauer spectral analysis of the sodic clinopyroxenes. Am.
Mineral., 67, 311 –327.
Dolling, G. and Cowley, R.A. (1966) Proc. Phys. Soc. (London), 88, 463–494.
Domanik, K.J. and Holloway, J.R. (1996) The stability and composition of phengitic muscovite and associated
phases from 5.5 to 11 GPa: implications for deeply subducted sediments. Geochim. Cosmochim. Acta,
60, 4133–4151.
Domeneghetti, M.C., Molin, G.M. and Tazzoli, V. (1995) A crystal-chemical model for Pbca orthopyroxene. Am.
Mineral., 80, 253 –267.
Doukhan, N., Doukhan, J.C., Ingrin, J., Jaoul, O. and Raterron, P. (1993) Early partial melting in pyroxenes. Am.
Mineral., 78, 1246–1256.
Dove, M.T. (1993) Introduction to Lattice Dynamics. Cambridge: Cambridge University Press, 258 pp.
Dove, M.T. (1997) Theory of displacive phase transitions in minerals. Am. Mineral., 82, 213–244.
Dove, M.T., Hagen, M.E., Harris, M.J., Powell, B.M., Steigenberger, U. and Winkler, B. (1992) Anomalous
inelastic neutron scattering from calcite. J. Phys.: Condens. Matter, 4, 2761–2774.
Dove, M.T., Heine, V. and Hammonds, K.D. (1995) Rigid unit modes in framework silicates. Mineral. Mag., 59,
629–639.
Dove, M.T., Thayaparam, S., Heine, V. and Hammonds, K.D. (1996) The phenomenon of low Al/Si ordering
temperatures in aluminosilicate framework structures. Am. Mineral., 81, 349–362.
Dove, M.T., Swainson, I.P., Powell, B.M. and Tennant, D.C. (1997) Structural studies of the orientational order-
disorder phase transition in calcite, CaCO3. Phys. Chem. Miner.
Downs, J.W. (1989) Possible sites for protonation in b-Mg2SiO4 from experimentally derived electrostatic
potential. Am. Mineral., 74, 1124–1129.
Downs, D.R. and Rossman, G.R. (1992) Water in Earth’s mantle: the role of nominally anhydrous minerals.
Science, 255, 1391–1396.
Downs, R.T. and Palmer, D.C. (1994a) The pressure behavior of a-cristobalite. Am. Mineral., 79, 9–14.
Downs, R.T. and Palmer, D.C. (1994b) The pressure behaviour of a-cristobalite. Z. Kristallogr., 121,
369–377.
Downs, R.T. and Somayazulu, M.S. (1998) Carbon dioxide at 1.0 GPa. Acta Crystallogr., 54,
897–898.
Downs, R.T., Hazen, R.M. and Finger, L.W. (1994) The high-pressure crystal chemistry of low albite and the
origin of the pressure dependency of Al–Si ordering. Am. Mineral., 79, 1042–1052.
Downs, R.T., Hazen, R.M. and Finger, L.W. (1995) Crystal chemistry of lead aluminosilicate hollandite: a new
high-pressure synthetic phase with octahedral Si. Am. Mineral., 80, 937–940.
Downs, R.T., Andalman, A. and Hudacsko, M. (1996) The coordination numbers of Na and K atoms in low albite
and microcline as determined from a procrystal electron-density distribution. Am. Mineral., 81,
1344– 1349.
Downs, R.T., Yang, H., Hazen, R.M., Finger, L.W. and Prewitt, C.T. (1999) Compressibility mechanism of alkali
feldspars: new data from reedmergnerite. Am. Mineral., 84, 333– 340.
Dowty, E. and Lindsley, D.H. (1973) Mössbauer spectra of synthetic hedenbergite–ferrosilite pyroxenes. Am.
Mineral., 58, 850 –868.
Drake, M.J. and Righter, K. (2002) Determining the composition of the Earth. Nature, 416, 39–44.
Drake, M.J., McFarlane, E.A., Gasparik, T. and Rubie, D.C. (1993) Mg-perovskite/silicate melt and majorite/
silicate melt partition coefficients in the system CaO–MgO– SiO2 at high temperatures and pressures.
J. Geophys. Res., 98, 5427–5431.
References 1123

Dreibus, G. and Wanke, H. (1985) Mars: a volatile-rich planet. Meteoritics, 20, 367– 382.
Drickamer, H.G. (1966) In: Seitz, F. and Turnbull, D. (eds.), Solid State Physics, Vol. 19. New York: Academic
Press, 135 pp.
Drickamer, H.G. and Frank, C.W. (1973) Electronic Transitions and the High Pressure Chemistry and Physics of
Solids. London: Chapman & Hall, 220 pp.
Drury, M.R. and Fitz Gerald, J.D. (1996) Grain boundary melt films in an experimentally deformed olivine-
pyroxene rock: implications for melt distribution in upper mantle rocks. Geophys. Res. Lett., 23,
401–407.
Dubrovskiy, V.A. and Pan’Kov, V.I. (1972) Izv. Earth Planet., 7, 48.
Dubrovinsky, L.S., Saxena, S.K. and Lazor, P. (1997) X-ray study of iron with in situ heating at ultra high
pressures. Geophys. Res. Lett., 24, 1835–1838.
Dubrovinsky, L.S., Saxena, S.K. and Lazor, P. (1998) High-pressure and high temperature in situ X-ray diffraction
study of iron and corundum to 68 GPa using internally heated diamond-anvil cell. Phys. Chem. Miner.,
25, 434 –441.
Dubrovinsky, L.S., Saxena, S.K., Dubrovinskaia, N.A., Rekhi, S. and Bihan, T.Le (2000) Grüneisen parameter of
1-iron up to 300 GPa from in-situ X-ray study. Am. Mineral., 85, 386– 389.
Dubrovinsky, L.S. et al. (2001) The hardest known oxide. Nature, 410, 653–654.
Duclos, R., Doukhan, N. and Escaig, B. (1978) High temperature creep behaviour of nearly stoichiometric
alumina spinel. J. Mater. Sci., 13, 1740–1748.
Duffy, T.S. and Ahrens, T.J. (1992) In: Syono, Y. and Maghnani, M.H. (eds.), High Pressure Research:
Application to Earth and Planetary Science, American Geophysical Union. pp. 353– 361.
Duffy, T.S. and Ahrens, T.J. (1993) Thermal expansion of mantle and core materials at very high pressures.
Geophys. Res. Lett., 20, 1103–1106.
Duffy, T.S. and Ahrens, T.T. (1994) Shock compression and release of polycrystalline MgO. In: Schmidt, S.C.
et al. (eds.), High Pressure Science & Technology. New York: AIP, pp. 1107–1110.
Duffy, T.S. and Ahrens, T.J. (1995) Compressional sound velocity, equation of state, and constitutive response of
shock-compressed magnesium oxide. J. Geophys. Res., 100, 529–542.
Duffy, T.S. and Anderson, D.L. (1989) Seismic velocities in mantle minerals and the mineralogy of the Earth as
determined from seismic profiles. J. Geophys. Rev., 94, 1895–1912.
Duffy, T.S. and Hemley, R.J. (1995) Temperature structure of the Earth. Rev. Geophys. (US Nat Rep to IUGG),
5–9.
Duffy, T.S., Ahrens, T.J. and Lange, M.A. (1991) The shock wave equation of state of brucite Mg(OH)2.
J. Geophys. Res., 96, 14319–14330.
Duffy, T.S., Vos, W.L., Zha, C.S., Hemley, R.J. and Mao, H.K. (1994) Science, 263, 1590.
Duffy, T.S., Hemley, R.J. and Mao, H.K. (1995a) Equation of state and shear strength and multimegabar
pressures: magnesium oxide to 227 GPa. Phys. Rev. Lett., 74(8), 1371–1374.
Duffy, T.S., Meade, C., Fei, Y., Mao, H. and Hemley, R.J. (1995b) High-pressure phase transition in brucite,
Mg(OH)2. Am. Mineral., 80, 222 –230.
Duffy, T.S., Shu, J., Mao, H.K. and Hemley, R.J. (1995c) Single-crystal X-ray diffraction of brucite to 14 GPa.
Phys. Chem. Miner., 22, 277 –281.
Duffy, T.S. et al. (1998) In MRS Symp. Proc.
Duffy, T.S., Shen, G., Heinz, D.L. et al. (1999) Lattice strains in gold and rhenium under nonhydrostatic
compression to 37 GPa. Phys. Rev. B, 60(22), 15063–15073.
Dupas, C., Doukhan, N., Doukhan, J.-C., Green, H.W. II, Young, T.E. (1994) Analytical electron microscopy of a
synthetic periodotite experimentally deformed in the bolivine stability field. J. Geophys. Res., 99(B8),
15821–15832.
Dupas-Bruzek, C., Tingle, T.J., Doukhan, N. and Doukhan, J.C. (1998) The rheology of olivine and spinel
magnesium germanate (Mg2GeO4): TEM study of defect microstructures. Phys. Chem. Miner., 25,
501–514.
Dupas-Bruzek, C., Sharp, T.G., Rubie, D.C. and Durham, W.B. (1998) Mechanism of transformation and
deformation in Mg0.8Fe0.2SiO4 olivine and wadsleyite under hydrostatic stress. Phys. Earth Planet.
Interiors, 108, 33 –48.
1124 References

Durben, D.J. and Wolf, G.H. (1991a) Raman spectroscopic study of the pressure-induced coordination change in
GeO2 glass. Phys. Rev. B, 43, 2355–2363.
Durben, D.J. and Wolf, G.H. (1991b) Thermally induced vitrification and internal stress in magnesium silicate
(MgSiO3) perovskite. Eos. Trans. Am. Geophys. Union, 72(44), 464, Abstracts.
Durben, D.J., McMillan, P.F. and Wolf, G.H. (1993) Raman study of the high-pressure behaviour of forsterite.
Am. Mineral., 78, 1143–1148.
Dyar, M.D., McGuire, A.V. and Ziegler, R.D. (1989) Redox equilibria and crystal chemistry of coexisting
minerals from spinel-lherzolite mantle xenoliths. Am. Mineral., 74, 969–980.
Dyar, M.D., McGuire, A.V. and Harrell, M.D. (1992) Crystal chemistry of iron in two styles of metasomatism in
the upper mantle. Geochim. Cosmochim. Acta, 56, 2579–2586.
Dyar, M.D., Mackwell, S.J., McGuire, A.V., Cross, L.R. and Robertson, J.D. (1993) Crystal chemistry of Fe3þ
and Hþ in mantle kaersutite: implications for mantle metasomatism. Am. Mineral., 78,
968–979.
Dyar, M., McCammon, D. and Schaeffermw, C. (eds.) (1996) A Tribute to Roger G. Burns. Houston, TX:
Geochemical Society.
Dziewonski, A. and Anderson, D.L. (1981) Preliminary reference Earth model. Phys. Earth Planet. Interiors, 25,
297–356.
Dziewonski, A.M. and Gilbert, F. (1971) Solidity of the inner core of the Earth inferred from normal modes.
Geophys. Res. Lett., 234, 465–466.
Eberhardt, P., Reber, M., Krankowsky, D. and Hodges, R.R. (1995) The D/H and 18O/16O ratios in water from
comet Halley. Astron. Astrophys., 302, 301–318.
Ecke, R.E., Hu, Y., Mainieri, R. and Ahlers, G. (1995) Excitation of spirals and chiral symmetry breaking in
Rayleigh–Benard convection. Science, 269, 1704.
Eckert, H., Yesinowski, J.P., Silver, L.A. and Stopler, E.M. (1988) Water in silicate glasses: quantitation
and structural studies by 1H solid echo and MAS – NMR methods. J. Phys. Chem., 92,
2055– 2064.
Edelson, E. (1991) Popular Sci., 52, 239.
Edgar, A.D. and Vukadinovic, D. (1993) Potassium-rich clinopyroxene in the mantle: an experimental
investigation of a K-rich lamproite up to 60 kbar. Geochim. Cosmochim. Acta, 57, 5063–5072.
Edwards, B. and Ashcroft, N.W. (1997) Nature, 388, 652–655.
Edwards, P.P. and Hensel, F. (1997) Will solid hydrogen ever be a metal? Nature, 388, 621–622.
Eggert, J.H., Mao, H.K. and Hemley, R.J. (1993) Phys. Rev. Lett., 70, 2301– 2304.
Eggler, D.H., Kushiro, I. and Holloway, J.R. (1976) Stability of carbonate minerals in a hydrous mantle. Carnegie
Inst. Washington Year Book, 75, 631– 636.
Eggleton, R.A., Boland, J.N. and Ringwood, A.E. (1978) High pressure synthesis of a new aluminum silicate:
Al5Si5O17(OH). Geochem. J., 12, 191 –194.
Eide, E., Liou, J.G. and McWilliams, M.O. (1994) 40Ar/39Ar constraints on high and ultrahigh pressure
metamorphism, Hubei Province, China. Geology, 22, 601–604.
Einstein, A. (1907) Planck’s theory of radiation and the theory of specific heat. Am. Phys., 22, 180.
Einstein, A. (1924) Sitzungsber. Kgl. Preuss. Akad. Wiss., 1924, 261.
Ekström, G. and Dziewonski, A.M. (1998) The unique anisotropy of the Pacific upper mantle. Nature, 394,
168–169.
Ellemaans, J.B.A., Vaanlaar, B., Vander-Veer, K.J.R. and Loopstar, B.O. (1971) The crystallographic and
magnetic structures of La12xBaxMn12yMyO3. J. Solid State Chem., 3, 238–242.
Elliot, S.R. (1991) Nature, 354, 445.
Emin, D. (1975) Transport properties of small polarons. J. Sol. State Chem., 12, 246–252.
Enami, M., Zang, Q. and Yin, Y. (1993) High pressure eclogites in northern Jiangsu-southern Shandong province,
eastern China. J. Metamorph. Geol., 11, 589–604.
Endo, S. and Ito, K. (1982) Electrical resistance of a-Fe2O3 under ultrahigh static pressure. In: Akimoto, S. and
Manghnani, M.H. (eds.), High Pressure Research in Geophysics. Tokyo: Center for Academic
Publishing, pp. 189–190.
References 1125

Engelhardt, W.V. and Stöffer, D. (1968) In: French, B.M. and Short, N.M. (eds.), Shock Metamorphism of Natural
Materials. Baltimore: Mono Book Corporation.
England, P. and Jackson, J. (1989) Annu. Rev. Earth Planet. Sci., 17, 197.
England, P. and McKenzie, D. (1982) J. R. Astron. Soc., 70, 295–321.
England, P. and Molnar, P. (1997) Active deformation of Asia: from kinematics to dynamics. Science, 278,
647–649.
England, P. and Molnar, P. (1997) Geophys. J. Int., 130, 551–582.
Eom, D. and Holstein, T. (1969) Studies of polaron motion, Part IV: small polaron adiabatic Hall effect. Am.
Phys., 53, 439–520.
Eremets, M. (1996) High Pressure Experimental Methods. New York: Oxford University Press.
Eremets, M.I., Hemsey, R.J., Mao, H.-K. and Gregoryanz, E. (2001) Semiconducting non-molecular nitrogen up
to 240 GPa and its low-pressure stability. Nature, 411, 170–173.
Eremets, M.I., Shimizu, K., Kobayashi, T.C. and Amaya, K. (1998) Metallic CsI at pressures of up to 220
Gigapascals. Science, 281, 1333–1335.
Erlich, H.E. (1995) Geomicrobiology. New York: Marcel Dekker, 719 pp.
Ernst, W.G. (1988) Tectonic history of subduction zones inferred from retrograde blueschist P – T paths. Geology,
16, 1081–1084.
Ernst, W.G. and Peacock, S.M. (1996) A thermotectonic model for preservation study of five Eurasian ulrahigh
pressure mineralogical relics in metamorphosed continental crust in subduction. In: Bebout, G.E. et al.
(eds.), American Geophysical Union Monograph, Vol. 96, pp. 171– 178.
Errington, J.R. and Debenedetti, P.G. (2001) Relationship between structural order and the anomalies of liquid
water. Nature, 409, 318 –321.
Erskine, D.J. and Nellis, W.J. (1992) J. Appl. Phys., 71, 1.
Esperanca, S. et al. (1997) Dating crust–mantle separation: Re–Os isotopic study of mafic xenoliths from central
Arizona. Geology, 25(7), 651– 654.
Essene, E. (1974) High-pressure transformations in CaSiO 3. Contrib. Mineral. Petrol., 45,
247–250.
Essene, E.J. and Fischer, D.C. (1986) Lightening strike fusion: extreme reduction metal-silicate liquid
immiscibility, Science, 234, 189–193.
Esser, B.K. (1991). Osmium isotope geochemistry of terrigenous and marine sediments. PhD Thesis, Yale
University, Connecticut, USA.
Evans, B.W., Johannes, W., Oterdoom, H. and Trommsdorf, V. (1976) Stability of chrysotile and antigorite in the
serpentine multisystem. Schweiz Mineral. Petrogre. Mitt., 56, 79–93.
Every, A.G. (1980) Phys. Rev., 22, 1746–1760.
Fabrichnaya, O.B. (1995) Thermodynamic data for phases in the FeO–MgO– SiO2 system and phase relations in
the mantle transition zone. Phys. Chem. Miner., 22, 323.
Fabrichnaya, O.B. and Sundman, B. (1997) The assessment of thermodynamic parameters in the Fe–O and
Fe–Si–O system. Geochim. Cosmochim. Acta, 61, 4539–4555.
Falzone, A.J. and Stacey, F.D. (1980) Second-order elasticity theory: explanation for the high Poisson’s ratio of
the inner core. EOS. Trans. Am. Geophys. Union, 67, 1100.
Fang, Z., Solovyev, I., Terakura, K., Sawada, H. and Miyazaki, T. (1999) First-principles study on electronic
structures and phase stability of MnO and FeO under high pressure. Phys. Rev., 59, 762.
Farber, D.L. and Williams, Q. (1996) Am. Mineral., 81, 273.
Farber, D.L., Williams, Q. and Ryerson, F.J. (1994) Nature, 371, 693–695.
Farmer, V.C. and Lazarev, A.N. (1974) Symmetry and crystal vibrations. In: Farmer, V.C. (ed.), The Infrared
Spectra of Minerals. London: Mineralogical Society, pp. 51–174.
Farrell, A. et al. (1991) Nature, 349, 770.
Fatseas, G.A., Dormann, J.L. and Blanchard, H. (1976) Study of the Fe3þ/Fe2þ ratio in natural chromites
(FexMg12x)(Cr12y 2 zFeyAlz)O4. J. Phys., 12, 787–790.
Faust, J. and Knittle, E. (1994) Static compression of chondrodite: implications for water in the upper mantle.
Geophys. Res. Lett., 21, 1935–1938.
1126 References

Faust, J. and Williams, Q. (1996) Infrared spectra of phase B at high pressures: hydroxyl bonding under
compression. Geophys. Res. Lett., 23, 427–430.
Fei, Y. (1996) Crystal chemistry of FeO at high pressure and temperature. Geochem. Soc., 5, 243–254, Special
issue.
Fei, Y. (1999) Effects of temperature and composition on the bulk modulus of (Mg, Fe)O. Am. Mineral., 84,
272–276.
Fei, Y. and Campbell, A.J. (1995) EOS Trans., 76, 607.
Fei, Y. and Mao, H.K. (1991) In situ X-ray diffraction study of wüstite (Fe0.947O) up to 70 GPa and 600 K. EOS,
72, 465, Abstract.
Fei, Y. and Mao, H.K. (1993) Static compression of Mg(OH)2 to 78 GPa at high temperature and constraints on
the equation of state of fluid H2O. J. Geophys. Res., 98, 11875–11884.
Fei, Y. and Mao, H.K. (1994) In-situ determination of the NiAs phase of FeO at high pressure and temperature.
Science, 266, 1678–1680.
Fei, Y. and Prewitt, C.T. (1996) High-pressure behavior of iron sulfide. Mater. Res. Soc. Symp. Proc., 410,
223–228.
Fei, Y. and Saxena, S.K. (1986) A thermochemical data base for phase equilibria in the system Fe– Mg–Si–O at
high pressure and temperature. Phys. Chem. Miner., 13, 311–324.
Fei, Y., Bertka, C.M., Prewitt, C.T. (2003) Structure type and bulk modulus of Fe3S, a new iron-sulfur compound.
Am. Min., 85, 1830–1833.
Fei, Y., Saxena, S.K. and Navrotsky, A. (1990) Internally consistent thermodynamic data and equilibrium phase
relations for compounds in the system MgO–SiO2 at high pressure and high temperature. J. Geophys.
Res., 95, 6915–6928.
Fei, T., Mao, H.K. and Mysen, B.O. (1991) Experimental determination of element partitioning and calculation of
phase relations in the MgO–FeO–SiO2 system at high pressure and high temperature. J. Geophys. Res.,
96(B2), 2157–2169.
Fei, Y., Mao, H.K., Shu, J., Parthasarathy, G., Bassett, W.A. et al. (1992) Simultaneous high-P, High-T X-ray
diffraction study of b-(Mg, Fe)2SiO4 to 26 GPa and 900 K. J. Geophys. Res., 97(B4), 4489–4495.
Fei, Y., Virgo, D., Mysen, B.O., Wang, Y. and Mao, H.K. (1994) Temperature dependent electron delocalization
in (Mg, Fe)SiO3 pervoskite. Am. Mineral., 79, 826 –837.
Fei, Y., Prewitt, C.T., Mao, H.K. and Bertka, C.M. (1995) Structure and density of FeS at high pressure and high
temperature and the internal structure of Mars. Science, 268, 1892–1894.
Fei, Y., Bertka, M. and Finger, L.W. (1997) High pressure iron–sulfur compounds, Fe3S2, and melting relations in
the Fe –FeS system. Science, 275, 1621–1622.
Fei, Y., Prewitt, C.T., Frost, D.J., Parise, J.B. and Brister, K. (1998) Structures of FeS polymrophs at high pressure
and temperature. High Press. Sci. Technol., 7, 55–58.
Fei, Y., Frost, D.J., Mao, H.K., Prewitt, C.T. and Häusermann, D. (1999) In situ structure determination of the
high-pressure phase of Fe3O4. Am. Mineral., 84, 203–206.
Fei, Y., Li, J., Bertka, C.M., Prewitt, C.T. (2003) Structure type and bulk modulus of Fe3S, a new iron-sulfur
compound. Am. Min., 85, 1830–1833.
Ferguson, C.C., Lloyd, G.E., Knipe, R.J. and Can, J. (1987) Earth Sci., 24, 544.
Fiebig, M., Miyyano, K., Tomioka, Y. and Tokura, Y. (1998) Science, 280, 1925–1928.
Finger, L.W. (1995) Minerals at high pressure. Nucl. Instrum. Meth. Phys. Res. B, 97, 55– 62.
Finger, L.W. and Hazen, R.M. (1978) Crystal structure and compression of ruby to 46 kb. J. Appl. Phys., 49,
5823– 5826.
Finger, L.W. and Hazen, R.M. (1980) Crystal structure and isochemical compression of Fe2O3, Cr2O3 and V2O3 to
50 kbars. J. Appl. Phys., 51, 5362– 5367.
Finger, L.W. and Hazen, R.M. (1987) Crystal structure of monoclinic ilvaite and the nature of the monoclinic-
orhorhombic transition at high pressure. Zeit. Kristallogr., 179, 415–430.
Finger, L.W. and Hazen, R.M. (1991) Crystal chemistry of six-coordinated silicon: a key to understanding the
Earth’s deep interior. Acta Crystallogr. B, 47, 561 –580.
Finger, L.W., Hazen, R.M. and Yagi, T. (1977) High pressure crystal structures of the spinel polymorphs of
Fe2SiO4 and Ni2SiO4. Carnegie Inst. Year Book Washington, 76, 504–505.
References 1127

Finger, L.W., Hazen, R.M. and Yagi, T. (1979) Crystal structures and electron densities of nickel and iron silicate
spinels at elevated temperature or pressure. Am. Mineral., 64, 1002– 1009.
Finger, L.W., Hazen, R.M. and Hofmeister, A.M. (1986) High pressure crystal chemistry of spinel (MgAl2O4) and
magnetite (Fe3O4): comparisons with silicate spinels. Phys. Chem. Miner., 13, 215–220.
Finger, L.W., Ko, H., Hazen, R.M., Gasparik, T., Hemley, R.J., Prewitt, C.T. and Weidner, D.J. (1989) Crystal
chemistry of phase B and new anhydrous magnesium silicate: implications for water storage in the upper
mantle. Nature, 341, 140–142.
Finger, L.W., Hazen, R.M. and Prewitt, C.T. (1991) Crystal structures of Mg12Si4O19(OH)2 (phase B) and
Mg14Si5O24 (phase AnhB). Am. Mineral., 76, 1–7.
Finger, L.W., Hazen, R.M., Zhang, J., Ko, J. and Navrotsky, A. (1993) The effect of Fe on the crystal structure of
wadsleyite b-(Mg12xFex)2SiO4, 0.00 # x # 0.40. Phys. Chem. Miner., 19, 361–368.
Finnerty, A.A. and Boyd, F.R. (1987) Thermobarometry for garnet peridotites: basis for the determination
of thermal and compositional structure of the upper mantle. In: Nixon, P.H. (ed.), Mantle Xenoliths.
New York: Wiley, pp. 381–402.
Fiquet, G. and Reynard, B. (1999) High-pressure equation of state of magnesite: new data and a reappraisal. Am.
Mineral., 84, 856 –860.
Fiquet, G., Guyot, F. and Itié, J.P. (1994) High pressure X-ray diffraction study of carbonates — MgCO3,
CaMg(CO3)2, and CaCO3. Am. Mineral., 79, 15–23.
Fiquet, G., Andrault, D., ltié, J.P., Gillet, P. and Richet, P. (1996) Phys. Earth Planet. Interiors, 95, 1.
Fiquet, G., Andrault, D., Dewade, A., Charpin, T., Kunz, M. and Hausermann, D. (1998) P– V–T equation of state
of MgSiO3 perovskite. Phys. Earth Planet. Interiors, 105, 21–31.
Fischer, J.E. and Heiney, P.A. (1993) Order and disorder in fullerene solids. J. Phys. Chem. Solids, 54,
1725– 1757.
Fisk, Z. and Pines, D. (1998) The ghost of magnetism. Nature, 394, 22– 23.
Fiske, P.S., Nellis, W.J., Lipp, M., Lorenzana, H., Kikuchi, M. and Syono, Y. (1995) Pseudotachylites generated
in shock experiments implication for impact cratering products and processes. Science, 270, 281–282.
Fitzgerald, J.D. and Ringwood, A.E. (1991) High pressure rhombohedral perovskite phase Ca2AlSiO5.5. Phys.
Chem. Miner., 18, 40– 46.
Fleet, M.E. (1981) The structure of magnetite. Acta Crystallogr., 37, 917–920.
Fleet, M.E., Herzberg, C.T., Henderson, G.S., Crozier, E.D., Osborne, M.D. and Scarfe, C.M. (1984) Coordination
of Fe, Ga and Ge in high pressure glasses by Mössbauer, Raman and X-ray absorption spectroscopy, and
geological implications. Geochim. Cosmochim. Acta, 48, 1455–1466.
Fletcher, G. and Addis, R. (1974) J. Phys. F, 4, 1951.
Flynn, J.J. and Boer, F.P. (1969) J. Am. Chem. Soc., 91, 5756–5761.
Fockenberg, T. (1995) New experimental results upto 100 kbar in the system MgO–Al2O3 –SiO2 –H2O (MASH):
preliminary stability fields of chloritiod, staurolite, MgMgAl– pumpellyite and pyrope. Bochumer Geol.
Geotech. Arbeiten., 44, 39 –44.
Fockenberg, T. (1998) An experimental study of the pressure-temperature stability of MgMgAl– pumpellyite in
the system MgO–Al2O3 –SiO2 –H2O. Am. Mineral., 83, 220–227.
Fockenberg, T., Wunder, B., Grevel, K.D. and Burchard, M. (1996) The equilibrium diaspore-corundum at high
pressure. Eur. J. Mineral., 8, 1293–1299.
Foit, F.F. and Peacor, D.R. (1973) The anorthite crystal structure at 410 and 8308C. Am. Mineral., 58, 665–675.
Fontana, M.D., Metrat, G., Servoin, J.L. and Gervais, F. (1984) J. Phys. C: Solid State Phys., 16, 483.
Forte, A.M. and Mitrovica, J.X. (2001) Deep mantle high-viscosity flow and thermochemical structure inferred
from seismic and geodynamic data. Nature, 410, 1049–1053.
Foster, J.G., Lambert, D.D., Frick, L.R. and Maas, R. (1996) Nature, 382, 703– 706.
Fowler, F. (1983) Experimental and theoretical nuclear astrophysics: the quest for the origin of the elements,
Nobel Lectures, December 8, 172 pp.
Fowler, P.W. and Monolopoulus, D.E. (1995) An Atlas of Fullerenes. Oxford: Clarendon Press.
Frank, S., Poncharal, P., Wang, Z.L. and de Heer, W.A. (1998) Science, 280, 1744.
Frankel, J.J. (1959) Uvarovite garnet and South African jade from Transvaal. Am. Mineral., 44, 565–591.
Franson, A. and Ross, R.G. (1983) J. Phys. C. Solid State Phys., 12, 219.
1128 References

Freed, A.M. and Lin, J. (2001) Nature, 411, 180–183.

Freeman, A.J. (1995) Electronic structure theory in the new age of computational materials science. Annu. Rev.
Mater. Sci., 25, 1– 5.
Freer, R. (1981) Diffusion in silicate minerals and glasses: a data digest and guide to the literature. Contrib.
Mineral. Petrol., 76, 440 –454.
Freiman, Yu.A. (1997) In: Manzehelii, V.G. and Freiman, Yu.A. (eds.), Physics of Cryocrystals. Woodbusy, NY:
American Institute of Physics, pp. 538– 596.
French, B.M. and Short, N.M. (1968) Shock Metamorphism of Natural Materials, Baltimore.
Frisilo, A.L. and Barsch, G.R. (1972) Measurement of single-crystal elastic constant of bronzite as a function of
pressure and temperature. J. Geophys Res., 77, 6360–6384.
Fritz, J.N., Marsh, S.P., Carter, W.J. and McQueen, R.G. (1971) The Hugoniot equation of state of sodium
chloride in NaCl structure. In: Lloyd, E.C. (ed.), Accurate Characterization of the High Pressure
Environment, US National Bureau of Standards Special Publication 326, pp. 201–208.
Fritzel, T.L.B. and Bass, J.D. (1997) Sound velocities of clinohumite, and implications for water in Earth’s upper
mantle. Geophys. Res. Lett., 24, 1023– 1026.
Frohlich, C. (1987) Kiyoo Wadati and early research on deep focus earthquakes: introduction to special section on
deep and intermediate focus earthquakes. J. Geophys. Res., 92, 13777– 13788.
Frolich, C. (1989) Annu. Rev. Earth Planet. Sci., 17, 250.
Frost, H.J. and Ashby, M.F. (1982) Deformation Mechanism Maps. Oxford: Pergamon Press.
Frost, D.J. and Fei, Y. (1998) Stability of phase D at high pressure and high temperature. J. Geophys. Res., 103,
7463– 7474.
Frost, D.J. and Fei, Y. (1999) Static compression of the hydrous magnesium silicate phase D to 30 GPa at room
temperature. Phys. Chem. Miner., 26, 415–418.
Frost, R.L. and Van Der Gaast, S.J. (1997) Kaolinite hydroxyl — a Raman microscopy study. Clay Miner., 45, 68 –72.
Frost, D.J. and Wood, B.J. (1997) Experimental measurements of the properties of H2O–CO2 mixtures at high
pressures and temperatures. Geochim. Cosmochim. Acta, 61(16), 3301–3309.
Frost, B.R., Lindsley, D.H. and Anderson, D.J. (1988) Fe†Ti oxide–silicate equilibria: assemblages with fayalite
olivine. Am. Mineral., 727–740.
Frost, R.L., Tran, T.H. and Kristof, J. (1997) Intercalation of an ordered kaolinite — a Raman microscopy study.
Clay Miner., 32, 587–596.
Frost, R.L., Kloprogge, J.T., Tran, T.H.T. and Kristof, J. (1998) The effect of pressure on the intercalation of an
ordered kaolinite. Am. Mineral., 83, 1182–1187.
Froster, A.L., Brown, G.E., Tingle, T.N. and Parks, G.A. (1998) Quantitative arsenic speciation in mine tailings
using X-ray absorption spectroscopy. Am. Mineral., 83, 553–568.
Fuji, T. (1978) Fe–Mg partitioning between olivine and spinel. Carnegie Inst. Geophys. Lab. Rep. 1976– 1977,
563–569.
Fuji, Y., Kowaka, M. and Onodera, A. (1985) J. Phys. C: Solid State Phys., 18, 789.
Fujino, K., Momoi, H., Sawamoto, H. and Kumazawa, M. (1986) Crystal structure and chemistry of MnSiO3
tetragonal garnet. Am. Mineral., 71, 781 –785.
Fujisawa, H., Fujii, N., Mizutani, H., Kanamori, H. and Akimoto, S. (1968) J. Geophys. Res., 75, 4727.
Fujishiro, I., Piermarini, G.J., Block, S. and Munro, R.G. (1981) Viscosities and glass transition pressure in the
methanol–ethanol–water system. Eighth AIRAPT Conference Abstracts, 608–611.
Fukai, Y. (1993) The Metal Hydrogen System. Berlin: Springer.
Fukao, Y., Obayashi, M., Inoue, H. and Nenbai, M. (1992) Subducting slabs stagnant in the mantle transition zone.
J. Geochim. Cosmochim. Acta, 49, 865–870.
Funamori, N. and Yagi, T. (1993) High pressure and high temperature in situ X-ray observation of MgSiO3
perovskite under lower mantle conditions. Geophys. Res. Lett., 20, 387–390.
Funamori, N., Yagi, T. and Uchida, T. (1996a) High-pressure and high-temperature in situ X-ray diffraction study
of iron to above 30 GPa using MA8-type apparatus. Geophys. Res. Lett., 23, 953–956.
Funamori, N., Yagi, T., Utsumi, W., Kondo, T., Uchida, T. and Funamori, M. (1996b) Thermoelastic properties of
MgSiO3 perovskite determined by in situ X-ray observations up to 30 GPa and 2000 K. J. Geophys. Res.,
101, 8257–8269.
References 1129

Furche, A. and Langer, K. (1998) Polarized electronic absorption spectra of Cr2SiO4 single crystals. Phys. Chem.
Miner., 25, 393–400.
Furukawa, N. (1993) Depth of the decoupling plate interface and thermal stresses under arcs. J. Geophys. Res., 98,
20005–20013.
Furukawa, N. (1994) J. Phys. Soc. Jpn, 63, 3214.
Gaetani, G.A., Grove, T.L. and Bryan, W.B. (1993) The influence of water on the petrogenesis of subduction-
related igneous rocks. Nature, 365, 332–334.
Gaherty, J.B., Puster, P., Jordan, T.H., Wang, Y. and Weidner, D.J. (1997) EOS, 78, F 486.
Galer, S.J.G. and Goldstein, S.L. (1991) Early mantle difterentiation and its thermal consequeuces. Geochim.
Cosmochim. Acta, 55, 227–239.
Galer, S.J.G. and Goldstein, S.L. (1996) Influence of accretion of lead in the Earth. In: Basu, A. and Hart, S. (eds.),
Earth Processes: Reading the Isotopic Code, AGU Monograph, Vol. 95. Washington, DC: AGU,
pp. 75 –98.
Galli, G., Martin, R., Car, R. and Parrinello, M. (1989) Phys. Rev. Lett., 63, 988.
Galli, G., Martin, R., Car, R. and Parrinello, M. (1990) Science, 250, 1547.
Gallup, R.F. and Coleman, L.B. (1990) Vibration in spectra and ferroelectric phase transition of colemanite. Phys.
Chem. Miner., 17, 271– 274.
Galoisy, L. and Calas, G. (1991) Spectrocopic evidence for five-coordinated Ni in CaNiSi2O6 glass. Am. Mineral.,
76, 1777–1781.
Galoisy, L. and Calas, G. (1992) Am. Mineral., 77, 677 –680.
Galvao de Silva, E., Aloras, A. and Speziali, N.Z. (1980) Mössbauer effect study of natural chromites of Brazilian
and Philippine origin. Appl. Phys., 22, 389 –392.
Ganguly, J., Cheng, W. and Chakraborty, S. (1998) Contrib. Mineral. Petrol., 131, 171.
Gao, L., Xue, Y.Y., Chen, F., Xiong, Q., Meng, R.L., Ramirez, D., Chu, C.W., Eggert, J.H. and Mao, H.K. (1994)
Phys. Rev. B, 50, 4260– 4263.
Garav, L.I., Anglin, I.R., Cirac, I.I. and Zoller, P. (2000) Sonic analog of Gravitational black holes in Bose-
Einstein condensates. Phys. Rev. Lett., 85, 4643–4647.
Garg, N. and Sharma, S.M. (2000) A molecular dynamical investigation of high pressure phase transformations in
berlinite (a-AlPO4). J. Phys., 12, 375–397.
Garnero, E.J. and Helmberger, D.V. (1995) A very slow basal layer underlying large-scale low-velocity anomalies
in the lower mantle beneath the Pacific: evidence from core phases. Phys. Earth Planet. Interiors, 91,
161–176.
Garnero, E.J. and Helmberger, D.V. (1996) Seismic detection of a thin laterally varying boundary layer at the base
of the mantle beneath the central Pacific. Geophys. Res. Lett., 23, 977– 980.
Garnero, E.J. and Helmberger, D.V. (1998) Seismic detection of a thin laterally varying boundary layer at the base
of the mantle beneath the central-Pacific. Geophys. Res. Lett., 19, 1049–1052.
Garnero, E.J. and Jeanloz, R. (2000) Earth’s enigmatic interface. Science, 289, 70–71.
Gasparik, T. (1977) Discovery of Na0.7Mg1.8Fe3þ 0.3 Si2.2O7: possible major mineral constituent of the upper mantle.
EOS, 78, Spring Meeting Supplement, S314.
Gasparik, T. (1984a) Two-pyroxene thermometry with new experimental data in the system CaO–MgO– Al2O3 –
SiO2. Contrib. Mineral. Petrol., 87, 87–97.
Gasparik, T. (1984b) Experimentally determined compositions of diopside– jadeite pyroxene in equilibrium
with albite and quartz at 1200–13508C and 15–34 kbar. Geochim. et Cosmochim. Acta, 49,
865–870.
Gasparik, T. (1984c) Two-pyroxene thermobarometry with new experimental data in the system CaO–MgO–
Al2O3 –SiO2. Contrib. Mineral. Petrol., 87, 87–97.
Gasparik, T. (1986) Experimental study of subsolidus phase relations and mixing properties of clinopyroxene in
the silica-saturated system CaO –MgO–Al2O3 –SiO2. Am. Mineral., 71, 686– 693.
Gasparik, T. (1989) Transformation of enstatite-diopside–jadeite pyroxenes to garnet. Contrib. Mineral. Petrol.,
102, 389 –405.
Gasparik, T. (1990) Phase relations in the transition zone. J. Geophys. Res., 95, 15751–15769.
1130 References

Gasparik, T. (1992) Enstatite–jadeite join and its role in the Earth’s mantle. Contrib. Mineral. Petrol., 111,
283–298.
Gasparik, T. (1993) The role of volatiles in the transition zone. J. Geophys. Res., 98, 4287–4299.
Gasparik, T. (1994) A petrogenetic grid for the system MgO.Al2O3.SiO2. J. Geol. Res., 102, 97 –109.
Gasparik, T. (1996a) Diopside–jadeite join at 16 –22 GPa. Phys. Chem. Miner., 23, 476–486.
Gasparik, T. (1996b) Melting experiments on the enstatite– diopside join at 70– 224 kbar, including the melting of
diopside. Contrib. Mineral. Petrol., 124, 139–153.
Gasparik, T. (2000) An internally consistant thermodynamic model for the system CaO–MgO–Al2O3 –SiO2
derived primarily from phase equilibrium data. J. Geol., 108, 103– 111.
Gasparik, T. and Drake, M.J. (1995) Partitioning of elements among two silicate perovskites, superphase B,
and volatile-bearing melt at 23 GPa and 1500–16008C. Earth Planet. Sci. Lett., 134, 307–3418.
Gasparik, T. and Litvin, A. (1997) Stability of Na2Mg2Si2O7 and melting relations on the forsterite–jadeite join at
pressures up to 22 GPa. Eur. J. Mineral., 9, 311–326.
Gasparik, T. and Newton, R.C. (1984) The reversed alumina contents of orthopyroxene in equilibrium with spinel
and forsterite in the system MgO–Al2O3 –SiO2. Contrib. Mineral. Petrol., 85, 186–196.
Gasparik, T., Parise, J.B., Eiben, B.A. and Hriljac, J.A. (1995) Stability and structure of a new high-pressure
silicate, Na18Ca1.1Si6O14. Am. Mineral., 80, 1269– 1276.
Gasparik, T., Parise, J.B., Reeder, R.J., Young, V.G. and Wilford, W.S. (1999) Composition stability and structure
of a new member of the senigmatite group, Na2Mg4þxFe3þ 222xSi62xO20 synthesized at 13–14 GPa. Am.
Mineral., 84, 257 –266.
Gasparik, T., Pripathi, A. and Parise, J.B. (2000) Structure of a new Al-phase [K, Na]0.9[Mg, Fe]2[Mg, Fe, Al,
Si]6O12, synthesized at 24 GPa. Am. Mineral., 85, 613–618.
Gasparik, T. (1994) A petrogenetic grid for the system MgO.Al2O3SiO2. J. Geol., 102, 97–109.
Geiger, C.A., Lottermoser, W. and Amthauer, G. (1990) A temperature dependent 57Fe Mössbauer study of
synthetic almandine– grossular and almandine–pyrope garnets: a comparison. Terra Abstr., 2, 75.
Geiger, C.A., Armbruster, W., Lager, G.A., Jiang, K., Lottermoser, W. and Amthauer, G. (1992) A combined
temperature dependent 57Fe Mössbauer and single crystal X-ray diffraction study of synthetic almandine:
evidence for the Gol’danskii-Karyagin effect. Phys. Chem. Miner., 19, 121–126.
Geisinger, K.L., Gibbs, G.V. and Navrotsky, A. (1985) A molecular orbital study of bond length and angle
variations in framework silicates. J. Non-Cryst. Solids, 68, 401–402.
Geissberger, A.E. and Galeener, F.L. (1983) Raman studies of vitreous SiO2, versus fictive temperature. Phys.
Rev., 28, 2366–3271.
Gelm, A.K., Dubonos, S.V., Lok, J.G.S., Henlmll, M. and Mann, J.C. (1998) Paramagnetic Meissner effect in
small superconductors. Nature, 396, 144– 147.
Gerald, J.D.F. and Ringwood, A.E. (1991) High pressure rhombohedral perovskite phase Ca2AlSiO5.5. Phys.
Chem. Miner., 18, 40– 46.
Gervais, F., Piriou, B. and Cabannes, F. (1973) Anharmonicity in silicate crystals: temperature dependencse of Au
type vibrational modes in ZrSiO4 and LiAlSi2O6. J. Phys. Chem. Solids, 43, 1785– 1796.
Gerward, L. and Olsen, J.S. (1997) J. Appl. Crystallogr., 30, 259.
Getting, I.C., Dutton, S.J., Burnley, P.C., Karato, S. and Spetzler, H.A. (1997) Shear attenuation and dispersion in
MgO. Phys. Earth Planet. Interiors, 99, 249–257.
Ghazibayat, B., Behruzi, M., Litterst, F.G., Lottermoser, W. and Amthauer, G. (1992) Crystallographic phase
transition and valence fluctuation in synthetic Mn-bearing ilvaite CaFe2þ 2þ 3þ
2x Mnx Fe [Si2O7/O/(OH)].
Phys. Chem. Miner., 18, 491 –496.
Ghiorso, M.S. (1990) Thermodynamic properties of hematite– ilmenite–geikielite solid solutions. Contrib.
Mineral. Petrol., 104, 645–667.
Ghose, S. (1985) Lattice dynamics, phase transitions and soft modes. Mineral. Soc. Am. Rev. Mineral., 14, 127–163.
Ghose, S., Wan, C., Okamura, F., Ohashi, H. and Weidver, J.R. (1975) Site references and crystal chemistry of
transition metal ions in pyroxenes and olivines. Acta Crystallogr., 431, 576.
Ghose, S., Tsukimura, K. and Hatch, D.M. (1989) Phase transitins in ilvaite a mixed-valence iron silicate, 2.
A single crystal X-ray diffraction study and Landau theory of the monoclinic to orthorhombic phase
transition induced by charge delocalization. Phys. Chem. Miner., 16, 483– 496.
References 1131

Ghosh, S., McMullan, R.K. and Weber, H.P. (1993) Neutron diffraction studies of the P1 ! I 1 phase transition in
anorthite, CaAl2Si2O8, and the crystal structure of the body-centered phase at 514 K. Z. Kristallogr., 204,
215–237.
Giauque, W.F. and Ashley, J.W. (1936) J. Am. Chem. Soc., 58, 1144–1150.
Gibbs, G.V. (1983) Molecules as models for bonding in silicates. Am. Mineral., 67, 421–450.
Giddy, A.P., Dove, M.T., Pawley, G.S. and Heine, V. (1993) The determination of rigid unit modes as potential
soft modes for displacive phase transitions in framework crystal structures. Acta Crystallogr. A, 49,
697–703.
Gieske, J.H. and Barsch, G.R. (1968) Pressure dependence of the elastic constants of single crystal crystalline
aluminium oxide. Phys. Stat. Sol., 29, 121– 131.
Gilder, S. and Glen, J. (1998) Magnetic properties of hexagonal closed-packed iron deduced from direct
observation in a diamond anvil cell. Science, 279, 72–74.
Gillet, P. (1993) Stability of magnesite (MgCO3) at mantle pressure and temperature conditions. A Raman
spectroscopic study. Am. Mineral., 7, 1328–1331.
Gillet, P. (1996) Raman spectroscopy at high pressure and high temperature. Phase transition and thermodynamic
properties of minerals. Phys. Chem. Miner., 23, 263–275.
Gillet, P. and Goffé, B. (1988) On the significance of aragonite occurrences in the Western Alps. Contrib. Mineral.
Petrol., 99, 70–81.
Gillet, P., Guyot, F. and Malézieux, J.M. (1989) High-pressure and high-temperature Raman
spectroscopy of Ca2GeO4: some insights on anharmonicity. Phys. Earth Planet. Interiors, 58, 141–154.
Gillet, P., Fiquet, G., Daniel, I. and Reynard, B. (1993a) Raman spectroscopy at mantle pressure and temperature
conditions. Experimental set-up and the example of CaTiO3 perovskite. Geophys. Res. Lett., 20,
1931– 1934.
Gillet, P., Guyot, F., Price, G.D., Tournerie, B. and Le Cléac’h, A. (1993b) Phase changes and thermodynamic
properties of CaTiO3, Spectroscopic data, vibrational modelling and some insights on the properties of
MgSiO3 perovskite. Phys. Chem. Miner., 20, 159–170.
Gillet, P., Badro, J., Varel, B. and McMillan, P. (1995) High-pressure behavior in alpha -AlPO4: Amorphization
and the memory-glass effect. Phys. Rev. B, 51, 11262–11269.
Gillet, P., Guyot, F. and Wang, Y. (1996) Microscopic anharmonicity and equation-of-state of MgSiO3
perovskite. Geophys. Res. Lett., 23, 3043–3046.
Gillet, P., Hemley, R.J. and Mc Millan, P.F. (1998) Vibrational properties at high pressure and temperatures.
In: Ultrahigh Pressure Mineralogy, Reviews in Mineralogy, Vol. 37. Mineralogical Society of America,
pp. 525 –590, Chapter 17.
Glatzmaier, G.A. and Roberts, P.H. (1996) Rotation and magnetism of Earth’s inner core. Science, 274, 1887– 1891.
Glazer, A.M. (1972) The classification of tilted octahedra in perovskites. Acta Crystallogr. B, 28,
3384– 3392.
Glazer, A.M. (1975) Acta Crystallogr. Sect., A31.
Gleitzer, C. and Goodenough, J.B. (1985) Struct. Bond., 61, 1.
Glinnemann, J. (1987). Pressure behaviours of quartz structures SiO2 and GeO2 and a new diamond anvil cell.
Dissertation, University of Aachen, Germany.
Glinnemann, J., King, H.E., Schulz, H., Hahn, Th., La Place, S.J. and Dacol, F. (1992) Crystal structures of the
low-temperature quartz-type phases of SiO2 and GeO2 at elevated pressure. Zeitschr. Kristallogr., 198,
177–212.
Goddat, A., Peyronneau, J. and Poirier, J.P. (1999) Dependence on pressure of conduction by hopping of small
polarons in minerals of the Earth’s lower mantle. Phys. Chem. Miner., 27, 81 –87.
Godwal, B.K., Sikka, S.K. and Chidambaram, R. (1983) Equation of state theories of condensed matter up to
about 10 TPa. Phys. Rep., 102, 121–197.
Gold, J.J., Bassett, W.A., Weathers, M.S. and Bird, J.M. (1984) Science, 225, 921.
Goldschimdt, V.N. (1926) Geochemische Verteilungs gesetze der Elemente VII Die Gesetzeder Krystallochemie.
Skr. Nor. Vidensk. Akad. Oslo I, Mat. -Natur, Kl., 2, 1– 117.
Goldsmith, J.R. and Jenkins, D.M. (1985) The high–low albite relations revealed by reversal of degree of order at
high pressures. Am. Mineral., 70, 911–923.
1132 References

Goncharov, A.F., Makarenko, I.N. and Stishov, S.M. (1989) Sov. Phys. JETP, 69, 380.
Goncharov, A.F., Struzhkin, V.V., Somayazulu, M.S., Hemley, R.J. and Mao, H.K. (1996) Compression of H2O
ice to 210 GPa: evidence for a symmetric hydrogen bonded phase. Science, 273, 218– 220.
Goncharov, A.F., Hemley, R.J., Mao, H.K. and Shu, J. (1998) New high-pressure excitations in parahydrogen.
Phys. Rev. Lett., 80, 101–104.
Goncharov, A.F., Mao, H.-K., Gillet, P. and Hemley, R.J. (2000) Raman spectroscopy of iron to 152 GPa:
implications for Earth’s inner core. Science, 288, 1626–1629.
Goncharov, A.F., Gregoryanz, E., Hemley, R.J. and Mao, H.K. (2001) Spectroscopic studies of the vibrational and
electronic properties of solid hydrogen to 285 GPa. PNAS, 98, 1434– 1437.
Gong, Z. and Cohen, R.E. (1992) Ferroelectrics, 136, 113.
Gonnard, P., Fetiveau, Y., Bauer, F. and Eyarand, L. (1972) C. R. Acad. Sci. Paris Ser. B, 275, 633–636.
Goodenough, J.B. (1955) Phys. Rev., 100, 564.
Goodenough, J.B. and Zhou, J.S. (1997) New forms of phase segregation. Nature, 386, 229–230.
Göpel, C., Manhes, G. and Allegre, C. (1994) Earth Planet. Sci. Lett., 121, 153.
Gordon, R.G. and Kim, Y.S. (1972) J. Chem. Phys., 56, 3122.
Goresey, A.El., Chen, M., Dubrovinsky, L., Gillet, P. and Graup, G. (2001) An ultradense polymorph of rutile
with seven-coordinated titanium from the Ries Crater. Science, 293, 1467–1469.
Goswami, J.N. and Vanhala, H.A.T. (2000) In: Mannings, V., Boss, A.P. and Russell, S.S. (eds.), Protostars and
Planets IV. Tucson, AZ: University of Arizona Press, pp. 963–994.
Goto, T., Ahrens, T.J. and Rossman, G.R. (1979) Absorption spectra of Cr3þ in Al2O3 under shock compression.
Phys. Chem. Miner., 4, 253 –263.
Goto, T., Ahrens, T.J., Rossman, G.R. and Syono, Y. (1980) Absorption spectrum of shock-compressed Fe2þ-
bearing MgO and radiative conductivity of the lower mantle. Phys. Earth Planet. Interiors, 22,
277–288.
Goto, T., Sato, J. and Syono, Y. (1982) Shock-induced spin-pairing transition in Fe2O3 due to the pressure effect
on the crystal field. In: Akimoto, S. and Manghani, M.H. (eds.), High Pressure Research in Geophysics.
Tokyo: Center for Academic Publishing, pp. 595– 609.
Grady, D.E. (1980) J. Geophys. Res., 85, 913.
Graham, E.K. (1973) On the compression of stishovite. Geophys. J. R. Astron. Res., 80, 4857–4861.
Grand, S.P. (1994) Mantle shear structure beneath the American and surrounding oceans. J. Geophys. Res., 99,
11591–11621.
Grand, S.P. and Helmberger, D.V. (1984) Upper mantle shear structure of North America, Geophys. J.R. Astr.
Soc., 76, 300–438.
Grand, S.P., Van der Hilst and Widiyantoro, S. (1997) GSA Today, 7, 1–10.
Grande, T., Holloway, J.R., McMillan, P.F. and Angell, C.A. (1994) Nature, 369, 43–45.
Gratz, A. (1984) J. Non-Cryst. Solids, 67, 543.
Gratz, A.J., DeLoach, L.D., Clough, T.N. and Nellis, W.J. (1993) Shock amorphization of cristobalite. Science,
259, 663 –666.
Green, A.G. (1976) Interpretation of Project MAGNET aeromagnetic profiles across Africa. Geophys. J. R.
Astron. Soc., 44, 203.
Green, H.W. II. (1994) Solving the paradox of deep earthquakes. Sci. Am., 271(3), 64–71.
Green, H.W. II. (2001) A graveyard for buoyant slabs? Science, 292, 2445–2446.
Green, H.W. and Dobrzhinetskaya, L. (1997) Response (The origin of ultra-high pressure lherzolites by Hacker
et al., 997). Science, 278, 704–707.
Green, T.H. (1994) Chem. Geol., 117, 1–36.
Greenwood, N.N. and Howe, A.T. (1972) J. Chem. Soc. Dalton. Trans., 116.
Green, D.H. and Ringwood, A.E. (1972) A comparison of recent experimental data on the gabbro-garnet
granulite–eclogite transition. J. Geol., 80, 277–288.
Green, H.W., Dobrzinetskaya, L., Riggs, E.M. and Jin, Z.M. (1997) Alpe Arami: a peridotite massif from the
mantle transition zone? Tectonophysics, 279, 1– 21.
Gregoryanz, E., Hemley, R.J., Mao, H.K. and Gillet, P. (2000) High-Pressure Elasticity of a-Quartz. Instability
and Ferroelastic Transition, 84(14), 3117– 3120.
References 1133

Gregoryanz, E., Struzhkin, V.V., Hemley, R.J. et al. (2002) Superconductivity in the chalcogens up to
multimegabar pressures. Phys. Rev. B, 65, 064504-1-6.
Grevel, K.D., Nowlan, E.U., Fasshauer, D.W. and Burchard, M. (2000) In situ X-ray diffraction investigation of
lawsonite and Zoisite at high pressures and temperatures. Am. Mineral., 85, 206–216.
Grewe, J., Schilling, J.S., Ikeda, K. and Gchneidner, K.A. Jr. (1989) Phys. Rev. B, 40, 9017.
Griffith, J.S. (1961) Theory of Transition Metal Ions. London: Cambridge University Press.
Griggs, D.T., Turner, F.J. and Heard, H.C. (1960) Deformation of rocks at 5008 to 8008C. Mem. Geol. Soc. Am.,
79, 39 –104.
Grimsditch, M., Popova, S., Brazkhin, V.V. and Voloshin, R.N. (1994) Temperature-induced amorphization of
SiO2 stishovite. Phys. Rev., 50, 12984–12986.
Grønvold, F. and Sveen, A. (1974) Heat capacity and thermodynamic properties of synthetic magnetite (Fe3O4)
from 300 to 1050 K. Ferromagnetic transition and zero-point entropy. J. Chem. Thermodyn., 6, 859–872.
Grønvold, F., Stølen, S., Tolmach, P. and Westrum, E.F. Jr. (1993) Heat capacities of the wüstites Fe0.9379O and
Fe0.9254O at temperatures from 5 to 350 K. Properties of Fe12gO to 1000 K. Thermodynamics of
formation of wüstite. J. Chem. Therm., 25, 1089–1117.
Grover, J. (1972) The stability of low-clinoenstatite in the system Mg2Si2O6 – CaMgSi2O6 (abstract). EOS Trans.
AGU, 53, 539.
Gruau, G., Chauvel, C., Arndt, N.T. and Cornichet, J. (1990) Aluminium depletion in komatite and garnet
fractionation in the early Archean mantle: hafnium isotope constrants. Geochim. Coschim. Acta, 54,
3095– 3101.
Grundy, W.M. and Buie, M.W. (2002) Icarus, 157, 128– 138.
Gubbins, D., Masters, T.G. and Jacobs, J.A. (1979) Geophys. J. R. Astron. Soc., 59, 57– 99.
Gudfinnsson, G.H. and Wood, B.J. (1998) The effect of trace element on the olivine–wadsleyite transformation.
Am. Mineral., 83, 1037–1044.
Guillot, T., Gautier, D., Chabrier, G. and Mosser, B. (1994) Nonadiabatic models of Jupiter and Saturn. Icarus,
112, 337.
Guinan, M.W. and Beshers, D.N. (1968) Pressure derivatives of the elastic constants of a iron to 10 kbs. J. Phys.
Chem. Solids, 29, 541– 549.
Guinan, M.W. and Steinberg, D.J. (1974) Pressure and temperature derivatives of the isotropic polycrystalline
shear modulus for 65 elements. J. Phys. Chem., 35, 1501–1512.
Gupta, H.K. (1983) Bull. Seismol. Soc. Am., 73, 679–682.
Gupta, Y.M. (ed.) (1986) Shock Waves in Condensed Matter. New York: Plenum Press.
Gupta, Y.M. (1992) In: Schmidt, S.C., Dick, R.D., Forbes, J.W. and Tasker, D.G. (eds.), Shock Waves in
Condensed Matter — 1991. Amsterdam: Elsevier, 15 pp.
Gupta, A. et al. (1996) Phys. Rev. B, 54, R15629–R15632.
Guy Master (1997) Nature, 386, 558.
Guyot, F., Richet, P., Courtial, P. and Gillet, P. (1993) High temperature heat capacity and phase transitions of
CaTiO3 perovskite. Phys. Chem. Miner., 20, 239–246.
Haas, J.R. Jr. and Hemmingway, B.S. (1992) Recommended standard electrochemical potentials and fugacities of
oxygen for the solid buffers and thermodynamic data in the systems iron–silicon–oxygen, nickel–
oxygen and copper–oxygen. US Geological Survey, Open File Report 92-267.
Haavik, C., Stølen, S., Fjellvåg, H., Hanfland, M. and Häusermann, D. (2000) Equation of state of magnetite and
its high-pressure modification: thermodynamics of the Fe–O system at high pressure. Am. Mineral., 85,
514–523.
Hacker, B.R., Sharp, T. et al. (1997) Determining the origin of ultrahigh pressure lherzolites. Science, 278,
702–707.
Hackwell, T.P. and Angel, R.J. (1995) Reversed brackets for the P1 , I 1 transition in anorthite at high pressures
and temperatures. Am. Mineral., 80(3–4), 239–246.
Hager, B.H. (1990) The viscosity profile of the mantle: a comparison of models on postglacial and convection
time scales. EOS Trans. AGU, 71, 1567.
Hager, B.H., Clayeton, R.W., Richards, M.A. et al. (1985) Lower mantle heterogeneity, dynamic topography and
geoid. Nature, 313, 541 –545.
1134 References

Haggerty, S.E. (1986) Nature, 320, 34 –38.

Haggerty, S.E. (1991) Oxide mineralogy of the upper mantle. Mineral. Soc. Am. Rev. Mineral., 25, 355–416.
Haggerty, S.E. (1999) A diamond triology: superplumes, supercontinents and supernova. Science, 285, 851–860.
Haggerty, S.E. and Sautter, V. (1990) Ultradeep (greater than 300 kilometers), ultramafic mantle xenoliths.
Science, 248, 993 –996.
Haggerty, S.E., Boyd, F.R., Bell, P.M., Finger, L.W. and Bryan, W.B. (1970) Opaque minerals and olivine in
lavas and breccias from Mare Tranquillitatis. Proceedings of the Apollo 11 Lunar. Geochim. Cosmochim.
Acta, 1, 513–538.
Haines, J. (1996) J. Phys. Condens. Matter, 8, 1631.
Haines, J. (1998) Nature, 634.
Halley, E. (1683) Philos. Trans. R. Soc., 13, 208– 221.
Halliday, A.N. (1999) Unmixing Hawaiian cocktails. Nature, 399, 733– 734.
Halliday, A.N. (2000) Earth Planet. Sci. Lett., 176, 17– 30.
Halliday, A.N. (2001) In the beginning…. Nature, 409, 144 –145.
Halliday, A.N. and Lee, D.-C. (1999) Geochim. Cosmochim. Acta, 63, 4157.
Hama, J. and Suito, K. (1996) The search for a universal equation of state correct up to very high pressure. J. Phys.:
Condens. Matter, 8, 67–81.
Hama, J. and Suito, K. (1998) High-temperature equation of state of CaSiO3 perovskite and its implications for the
lower mantle. Phys. Earth Planet. Interiors, 105, 33–46.
Hamada, N., Sawada, H. and Terakuka, T. (1995) In: Fujimori, A. and Tokura, Y. (eds.), Spectroscopy of Mott
Insulator and Correlated Metals. Berlin: Springer.
Hamley, A. (1994) Calculation of the pressure induced insulator-metal transition of nitrogen. J. Phys. Condens.
Matter, 6, 985– 988.
Hammersley, A.P., Svensson, S.O., Hanfland, M., Fitch, A.N. and Häusermann, D. (1996) Two-dimensional
detector software: from real detector to idealised image or two-theta scan. High Press. Res., 14,
235–248.
Hammonds, K.D., Dove, M.T., Giddy, A.P. and Heine, V. (1994) Crush: a FORTRAN program for the analysis of
the rigid unit mode spectrum of a framework structure. Am. Mineral., 79, 1207– 1209.
Hammonds, K.D., Dove, M.T., Giddy, A.P., Heine, V. and Winkler, B. (1996) Rigid unit phonon modes and
structural phase transitions in framework silicates. Am. Mineral., 81, 1057–1079.
Hanan, B. and Graham, D. (1996) Science, 272, 991–995.
Hanfland, M. and Syassen, K. (1985) J. Appl. Phys., 57, 2752.
Hanfland, M., Hemley, R.J., Mao, H.K. and Williams, G.P. (1992) Synchrotron infrared spectroscopy at megabar
pressures: vibrational dynamics of hydrogen to 180 GPa. Phys. Rev Lett., 69, 1129–1132.
Hanfland, M., Hemley, R.J. and Mao, H.K. (1993) Novel infrared vibron absorption in solid hydrogen at megabar
pressures. Phys. Rev. Lett., 70, 3760–3763.
Hanson, R.C. and Jones, L.H. (1981) Infrared and Raman studies of pressure effects on the vibrational modes of
solid CO2. J. Chem. Phys., 75, 1102–1112.
Harder, H. and Christensen, U.R. (1996) A one-plume model of martian mantle convection. Nature, 380, 507–509.
Harder, H. and Schubert, G. (2001) Sulfer in Mercury core? Icarus, 151, 118– 122.
Hariya, Yu. (1984) H2O in the Earth’s interior. In: Sunagawa, I. (ed.), Materials Science of the Earth’s Interior.
Tokyo: Terra Scientific, pp. 463–475.
Harley, S.L. (1984) An experimental study of the partitioning of Fe and Mg between garnet and orthopyroxene.
Contrib. Mineral. Petrol., 86, 359 –373.
Harlov, D.H. and Newton, R.C. (1993) Reversal of the metastable kyanite þ corundum þ quartz and
andalusite þ corundum þ quartz equilibria and the enthalpy of formaion of kyanite and andalusite.
Am. Mineral., 78, 594 –600.
Harlow, G.E. (1990) Am. Mineral., 82, 259 –269.
Harlow, G.E. (1996) Structure refinement of a natural K-rich diopside: the effect of K on the average structure.
Am. Mineral., 81, 632 –638.
Harlow, G.E. (1997) K in clinopyroxene at high pressure and temperature: an experimental study. Am. Mineral.,
82, 259 –269.
References 1135

Harlow, G.E. and Veblen, D.R. (1991) Potassium in clinopyroxene inclusions from diamonds. Science, 251,
652–655.
Harrington, S., Zhang, R., Poole, P.H., Sciortino, F. and Stanley, H.E. (1997) Phys. Rev. Lett., 78, 2409.
Harris, M. (1999) Taking the frustration out of ice. Nature, 399, 311–312.
Harris, D.M. Jr. and Anderson, A.T. (1984) Volatiles H2O, CO2, and Cl in a subduction related basalt. Contrib.
Mineral. Petrol., 87, 120 –128.
Harris, J., Hutchison, M.T., Hursthouse, M., Light, M. and Harte, B. (1997) A new tetragonal silicate mineral
occurring as inclusions in lower-mantle diamonds. Nature, 387, 486–488.
Harrison, W.A. (1980) Electronic Structure and the Properties of Solids: The Physics of the Chemical Bond.
San Francisco: W.H. Freeman and Company, 582 pp.
Harrison, R.J. and Putnis, A. (1997) The coupling between magnetic and cation ordering: a macroscopic approach.
Eur. J. Mineral., 9, 1115–1130.
Harrison, R.J. and Salje, E.K.H. (1994) X-ray diffraction study of the displacive phase transition in anorthoclase,
grain-size effects and surface relaxations. Phys. Chem. Miner., 21(5), 325–329.
Harrison, R.J., Dove, M.T., Knight, K.S. and Putnis, A. (1999) In-situ neutron diffraction study of non-
convergent cation ordering in the (Fe3O4)12x(MgAl2O4)x spinel solid solution. Am. Mineral., 84,
555–563.
Harte, B. and Harris, J.W. (1994) Mineral. Mag., 58A, 384.
Harvey, R.P. and McSween, H.Y. (1992) The parent magma of the nakhlite meteorites: clues from melt
inclusions. Earth Planet. Sci. Lett., 111, 467– 482.
Hashin, Z. and Shtrikman, S. (1962) A variational approach to of the elastic behaviour of polycrystals. J. Mech.
Solids, 10, 343–352.
Hatch, D.M. and Ghose, S. (1989) Symmetry analysis of the phase transition and twining in MgSiO3 garnet:
implications to mantle mineralogy. Am. Mineral., 74, 1221–1224.
Hatch, D.M. and Ghose, S. (1991) The a–b phase transition in cristobalite, SiO2: symmetry analysis, domain
structure, and the dynamic nature of the b-phase. Phys. Chem. Miner., 17, 554–567.
Hatch, D.M. and Griffen, D.T. (1989) Phase transitions in the grandite garnets. Am. Mineral., 74, 151– 159.
Hatch, D.M., Ghose, S. and Bjorkstam, J.L. (1994) The a–b phase transition in AlPO4 cristobalite: symmetry
analysis, domain structure and transition dynamics. Phys. Chem. Miner., 21, 67–77.
Hattori, T., Matsuda, T., Tsuctiya, T., Nagai, T. and Yamanaka, T. (1999) Clinopyroxene–perovskite phase
transition of FeGeO3 under pressure and room temperature. Phys. Chem. Miner., 26, 212–216.
Hau, L.V., Harris, S.E., Dutton, Z. and Behroozi, C.H. (1999) Nature, 397, 594–598.
Hauri, E.H. and Hart, S.R. (1993) Earth Planet. Sci. Lett., 114, 353–371.
Hauri, E.H. and Hart, S.R. (1997) Rhenium abundances and systematics in oceanic basalts. Chem. Geol., 139,
185–205.
Hauri, E.H. (2003) Osmium isotopes and mantle convection. Phil. Trans. R. Soc. Lond., A360, 2371– 2382.
Hawking, S.W. (1976) Phys. Rev. D, 13, 191 –197.
Hawthorne, F.C., Ungaretti, L., Oberti, R., Caucua, F. and Callegari, A. (1993) The crystal chemistry of staurolite.
2. Order-disorder and the monoclinic-orthorhmbic phase transition. Can. Mineral., 31, 583–595.
Hayes, W. (1978) Superionic conductors. Contemp. Phys., 19, 469–486.
Hazen, R.M. (1976) Effect of temperature and pressure on the crystal structure of forsterite. Am. Mineral., 61,
266–271.
Hazen, R.M. (1981) Systematic variation of bulk modulus of wüstite with stoichiometry. Carnegie Inst.
Washington Yearbook, 80, 277 –280.
Hazen, R.M. (1988) Perovskites. Scientific American, June, 74–81.
Hazen, R.M. (1990) Crystal structures of high temperature superconductors. In: Ginsberg, D.M. (ed.),
Physical Properties of High Temperature Superconductors II. New Jersey: World Scientific,
pp. 121 –198.
Hazen, R.M. (1993) Comparative compressibilities of silicate spinels: anomalous behavior of (Mg, Fe)2SiO4.
Science, 259, 206 –209.
Hazen, R.M. and Finger, L.W. (1978) Crystals structure and compressibilities of pyrope and grossular to 60 kbar.
Am. Mineral., 63, 297 –303.
1136 References

Hazen, R.M. and Finger, L.W. (1979) Crystal structure and compressibility of zircon at high pressure. Am.
Mineral., 64, 196 –201.
Hazen, R.M. and Finger, L.W. (1981a) Bulk moduli and high-pressure crystal structures of rutile-type
compounds. J. Phys. Chem. Solids, 42, 143 –151.
Hazen, R.M. and Finger, L.W. (1981b) Phys. Chem. Solids, 42, 143.
Hazen, R.M. and Finger, L.W. (1982) Comparative Crystal Chemistry. New York: Wiley, 231 pp.
Hazen, R.M. and Finger, L.W. (1989) High-pressure crystal chemistry of andradite and pyrope: revised
procedures for high-pressure diffraction experiments. Am. Mineral., 74, 352–359.
Hazen, R.M. and Finger, L.W. (1993) Effects of pressure on Mg–Fe ordering in orthopyroxene synthesized at
11.3 GPa and 16008C. Am. Mineral., 78, 1336–1339.
Hazen, R.M. and Jeanloz, R. (1984) Wüstite (Fe12xO): a review of its defect structure and physical properties.
Rev. Geophys. Space Phys., 122, 37–46.
Hazen, R.M. and Navrotsky, A. (1996) Effects of pressure on order–disorder reactions. Am. Mineral., 81,
1021– 1035.
Hazen, R.M. and Prewit, C.T. (1977) Effects of temperature and pressure on interatomic distances in oxygen-
based minerals. Am. Mineral., 62, 309 –315.
Hazen, R.M., Finger, L.W. and Mariathasan, J.W.E. (1985) High-pressure crystal chemistry of scheelite-type
tungstates and molybdates. J. Phys. Chem. Solids, 46, 253–263.
Hazen, R.M., Finger, L.W., Hemley, R.J. and Mao, H.K. (1989) Solid State Commun., 72, 507.
Hazen, R.M., Zang, J. and Ko, J. (1990) Phys. Chem. Miner., 17, 416.
Hazen, R.M., Finger, L.W. and Ko, J. (1992) Crystal chemistry of Fe-bearing anhydrous phase B: implications for
transition zone mineralogy. Am. Mineral., 77, 217 –220.
Hazen, R.M., Downs, R.T., Finger, L.W. and Ko, J. (1993) Crystal chemistry of ferromagnesian silicate spinels:
evidence for Mg–Si disorder. Am. Mineral., 78, 1320–1323.
Hazen, R.M., Downs, R.T., Finger, L.W., Conrad, P.G. and Gasparik, T. (1994a) Crystal chemistry of calcium-
bearing majorite. Am. Mineral., 79, 581– 584.
Hazen, R.M., Downs, R.T., Conrad, P.G., Finger, L.W. and Gasparik, T. (1994b) Comparative compressibilities
of majorite-type garnets. Phys. Chem. Miner., 21(5), 344–349.
Hazen, R.M., Downs, R.T. and Finger, L.W. (1996) High pressure crystal chemistry of LiScSiO4: an olivine with
nearly isotropic compression. Am. Mineral., 81, 327–334.
Hazen, R.M., Yang, H. and Prewitt, C.T. (2000a) High- pressure crystal chemestry of Fe3þ-wadsleyite,
b-Fe2.33Si0.67O4.. Am. Mineral., 85, 778–783.
Hazen, R.M., Weinberger, M.B. et al. (2000b) Comparative high-pressure crystal chmistry of wadsleyite,
b-(Mg12xFex) SiO4, with x ¼ 0 and 0.25. Am. Mineral., 85, 770– 777.
Hazen, R.M. (2001) Life’s rocky start. Sci. Am., April, 63–71.
Hearn, E.H. (2001) Shock delay. Nature, 411, 150– 151.
Heath, J.R. (1998) C60’s smallest cousin. Nature, 393, 730.
Hebard, A.F. et al. (1991) Nature, 350, 600.
Hedin, L. and Lundqvist, B.I. (1971) J. Phys., 4, 2064.
Heer, W.D. (2000) A question of dimension. Science, 289, 1702–1703.
Heinemann, S., Sharp, T.G., Seifert, F. and Rubie, D.C. (1997) The cubic-tetragonal phase transition in the system
majorite (Mg4Si4O12) –pyrope (Mg3Al2Si3O12), and garnet symmetry in the Earth’s transition zone.
Phys. Chem. Miner., 24, 206 –221.
Heinz, D.L. (1990) Geophys. Res. Lett., 17, 1161.
Heinz, D.L. and Jeanloz, R. (1987) Measurement of the melting curve of Mg0.1 Fe0.9 SiO3
perovskite at lower mantle conditions and its geophysical implications. J. Geophys. Res., 92
11437–11444.
Heitweil, E.J. (1999) Ultrafast glimpses at water and ice. Science, 283, 1467–1468.
Helffrich, G.R., Stein, S. and Wood, B.J. (1989) J. Geophys. Res., 94, 753.
Hellemans, A. (1999) Getting to the bottom of water. Science, 283, 614– 615.
Heller-Kellai, L., Yarvi, S.H. and Gross, S. (1975) Hydroxyl stretching frequencies of serpentine minerals.
Mineral. Mag., 40, 197–200.
References 1137

Helmberger, D.V., Win, L. and Ding, X. (1998) Seismic evidence that the source of the Iceland hotspot lies at the
core– mantle boundary. Nature, 396, 251–252.
Hemingway, B.S., Bohlen, S.R., Hankins, W.B., Westurn, E.F. Jr. and Kuskov, O.L. (1998) Heat capacity and
thermodynamic properties for coesite and jadeite, reexamination of the quartz–coesite equilibrium
boundary. Am. Mineral., 83, 409–418.
Hemley, R.J. (1987a) Nature, 330, 737–740.
Hemley, R.J. (1987b) Pressure dependence of Raman spectra of SiO2 polymorphs: a-quartz, coesite,
and stishovite. In: Manghnani, M.H. and Syono, Y. (eds.), High-pressure Research in
Mineral Physics. Tokyo/Washington, DC: Terra Scientific/American Geophysical Union,
pp. 347 –359.
Hemley, R.J. (1988) EOS Trans. Am. Geophys. Union, 69, 159.
Hemley, R.J. (1995) Turning off the water. Nature, 378, 14 –15.
Hemley, R.J. (1998a) Matter properties at high pressure and temperature. In: Good, G.A. (ed.), Science of
the Earth: An Encyclopaedia of Events, People, and Phenomena. New York: Gerlands Publishing,
pp. 523 –535.
Hemley, R.J. (ed.) (1998b) Ultrahigh Pressure Mineralogy: Physics and Chemistry of the Earth’s Deep Interior.
Reviews in Mineralogy, Mineralogical Society of America, Vol. 37, p. 671.
Hemley, R.J. (eds.) (1998) Spectroscopic studies of p-H2 to above 200 GPa. J. Low Temp. Phys., 110(1–2),
75–88.
Hemley, R.J. (1998d) Superconductivity in a grain of salt. Science, 281, 1296–1297.
Hemley, R.J. (2000) The element of uncertainty. Nature, 404, 240– 241.
Hemley, R.J. and Aschroft, N.W. (1998) The revealing role of pressure in the condensed matter sciences. Phys.
Today Am. Inst. Phys., 26–32.
Hemley, R.J. and Cohen, R.E. (1996) Structure and bonding in the deep mantle and core. Phil. Trans. R. Soc.
Lond. A, 354, 1461–1479.
Hemley, R.J. and Mao, H.K. (1988a) New phenomena in low-Z materials at megabar pressures. J. Phys. Condens.
Matter, 10, 11157–11167.
Hemley, R.J. and Mao, H.K. (1988b) Phys. Rev. Lett., 61, 857.
Hemley, R.J. and Mao, H.K. (1997) In: Trigg, L. (ed.), Encyclopadia of Applied Physics, Vol. 18. New York:
VCH, 555 pp.
Hemley, R.J. and Mao, H.K. (1998) Static compression experiments on low-Z planetary materials, Properties of
Earth and Planetary Materials at High Pressure and Temperature, Vol. 101. Washington, DC: American
Geophysics Union, pp. 173-183.
Hemley, R.J. and Mao, H.K. (2002) New windows on earth and planetary interiors, Min. Mag., Vol. 66(5),
pp. 791 –811.
Hemley, R.J., Jackson, M.D. and Gordon, R.G. (1985) Lattice dynamics and equations of state of high-
pressure mineral phases studied with electron-gas theory. EOS Trans. Am. Geophys. Union, 66, 357,
Abstract.
Hemley, R.J., Mao, H.K., Bell, P.M. and Mysen, B.O. (1986) Raman spectroscopy of SiO2 glass at high pressure.
Phys. Rev. Lett., 57, 747–750.
Hemley, R.J., Jackson, M.D. and Gordon, R.G. (1987a) Theoretical study of the structure, lattice dynamics, and
equations of state of perovskite-type MgSiO3 and CaSiO3. Phys. Chem. Miner., 14, 2– 12.
Hemley, R.J., Bell, P.M. and Mao, H.K. (1987b) Laser techniques in high pressure geophysics. Science, 237,
605–612.
Hemley, R.J., Jephcoat, A.P., Mao, H.K., Ming, L.C. and Manghnani, M.H. (1988) Pressure-induced
amorphisation of crystalline silica. Nature, 334, 52–54.
Hemley, R.J., Cohen, R.E., Yeganeh-Haeri, A., Mao, H.K., Weidner, D.J. and Ito, E. (1989a) Raman spectroscopy
and lattice dynamics of MgSiO3 perovskite at high pressure. In: Navrotsky, A. and Weidner, D.J. (eds.),
Perovskite: A Structure of Great Interest Geophysics and Material Science, Geophys. Monograph, Vol.
45. Washington, DC: American Geophysical Union, pp. 35–44.
Hemley, R.J., Chen, L.C. and Mao, H.K. (1989b) New transformation between crystalline and amorphous ice.
Nature, 338, 638–640.
1138 References

Hemley, R.J., Mao, H.K. and Shu, J.F. (1990) Low-frequency vibrational dynamics and structure of hydrogen at
megabar pressures. Phys. Rev. Lett., 65, 2670–2673.
Hemley, R.J., Hanfland, M. and Mao, H.K. (1991) High-pressure dielectric measurements of hydrogen to
170 GPa. Nature, 350, 488– 491.
Hemley, R.J., Stixrude, L., Fei, Y. and Mao, H.K. (1992) Constraints on lower mantle composition from P–V –T
measurements of (Fe, Mg)SiO3 perovskite and (Fe, Mg)O. In: Syono, Y. and Manghnani, M.H. (eds.),
High Pressure Research: Applications to Earth and Planetary Science. Tokyo/Washington, DC: Terra
Scientific/AGU.
Hemley, R.J., Prewitt, C.T. and Kingma, K.J. (1994) High-pressure behaviour of silica. In: Heaney, P.J., Prewitt,
C.T. and Gibbs, G.V. (eds.), Silica: Physical Behaviour, Geochemistry and Materials Applications, Rev.
Mineral., Vol. 29. Washington, DC: Mineralogical Society of America.
Hemley, R.J., Mao, H.K., Duffy, T.S., Eggert, J.H., Goncharov, A.F., Hanfland, M., Li, M., Somayazulu, M., Vos,
W.L. and Zha, C.S. (1995) Dense hydrogen in the outer solar system: implications from recent high
pressure experiments. Am. Inst. Phys., 250 –260.
Hemley, R.J., Mao, H.K., Goncharov, A.F., Hanfland, M. and Struzhkin, V.V. (1996) Synchrotron infrared
spectroscopy to 0.15 eV of H2 and D2 at megabar pressures. Phys. Rev. Lett., 76, 1667–1670.
Hemley, R.J., Mazin, I.I., Goncharov, A.F. and Mao, H.K. (1997a) Vibron effective charges in dense hydrogen.
Eurphys. Lett., 37, 403 –407.
Hemley, R.J.X-r. et al. (1997b) ray imaging of stress and strain of diamond, iron, and tungsten at megabar
pressures. Science, 276, 1242– 1245.
Hemley, R.J., Mao, H., Shen, G., Badro, J., Gillet, P., Hanfland, M. and Häusermann, D. (1997c) X-ray
imaging of stress and strain of diamond. Iron, and Tungsten at megabar pressures. Science, 276,
1242– 1245.
Hemley, R.J., Somayazulu, M.S., Goncharov, A.F. and Mao, H.K. (1998a) High-pressure Raman spectroscopy of
Ar –H2 and CH4 –H2 van der Waals compounds. Asian J. Phys., 7(2), 319–322.
Hemley, R.J., Goncharov, A.F., Struzhkin, V.V., Li, M. and Mao, H.K. (1998b) High presssure
synchrotron infrared spectrscopy at the National Synchrotron Light Source, Laboratory and Center
for High Pressure Research, Carnegie Institution of Washington. Nuovo Cimento, 20(4),
539 –551.
Hemley, R.J., Goncharov, A.F., Mao, H.K., et al. (1998c) Spectroscopic studies of p-H2 to above 200 GPa. J. Low
temp. Physics, 110(1/2), 75 –88.
Hemley, R.J., Mao, H. and Cohen, R.E. (1998d) High-pressure electronic and magnetic properties. In: Ultrahigh
Pressure Mineralogy. Physics and Chemistry of the Earth’s Deep Interior, Reviews in Mineralogy, 37,
Mineralogical Society of America, pp. 591–638, Chapter 18.
Hemley, R.J., Mao, H.K. and Gramsch, S.A. (2000) Pressure-induced transformations in deep mantle and core
minerals. Min. Mag., 64(2), 157–184.
Hemley, R.J., Shu, J., Carpenter, M.A., Hu, J., Mao, H.K. and Kingma, K.J. (2000a) Strain/order parameter
coupling in the ferroelastic transition in dense SiO2. Solid State Commun., 114, 527–532.
Hemley, R.J., Mao, H.K. and Gramsch, S.A. (2000b) Pressure-induced transformations in deep mantle and core
minerals. Mineral. Mag., 64(2), 157– 184.
Henning, Th. and Salama, F. (1998) Carbon in the Universe. Science, 282, 2204–2210.
Herzberg, C.T. (1992) Depth and degree of melting of komatiites. J. Geophys. Res., 97, 4521–4540.
Herzberg, C. and Gasparik, T. (1991) Garnet and pyroxenes in the mantle: a test of the majorite fractionation
hypothesis. J. Geophys. Res., 96, 16263– 16274.
Herzberg, C.T. and Zhang, J. (1996) Melting experiments on anhydrous peridotite KLB-1: compositions of
magmas in the upper mantle and transition zone. J. Geophys. Res., 101, 8271–8295.
Herzberg, C. and Zhang, J. (1997) Melting experiments of komatiite analog compositions at 5 GPa. Am. Mineral.,
82, 354 –367.
Hesson, S.E. and Ringwood, A.E. (1989) Slab–mantle interactions 2. The formation of diamonds. Chem. Geol.,
78, 97 –118.
Hewins, R.H. and Connolly, H.C. Jr. (1996) Chondrules and the Protoplanetary Disk. Cambridge: Cambridge
University Press, 197 pp.
References 1139

Heymann, D., Chibante, L.P.F., Brooks, R.R., Wolbach, W.S. and Smalley, R.E. (1994) Fullerene s in the K/T
boundary layers. Science, 265, 645–647.
Hickman, S.H., Zoback, M.D. and Healy, J.H. (1988) Continuation of a deep borehole stress measurement profile
near the San Andreas Fault, 1. Hydraulic fracturing stress measurements at Hi Vista, Mojave Desert,
California. J. Geophys. Res., 93, 15183.
Hiki, Y. and Granato, A. (1966) Phys. Rev., 144, 411.
Hill, R.J., Craig, J.R. and Gibbs, G.V. (1979) Systematics of the spinel structure type. Phys. Chem. Miner., 4,
317–339.
Hillgren, V.J., Drake, M.J. and Rubie, D.C. (1994) High-pressure and high-temperature experiments on core –
mantle segregation in the accreting Earth. Science, 264, 1442–1444.
Hillgren, V.J., Drake, M.J. and Rubie, D.C. (1996) High pressure and high temperature metal-silicate partitioning
of siderophile elements: the importance of silicate liquid composition. Geochim. et Cosmochim. Acta,
60(12), 2257– 2263.
Hilsch, R. and Pohl, R.W. (1928) Z. Phys., 57, 145.
Hinckley, D.N. (1963) Variability in crystallinity values among the kaolin deposits of the coastal plain of Georgia
and South Carolina. Clays Clay Miner., 11, 229– 235.
Hirai, H. and Kondo, K.I. (1991) Science, 253, 772.
Hirose, K., Fei, Y., Ma, Y. and Mao, H. (1999) The fate of subducted basaltic crust in the Earth’s lower mantle.
Nature, 397, 53–54.
Hirose, K. et al. (2001) Earth Planet. Sci. Lett., 184, 523 –532.
Hirsch, L.M., Shankland, T.J. and Duba, A.G. (1993) Electrical conductivity and polaron mobility in Fe-bearing
olivine. Geophys. J. Int., 114, 36 –44.
Hirschmann, M.M. and Stolper, E.M. (1996) Possible role for garnet pyroxenite in the origin of the garnet
signature in MORB. Contrib. Mineral. Petrol., 124, 185–209.
Hirth, G. and Kohlstedt, D.L. (1996) Water in the oceanic upper mantle: implications for rheology, melt extraction
and the evolution of the lithosphere. Earth Planet. Sci. Lett., 144, 93 –108.
Hixson, R.S., Winkler, M.A. and Hodgon, M.L. (1990) Sound speed and thermophysical properties of liquid iron
and nickel. Phys. Rev. B, 42, 6484– 6491.
Hobbs, P.V. (1974) Ice Physics. Oxford: Clarendon Press.
Hoffman, A.W. and White, W.M. (1982) Mantle plumes from ancient oceanic crust. Earth Planet. Sci. Lett., 57,
421–436.
Hofmann, A.W. (1997) Mantle geochemistry: the message from oceanic volcanism. Nature, 385,
219–229.
Hofmann, A.W., Jochum, K.P., Seifert, M. and White, W.M. (1986) Earth Planet. Sci. Lett., 79, 33–45.
Hofmann, A.W. (1988) Chemical differentiation of the Earth: the relationship between mantle, continental crust
and oceanic crust. Earth Planet. Sci. Lett., 990, 297–314.
Hofmeister, A.M. (1999) Mantle values of thermal conductivity and the geotherm from phonon lifetimes. Science,
283, 1699–1706.
Hofmeister, A.M. and Chopelas, A. (1991) Vibrational spectroscopy of end-member silicate garnets. Phys. Chem.
Miner., 17, 503–526.
Hofmeister, A.M., Xu, J., Mao, H.K., Bell, P.M. et al. (1989) Thermodynamics of Fe –Mg olivines at
mantle pressures: mid- and far-infrared spectroscopy at high pressure. Am. Mineral., 74, 281– 306.
Hofmeister, A.M. (1996) In: Dyar, M.D. et al. (eds.), Mineral Spectroscopy: A Tribute to Roger R. Burns.
Houston, TX: Geochemical Society, pp. 215 –227.
Hofmeister, A.M., Cynn, H., Burnley, P.C. and Meade, C. (1999) Vibrational spectra of dense, hydrous magnesium
silicates at high pressure: importance of the hydrogen bond angle. Am. Mineral., 84, 454– 464.
Hogrefe, A., Rubie, D.C., Sharp, T.G. and Seifert, F. (1994) Metastability of enstatite in deep subducting
lithosphere. Nature, 372, 351–353.
Holian, B.L. and Lomdahl, P.S. (1998) Plasticity induced by shock waves in nonequilibrium molecular-dynamics
simulations. Science, 280, 2085.
Holland, J.H. (1998) Emergence: From Chaos to Order. Helix Books.
1140 References

Holland, T.J.B. (1980) The reaction albite ¼ jadeite þ quartz, determined experimentally in the range 600–
12008C. Am. Mineral., 65, 129 –134.
Holland, T.J.B. (1983) The experimental determination of activities in disordered and short-range ordered jadeitic
pyroxenes. Contrib. Mineral. Petrol., 82, 214 –220.
Holland, K.G. and Ahrens, T.J. (1997) Melting of (Mg, Fe)2SiO4 at the core–mantle boundary of the Earth.
Science, 275, 1623–1625.
Holland, T.J.B. and Powell, R. (1990) An enlarged and updated internally consistent thermodynamics dataset with
uncertainties and correlations: the system K2O–Na2O –CaO– MgO–MnO–FeO–Fe2O3 –Al2O3 –TiO2 –
SiO2 –C–H2 – O2. J. Metamorph. Geol., 8, 89 –124.
Holland, T.J.B., Redfern, S.A.T. and Pawley, A.R. (1996) Volume behaviour of hydrous minerals at high pressure
and temperature: II. Compressibilities of lawsonite, zoisite, clinozoisite, and epidote. Am. Mineral., 81,
341–348.
Hollerbach, R. and Jones, C.A. (1999) Nature, 365, 541 –543.
Holme, R. (2000) Phys. Earth Planet. Interiors, 117, 329.
Holmes, N.C., Moriarty, J.A., Gathers, G.R. and Nellis, W.J. (1989) J. Appl. Phys., 66, 2962–2967.
Holmes, N.C., Ross, M. and Nellis, W.J. (1995) Phys. Rev. B, 52, 15835.
Holtz, F., Behrens, H., Dingwell, D.B. and Jones, W. (1995) H2O solubility in haplogranitic melts: compositional,
pressure, and temperature dependence. Am. Mineral., 80, 94–108.
Holzapfel, W.B. (1972) J. Chem. Phys. 56, 712.
Holzapfel, W.B. (1996) Physics of solids under strong compression. Rep. Prog. Phys., 59, 29–90.
Holzapfel, W.B. (1972) J. Chem. Phys., 712.
Holzheid, A., Borisov, A. and Palme, H. (1994) The effect of oxygen fugacity and temperature on
solubilities of Ni, Co and Mo in silicate melts. Geochim. et Cosmochim. Acta, 58,
1975–1981.
Honda, M., McDougall, I., Patterson, D.B., Doulgreis, A. and Clague, D.A. (1991) Nature, 349, 149.
Honda, S., Yuen, D.A., Balachandan, S. and Reuteier, D. (1993) 3-dimensional instabilities of mantle convection
with multiple phase-transition. Science, 259, 1308– 1311.
Hone, J. et al. (2000) Science, 289, 1730.
Hoover, W.G. (1979) Phys. Rev. Lett., 42, 1531.
Hopper, J.R. and Buck, W.R. (1993) The initiation of rifting at constant tectonic force: role of diffusion creep.
J. Geophys. Res., 98, 16213–16221.
Horai, K. (1971) J. Geophys. Res., 76, 1278.
Horai, K. and Sasaki, J. (1989) Phys. Earth Planet. Interiors, 55, 292.
Horiuchi, H., Hirano, M., Ito, E. and Matsui, Y. (1982) MgSiO3 (ilmenite type): single crystal X-ray diffraction
study. Am. Mineral., 67, 788 –793.
Horowood, J.L., Townsend, M.G. and Webster, H. (1976) J. Solid State Chem., 17, 35.
Horz, F. and Quaide, W.L. (1973) Debye–Scherrer investigations of experimentally shocked silicates. The Moon,
6, 45 –72.
Hough, R.M., Gllmour, I., Pillinger, C.T., Arden, J.W., Gilkes, K.W.R., Yuan, J. and Milledge, H.J.
(1995) Diamond and silicon carbide in impact melt rock from the Ries impact crater. Nature,
378, 41.
Houser, B., Alberding, N., Ingalls, R. and Rozier, E.D. (1988) Phys. Rev. B, 37, 65513.
Hu, J., Mao, H.K., Shu, J. and Hemley, R.J. (1993) High pressure energy dispersive X-ray diffraction technique
with synchrotron radiation, High Pressure Science and Technology — 1993, Vol. 1. New York: AIP
Press, 1994, 441 pp.
Huang, E. and Bassett, W.A. (1986) Rapid determination of Fe3O4 phase diagram by synchrotron radiation.
J. Geophys. Res., 91, 4697–4703.
Huang, E. and Wyllie, P.J. (1975) Melting and subsolidus phase relations for CaSiO3 to 35 kb pressure. Am.
Mineral., 60, 213 –217.
Huang, E., Lin, J.F., Xu, J.A. and Yu, S.C. (1995) Raman spectroscopic study of diaspore up to 25 GPa. J. Geol.
Soc. China, 38, 25–36.
Hubbard, W.B. (1981) Science, 214, 145.
References 1141

Hubbard, W.B. (1997) Science, 275, 1279.

Hubbard, W.B., Poodolak, M. and Stevenson, D.J. (1995) In: Cruikshank, D.P. (ed.), Neptune and Triton. Tucson,
AZ: University of Arizona Press, pp. 109–138.
Hubbard, W.B. et al. (1997) Phys. Plasmas, 4, 2011.
Huebner, J.S., Duba, A. and Wiggins, L.B. (1979) Electrical conductivity of pyroxene which contains trivalent
cations: laboratory measurements and the lunar temperature profile. J. Geophys. Res., 84, 4652–4656.
Hugh-Jones, D. and Angel, R.J. (1979) A compressional study of MgSiO3 orthoenstatite up to 8.5 GPa. Am.
Mineral., 79, 405 –410.
Hugh-Jones, D.A., Woodland, A.B. and Angel, R.J. (1994) The structure of high-pressure C2=c ferrosilite and
crystal chemistry of high-pressure C2=c pyroxenes. Am. Mineral., 79, 1032–1041.
Hugh-Jones, D., Sharp, T., Angel, R.J. and Woodland, A.B. (1996) The transformation of orthoferrosilite to high
pressure C2=c clinoferrosilite at ambient temperature. Eur. J. Mineral., 8, 1337.
Hugh-Jones, D., Chopelas, A. and Angel, R.J. (1997) Tetrahedral compression in (Mg, Fe)SiO3 orthopyroxenes.
Phys. Chem. Miner., 24, 301 –310.
Humbert, P. and Plicque, F. (1972) Properiétés élastiques des carbonates rhombohédriques monocristallins:
calcite, magnésite et dolomite. C.R. Acad. Sci. Paris, 275, 391–394.
Hung, W.L. and Wyllie, P.J. (1975b) Melting and subsolidus phase relations for CaSiO3 to 35 Kb pressure.
Am. Mineral., 60, 213 –217.
Huss, G.R. and Lewis, R.S. (1995) Geochim. Cosmochim. Acta, 59, 115.
Hwang, H.Y., Cheong, S.W., Radaelli, P.G., Marezio, M. and Batlogg, B. (1995) Phys. Rev. Lett., 75, 914.
Hwang, S.-L., Shen, P., Chu, H.-T. and Yui, T.-F. (2000) Science, 288, 321.
Ichimaru, S. and Ogata, S. (eds.) (1995) Elementary Processes in Dense Plasmas. Reading, MA: Addison-Wesley.
Ida, Y. (1987) Structure of mantle wedge and volcanic activities in the island arcs. In: Manghani, M.H. and Syono,
Y. (eds.), High Pressure in Mineral Physics. Tokyo/Washington, DC: Terra Scientific/AGU,
pp. 473 –480.
Iedema, M.J. et al. (1998) J. Phys. Chem. B, 102, 9203–9214.
Ienniskens, P. and Blake, D.F. (1994) Science, 265, 753–756.
Iidaka, T. and Suetsugu, D. (1992) Seismological evidence for metastable olivine inside a subducting slab. Nature,
356, 140 –141.
Iishi, K. and Kitayama, K. (1995) Stability of clinoenstatite. N. Jb. Miner. Mh., 2, 65 –74.
Inbar, I. and Cohen, R.E. (1994) Comparison of the electronic structures and energetics of LiTaO3 and LiNbO3.
Phys. Rev. B, 53, 1193– 1204.
Inbar, I. and Cohen, R.E. (1995a) Geophys. Res. Lett., 22, 1533.
Inbar, I. and Cohen, R.E. (1995b) Origin of ferroelectricity in LiTaO3 and LiNbO3; LAPW total energy
calculations. Ferroelectrics, 164, 45–55.
Ingrin, J. and Skogby, H. (2000) Hydrogen in nominally anhydrous upper-mantle minerals: concentration levels
and implications. Eur. J. Mineral., 12, 543–570.
Inoe, T. and Sawamoto, H. (1992) In: Syono, Y. and Manghnani, M.H. (eds.), High Pressure Research:
Application to Earth and Planetary Sciences. Tokyo: Terra Scientific.
Inoue, T. (1994) Effect of water on melting phase relations and melt composition in the system Mg2SiO4 –
MgSiO3 – H2O up to 15 GPa. Phys. Earth Planet. Interiors, 85, 237– 263.
Inoue, T., Ringwood, A.E., Yuromoto, H. and Miyagi, I. (1995a) Decomposition of K-amphibole at high pressure:
implications for the origin of the third chain volcanism (abstr). EOS, 76, F711.
Inoue, T., Yurimoto, H. and Kudoh, Y. (1995b) Hydrous modified spinel, Mg1.75SiH0.5O4: a new water reservoir
in the mantle transition region. Geophys. Res. Lett., 22, 117.
Inoue, T., Weidner, D.J., Northrup, P.A. and Parise, J.B. (1997) Elastic properties of hydrous ringwoodite
(g-phase) in Mg2SiO4. Earth Planet. Sci. Lett., 160, 107–113.
Inoue, T., Irifune, T., Yurimoto, H. and Miyagi, I. (1998) Decomposition of K-amphibole at high pressures and
implications for subduction zone volcanism. Phys. Earth Planet. Interiors, 107, 221–231.
Ionov, D.A. and Hofmann, A.W. (1995) Nb– Ta-rich mantle amphiboles and micas. Implications for subduction-
related metasomatic trace element fractionations. Earth Planet. Sci. Lett., 131, 341–356.
1142 References

Iota, V., Yoo, C.S. and Cynn, H. (1999) Quartz-like carbon dioxide: an optically nonlinear extended solid at high
pressure and temperatures. Science, 283, 1510–1513.
Irifune, T. (1994) Absence of an aluminous phase in the upper part of the earth’s lower mantle. Nature, 370,
131–133.
Irifune, T. (1996) Amorphization of serpentine at high pressure and high temperature: implications for the
mechanism of deep-focus earthquakes. J. Crystallogr. Soc. Jpn, 38, 396–399.
Irifune, T. and Hiraya, Y. (1983) Phase relationships in the system Mg3Al2Si3O12 –Mg3Cr2Si3O12 at high
pressure and some mineralogical properties of synthetic garnet solid solutions. Mineral. J., 11,
269–281.
Irifune, T. and Issuhiki, M. (1998) Iron partitioning in a pyrolite mantle and the nature of the 410-km seismic
discontinuity. Nature, 392, 702.
Irifune, T., Ohtani, E., Kumazawa, M. (1982) Stability field of knorringite Mg3Cr Si3O12 at high pressure and its
implication to the occurrence of Cr-rich pyrope in the upper mantle. Phys. Earth Planet Interior,
263–272.
Irifune, T. and Ringwood, A.E. (1987a) Phase transformation in a harzburgite composition to 26 GPa:
implication for dynamics behaviour of the subducting slab. Earth Planet. Sci. Lett., 86,
365–376.
Irifune, T. and Ringwood, A.E. (1987b) In: Manghnani, M.H. and Syono, Y. (eds.), High-Pressure Research in
Mineral Physics. Tokyo/Washington, DC: Terra Scientific/AGU, pp. 231–342.
Irifune, T. and Ringwood, A.E. (1993) Phase transformation in subducted oceanic crust and buoyancy relationship
at depths of 600– 800 km in the mantle. Earth Planet. Sci. Lett., 117, 101–110.
Irifune, T., Susaki, J., Yagi, T. and Sawamoto, H. (1989) Phase transformations in diopside CaMgSi2O6 at
pressures up to 25 GPa. Geophys. Res. Lett., 16, 187–190.
Irifune, T., Fujino, K. and Ohtani, E. (1991) A new high-pressure form of MgAl2O4. Nature, 349,
409–411.
Irifune, T., Ringwood, A.E. and Hibberson, W.O. (1994) Subduction of continental crust and terrigenous and
pelagic sediments: an experimental study. Earth Planet. Sci. Lett., 126, 351–368.
Irifune, T., Kuroda, K., Minigawa, T. and Unemoto, M. (1995) Experimental study of the decomposition of
kyanite at high pressure and high temperature. In: Yukutake, T. (ed.), The Earth’s Central Part: Its
Structure and Dynamics. Tokyo: Terra Scientific, pp. 35–44.
Irifune, T., Koizumi, T. and Ando, J. (1996) An experimental study of the garnet–perovskite transformation in the
system MgSiO3 –Mg3Al2Si3O12. Phys. Earth Planet. Interiors, 15, 90–106.
Irifune, T., Kubo, N., Isshiki, M. and Yamasaki, Y. (1998a) Phase transformations in serpentine and transportation
of water into the lower mantle. Geophys. Res. Lett., 25, 203–206.
Irifune, T., Nishiyama, N., Kuroda, K., Inoue, T., Isshiki, M., Utsumi, W., Funakoshi, K., Urakawa, S., Uchida, T.,
Katsura, T. and Ohtaka, O. (1998b) The post spinel phase boundary in Mg2SiO4 determined by in situ X-
ray diffraction. Science, 279, 1698–1700.
Irvine, T.N. (1975) Crystallization sequence in the Muskox intrusion and other layered intrusions II. Origin of
chromitite layers and similar deposits of other magmetic ores. Geochim. Cosmochim. Acta, 39, 991– 1020.
Isaak, D.G. (1992) J. Geophys. Res., 97, 1871–1885.
Isaak, D.G. and Masuda, K. (1995) Elastic and viscoelastic properties of a iron at high temperatures. J. Geophys.
Res., 100, 17689–17698.
Isaak, D.G., Anderson, O.L. and Goto, T. (1989) Measured elastic moduli of single-crystal MgO up to 1800 K.
Phys. Chem. Miner., 16, 704 –713.
Isaak, D.G., Cohen, R.E. and Mehl, M.J. (1990) Calculated elastic and thermal properties of MgO at high
pressures and temperatures. J. Geophys. Res., 95, 7055–7067.
Isaak, D.G., Anderson, O.L. and Cohen, R.E. (1992) The relationship between shear and compressional velocities
at high pressures: reconciliation of seismic tomography and mineral physics. Geophys. Res. Lett., 19,
741–744.
Isaak, D.G., Cohen, R.E., Mehl, M.J. and Singh, D.J. (1993) Phys. Rev. B, 47, 7720.
Ita, J. and Cohen, R.E. (1998) Diffusion in MgO at high pressure: implications for lower mantle rheology.
Geophys. Res. Lett., 25, 1095–1098.
References 1143

Ita, J. and Stixrude, L. (1987) J. Geophys. Res., 97, 6849.

Ita, J. and Stixrude, L. (1992) Petrology, elasticity, and composition of the mantle transition zone. J. Geophys.
Res., 97, 6849–6866.
Ita, J. and Stixrude, L. (1993) Density and elasticity of model upper mantle composition and their implications for
whole mantle structure. Geophys. Monogr., 74, 111–130.
Itie, J.P., Polian, A., Calas, G., Petiau, J., Fontaine, A. and Tolentino, H. (1990) High Press. Res., 5, 717.
Ito, E. (1984) Ultrahigh pressure phase relations in the system MgO–FeO–SiO2 and their geophysical
implications. In: Sunagawa, I. (ed.), Material Science of the Earth’s Interior. Tokyo: Terra Scientific,
pp. 387 –394.
Ito, E. (1989) Stability relations of silicate perovskite under subsolidus conditions. In: Navrotsky, A. and Weidner,
D.J. (eds.), Perovskite: A structure of Great Interest Geophysics and Material Science, Geophysical
Monograph, Vol. 45. Washington, DC: AGU, pp. 27– 32.
Ito, E. and Katsura, T. (1989) A temperature profile of the mantle transition zone. Geophys. Res. Lett., 16,
425–428.
Ito, E. and Katsura, T. (1992) Melting of ferromagnesian silicates under the lower mantle conditions. In: Syono, Y.
and Manghnani, M.H. (eds.), High Pressure Research: Applications to Earth and Planetary Science.
Tokyo/Washington, DC: Terra Scientific/AGU.
Ito, K. and Kennedy, G.C. (1971) An experimental study of the basalt–garnet granulite– eclogite transition. In:
Heacock, J.G. (ed.), The Structure and Physical Properties of the Earth’s Crust, Geophysical Monograph
Series 14. Washington, DC: AGU, pp. 303–314.
Ito, E. and Matsui, Y. (1978) Synthesis and crystal chemical characterization of MgSiO3 perovskite. Earth Planet.
Sci. Lett., 38, 443 –450.
Ito, E. and Matsui, Y. (1979) High-pressure transformation in silicates, germanates, and titanates with ABO3
stoichiometry. Phys. Chem. Miner., 4, 265 –274.
Ito, E. and Navrotsky, A. (1985) MgSiO3 ilmenite: calorimetry, phase equilibria, and decomposition at
atmospheric pressure. Am. Mineral., 70, 1020–1026.
Ito, E. and Sato, H. (1991) Aseismicity in the lower mantle by superplasticity of the descending slab. Nature, 351,
140–141.
Ito, E. and Sato, H. (1992) Effects of phase transitions on the dynamics of the descending slab. In: Syono,
Y. and Manghani, M.H. (eds.), High Pressure Research in Mineral Physics: Application to Earth
and Planetary Science. Tokyo/Washington, DC: Terra Scientific/AGU, pp. 257–262.
Ito, E. and Takahashi, E. (1987a) Ultrahigh-pressure phase transformations and the constitution of the deep
mantle. In: Manghnani, M.H. and Syono, Y. (eds.), High Pressure Research in Mineral Physics.
TERRAPUB/AGU, Tokyo/Washington, 221–229.
Ito, E. and Takahashi, E. (1987b) Melting of peridotite at uppermost lower-mantle conditions. Nature, 328,
514–516.
Ito, E. and Takahashi, E. (1989) Post-spinel transformation in the system Mg2SiO4 –Fe2SiO4 and some
geophysical implications. J. Geophys. Res., 94, 10637–10646.
Ito, E. and Weidner, D.J. (1986) Geophys. Res. Lett., 13, 464–466.
Ito, E. and Yamada, H. (1982) Stability relations of silicate spinels, ilmenites and perovskites. In: Akimoto, S. and
Manghnani, M.H. (eds.), High-Pressure Research in Geophysics. Tokyo: Center for Academic
Publications, pp. 405–419.
Ito, H., Kawada, K. and Akimoto, S. (1974a) Thermal expansion of stishovite. Phys. Earth Planet. Interiors, 8,
277–281.
Ito, H., Kawada, K. and Akimoto, S. (1974b) Erratum. Phys. Earth Planet. Interiors, 9, 371.
Ito, E., Matsui, Y. and Yamada, H. (1982) Stability relations of silicate ilmenites and perovskites. In: Akimoto, S.
and Manghnani, M.H. (eds.), High Pressure Research in Geophysics. Tokyo: Center for Academic
Publications.
Ito, E., Takahashi, E. and Matsui, Y. (1984) The mineralogy and chemistry of lower mantle; an implication of the
ultrahigh-pressure phase relations in the system MgO–FeO– SiO2. Earth Planet. Sci. Lett., 67, 238–248.
Ito, E., Akaogi, M., Topor, L. and Navrotsky, A. (1990) Negative pressure–temperature slopes for reactions
forming MgSiO3 perovskite from calorimetry. Science, 249, 1275–1278.
1144 References

Ito, E., Morooka, K., Ujike, O. and Katsura, T. (1995) Reactions between molten iron and silicate melts at high
pressure: implications for the chemical evolution of the Earth’s core. J. Geophys. Res., 100,
5901– 5910.
Iton, L.E., Curties, L.A. and Hopfinger, A.J. (1992) Ab initio molecular orbital study of the effects of basis set size
on the calculated structure and acidity of hydroxyl groups in framework molecular sieves. J. Phys.
Chem., 96, 10247–10257.
Iwamori, H. (1993) Nature, 366, 734–737.
Iwamori, H. (2000) Earth Planet. Sci. Lett., 189, 19.
Jackson, I. (1998) Elasticity, composition and temperature of the Earth’s lower mantle: a reappraisal. Geophys. J.
Int., 134, 291–311.
Jackson, I. and Niesler, H. (1982) The elasticity of periclase and some geophysical implications. In: Akimoto, S.
and Maghnani, M. (eds.), High-Pressure Research in Geophysics. Tokyo: Academic Publications,
pp. 93 –133.
Jackson, I. and Rigden, S.M. (1996) Analysis of P – V – T data: constraints on the thermoelastic properties of high-
pressure minerals. Phys. Earth Planet. Interiors, 96, 85– 112.
Jackson, W.E., Knittle, E., Brown, G.E. and Jeanloz, R. Jr. (1987) Partitioning of Fe within high-pressure
silicate perovskite: evidence for unusual geochemistry in the lower mantle. Geophys. Res. Lett., 14,
224–226.
Jackson, I., Khanna, S.K., Revcolevschi, A. and Berthon, J. (1990) Elasticity, shear-mode softening and high-
pressure polymorphism of wüstite (Fe12xO). J. Geophys. Res., 95, 21671– 21685.
Jackson, S.M., Neild, V.M., Whitworth, R.W., Oguro, M. and Wilson, C.C. (1997) J. Phys. Chem. B, 101,
6142– 6145.
Jacobs, J.A. (1987) Geomagnetism. New York: Academic Press.
Jacobsen, S.B. (2001) Nature, 411, 69–73.
Jambon, A. (1994) Earth degassing and large scale geochemical cycling of volatile elements. Rev. Mineral., 30,
479–517.
James, F. and Ross, M. (1975) Minuit — a system for function minimization and analysis of the parameter errors
and correlations. Phys. Commun., 10, 343–367.
Jamieson, J.C. (1977) Phase transitions in rutile type structures. In: Manghnani, M.H. and Akimoto, S. (eds.), High
Pressure Research: Applications in Geophysics. Orlando, FL: Academic Press, pp. 209–218.
Jamieson, J.C., Fritz, J.N. and Manghnani, M.H. (1982) Pressure measurement at high temperature in X-ray
diffraction studies: gold as a primary standard. In: Akimoto, S. and Manghnani, M.H. (eds.), High
Pressure Research in Geophysics. Tokyo: Center for Academic Publications, pp. 27–48.
Jana, D. and Walker, D. (1997) The influence of sulfur on partitioning of siderophile elements. Geochim.
Cosmochim. Acta, 61(24).
Jana, D. and Walker, D. (1999) Core formation in the presence of various C –H–O species. Geochim. Cosmochim.
Acta, 63, 2299– 2310.
Jaoul, O., Poumellec, M., Froidevaux, C. and Havette, A. (1981) Silicon diffusion in forsterite: a new constraint
for understanding mantle deformation. In: Stacey, F.D. et al. (eds.), Anelasticity in the Earth.
Washington, DC: AGU, pp. 95–100.
Jault, G., Hulot, G. and LeMouel, J.L. (1996) Phys. Earth Planet. Interiors, 98, 1987.
Jayaraman, A. (1984) The diamond-anvil high pressure cell. Sci. Am., 250(4), 54 –62.
Jayaraman, A. and Ballman, A.A. (1986) J. Appl. Phys., 60, 1208.
Jeanloz, R. and Hemley, R.J. (1994) Thermoelasticity of perovskite: an emerging consensus. EOS, 75(14),
476–477.
Jeanloz, R. (1981a) Finite-strain equation of state for high-pressure phase. Geophys. Res. Lett., 8, 1219–1222.
Jeanloz, R. (1981b) Majorite: vibrational and compressional properties of a high-pressure phase. J. Geophys. Res.,
86, 6171–6179.
Jeanloz, R. (1983) Mineral and melt physics. Rev. Geophys. Space Phys., 21, 1487–1503.
Jeanloz, R. (1987) J. Geophys. Res., 92, 10352.
Jeanloz, R. (1990a) Annu. Rev. Phys. Chem., 40, 237 –259.
Jeanloz, R. (1990b) The nature of the earth’s core. Annu. Rev. Earth Planet. Sci., 18, 357–386.
References 1145

Jeanloz, R. (1991) Effects of phase transitions and possible compositional changes on the seismological structure
near 650 km depth. Geophys. Res. Lett., 18, 1743–1746.
Jeanloz, R. (1993) Relating Geophysical Structures and Process, The Jeffreys Volume, Geophysical Monograph,
Vol. 76. Washington, DC: AGU, pp. 121–127.
Jeanloz, R. (1994) Proceedings of the International Symposium on Advanced Materials, Tsukuba, Japan.
Jeanloz, R. and Ahrens, T.J. (1980) Equations of state of FeO and CaO. Geophys. J. R. Astron. Soc., 62, 505–528.
Jeanloz, R. and Hazen, R.M. (1983) Nature, 304, 620.
Jeanloz, R. and Knittle, E. (1989) Density and composition of the lower mantle. Phil. Trans. R. Soc. Lond. A, 328,
377–389.
Jeanloz, R. and Morris, S. (1986) Temperature distribution in the crust and mantle. Annu. Rev. Earth Planet. Sci.,
14, 377 –415.
Jeanloz, R. and Richter, F.M. (1979) Convection, composition, and the thermal state of lower mantle. J. Geophys.
Res., 84, 5497–5504.
Jeanloz, R. and Romanowics, B. (1997) Phys. Today, August, 22.
Jeanloz, R. and Thompson, A.B. (1983) Phase transitions and mantle discontinuities. Rev. Geophys., 21, 51 –74.
Jeanloz, R., Ahrens, T.J., Lally, J.S., Nord, G.L., Christie, J.M. and Hauer, A.H. (1977) Shock-produced olivine
glass; first observation. Science, 197, 457– 459.
Jeffreys, H. (1963) On the hydrostatic theory of the figure of the Earth. Geophys. J., 8, 196.
Jephcoat, A.P. (1998) Rare-gas solids in the Earth’s interior. Nature, 393, 355–358.
Jephcoat, A.P. and Besedin, S.P. (1996) Temperature measurement and melting determination in the laser heated
diamond anvil cell. Phil. Trans. R. Soc. Lond. A, 354, 1333–1360.
Jephcoat, A.P. and Olson, P. (1987) Is the inner core of the Earth pure iron? Nature, 389, 371–374.
Jephcoat, A.P., Mao, H.K. and Bell, P.M. (1986a) The static compression of iron to 78 GPa with rare gas solids as
pressure-transmitting media. J. Geophys. Res. B, 91, 4677–4684.
Jephcoat, A.P., Mao, H.K., Zha, C.S., Bell, P.M., Finger, L.W. and Cox, D.E. (1986b) Energy-dispersive
diffraction above 100 GPa. Nat. Synchrotron Light Source Annu. Rep., 1986, 323–324.
Jephcoat, A.P., Mao, H.K. and Bell, P.M. (1987) Operation of the megabar diamond-anvil cell. In: Ulmer, G.C.
and Barnes, H.L. (eds.), Hydrothermal Experimental Techniques. New York: Wiley Interscience,
pp. 469 –506.
Jephcoat, A.P., Hemley, R.J. and Mao, H.K. (1988) X-ray diffraction of ruby (Al2O3:Cr3þ) to 175 GPa. Physica B,
150, 115 –121.
Jin, W., Kalia, R.K., Vashishta, P. and Rino, J.P. (1994a) Structural transformation in densified silica: a molecular
dynamic study, Phys. Rev., (B50), 118 –131.
Jin, S., Tiefel, T.H., McCormack, M., Fastnacht, R.A., Ramesh, R. and Chen, L.H. (1994a)
Structural transformation in densified silica glass: a molecular dynamics study. Phys. Rev., (B) 50,
118–131.
Jin, S., Tiefel, T.H., McCormack, M., Fastnacht, R.A., Ramesh, R. and Chen, L.H. (1994b) Thousand fold change
in resistivity in magnetoresistive La–Ca –Mn–O films, Science 264, 413.
Jin, D.H., Karato, S. and Obata, M. (1998) Mechanism of shear dislocation in the continental lithosphere:
inference from the deformation microstructure in peridotites from the Ivrea zone, North Western Italy,
J. Struct. Geol., 20, 195–209.
Johnson, K.A. and Ashcroft, N.W. (2000) Structure and band gap closer in dense hydrogen, Nature, 403,
632–635.
Johnson, W.L. and Fecht, H.J. (1988) In: Schwartz, R.B. and Johnson, W.L. (eds.), Solid-State Amorphizing
Transformations. The Netherlands: Elsevier Sequoia, 63 pp.
Johnson, W.A. and Mehl, R.F. (1939) Trans. Am. Inst. Min. Metall. Petrol. Eng., 135, 416.
Johnson, M.C. and Walker, D. (1993) Brucite [Mg(OH)2] dehydration and the molar volume of H2O to 15 GPa.
Am. Mineral., 78, 271 –284.
Johnson, M.C., Rutherford, M.J. and Hess, P.C. (1991) Chassigny petrogenesis: melt composition, intensive
parameters, and water contents of Martain(?) magmas. Geochim. et Cosmochim. Acta, 55(1), 349–367.
Jonas, J. (1982) Nuclear magnetic resonance at high pressure. Science, 216, 1179–1184.
Jones, L.E.A. (1977) High temperature elasticity of rutile-structure MgF2. Phys. Chem. Miner., 1, 179–180.
1146 References

Jones, J.H., Capobianco, C.J. and Drake, M.J. (1992) Science, 257, 1281.
Jouniaux, L. and Pozzi, J.P. (1995) J. Geophys. Res., 100, 10197–10209.
Jung, H. and Karato, S. (2001) Water-induced fabric transitions in olivine. Science, 293, 1460– 1463.
Jurewics, A.J.G. and Watson, E.B. (1988) Cations in olivine, Part 1. Calcium partitioning and calcium-magnesium
distribution between olivines and coexisting melts, with petrologic applications. Contrib. Mineral.
Petrol., 99, 176–185.
Kafalas, J.A., Menyuk, N., Dwight, K. and Longo, J.M. (1971) J. Appl. Phys., 42, 11497.
Kagi, H., Inoue, T., Weidner, D.J., Lu, R. and Rossman, G.R. (1997) Speciation of hydroxides in hydrous
ringwoodite. EOS Trans. Am. Geophys. Union, 78, S312.
Kagi, H., Lu, R., Davidson, P.M., Goncharov, A.F., Mao, H.K. and Hemley, R.J. (2000a) Evidence for ice IV as
an inclusion of cuboid diamonds from high P – T near infrared spectroscopy. Mineral. Mag., 64,
1089– 1097.
Kagi, H., Parise, J.B., Cho, H., Rossman, G.R. and Loveday, J.S. (2000b) Hydrogen bonding interactions
in phase A [Mg7Si2O8 (OH)6] at ambient and high pressure. Phys. Chem. Miner., 27(4),
225–233.
Kajiyoshi, K. (1986) High Temperature Equation of State for Mantle Minerals and Their Anharmonic Properties.
Okayama, Japan: Okayama University.
Kakol, Z., Sabol, J., Stickler, J. and Honig, J.M. (1992) Effect of low level titanium (IV) doping on the resistivity
of magnetite near the Verwey transition. Phys. Rev. B, 46, 1575–1578.
Kamb, B. (1973) In: Whalley, E., Jones, S.J. and Gold, L.W. (eds.), Physics and Chemistry of Ice. Ottawa: Royal
Society of Canada, pp. 28–41.
Kanamori, J. (1959) J. Phys. Chem. Solids, 10, 87.
Kandelin, J. and Weidner, D.J. (1988) Elastic properties of hedenbergite. J. Geophys. Res., 93, 1063–1072.
Kaneoka, I. (2000) Earth’s history trapped in the mantle. Science, 288, 988–989.
Kanzaki, M. (1989) High pressure phase relations in the system MgO– SiO2 –H2O. EOS, 70, 508.
Kanzaki, M. (1991a) Ortho-clinoenstatite transition. Phys. Chem. Miner., 17, 726–730.
Kanzaki, M. (1991b) Stability of hydrous magnesium silicates in the mantle transition zone. Phys. Earth Planet.
Interiors, 66, 307 –312.
Kanzaki, M., Stebbins, J.F. and Xue, X. (1991) Characterization of quenched high pressure phases in CaSiO3
system by XRD and 29Si NMR. Geophys. Res. Lett., 18, 463–466.
Kanzaki, M., Stebbins, J.F. and Xue, X. (1992) Characterization of crystalline and amorphous silicates quenched
form high pressure by 29Si MAS NMR spectroscopy. In: Syono, Y. and Manghnani, M.H. (eds.), High-
Pressure Research: Application to Earth and planetary Science. Washington, DC: AGU, pp. 89–100.
Kanzaki, M., Matsui, Y. and Matsui, M. (1997) A new high-pressure silica phase obtained by molecular
dynamics — discussion. Am. Mineral., 82, 1042.
Kapitza, P. (1938) Nature, 141, 74.
Kapp, R.P. and Watson, E.B. (1995) Dehydration melting of metabasalt at 8 –32 kbar: implication for continental
growth and crust –mantle recycling. J. Petrol., 36(4), 891–931.
Karato, S. (1981) Rheology of the lower mantle. Phys. Earth Planet. Interiors, 24, 1–14.
Karato, S. (1984) Grain-size distribution and rheology of the upper mantle. Tectonophysics, 104,
155–176.
Karato, S. (1987) Seismic anisotropy due to lattice preferred orientation of minerals: kinematic or dynamic (?).
In: Manghnani, M.H. and Syono, Y. (eds.), High Pressure Research in Mineral Physics. Tokyo:
Terra Scientific, pp. 455 –471.
Karato, S. (1989) Plasticity–crystal structure systematics in dense oxides and its implications for the creep
strength of the Earth’s deep interior: a preliminary result. Phys. Earth Planet. Interiors, 55,
234–240.
Karato, S. (1990) The role of a hydrogen in the electrical conductivity of the upper mantle. Nature, 347,
272–273.
Karato, S. (1992) On the Lehmann discontinuity. Geophys. Res. Lett., 19, 2255– 2258.
Karato, S. (1993) Dislocation recovery in olivine under deep upper mantle conditions: implications for creep and
diffusion. J. Geophys. Res., 98, 9761–9768.
References 1147

Karato, S. (1998) Seismic anisotropy in the deep mantle, boundary layers and the geometry of mantle convection.
Pure Appl. Geophys., 151, 565–587.
Karato, S. (1999) Seismic anisotropy of the Earth’s core caused by the Maxwell stress-induced flow. Nature, 402,
871–873.
Karato, S. and Ogawa, M. (1982) High-pressure dislocation recovery in olivine: implications for creep
mechanisms and creep activation volume. Phys. Earth Planet. Interiors, 28, 102– 117.
Karato, S. and Rubie, D.C. (1997) Toward an experimental study of deep mantle rheology: a new multianvil
sample assembly for deformation studies under high pressures and temperatures. J. Geophys. Res., 102,
20111–20120.
Karato, S. and Wu, P. (1993) Rheology of the upper mantle: a synthesis. Science, 262, 1708.
Karato, S., Paterson, M.S. and FiltzGerald, J.D. (1986) Rheology of synthetic olivine aggregates: influence of
grain size and water. J. Geophys. Res., 91, 8151–8176.
Karato, S., Fujino, K. and Ito, E. (1990) Plasticity of MgSiO3 perovskite: the results of microhardness tests on
single crystals. Geophys. Res. Lett., 17, 13 –16.
Karato, J., Rubie, D.C. and Yan, H. (1993) Dislocation recovering in olivine under deep upper mantle conditions:
implications for creep and diffusion. J. Geophys. Res., 98(B6), 9761– 9768.
Karato, S., Zhang, S. and Wenk, H.R. (1995) Superplasticity in Earth’s lower mantle: evidence from seismic
anisotropy and rock physics. Science, 270, 458 –461.
Karato, S., Bruzek, C.D. and Rubie, D.C. (1998) Plastic deformation of silicate spinel under the transition-zone
conditions of the Earth’s mantle. Nature, 395, 266– 269.
Karki, B.B. and Crain, J. (1998) First-principles determination of elastic properties of CaSiO3 perovskite at lower
mantle pressures. Geophys. Res. Lett., 25, 2741–2744.
Karki, B.B. and Stixrude, L. (1999) Seismic velocities of major silicate and oxide phases of the lower mantle.
J. Geophys. Res., 104, 13025–13033.
Karki, B.B., Warren, M.C., Stixrude, L., Ackland, G.J. and Crain, J. (1997a) Ab initio studies of high-pressure
structural transformations in silica. Phys. Rev. B, 55, 3465–3472.
Karki, B.B., Stixrude, L., Clark, S.J., Warren, M.C., Ackland, G.J. and Crain, J. (1997b) Structure and elasticity of
MgO at high pressure. Am. Mineral., 82, 51 –60.
Karki, B.B., Stixrude, L., Clark, S.J., Warren, M.C., Ackland, G.J. and Carin, J. (1997c) Elastic
properties of orthorhombic MgSiO3 perovskite at lower mantle pressures. Am. Mineral., 82,
635–638.
Karki, B.B., Duan, W., Da Silva, C.R.S. and Wentzcouitch, R.M. (2000) Ab initio structure of MgSiO3 ilmenite at
high pressure. Am. Mineral., 85, 317–320.
Kastner, M.A., Birgeneau, R.J., Shirane, G. and Endoh, Y. (1998) Rev. Mod. Phys., 70, 3686–3689.
Kato, T. (1986) Stability relation of (Mg,Fe)SiO3 garnets, major constituents in the Earth’s interior. Earth Planet.
Sci. Lett., 77, 399 –408.
Kato, T. and Kumazawa, M. (1985) Stability of phase B, a hydrous magnesium silicate to 23008C at 20 GPa.
Geophys. Res. Lett., 12, 534–535.
Kato, T. and Ringwood, A.E. (1984) Earth Planet. Sci. Lett., 71, 85.
Kato, T. and Ringwood, A.E. (1989) Melting relationships in the system Fe–FeO at high pressures: implications
for the composition and formation of the Earth’s core. Phys. Chem. Miner., 16, 524–538.
Kato, T., Ringwood, A.E. and Irifune, T. (1988a) Experimental determination of element partitioning between
silicate perovskites, garnets, and liquids: constraints on early differentiation of the mantle. Earth Planet.
Sci. Lett., 89, 123 –145.
Kato, T., Ringwood, A.E. and Irifune, T. (1988b) Constraints on element partitioning coefficients between
MgSiO3 perovskite and liquid determined by direct measurements. Earth Planet. Sci. Lett., 90, 65 –68.
Kato, I. et al. (1997) Rev. High Press. Sci. Technol., 6, 718, Spec. Issue.
Katsura, T. (1995) Geophys. J. Int., 122, 67.
Katsura, T. (1997) Phys. Earth Planet. Interiors, 101, 73.
Katsura, T. and Ito, E. (1989) The system Mg2SiO4 –Fe2SiO4 at high pressures and temperatures: precise
determination of stabilities of olivine, modified spinel and spinel. J. Geophys. Res., 94,
15663–15670.
1148 References

Katsura, T., Tsuchida, Y., Ito, E., Yagi, T., Utsumi, W. and Akimoto, S. (1991) Stability of magnesite under the
lower mantle conditions. Proc. Jpn. Acad., 67, 57– 60.
Kaula, A. and Lenardic, W.M. (1995) Mantle dynamics and the heat flow into the Earth’s continents. Nature, 378,
709–710.
Kavner, A., Li, X. and Jeanloz, R. (1995) Electrical conductivity of a natural (Mg, Fe)SiO3 majorite garnet.
Geophys. Res. Lett., 22.
Kawai, N. and Nishiyama, A. (1974) Proc. Jpn. Acad. Sci., 5, 72.
Kawai, T., Kannai, M. and Tabata, H. (1996) Mater. Sci. Eng., 41, 123.
Kawamoto, T., Hervig, R.L. and Holloway, J.R. (1996) Experimental evidence for a hydrous transition zone in
Earth’s early mantle. Earth Planet. Sci. Lett., 142, 587–592.
Kellogg, L.H. (1994) Fast cleaning in the deep. Nature, 371, 656.
Kelly, A. and Macmillan, N.H. (1986) Strong Solids. New York: Oxford University Press.
Kelso, P.R. and Banerjee, S.K. (1995) Effect of hydrostatic pressure on viscous remanent magnetization in
magnetite-bearing specimens. Geophys. Res. Lett., 22, 1953–1956.
Kendall, J.M. (1994) Geophys. Res. Lett., 21, 301–304.
Kendall, J.M. and Silver, P.G. (1996) Constraints from seismic anisotropy on the nature of the lowermost mantle.
Nature, 381, 409–412.
Kendall, J.M. and Silver, P.G. (1998) Investigating causes of D00 anisotropy. Geodynamics, 28, 97– 118.
Kennett, B.L.N., Engdahl, E.R. and Buland, R. (1995) Geophys. J. Int., 122, 108.
Keppler, H. (1992) Am. Mineral., 77, 62 –75.
Keppler, H. and Rauch, M. (2000) Water solubility in nominally anhydrous minerals measured by FTIR and 1H
MAS NMR: the effect of sample preparation. Phys. Chem. Miner., 27, 371–376.
Keppler, H. and Rubie, D.C. (1993) Pressure-induced coordination changes of transition-metal ions in silicate
melts. Nature, 364, 54–55.
Keppler, H., McCammon, C.A. and Rubie, D.C. (1994) Crystal-field and charge-transfer spectra of (Mg, Fe)SiO3
perovskite. Am. Miner., 79, 215–1218.
Kerley, G.I. (1980) A theoretical equation of state for deuterium (Los Alamos Laboratory Rep. LA-4776, Los
Alamos, NM, 1972). J. Chem. Phys., 73, 460.
Kerr, R.A. (1998) A deep root for Iceland. Science, 279, 806.
Kerr, R.A. (1999) A deep look beneath tall mountains. Science, 284, 124.
Kerr, R.A. (2000) Looking back to early Mars deep into Earth. Science, 287, 218–219.
Kerrick, D.M. (1990) The Al2SiO5 polymorphs. Mineral. Soc. Am. Rev. Mineral., 22, 1–406.
Kerrick, D. (2002) Serpentinite seduction. Science, 298, 1344–1345.
Kerrick, D.M. and Connolly, J.A. (2001) Metamorphic devolatilization of subducted marine sediments and the
transport of volatiles into the Earth’s mantle. Nature, 411, 293–296.
Kerschhofer, L., Sharp, T.G. and Rubie, D.C. (1996) Intracrystalline transformation of olivine to wadsleyite and
ringwood under subduction zone conditions. Science, 274, 79–81.
Keskar, N.R., Chelikowsky, J.R. and Wentzcovitch, R.M. (1994) Mechanical instabilities in AlPO4. Phys. Rev.,
50, 9072–9078.
Kesson, S.E. and FitzGerald, J.D. (1991) Partitioning of MgO, FeO, NiO, MnO and Cr2O3 between magnesian
silicate perovskites and magnesiowüstite: implications for the origin of inclusions in diamond and the
composition of the lower mantle. Earth Planet. Sci. Lett., 111, 229–240.
Kesson, S.E. and White, T.J. (1986) [BaxCsy][Ti, Al3þ 42
2xþyTiS22x2y]O16 Synroctype hollandites. II. Structural
chemistry. Proc. R. Soc. Lond. Ser., 408, 295 –319.
Kesson, S.E., FitzGerald, J.D. and Shelley, J.M. (1994) Mineral chemistry and density of subducted basaltic crust
at lower mantle pressures. Nature, 372, 767–769.
Kesson, S.E., Fitzgerald, J.D., Shelley, J.M.G. and Withers, R.L. (1995) Phase relations, structure and
crystal chemistry of some aluminous silicate perovskites. Earth Planet. Sci. Lett., 134,
187–201.
Kesson, S.E., Fitzgerald, J.D. and Shelley, J.M. (1998) Mineralogy and dynamics of a pyrolite lower mantle.
Nature, 393, 252–255.
Kestenbaum, D. (1998) Under pressure deuterium gets into quite a state. Science, 281, 1136.
References 1149

Khan, O. (1999) The magnetic turn about. Nature, 399, 21–23.

Khazeni, K., Jia, Y.X., Lu, L., Crespi, V.H., Cohen, M.L. and Zettl, A. (1996) Effect of pressure on the
magnetoresistance of single crystal Nd0.5Sr0.36Pb0.14MnO32d. Phys. Rev. Lett., 76, 295–298.
Kiefer, B., Stixrude, L. and Wentzcovitch, R.M. (1997) Calculated elastic constants and anisotropy of Mg2SiO4
spinel at high pressure. Geophys. Res. Lett., 24(22), 2841–2844.
Kiefer, B., Stixrude, L. and Wentzcovitch, R. (1999) Normal and inverse ringwoodite at high pressures. Am.
Mineral., 84, 288 –293.
Kieffer, S.W. (1976) J. Geophys. Res., 81, 3025.
Kieffer, S.W. (1980) Thermodynamics and lattice vibrations of minerals: 4. Application to chain and sheet
silicates and orthosilicates. Rev. Geophys. Space Phys., 18, 862–886.
Kieffer, S.W. (1982) Thermodynamics and lattice vibrations of minerals: 5. Applications to phase equilibria,
isotopic fractionation, and high pressure thermodynamic properties. Rev. Geophys. Space Phys., 20,
827–849.
Kim, Y.H., Ming, L.C., Manghnani, M.H. and Ko, J. (1991) Phase transformation studies on a synthetic
hedenbergite up to 26 GPa at 12008C. Phys. Chem. Miner., 17, 540–544.
King, H. and Finger, L.W. (1979) Diffracted beam crystal centering and its application to high-pressure
crystallography. J. Appl. Crystallogr., 12, 374 –378.
King, S.D. and Mesers, G. (1982) Geophys. Res. Lett., 19, 1551–1554.
King, H.E. Jr., Virgo, D. and Mao, H.K. (1978) Carnegie Inst. Washington Yearb., 77, 830.
Kingma, K.J. (1994) Pressure induced transition in SiO2. PhD Thesis. Baltimore, MD: The John Hopkins
University.
Kingma, K.J., Meade, C., Hemley, R.J., Mao, H.K. and Veblen, D.R. (1993a) Science, 259, 666.
Kingma, K.J., Hemley, R.J., Mao, H.K. and Veblen, D.R. (1993b) New high-pressure transformation in a-quartz.
Phys. Rev. Lett., 70, 3927–3930.
Kingma, K.J., Meade, C., Hemley, R.J., Mao, H.K. and Veblen, D.R. (1993c) Microstructural observations of
a-quartz amorphization. Science, 259, 666–669.
Kingma, K.J., Cohen, R.E., Hemley, R.J. and Mao, H.K. (1995) Nature, 374, 243–245.
Kingsley, R.H. and Schilling, J.G. (1995) Carbon in mid-Atlantic ridge basalt glasses from 288N to 638N:
evidence for a carbon enriched Azores mantle plume. Earth Planet. Sci. Lett., 129, 31–53.
Kinomura, N., Kume, S. and Koizumi, M. (1975) Stability of K2Si4O9 with wadeite type structure. Proc. 4th Int.
Conf. High Press., 211– 214.
Kirby, S.H., Durham, W.B. and Stern, L.A. (1991) Mantle phase changes and deep-earthquake faulting in
subducting lithosphere. Science, 252, 216.
Kirby, S.H., Engdahl, E.R. and Denlinger, R. (1996a) In: Bebout, G.E., Scholl, D.W., Kirby, S.H. and Platt, J.P.
(eds.), Subduction Top to Bottom, Monograph 96. Washington, DC: AGU, pp. 195–214.
Kirby, S.H., Stein, S., Okal, E.A. and Rubie, D.C. (1996b) Metastable mantle phase transformations and deep
earthquakes in subducting oceanic lithosphere. Rev. Geophys., 34, 261–306.
Kirk, R.L. and Stevenson, D.J. (1987) Astrophys. J., 316, 836.
Kirkpatrick, R.J., Howell, D., Phillips, B.L., Cong, X.D., Ito, E. and Navrotsky, A. (1991) MAS NMR
spectroscopic study of Mg29SiO3 with the perovskite structure. Am. Mineral., 76, 673–676.
Kirkwood, J.G. (1950) J. Chem. Phys., 18, 380 –382.
Kirtley, J.R.J. et al. (1998) Phys. Condens. Matter., 10, L97–L103.
Kita, N.T. et al. (2000) Geochim. Cosmochim. Acta, 64, 3913.
Kitamura, M., Kondoh, S., Morimoto, N., Miller, G.H., Rossman, G.R. and Putnis, A. (1987) Planar OH-bearing
defects in mantle olivine. Nature, 328, 143–144.
Kitmura, H., Tsuneyuki, S. and Miyake, T. (2000) Nature, 404, 259–262.
Kittel, C. (1996) Introduction to Solid State Physics, 7th ed. New York: Wiley.
Klepeis, J.K., Schafer, K.J., Barbee, T.W. and Ross, M. (1991) Science, 254, 986– 988.
Klug, D.D., Handa, Y.P., Tse, J.S. and Whalley, E. (1989) J. Chem. Phys., 90, 2390.
Knedall, J.M. (1994) Geophys. Res. Lett., 21, 301–304.
Knittle, E. (1995) Static compression measurements of equations of state. In: Ahrens, T.J. (ed.), Mineral Physics
and Crystallography: A Handbook of Physical Constants. Washington, DC: AGU, pp. 98 –142.
1150 References

Knittle, E. and Jeanloz, R. (1986a) Geophys. Res. Lett., 13, 1541.

Knittle, E. and Jeanloz, R. (1986b) High pressure electrical resistivity measurements of a-Fe2O3: comparison of
static compression and shockwave experiments to 61 GPa. Solid State Commun., 58(2), 129–131.
Knittle, E. and Jeanloz, R. (1987a) Science, 235, 668.
Knittle, E. and Jeanloz, R. (1987b) The activation energy of the back transformation of silicate perovskite to
enstatite. In: Manghnani, M.H. and Syono, Y. (eds.), High-Pressure Research in Mineral Physics.
Washington, DC: AGU, pp. 243–250.
Knittle, E. and Jeanloz, R. (1989) Geophys. Res. Lett., 16, 421–424; 609– 612.
Knittle, E. and Jeanloz, R. (1991a) Earth’s core–mantle boundary: results of experiments at high pressure and
temperatures. Science, 251, 1438–1443.
Knittle, E. and Jeanloz, R. (1991b) The high-pressure phase diagram of Fe0.94O: a possible constituent of the
Earth’s core. J. Geophys. Res., 96, 16169–16180.
Knittle, E., Jeanloz, R. and Smith, G.L. (1985) The thermal expansion of silicate perovskite and stratification of
the Earth’s mantle. Nature, 319, 214.
Knittle, E., Jeanloz, R. and Smith, G.L. (1986) Thermal expansion of silicate perovskite and stratification of the
Earth’s mantle. Nature, 319, 214–216.
Knittle, E., Hathorne, A., Davis, M. and Williams, Q. (1992) A spectroscopic study of the high-pressure behaviour
of the O4H4 substitution in garnet. In: Syono, Y. and Manghnani, M.H. (eds.), High Pressure Research:
Application to Earth and Planetary Sciences. Tokyo/Washington, DC: Terra Scientific/AGU, pp.
297–304.
Knoche, R., Webb, S.L. and Rubie, D.C. (1997) Experimental determinations of acoustic wave velocities at Earth
mantle conditions using a multianvil press. Phys. Chem. Earth, 22, 125–130.
Knuth, E.L., Schünemann, F. and Toennies, J.P. (1995) J. Chem. Phys., 102, 6258.
Koebert, C. (1995) Diamonds everywhere. Nature, 378, 17–18.
Koebert, C. et al. (1995) Diamonds from the Popigai impact structure, Russia. Lunar Planet. Conf., XXVI,
777–778.
Kohlsted, D.L., Keppler, H. and Rubie, D.C. (1996) Solubility of water in the a-, b- and g-phases of
(Mg, Fe)2SiO4. Contrib. Mineral. Petrol., 123, 345 –357.
Kohn, W. and Sham, L.J. (1965) Self-consistent equations, including exchange and correlation effects. Phys. Rev.
A, 140, 1133–1140.
Kokubo, E. and Ida, S. (2000) Icarus, 143, 15.
Kondo, T., Yagi, T., Syono, Y. et al. (1998) High Press. Sci. Technol., 7, 148.
Konodo, K., Mashimo, T. and Sawaoko, A. (1980) J. Geophys. Res., 85, 977.
Koyabashi, Y. (1974) J. Phys. Earth, 22, 359.
Koyabashi, Y.J. et al. (1997) J. Phys. Condens. Matter, 9, 515.
Kraft, S., Knittle, E. and Williams, Q. (1991) Carbonate-silicate reactions at high pressures and possible presence
of dolomite and magnesite in the upper mantle. Earth Planet. Sci. Lett., 28, 116– 120.
Kramers, H.A. (1934) Physica, 1, 182.
Kratschmer, W., Lamb, L.D., Fostiropoulos, K. and Huffman, D.R. (1990) Nature, 347, 354–356.
Krog, E.J. and Carswell, D.A. (1995) In: Coleman, R.G. and Wang, X. (eds.), Ultrahigh Pressure Metamorphism.
Cambridge: Cambridge University Press, pp. 244 –298.
Krogh, J. (1988) Temperature estimates from the garnet–clinopyroxene thermometer. Contrib. Mineral. Petrol.,
99, 44.
Kroll, H. and Ribbe, P.H. (1980) Determinative diagrams for Al, Si order in plagioclases. Am. Mineral., 65,
449–457.
Kroll, H., Lueder, T., Schlenz, H., Kirfel, A. and Vad, T. (1997) The Fe2þ, Mg distribution in orthopyroxene: a
critical assessment of its potential as a gesopeedometer. Eur. J. Mineral, 9(4), 705–733.
Kronenberg, A.K. and Kirby, S.H. (1991) Probing the Earth’s strength: can we measure small stress at high
pressure? EOS Trans. AGU, 72, 453–454.
Kröner, A. (1985) Evolution of the Archean continental crust. Annu. Rev. Earth Planet. Sci., 13, 49–74.
Kroto, H.W., Heath, J.R., O’Brien, S.C., Curl, R.F. and Smalley, R.E. (1985) Nature, 318, 162–163.
Kruger, M.B. and Jeanloz, R. (1990) Science, 249, 647.
References 1151

Kruger, M.B., Williams, Q. and Jeanloz, R. (1989) Vibrational spectra of Mg(OH)2 and Ca(OH)2 under pressure.
J. Chem. Phys., 91, 5910– 5915.
Kruger, M.B., Jeanloz, R., Pasternak, M.P., Taylor, R.D., Snyder, B.S., Stacy, A.M. and Bohlen, S.R. (1992)
Science, 255, 703.
Kubicki, J.D. and Lasaga, A.C. (1991) Molecular dynamics simulations of pressure and temperature effects on
MgSiO3 and MgSiO4 melts and glasses. Phys. Chem. Miner., 17, 661–673.
Kubicki, D., Xiao, Y. and Lasaga, A.C. (1993) Geochim. Cosmochim. Acta, 57, 3847.
Kubo, K. and Ohata, N. (1972) A quantum theory of double exchange. J. Phys. Soc. Jpn, 33, 21–32.
Kubo, A., Suzuki, T. and Akaogi, M. (1997) High pressure phase equilibria in the system CaTiO3 –CaSiO3:
stability of perovskite solid solutions. Phys. Chem. Miner., 24, 488–491.
Kubo, T., Ohtani, E., Kato, T., Shinmel, T. and Fujino, K. (1998) Effects of water on the a–b transformation
kinetics in San Carlos olivine. Science, 281, 85.
Kudoh, Y. and Inoue, T. (1999) Mg-vacant structural modules and dilution of the symmetry of hydrous
wadsleyite, b-Mg22xSiH2xO4 with 0.00 # x # 0.25. Phys. Chem. Miner., 26, 382–388.
Kudoh, Y. and Kanazaki, M. (1998) Crystal chemical characteristics of a-CaSi2O5, a new high pressure calcium
silicate with five-coordinated silicon synthesized at 15008C and 10 GPa. Phys. Chem. Miner., 25,
429–433.
Kudoh, Y. and Takeda, H. (1986) Single crystal X-ray diffraction study on the bond compressibility of fayalite
Fe2SiO4, and rutile TiO2 under high pressure. Physica, 139–140, 333–336.
Kudoh, T. and Takeuchi, Y. (1985) The crystal structure of forsterite Mg2SiO4 under high pressure up to 149 kb.
Z. Kristallogr., 171, 291–302.
Kudoh, Y., Prewitt, C.T., Finger, L.W., Darovskikh, A. and Ito, E. (1992) Effect of iron on the crystal structure of
(Mg, Fe)SiO3 perovskite. Geophys. Res. Lett., 17, 1481– 1484.
Kudoh, Y., Finger, L.W., Hazen, R.M., Prewitt, C.T., Kanzaki, M. and Vebles, D.R. (1993) Phase E: a high
pressure hydrous silicate with unique crystal chemistry. Phys. Chem. Miner., 19, 357–360.
Kudoh, Y., Inoue, T. and Arashi, H. (1996) Structure and crystal chemistry of hydrous wadsleyite,
Mg1.75Si0.5O4: possible hydrous magnesium silicate in the mantle transition zone. Phys. Chem. Miner.,
23, 461 –469.
Kudoh, Y., Nagase, T., Sasaki, S., Tanaka, M. and Kanzaki, M. (1997) Phase F, a new hydrous magnesium silicate
synthesised at 10008C and 1700 GPa: crystal structure and estimated bulk modulus. Phys. Chem. Miner.,
24, 601.
Kumazawa, M. and Anderson, O.L. (1969) Elastic moduli, pressure derivatives, and temperature derivatives of
single-crystal olivine and single-crystal forsterite. J. Geophys. Res., 74, 5961–5972.
Kung, W. and Bloxham, J. (1997) An Earth-like numerical dynamo model. Nature, 389, 371.
Kung, J. and Rigden, S.M. (1997) EOS Trans. Am. Geophys. Union, Fall Meeting Suppl., 78, F803.
Kung, T.E., Grenn, H.W.H., Hofmeister, A.M. and Walker, D. (1993) Infrared spectroscopic investigation of
hydroxyl in b-(Mg, Fe)2SiO4 and coexisting olivine: implications for mantle evolution and dynamics.
Phys. Chem. Miner., 19, 409 –422.
Kunz, J. (1999) Is there solar argon in the earth’s mantle? Nature, 399, 649–650.
Kunzmann, T. (1999) The aenigmatite–rhönite mineral group. Eur. J. Mineral., 11, 743–756.
Kurimoto et al. (1986) Study of a-Fe2O3 under ultra-high pressure, Mineralogical Society of America.
Kuroda, K. and Irifune, T. (1998) Observation of phase transformations in serpentine at high pressure and high
temperature by in situ X-ray diffraction measurements. In: Manghnani, M.H. and Syono, Y. (eds.),
Properties of Earth and Planetary Materials at High Pressure and Temperature. Washington, DC: AGU,
pp. 545 –554.
Kuroda, K. et al. (2000) Phys. Chem. Miner., 27, 523–532.
Kurosawa, M., Yurimoto, H. and Sueno, S. (1997) Patterns in the hydrogen and trace element compositions of
mantle olivine. Phys. Chem. Miner., 24, 385– 395.
Kusaba, K., Syono, Y., Kikegawa, T. and Shimomura, O. (1997) J. Phys. Chem. Solids, 58, 241.
Kushiro, I. (1976) J. Geophys. Res., 81, 6347–6350.
Kushiro, I. (1990) Partial melting of mantle wedge and evolution of island arc crust. J. Geophys. Res., 95,
15929–15939.
1152 References

Kuskov, O.L. and Panferov, A.B. (1991) Phase diagrams of the FeO– MgO–SiO2 system and the structure of the
mantle discontinuities. Phys. Chem. Miner., 17, 642 –653.
Kuster, R.M., Singleton, J., Keen, D.A., McGreevy, R. and Hayes, W. (1989) Physica, 155B, 362.
Kusznir, N.J. and Bott, M.H.P. (1977) Stress concentration in the upper lithosphere caused by underlying visco-
elastic creep. Tectonophysics, 43, 247–256.
Kvick, A., Pluth, J.J., Richardson, J.W. Jr. and Smith, J.V. (1988) The ferric ion distribution and hydrogen
bonding in epidote: a neutron diffraction study at 15 K. Acta Crystallogr., 44, 351–355.
Labrosse, S., Poirier, J.P. and Le Mouel, L. (1997) Phys. Earth Planet. Interiors, 99, 1.
Ladbury, R. (1994) Phys. Today, 47, 17.
Laeter, De., J.R. (1999) Cosmochemical applications using mass spectrometry. In: Encyclopaedia of Physics.
Academic Press, 359 –367.
Lager, G.A., Armbruste, Th., Rotella, F.J. and Rossman, G.R. (1989) The OH substitution in garnet: X-ray and
neutron diffraction, infrared and geometric-modeling studies. Am. Mineral., 74, 840–851.
Lameyre, J. and Bowden, P. (1982) Plutonic rock type series: discrimination of various granitoid series and related
rocks. J. Volcanol. Geotherm. Res., 14, 169–186.
Langer, K. (1990) High pressure spectroscopy. In: Mottana, A. and Burragato, F. (eds.), Absorption Spectroscopy
in Mineralogy. Amsterdam: Elsevier, pp. 227–284.
Langer, K. and Fentrup, K.R. (1979) Automated microscope-absorption-spectrophotometry of rock-forming
minerals in the range 40,000–50,000 cm21 (250–2000 mm). J. Microsc., 116, 311–320.
Langer, K., Taran, M.N. and Platonov, A.N. (1997) Compression moduli of Cr3þ-centered octahedra in a variety
of oxygen-based rock-forming minerals. Phys. Chem. Miner., 24, 109–114.
Langereis, C.G. (1999) Excursions in geomagnetism. Nature, 399, 207–209.
Langereis, C.G. et al. (1997) Geophys. J. Int., 129, 75.
Larson, A. and Von Dreele, R.B. (1986) GSAS — General Structure Analysis System. LA-UR 86-748,
Los Alamos National Laboratory.
Larson, R.L. (1993) Mantle plumes and magnetic reversals: a surprising link. EOS, 74(4), 46.
Larson, R.L. and Olson, P. (1991) Mantle plumes control magnetic reversal frequency. Earth Planet. Sci. Lett.,
107, 437 –447.
Lasaga, A.C. and Gibbs, G.V. (1990) Ab-initio quantum mechanical calculations for water–rock interactions:
adsorption and hydrolysis reactions. Am. J. Sci., 290, 263–295.
Laudon, T.C., Mathur, N.D. and Midgley, P.A. (2002) Charge ordered ferromagnetic phase in La0.5Ca0.5MnO3.
Nature, 420, 797–799.
Lay, T. (1995) Seismology of the lower mantle and core–mantle boundary. Rev. Geophys. Suppl., 325–328.
Lay, T. and Young, C.J. (1990) The stably-stratified outermost core revisited. Geophys. Res. Lett., 17,
1001– 1004.
Lay, T. and Young, C.J. (1991) Analysis of seismic SV waves in the core’s penumbra. Geophys. Res. Lett., 18,
1373– 1376.
Lay, T., Garnero, E.J., Young, C.J. and Gaherty, J.B. (1997) Scale-lengths of shear velocity heterogeneity at the
base of the mantle from S wave differential travel times. J. Geophys. Res., 102, 9887–9910.
Lay, T., Williams, Q. and Garnero, E.J. (1998) The core– mantle boundary layer and deep Earth dynamic. Nature,
392, 461.
Le Cl’eac’h, A. and Gillet, P. (1990) IR and Raman spectroscopy study of natural lawsonite. Eur. J. Mineral., 2,
43–53.
Lecuyer, C., Gillet, P. and Robert, F. (1998) The hydrogen isotope composition of seawater and the global water
cycle. Chem. Geol., 145, 249 –261.
Lee, D.C. and Halliday, A.N. (1995) Hafnium–tungsten chronometry and the timing of terrestrial core formation.
Nature, 378, 771–774.
Lee, C.Y., Yin, Q., Rudnik, R.L. and London, F. (1938) Nature, 141, 643.
Lee, S.H., Conradi, M.S. and Norberg, R.E. (1992) Rev. Sci. Instrum., 63, 3674–3676.
Lee, C., Vanderbilt, D., Laasonen, K., Carr, R. and Parrinello, M. (1993) Ab initio studies on the structural and
dynamical properties of ice. Phys. Rev. B, 47, 4863–4872.
Leger, J.M. (1996) Nature, 383, 401.
References 1153

Leger, J.M., Redon, A.M. and Chateau, C. (1990) Compression of synthetic pyrope, spessartine and uvarovite
garnets up to 25 GPa. Phys. Chem. Miner., 17, 161– 167.
Leger, J.M., Haines, J., Schmidt, M.W., Petitet, J.P., Pereira, A.S. and da Jornada, J.A.H. (1996) Discovery of
hardest known oxide. Nature, 383, 401.
Lehman, J. (1983) Diffusion between olivine and spinel: application to geothermometry. Earth Planet. Sci. Lett.,
64, 123 –138.
Lehmann, G. (1970) Ligand field and charge transfer spectra of Fe(III)–O complexes. Z. Phys. Chem. (Neue
Folge), 72, 279–297.
Lehmann, M.S. and Larsen, M.K. (1974) A method for location of peaks in step-scan-measured Bragg reflections.
Acta Crystallogr. A, 30, 580–584.
Leider, H.R. and Pipkorn, D.N. (1968) Mössbauer effect in MgO:Fe3þ; low temperature quadrupole splittings.
Phys. Rev., 165, 494–500.
Leidner, B.J., Weidner, D.J. and Lieberman, R.G. (1980) Elasticity of single crystal pyrope and implications for
garnet solid solution series. Phys. Earth Planet. Interiors, 22, 111–121.
Leinenweber, K., Wang, Y., Yagi, T. and Yusa, H. (1994) An unquenchable perovskite phase of MgGeO3 and
comparison with MgSiO3 perovskite. Am. Mineral., 79, 197–199.
Leinenweber, K., Linton, J., Navrotsky, A., Fei, Y. and Parise, J.B. (1995) High pressure perovskites on the join
CaTiO3 –FeTiO3. Phys. Chem. Miner., 22, 251– 258.
Leitner, B.J., Weidner, D.J. and Liebermann, R.C. (1980) Elasticity of single crystal pyrope and implications for
garnet solid solution series. Phys. Earth Planet Inter., 22, 111–121.
Leninweber, K., Utsumi, W., Tschids, Y., Yagi, T. and Kurita, K. (1991) Unquenchable high-pressure polymorphs
of MnSnO3 and FeTiO3. Phys. Chem. Miner., 18, 244– 250.
Leonhardt, U. and Piwnicki, P. (2000) Phys. Rev. Lett., 84, 822–825.
Leroux, H., Doukhan, J.C. and Langenhorst, F. (1994) Microstructural defects in experimentally shocked
diopside: a TEM characterization. Phys. Chem. Miner., 20, 521–530.
LeToullec, R., Loubeyre, P., Pinceaux, J.P., Mao, H.K. and Hu, J. (1992) A system for doing low temperature –
high pressure single crystal X-ray diffraction with a synchrotron source. High Press. Res., 6,
379–388.
Levein, L. and Prewitt, C.T. (1981a) High-pressure structural study of diopside. Am. Mineral., 66, 315–323.
Lever, A.B.P. (1984) Inorganic Electronic Spectroscopy, 2nd ed. Amsterdam: Elsevier.
Levien, L. and Prewitt, C.T. (1981b) High pressure crystal structure and compressibility of coesite. Am. Mineral.,
66, 324 –333.
Levien, L., Prewitt, C.T. and Weidner, D.J. (1979) Compression of pyrope. Am. Mineral., 64, 805–808.
Lewis, J. (1998) A short cut to the nucleus. Nature, 393, 304.
Lewis, J.S. and Prinn, R.G. (1984) Planets and their Atmosphere: Origin and Evolution. Orlando, FL: Academic
Press.
Lewis, S. and Sherman, W.F. (1979) Pressure-scanned Fermi resonance in the vibrational spectra of the sulphate
ion. Spectrochim. Acta, 35A, 613 –624.
Li, J. and Agee, C. (1996) Nature, 381, 686– 689.
Li, X. and Jeanloz, R. (1987) Electrical conductivity of (Mg, Fe)SiO3 perovskite and a perovskite-dominated
assemblage at lower mantle conditions. Geophys. Res. Lett., 14, 1075–1078.
Li, X. and Jeanloz, R. (1990) Laboratory studies of the electrical conductivity of silicate perovskites at high
pressures and temperatures. J. Geophys. Res., 95, 5067–5078.
Li, X. and Jeanloz, R. (1991) Phases and electrical conductivity of a hydrous silicate assemblage at lower-mantle
conditions. Nature, 350, 332 –334.
Li, B. and Liebermann, R.C. (2000) Sound velocities of wadsleyite b-(Mg0.88Fe0.12)2SiO4 to 10 GPa. Am.
Mineral., 85, 292 –295.
Li, X. and Mao, H.K. (1994) Solid carbon at high pressure: electrical resistivity and phase transition. Phys. Chem.
Miner., 21(1– 2), 1–5.
Li, S., Xiao, Y., Liou, D., Chen, Y., Ge, N., Zhang, Z., Sun, S., Cong, B., Zhang, R., Hart, S.R. and Wang, S.
(1993) Collision of the north China and Yangtze blocks and formation of coesite-bearing eclogites:
timing and processes. Chem. Geol., 109, 89–111.
1154 References

Li, S., Wang, S. et al. (1994) Excess Ar in phengite from eclogite: evidence from dating of eclogite minerals by
Sm–Nd, Rb –Sr and 40Ar/39Ar methods. Chem. Geol., 112, 343–350.
Li, B., Liebermann, R.C. and Gwanmesia, G.D. (1995a) EOS Trans. Am. Geophys. Union (Fall Meeting), 76,
F619.
Li, J.P., O’Neill, H.St.C. and Seifert, F. (1995b) Subsolidus phase relations in the system MgO–SiO2 –Cr –O in
equilibrium with metallic Cr, and their significance for the petrochemistry of chromium. J. Petrol., 36,
107–132.
Li, B., Gwanmesia, G.D. and Liebermann, R.C. (1996a) Sound velocities of olivine and beta polymorphs of
Mg2SiO4 at Earth’s transition zone pressures. Geophys. Res. Lett., 23, 2259–2262.
Li, P., Shun-ichiro, K. and Wang, Z. (1996b) High-temperature creep in fine-grained polycrystalline CaTiO3, and
analogue material of (Mg, Fe)SiO3 perovskite. Phys. Earth Planet. Interiors, 95, 19 –36.
Li, B., Jackson, I., Gasparik, T. and Liebermann, R.C. (1997a) Elastic wave velocity measurement in multi-anvil
apparatus to 10 GPa using ultrasonic interferometry. Phys. Earth Planet. Interiors, 98, 79 –91.
Li, W., Lu, R., Yang, H., Prewitt, C.T. and Fei, Y. (1997b) Hydrogen in synthetic coesite crystals. EOS Trans. Am.
Geophys. Union, 78, F736.
Li, B., Liebermann, R.C. and Weidner, D.J. (1998) Elastic moduli of wadsleyite (b-Mg2SiO4) to 7 gigapascals and
873 kelvin. Science, 281, 675– 676.
Libermann, R.C. and Schreiber, E. (1968) Elastic constants of polycrystalline hematite as a function of pressure to
3 kilobars. J. Geophys. Res., 73, 6585–6590.
Libowitzky, E. and Armbruster, T. (1995) Low-temperature phase transitions and the role of hydrogen bonds in
lawsonite. Am. Mineral., 80, 1277–1285.
Libowitzky, E. and Beran, A. (1995) OH defects in forsterite. Phys. Chem. Miner., 22, 387– 392.
Libowitzky, E. and Rossman, G.R. (1996) FTIR spectroscopy of laweronite between 82 and 325 K. Am. Mineral.,
81, 1080–1091.
Liebau, F. (1985) Structural Chemistry of Silicates. Berlin: Springer, 104 pp.
Liebermann, R.C. (1976) Elasticity of ilmenites. Phys. Earth Planet. Interiors, 12, 5–10.
Lin, C.C. and Liu, L.G. (1997) High pressure phase transformations in aragonite-type carbonates. Phys. Chem.
Miner., 24, 149–157.
Lindsley, D.H. (1965) Ferrosilite. Carnegie Institution Washington Year Book, 65, 148– 149.
Lindsley, D.H. (1967) Pressure–temperature relations in the system FeO– SiO2. Carnegie Inst. Washington
Yearb., 65, 226– 230.
Lines, M.E. and Glass, A.M. (1977) Principles and Applications of Ferroelectrics and Related Materials. Oxford:
Clarendon Press.
Linton, J.A., Fei, Y. and Navrotsky, A. (1997) Complete Fe–Mg solid solution in lithium niobate and perovskite
structures in titanates at high pressures and temperatures. Am. Mineral., 82, 639– 642.
Linton, J.A., Fei, Y. and Navrotsky, A. (1999) The MgTiO3 –FeTiO3 join at high pressure and temperature. Am.
Mineral., 84, 1595–1603.
Liou, J.G. and Zhang, R.Y. (1996) Petrogenesis of ultrahigh-P garnet bearing ultramafic body from Maowu, the
Dabie Mountains, Central China. Island Arc, 7, 115 –134.
Liou, J.G. and Zhang, R.Y. (1998) Island Arc, 7, 115.
Liou, J.G., Zhang, R.Y., Wang, X., Eide, E.A., Ernst, W.G. and Maruyama, S. (1996) Metamorphism and
tectonics of high-pressure belts in the Dabie-Sulu region, China. In: Yin, A. and Harrison, T.M. (eds.),
The Tectonic Evolution of Asia. Cambridge: Cambridge University Press, 666 pp.
Liou, J.G., Zhang, R.Y. and Ernst, W.G. (1998) High-pressure minerals from deeply subducted metamorphic
rocks. In: Rumble, D. III and Maruyama, S. (eds.), Reviews in Mineralogy, Ultrahigh Pressure
Mineralogy, Vol. 37. Washington, DC: Mineralogical Society of America, pp. 34–88.
Liou, J.G., Hacker, B.R. and Zhang, R.Y. (2000) Into the forbidden zone. Science, 287, 1215–1216.
Lipp, M.J. and Daniels, W.B. (1991) Electronic structure measurements in KI high pressure using three-photon
spectroscopy. Phys. Rev. Lett., 67, 2810–2813.
Lipp, M.J. and Daniels, W.B. (1993) Electronic band gap, excitonic binding energy, and electron-hole exchange
energy of KI under high pressure. Phys. Rev., 47, 6931–6936.
Lipp, M.J., Blechschmidt, J. and Daniels, W.B. (1993) Phys. Rev. B, 48, 687– 694.
References 1155

Littlewood, P. (1999) Phases of resistance. Nature, 399, 529 –531.

Liu, C., Dutton, Z., Behroozi, C.H. and Hau, L.V. (2001) Nature, 409, 490–493.
Liu, L. (1974) Silicate perovskite from phase transformations of pyrope–garnet at high pressure and temperature.
Geophys. Res. Lett., 1, 277–280.
Liu, L. (1975) Post-oxide phases of olivine and pyroxene and mineralogy of the mantle. Nature, 258,
510–512.
Liu, L. (1976a) Orthorhombic perovskite phase observed in olivine, pyroxene and garnet at high pressures and
temperatures. Phys. Earth Planet. Interiors, 11, 289– 298.
Liu, L. (1976b) The high pressure phases of FeSiO3 with implications for Fe2SiO4 and FeO. Earth Planet. Sci.
Lett., 33, 101–106.
Liu (1977a) The post-spinel phases of 12 silicates and germanates. In: Maghnani, M.H. and Akimoto, S. (eds.),
High Pressure Research, Applications in Geophysics. New York: Academic Press, pp. 245– 253.
Liu, H.S. (1977b) Convection pattern and stress system under the African plate. Phys. Earth Planet. Interiors,
15, 60 –68.
Liu, L. (1977c) Earth Planet. Sci. Lett., 35, 161 –168.
Liu, L. (1977d) The system enstatite-pyrope at high pressures and temperatures and the mineralogy of the Earth’s
mantle. Earth Planet. Sci. Lett., 36, 237–245.
Liu, L.G. (1977e) High pressure NaAlSiO4: the first silicate Ca-ferrite isotype. Geophys. Res. Lett., 4, 183–186.
Liu, L. (1978a) Earth Planet. Sci. Lett., 37, 438 –444.
Liu, L. (1978b) Phys. Chem. Miner., 3, 291–299.
Liu, L. (1978c) A new high pressure phase of spinel. Earth Planet. Sci. Lett., 41, 398–404.
Liu, H.S. (1978d) Mantle convection pattern and subcrustal stress field under Asia. Phys. Earth Planet. Interiors,
16, 247 –256.
Liu, L. (1979) On the 650 km seismic discontinuity. Earth Planet. Sci. Lett., 42, 202–208.
Liu, L. (1980) The equilibrium boundary of spinel–corundum þ periclase: a calibration curve for pressure above
100 kbar. High Temp. High Press., 12, 217 –220.
Liu, L. (1982) Compression of ice VIII to 500 kbar. Earth Planet. Sci. Lett., 61, 359–364.
Liu, L. (1987a) Effects of H2O on the phase behaviour of the forsterite–enstatite system at high pressures and
temperatures and implications for the Earth. Phys. Earth Planet. Interiors, 49, 142–167.
Liu, L. (1987b) New silicate perovskites. Geophys. Res. Lett., 14, 1079–1082.
Liu, L. (1989) Silicate perovskites: a review. Surv. Geophys., 10, 63 –81.
Liu, L. (1993) Effects of H2O on the phase behavior of the forsterite–enstatite system at high pressures and
temperatures: revisited. Phys. Earth Planet. Interiors, 76, 209–218.
Liu, L. (1994a) Silicate perovskite from phase transformations of pyrope garnet at high pressure and temperature.
Geophys. Res. Lett., 1, 277–280.
Liu, M. (1994b) Asymmetric phase effects and mantle convection patterns. Science, 264.
Liu, L. and Bassett, W.A. (1986) Elements, Oxides and Silicates: High-Pressure Phases with Implications for the
Earth’s Interior. New York: Oxford University Press.
Liu, X. and Lieberman, R.C. (1993) X-ray powder diffraction study of CaTiO3 perovskite at high temperature.
Phys. Chem. Miner., 20, 171 –175.
Liu, L. and Lin, C.C. (1995) High pressure phase transformation of carbonates in the system CaO–MgO– SiO2 –
CO2. Earth Planet. Sci. Lett., 134(3–4), 297 –305.
Liu, L. and Lin, C.C. (1997) A calcite ! aragonite-type phase transition in CdCO3. Am. Mineral., 82, 643–646.
Liu, M. and Liu, L. (1987) Bulk moduli of wüstite and periclase: a comparative study. Phys. Earth Planet.
Interiors, 45, 273 –279.
Liu, L. and Mernagh, T.P. (1993) Raman spectra of forsterite and fayalite at high pressures and room temperature.
High Press. Res., 11, 241 –256.
Liu, L. and Ringwood, A.E. (1975) Synthesis of perovskite type polymorph CaSiO3. Earth Planet. Sci. Lett.,
28, 209 –211.
Liu, J. and Vohra, Y.K. (1994) Appl. Phys. Lett., 64, 33.
Liu, J. and Vohra, Y.K. (1996) Photoluminescence and X-ray diffraction studies on Sm-doped yttrium aluminium
garnet to ultrahigh pressure of 338 GPa. J. Appl. Phys., 79, 7978–7982.
1156 References

Liu, F., Garofalini, S.H., Smith, D.K. and Vanderbilt, D. (1994a) Phys. Rev. B, 49, 12558.
Liu, L., Mernagh, T.P. and Irifune, T. (1994b) Raman spectra of pyrope and MgSiO3 –10Al2O3 garnet at various
pressures and temperatures. High Temp. High Press., 26, 363– 374.
Liu, J. et al. (1995a) Appl. Phys. Lett., 66, 3218–3220.
Liu, L., Mernagh, T.P. and Irifune, T. (1995b) Raman spectra of MgSiO3 10% Al2O3 – perovskite at various
pressures and temperatures. Phys. Chem. Miner., 22, 511–516.
Liu, B., Kern, J. and Popp, T. (1997a) Attenuation and velocities of P- and S-waves in dry and saturated crystalline
and sedimentary rocks at ultrasonic frequencies. Phys. Chem. Earth, 22(1 –2), 75–79.
Liu, L., Lin, C.C., Mernagh, T.P. and Irifune, T. (1997b) Raman spectra of phase B at various pressures and
temperatures. J. Phys. Chem. Solids, 59, 2023–2030.
Liu, L., Mernagh, T.P., Lin, C.C. and Irifune, T. (1997c) Raman spectra of phase E at various pressures and
temperatures with geophysical implications. Earth Planet. Sci. Lett., 149, 57 –65.
Liu, L., Mernagh, T.P. and Hibberson, W.O. (1997d) Raman spectra of high pressure polymorphs of SiO2 at
various temperatures. Phys. Chem. Miner., 24, 396 –402.
Liu, C., Dutton, Z., Behroozi, C.H. and Hau, L.V. (2001) Nature, 409, 490–493.
Lobban, C., Finey, J.L. and Kuhs, W.F. (1998) Nature, 391, 268.
London, F. (1938) Nature, 141, 643.
Long, D.A. (1977) Raman Spectroscopy. New York: McGraw-Hill.
Longhi, J., Knittle, E., Holloway, J.R. and Wäanke, H. (1992) In: Kieffer, H.H., Jakosky, B.M., Snyder, C.W. and
Matthews, M.S. (eds.), Mars. Tucson, AZ: University of Arizona Press, pp. 184–208.
Loper, D.E. and Lay, T. (1995) The core –mantle boundary region. J. Geophys. Res., 100, 6397–6420.
Lorand, J.P. (1989) The Cu– Fe–Ni sulfide component of the amphibole-rich veins from Lherz and Freychinede
spinel peridotite massifs (NE Pyrenees, Fra.), comprising mantle-derived megacrysts from alkali basalts.
Lithos, 23.
Lorenzana, H.E., Silvera, I.F., Mazin, I.I., Hemley, R.J. and Mao, H.K. (1996) Phys. Rev., 54, 15590.
Loubeyre, P. (1991) In: Pucci, R. and Puccitto, G. (eds.), Molecular System Under Pressure. Amsterdam:
Elsevier/North Holland, 245 pp.
Loubeyre, P. and LeToullec, R. (1995) Nature, 378, 44–46.
Loubeyre, P., Jean-Louis, M., LeToullec, R. and Charon-Gerard, L. (1993) Phys. Rev. Lett., 70, 178.
Loubeyre, P. et al. (1996) X-ray diffraction and equation of state of hydrogen at megabar pressures. Nature, 383,
702–704.
Loveday, J.S. et al. (1996) Phys. Rev. Lett., 76, 74.
Lpers, C.N., Nordstrom, D.K. and Ball, J.W. (1989) Solubility of jarosite solid solutions precipitated from acid
mine waters, Iron Mountain, California, USA. Sci. Geol. Bull., 42, 281–298.
Lu, R. and Hofmeister, A.M. (1994) Infrared spectroscopy of CaGeO3 perovskite to 24 GPa and thermodynamic
implications. Phys. Chem. Miner., 21, 78– 84.
Lu, R., Hofmeister, A. and Wang, Y. (1994) Thermodynamic properties of ferromagnesium silicate perovskites
from vibrational spectroscopy. J. Geophys. Res., 99, 11795–11804.
Lu, R., Goncharov, A.F., Mao, H.K. and Hemley, R.J. (1999) Synchrotron infrared microspectroscopy:
applications to hydrous minerals. In: Schulze, D.G., Stucki, J.W. and Bertsch, P.M. (eds.),
Synchrotron Methods in Clay Science, Vol. 9. Boulder, CO: The Clay Minerals Society,
pp. 164 –182.
Lugmair, G.W. and Galer, S.J.G. (1992) Geochim. Cosmochim. Acta, 56, 1673.
Lugmair, G.W. and Maclsaac, Ch. (1995) Lunar Planet. Sci., XXVI, 879–880.
Lugmair, G.W. and Shukolyukov, A. (1998) Geochim. Cosmochim. Acta, 62, 2863–2886.
Lunine, J.I. (1993) Rev. Geophys., 31, 133.
Luo, H. and Ruoff, A.L. (1993) X-ray diffraction study of sulfur to 32 GPa: Amorphization at 25 GPa. Phys. Rev.
B, 48, 569–572.
Luo, H., Desegreniers, S., Vohra, Y.K. and Ruoff, A.L. (1991) High-pressure optical studies on sulfur to 121 GPa:
Optical evidence for metallization. Phys. Rev. Lett., 67, 2998–3001.
Luo, H., Greene, R.G. and Ruoff, A.L. (1993) B–Po phase of sulfur at 162 GPa: X-ray diffraction study to
212 GPa. Phys. Rev. Lett., 71, 2943–2946.
References 1157

Luth, R.W. (1992) Potassium in clinopyroxene at high pressure: experimental constraints (abs.). EOS,
73, 608.
Luth, R.W. (1995) Is phase A relevant to the Earth’s mantle? Geochim. Cosmochim. Acta, 59, 679–682.
Luth, R.W. (1997) Experimental study of the system phlogopite–diopside from 3.5 to 17 GPa. Am. Mineral., 82,
1198– 1209.
Luth, R.W., Virgo, D., Boyd, F.R. and Wood, B.J. (1990) Ferric iron in mantle-derived garnets, implications for
thermobarometry and for the oxidation state of the mantle. Contrib. Mineral. Petrol., 104, 56–72.
Lutz, H.D. and Jung, C. (1997) Water molecules and hydroxide ions in condensed materials; correlation of
spectroscopic and structural data. J. Mol. Struct., 404, 63–66.
Lutz, H.D., Jung, C., Twomel, M. and Losel, J. (1995) Brown’s bond valences, a measure of the strength of
hydrogen bonds. J. Mol. Struct., 351, 205–209.
Lyapin, A.G. and Brazhkin, V.V. (1996) Pressure-induced lattice instability and solid state amorphization. Phys.
Rev. B, 54, 12036–12048.
Lysne, P.C. and Percival, C.M. (1975) J. Appl. Phys., 46, 1519–1525.
Lyubutin, I.S. and Dodokin, A.P. (1971) Temperature dependence of the Mössbauer effect for Fe2þ in
dodecahedral coordination in garnet. Sov. Phys. Crystallogr., 15, 1091–1092.
Lyubutin, I.S., Dodokin, A.P. and Belyaev, L.M. (1970) Temperature dependence of the Mössbauer effect for
octahedral iron atoms in garnet. Sov. Phys. Solid State, 12, 1100–1101.
MacChesney, J.B. et al. (1965) J. Chem. Phys., 43, 1907.
MacGregor, I.D. (1974) The system MgO– Al2O3 –SiO2: solubility of Al2O3 in enstatite for spinel and garnet
peridotite compositions. Am. Mineral., 61, 603 –615.
Machetel, P. and Weber, P. (1991) Intermittent layered convection in a model mantle with an endothermic phase
change at 670 km. Nature, 350, 55–57.
Machone, J.F., Nieman, R.A. and Lewis, C.F. (1989) Science, 243, 1182–1184.
Mackay, A. (1998) Some are less equal than others. Nature, 391, 334–335.
Mackenzie, D. (1998) Polyhedra can bend but not breathe. Science, 279, 1637.
Mackwell, S.J. and Kohlstedt, D.L. (1990) Diffusion of hydrogen in olivine: implications for water in the mantle.
J. Geophys. Res., 95, 5079–5088.
Mackwell, S.J., Kohlstedt, D.L. and Paterson, M.S. (1985) The role of water in the deformation of olivine single
crystal. J. Geophys. Res., 90, 11319–11333.
Madon, M. and Poirier, J.P. (1983) Transmission electron microscopy observation of a, b and g (Mg,Fe)2SiO4
in shocked meteorites: planar defects and polymorphic transitions. Phys. Earth Planet. Interiors,
33, 31–44.
Madon, M. et al. (1989) Nature, 342, 422– 425.
Madon, M. (1992) Mantle. Encyclopaedia of Earth System Science, 3, 85 –98.
Mailhiot, C., Yang, L.H. and McMahan, A.K. (1992) Polymeric nitrogen. Phys. Rev. B, 46, 14419–14435.
Maines, J. (1998) Continental mechanics. Nature, 391, 634.
Malin, M.C. and Edgett, K.S. (2000) Evident for recent ground water seepage and surface runoff on Mars. Science,
288, 2330–2335.
Manga, M. and Jeanloz, R. (1996) Implications of a metal-bearing chemical boundary layer in D00 for mantle
dynamics. Geophys. Res. Lett., 23, 3091–3094.
Manghnani, M. (1969) Elastic constants of single-crystal rutile under pressures to 7.5 kilobars. J. Geophys. Res.,
74, 4317–4328.
Manghnani, M. and Ramananantoandro, R. (1974) J. Geophys. Res., 79(35), 609–628.
Manghnani, M.H., Syono, Y. (eds.) (1987) High-Pressure Research in Mineral Physics. Tokyo/Washington,
DC: Terra Scientific/AGU.
Manning, C.E. (1994) The solubility of quartz in H2O in the lower crust and upper mantle. Geochim. Cosmochim.
Acta, 58, 4831– 4840.
Manning, P.G. and Tricker, M.J. (1977) A Mössbauer spectral study of ferrous and ferric ion distributions in
grossular crystals: evidence for local crystal disorder. Can. Mineral., 15, 81 –86.
Manolopous, D.E. and Fowler, P.W. (1996) In: Andreoni, W. (ed.), The Chemical Physics of Fullerenes.
Dordrecht: Kluwer, pp. 51– 70.
1158 References

Mao, H.K. (1974) Charge-transfer processes at high pressure. In: Strens, R.G.J. (ed.), Phys. Chem. Minerals
Rocks. New York: Wiley, pp. 573– 581.
Mao, H.K. (1993) Electrical and optical properties of the olivine series at high pressure. Observations of optical
absorption and electrical conductivity in magnesiowüstite at high pressure. Thermal and optical propeties
of the Earth’s mantle. Washington Geophys. Lab. Yearb., 72, 52–64.
Mao, H.K. and Bell, P.M. (1972) Electrical conductivity and the red shift of absorption in olivine and spinel at
high pressure. Science, 176, 403–406.
Mao, H.K. and Bell, P.M. (1977) Disproportionation equilibrium in iron-bearing systems at pressures above
100 kbar with applications to chemistry of the earth’s mantle. In: Saxena, S.K. and Bhattacharjee, S.
(eds.), Energetics of Geological Processes. New York: Springer, pp. 236–249.
Mao, H.K. and Bell, R.M. (1975) Contribution of anionic complexes to charge-transfer and associated
optical, electrical, and thermal effects at high pressure. Annu. Rep. Geophys. Lab. Yearb., 74,
559–611.
Mao, A.K. and Bell, P.M. (1978) Science, 200, 1145.
Mao, H.K. and Bell, P.M. (1979) Science, 203, 1004.
Mao, H.K. and Hemley, R.J. (1989) Optical observations of hydrogen above 200 GPa: evidence for metallization
by band overlap. Science, 244, 1462–1465.
Mao, H.K. and Hemley, R.J. (1991) Optical transition in diamond at ultra high pressure. Nature, 351, 721–724.
Mao, H.K. and Hemley, R.J. (1994) Rev. Mod. Phys., 66, 671 –692.
Mao, H.K. and Hemley, R.J. (1996a) Energy dispersive X-ray diffraction of micro-crystals at ultrahigh pressures.
High Press. Res., 14, 257 –267.
Mao, H.K. and Hemley, R.J. (1996b) Experimental studies of Earth’s deep interior: accuracy and versatility of
diamond-anvil cell. Phil. Trans. R. Soc. Lond. A, 354, 1315–1332.
Mao, H.K. and Hemley, R.J. (1998) New windows on the earth’s deep interior. In: Hemley, R.J. (ed.), Ultrahigh-
Pressure Mineralogy: Physics and Chemistry of the Earth’s Deep Interior, Mineralogical Society of
America, pp. 1 –27.
Mao, H.K., Bassett, W.A. and Takahashi, T. (1967) Effect of pressure on crystal structure and lattice parameters of
iron up to 300 kbar. J. Appl. Phys., 38, 272–276.
Mao, H.K., Takahashi, T., Bassett, W.A., Kinsland, G.L. and Merrill, L. (1974) Isothermal compression of
magnetite to 320 kbar and pressure-induced phase transformation. J. Geophys. Res., 79, 1165– 1170.
Mao, H.K., Virgo, D. and Bell, P.M. (1977a) High-pressure 57Fe Mössbauer data on the phase and magnetic
transitions of magnesioferrite (MgFe2O4), magnetite (Fe3O4) and hematite (Fe2O3). Carnegie Inst.
Washington Yearb., 76, 522 –525.
Mao, H.K., Yagi, T. and Bell, P.M. (1977b) Mineralogy of the earth’s deep mantle: quenching experiments on
mineral compositions at high pressures and temperature. Carnegie Inst. Washington Yearb., 76,
502–504.
Mao, H.K., Bell, P.M., Shaner, J.W. and Steinberg, J. (1978) Specific volume measurement of Cu, Mo, Pb and Ag
and calibration of the ruby R1 fluorescence pressure range from 0.06 to 1 Mbar. J. Appl. Phys.,
49, 3276–3283.
Mao, H.K., Bell, P.M., Dunn, K.J., Chrenko, R.M. and Devies, R.C. (1979) J. Appl. Phys., 50,
1002– 1009.
Mao, H.K., Zou, G. and Bell, P.B. (1981) Carnegie Inst. Washington Yearb., 77, 803.
Mao, H.K., Bell, P.M. and Yagi, T. (1982) Iron– magnesium fractionation model for the Earth. In: Akimoto, S.
and Manghnani, M.H. (eds.), High-Pressure Research in Geophysics. Tokyo: Academic Publications,
pp. 319 –325.
Mao, H.K., Xu, J. and Bell, P.M. (1986) Calibration of the ruby pressure gauge to 800 kbar under quasihydrostatic
conditions. J. Geophys. Res., 91, 4673–4676.
Mao, H.K., Bell, P.M. and Hadidiacos, C. (1987) Experimental phase relations of iron to 360 kbar and 14008C,
determined in an internally heated diamond anvil apparatus. In: Manghnani, M.H. and Syono, Y. (eds.),
High Pressure Research in Mineral Physics. Tokyo: Terra Scientific, pp. 135–138.
Mao, H.K., Chen, L.C., Hemley, R.J., Jephcoat, A.P., Wu, Y. and Bassett, W.A. (1989a) Stability and equation of
state of CaSiO3 perovskite to 134 GPa. J. Geophys. Res., 94, 17889–17894.
References 1159

Mao, H.K., Hemley, R.J., Shu, J., Chen, L., Jephcoat, A.P. et al. (1989b) The effect of pressure, temperature, and
composition on lattice parameters and density of (Fe, Mg)SiO3 –perovskites to 30 GPa. Annu. Rep. Dir.
Geophys. Lab., 1988-1989, 82 –89.
Mao, H.K., Wu, Y., Hemley, R.J., Chen, L.C., Shu, J.F. and Finger, L.W. (1989c) X-ray diffraction to 302
gigapascals: high-pressure crystal structure of cesium iodide. Science, 246, 649.
Mao, H.K. et al. (1990a) Phys. Rev. Lett., 64, 1749.
Mao, H.K., Wu, Y., Chen, L.C., Shu, J.F. and Jephcoat, A.P. (1990b) Static compression of iron to 300 GPa and
Fe0.8Ni0.2 alloy to 260 GPa: implications for composition of the core. J. Geophys. Res., 96, 8069– 8079.
Mao, H.K., Wu, Y., Chen, L.C., Shu, J.F. and Jephcoat, A.P. (1990c) Static compression of iron to 300 GPa and
Fe0.8Ni0.2 alloy to 260 GPa: implications for composition of the core. J. Geophys. Res., 95,
21737–21742.
Mao, H.K., Hemley, R.J., Fei, Y., Shu, J.F., Chin, L.C., Jephcoat, A.P., Wu, Y. and Bassett, W.A. (1991) Effect of
pressure, temperature and composition on the lattice parameters and density of (Mg,Fe)SiO3 perovskite
to 30 GPa. J. Geophys. Res., 96, 8069–8079.
Mao, H.K., Shu, J.F., Fei, Y., Hu, H. and Hemley, R.J. (1996) The wüstite enigma. Phys. Earth Planet. Interiors,
96, 135 –145.
Mao, H.K., Hemley, R.J. and Mao, A.L. (1997) Diamond-cell research with synchrotron radiation. High Press.
Res. Condens. Matter, 12–19.
Mao, H.K., Shu, J., Shen, G., Hemley, R.J., Li, B., Singh, A.K. (1998) Elasticity and rheology of iron above
220 GPa and the nature of the Earth’s inner core. Nature, 396, 741–743.
Mao, H.K., Shu, J., Shen, G., Hemley, R.J., Li, B. and Singh, A.K. (1999) Elasticity and rheology of iron above
220 GPa and the nature of the Earth’s inner core. Nature, 399, 280.
Mao, W.L., Mao, H.K. et al. (2002) Hydrogen clusters in clathrate hydrate. Science, 297, 2247–2249.
Maradudin, A., Montroll, E.W., Weiss, G.H. and Ipatova, I.P. (1971) Theory of Lattice Dynamics in the Harmonic
Approximation. Orlando, FL: Academic Press.
Margulies, G., Winther, G. and Poulsen, H.F. (2000) Science, 291, 2392.
Markl, G. and Bucher, K. (1998) Composition of fluids in the lower crust inferred from metamorphic salt in lower
crustal rocks. Nature, 394, 781–783.
Martin, H. (1986) Effect of steeper archean geothermal gradient on geochemistry of subduction zone magmas.
Geology, 14, 753–756.
Martin, R.M. (1999) Simple metals under pressure. Nature, 400, 117–118.
Martin, R.F. and Donnay, G. (1972) Hydroxyl in the mantle. Am. Mineral., 57, 554–570.
Martin, H., Chauvel, C. and Jahn, B.M. (1983) Major and trace element geochemistry of Archean granodioritic
rocks from eastern Finland. Precambr. Res., 21, 159– 180.
Martinez, I., Zhang, J. and Reeder, R.J. (1996) In situ X-ray diffraction study of aragonite and dolomite at high
pressure and temperature: evidence for dolomite breakdown to aragonite and magnesite. Am. Mineral.,
81, 611 –624.
Martini, J.E.J. (1978) Nature, 272, 715.
Martins, J.L. and Cohen, M.L. (1988) Diagonalization of large matrices in pseudopotential band-structure
calculations: dual-space formalism. Phys. Rev., 37, 6134–6138.
Marton, F.C. and Cohen, R.E. (1994) Prediction of a high-pressure phase transition in Al2O3. Am. Mineral., 789.
Marzocchi, W. and Mulargia, E. (1992) The periodicity of geomagnetic reversals. Phys. Earth Planet. Interiors,
73, 222 –228.
Masaitis, V. (1992) Meteoritics, 27, 21–27.
Mashimo, T., Kondo, K.I., Sawaoka, A., Syono, Y., Takei, H. and Ahrens, T.J. (1976) J. Geophys. Res., 94, 17889.
Mashimo, T., Kondo, K.I., Sawaoka, A., Syono, Y., Takei, H. and Ahrens, T.J. (1980) Electrical conductivity
measurements of fayalite under shock compression up to 56 GPa. J. Geophys. Res., 85, 1876–1881.
Mason, T.E., Aeppli, G. and Mook, H.A. (1992) Magnetic dynamics of superconducting La1.86Sr0.14CuO4. Phys.
Rev. Lett., 68, 1414–1417.
Massonne, H.J. (1995) Experimental and petrogenetic study of UHPM. In: Coleman, R.G. and Wang, X. (eds.),
Ultrahigh Pressure Metamorphism. Cambridge: Cambridge University Press, pp. 33–95.
Masters, G. (1979) Geophys. J. R. Astron. Soc., 57, 507–534.
1160 References

Masters, G. and Gilbert, F. (1981) Structure of the inner core inferred from observations of its spheroidal shear
modes. Geophys. Res. Lett., 8, 569 –571.
Masters, T.G., Johnson, S., Leake, G. and Bolton, H. (1996) A shear–velocity model of the mantle. Phil. Trans. R.
Soc. Lond. A, 354, 1385–1411.
Mathoniere, C., Nuttal, C.J., Carling, S.G. and Day, P. (1996) Inorg. Chem., 35, 1201–1206.
Mathur, N. (1999) Magnetic phases to order. Nature, 400, 405– 406.
Mathur, N. et al. (1998) Nature, 394, 39–43.
Matsui, T. and Abe, Y. (1986) Nature, 322, 526–528.
Matsui, M. and Anderson, O.L. (1997) The case for a body-centered cubic phase (a0 ) for iron at inner core
conditions. Phys. Earth Planet. Interiors, 103, 55–62.
Matsui, Y. and Kawamura, K. (1987) In: Manghnani, M.H. and Syono, Y. (eds.), High-Pressure Research in
Mineral Physics. Tokyo/Washington, DC: Terra Scientific/AGU, 305 pp.
Matsui, M. and Price, G.D. (1991) Simulation of the pre-melting behavior of MgSiO3 perovskite at high pressures
and temperatures. Nature, 351, 735– 737.
Matsui, M. and Price, G.D. (1992) Computer simulation of the MgSiO3 polymorphs. Phys. Chem. Miner.,
18, 365 –372.
Matsui, Y., Onuma, N., Nagasawa, H., Higuchi, H. and Banno, S. (1977) Crystal structure control in trace element
partition between crystal and magma. Bull. Soc. Fr. Mineral. Cristallogr., 100, 315– 324.
Matsui, M., Akaogi, M. and Matsumoto, T. (1987) Computational model of the structural and elastic properties of
the ilmenite and perovskite phases of MgSiO3. Phys. Chem. Miner., 14, 101–106.
Matsui, M., Price, G.D. and Patel, A. (1994) Comparison between the lattice dynamics and molecular dynamics
methods: calculation results for MgSiO3 perovskite. Geophys. Res. Lett., 21, 1659–1662.
Matsui, M., Parker, S.C. and Leslie, M. (2000) The MD simulation of the equation of state of MgO:
application as a pressure calibration standard at high temperature and high pressure. Am. Mineral., 85,
312–316.
Mattson, S.M. and Rossman, G.R. (1987) Identifying characteristics of charge transfer transitions in minerals.
Phys. Chem. Miner., 14, 94 –99.
Mayburg, H. (1950) Effect of pressure on the low frequency dielectric constant of ionic crystals. Phys. Rev.,
79, 375 –382.
Maynard, J. (1996) Resonant ultrasound spectroscopy. Phys. Today, 26 –31.
Mazin, I.I. and Cohen, R.E. (1995) Phys. Rev., 52, 8797.
Mazin, I.I., Hemley, R.J., Goncharov, A.F., Hanfland, M. and Mao, H.K. (1997) Quantum and classical
orientational ordering in solid hydrogen. Phys. Rev. Lett., 78, 1066– 1069.
Mazin, I.I., Fei, I.Y., Downs, R. and Cohen, R.E. (1998) Possible polymorphism in FeO at high pressures.
Am. Mineral., 83, 451 –457.
McCammon, C.A. (1991) Static compression of a-MnS at 298 K to 21 GPa. Phys. Chem. Miner., 17, 636–641.
McCammon, C.A. (1992) J. Magn. Magn. Mater., 104– 107, 1937.
McCammon, C.A. (1993a) Composition limits of FexO and the Earth’s lower mantle. Science, 261, 923–925.
McCammon, C.A. (1993b) Effect of pressure on the composition of the lower mantle end member FexO. Science,
259, 66 –68.
McCammon, C.A. (1995) Mössbauer spectroscopy of minerals. In: Ahrens, T.J. (ed.), Mineral Physics and
Crystallography: A Handbook of Physical Constants, AGU Reference Shelf 2. Washington, DC: AGU,
pp. 332 –347.
McCammon, C.A. (1995b) Equation of state, bonding character, and phase transition of cubanite, CuFe2S3,
studied from 0– 5 GPa. Amer. Miner., 80, 1–8.
McCammon, C.A. (1996) Crystal chemistry of iron-containing perovskites. Phase Transit., 58, 1–26.
McCammon, C.A. (1997) Perovskite as a possible sink for ferric iron in the lower mantle. Nature, 387,
694–696.
McCammon, C.A. and Liu, L. (1984) Phys. Chem. Miner., 10, 106.
McCammon, C.A. and Tennant, C. (1996) High-pressure Mössbauer study of synthetic clinoferrosilite, FeSiO3.
In: Dyar, M.D., McCammon, C. and Schaefer, M.W. (eds.), Mineral Spectroscopy: A Tribute to Roger G.
Burns, The Geochemical Society, Special Publication 5, pp. 281–288.
References 1161

McCammon, C., Rubie, D.C., Ross, C.R. II, Seifert, F. and O’Neill, H.St.C. (1992) Mössbauer spectra of
57
Fe0.05Mg0.95SiO3 perovskite at 80 K and 298 K. Am. Mineral., 77, 894–897.
McCammon, C.A., Harris, J.W., Harte, B. and Hutchison, M.T. (1996) Partitioning of ferric iron between lower
mantle phases: results from natural and synthetic samples. J. Conf. Abstr., 1, 390.
McCammon, C.A., Hutchison, M.T., Harte, B. and Harris, J.W. (1997) Ferric iron and the electrical conductivity
of the lower mantle. EOS, American Geophysical Union.
McCauley, T.S. and Vohra, Y.K. (1995) Appl. Phys. Lett., 66, 1486–1488.
McCord, T.B., Hansen, G.B., Matson, D.L., Johnson, T.V., Crowley, J.K. et al. (1999) Hydrated salt minerals on
Europa’s surface from the Galileo near-infrared mapping spectrometer (NIMS) investigation. J. Geophys.
Res., 104, 11827–11851.
McCormick, T.C., Hazen, R.M. and Angel, R.J. (1989) Compressibility of omphacite at 60 kbar: role of
vacancies. Am. Mineral., 74, 1287–1292.
McDonough, W.F. and Rudnick, R.L. (1998) Mineralogy and composition of the upper mantle. Rev. Mineral.,
37, 139 –164.
McDonough, W.F. and Sun, S.s. (1995) The composition of the Earth. Chem. Geol., 120, 223–253.
McEwen, A.S., Keszthelyi, L., Spencer, J.R., Schubert, G., Matson, D.L., Gautier, R.L., Kalaasen, K.P., Johnson,
T.V., Head, J.W., Geissier, P., Fagents, S., Davies, A.G., Carr, M.H., Breneman, H.H. and Belton, M.J.S.
(1998) High-temperature silicate volcanism on Jupiter’s moon Io. Science, 281, 87.
McFariane, E.A. and Drake, M.J. (1990) In: Newsome, H.E. and Jones, J.H. (eds.), Origin of the Earth. New York:
Oxford University Press, pp. 135 –150.
McKinnon, W.B. (2002) Out on the edge. Nature, 418, 135– 137.
McMahan, A.K. (1978) Phys. Rev., 17, 1521–1527.
McMillan, P.F. (1989) Raman spectroscopy in mineralogy and geochemistry. Annu. Rev. Earth Planet. Sci.,
17, 255 –283.
McMillan, P. and Akaogi, M. (1987) Am. Mineral., 72, 361.
McMillan, P.F. and Hofmeister, A.M. (1988) Spectroscopic methods in mineralogy and geology. In: Hawthorne,
F.C. (ed.), Reviews of Mineralogy, Vol. 18. Washington, DC: MSA, 99 pp.
McMillan, P. and Piriou, B. (1982) The structures and vibrational spectra of crystals and glasses in the silica–
alumina system. J. Non-Cryst. Solids, 53, 279 –298.
McMillan, P.F. and Wolf, G.H. (1995) Vibrational spectroscopy of silicate liquids. In: Stebbins, J.F., Dingwell,
D.B. and McMillan, P.F. (eds.), Structure and Dynamics of Silicate Mantle. Washington, DC:
Mineralogical Society of America.
McMillan, P.F., Akaogi, M., Ohtani, E., Williams, Q., Nieman, R. and Sato, R. (1989) Cation disorder in garnets
along the Mg3Al2Si3O12 –Mg4Si4O12 join: an infrared, Raman and NMR study. Phys. Chem. Miner.,
16, 428 –435.
McMillan, P.F., Poe, B.T., Gillet, P. and Reynard, B. (1994) A study of SiO2 glass and supercooled liquid
to 1950 K via high temperature Raman spectroscopy. Geochim. Cosmochim. Acta, 58, 3653–3664.
McMillan, P.F., Hemley, R.J. and Gillet, P. (1996) Vibrational spectroscopy of mantle minerals. Geochem. Soc.,
5, 175 –213.
McPherson, W.R. and Schloessin, H.H. (1982) Phys. Earth Planet. Interiors, 29, 58.
McSween, H.Y. Jr. (1994) What have we learned about Mars from SNC meteorites? Meteoritics, 29,
757–779.
McSween, H.Y. and Harvey, R.P. (1993) Outgassed water on Mars: constraints from melt inclusions in SNC
meteorites. Science, 259, 1890–1892.
McSween, H.Y. Jr et al. (2001) Geochemical evidence for magmatic water within Mars from pyroxenes in the
Shergotty meteorite. Nature, 409, 487– 490.
Meade, C. and Hemley, R.J. (1991c) Geophys. Lab., Carnegie Inst. Washington, Annu. Rep., 1990–1991, 135.
Meade, C. and Jeanloz, R. (1988) Yield strength of MgO to 40 GPa. J. Geophys. Res., 93, 3261–3269.
Meade, C. and Jeanloz, R. (1989) Acoustic emissions and shear instabilities during phase transformations in Si
and Ge at ultrahigh pressure. Nature, 339, 616–618.
Meade, C. and Jeanloz, R. (1990) The strength of mantle silicates at high pressures and room temperature:
implications for the viscosity of the mantle. Nature, 348, 533–535.
1162 References

Meade, C. and Jeanloz, R. (1991a) Deep focus earthquakes and recycling of water into the Earth’s mantle.
Science, 252, 68 –72.
Meade, C. and Jeanloz, R. (1991b) Linear finite strain theory for anisotropic materials. EOS Trans. Am. Geophys.
Union, 72(12), 282, Abstr.
Meade, C., Hemley, R.J. and Mao, H.K. (1992a) Phys. Rev. Lett., 65, 1387.
Meade, C., Jeanloz, R. and Hemley, R.J. (1992b) Spectroscopic and X-ray diffraction studies of metastable
crystalline–amorphous transitions in Ca(OH)2 and serpentine. In: Syono, Y. and Manghani, M.H. (eds.),
High Pressure Research in Minerals Physics: Applications to Earth and Planetary Sciences. Tokyo:
Terra Scientific, pp. 485 –492.
Meade, C., Reffner, J.A. and Ito, E. (1994) Synchrotron infrared absorbance of hydrogen in MgSiO3 perovskite.
Science, 264, 1558–1560.
Meade, C., Mao, H.K. and Hu, J. (1995a) High-temperature phase transition and dissociation of (Mg,Fe)SiO3
perovskite at lower mantle pressures. Science, 268, 1743–1745.
Meade, C., Silver, P.G. and Kaneshima, S. (1995b) Laboratory and seismological observations of lower mantle
isotropy. Geophys. Res. Lett., 22, 1293–1296.
Mechie, J., Gorkin, A.V., Fuchs, K., Ryberg, T., Solodilov, L. and Wenzel, F. (1993) P-wave mantle velocity
structure beneath northern Eurasia from long-range recordings along profile quartz. Phys. Earth Planet.
Interiors, 79, 269 –286.
Megaw, H.D. (1968) A note on the structure of lithium niobate, LiNbO3. Acta Crystallogr., 24,
583–588.
Megaw, H.D. (1973) Crystal Structures: A Working Approach. Philadelphia, PA: W.B. Saunders.
Mehl, M.J., Cohen, R.E. and Krakaner, H. (1988) Linearized augmented plane wave electronic structure
calculations for MgO and CaO. J. Geophys. Res., 93, 8009– 8022.
Mehl, M.J., Osburn, J.E., Papaconstantopoulos, D.A. and Klein, B.M. (1990) Phys. Rev. B, 41, 10311.
Mehta, A., Navrotsky, A., Kumada, N. and Kinomura, N. (1993) J. Solid State Chem., 102, 213.
Mehta, A., Leinenweber, K., Navrotsky, A. and Akaogi, M. (1994) Calorimetric study of high pressure
polymorphism in FeTiO3: stability of the perovskite phase. Phys. Chem. Miner., 21, 207–212.
Meibom, A. and Frei, R. (2002) Science, 296, 516.
Meisel, T., Walker, R.J. and Morgan, J.W. (1996) The osmium isotopic composition of the earth’s primitive upper
mantle. Nature, 383, 517–520.
Melosh, H.J. (1990) In: Newsom, H.E. and Jones, J.H. (eds.), Origin of the Earth. New York: Oxford University
Press, pp. 69–83.
Menegazzo, G., Carbonin, S. and Della Giusta, A. (1997) Cation and vacancy distribution in an artificially
oxidized natural spinel. Mineral. Mag., 61, 411– 421.
Meng, Y., Weidner, D.J. and Fei, Y. (1993a) Deviatoric stress in a quasi-hydrostatic diamond anvil cell: effect on
the volume-based pressure calibration. Geophys. Res. Lett., 20, 1147–1150.
Meng, Y., Weidner, D.J., Gwanmesia, G.D., Liberman, R.C., Vaughan, M.T., Wang, Y., Leinenweber, K., Pacalo,
R.E., Yeganeh-Haeri, A. and Zhao, Y. (1993) In situ high P – T X-ray diffraction studies on three
polymorphs (a, b, g) of Mg2SiO4. J. Geophysics, Res, 98, 22, 199–222, 204.
Meng, Y., Fei, T., Weidner, D.J., Gwanmesia, G.D. and Hu, J. (1994) Hydrostatic compression of g-Mg2SiO4
mantle pressures and 700 K: thermal equation of state and related thermoelastic properties. Phys. Chem.
Miner., 21, 407–412.
Mercier, J.-C.C. and Carter, N.L. (1975) Pyroxene geotherm. J. Geophys. Res., 80, 3349–3362.
Mercier, J. and Nicolas, A. (1975) J. Petrol., 16, 454.
Merkel, S., Goncharov, A.F., Mao, H.K., Gillet, P., Hemley, R.J. (2000) Raman spectroscopy of iron to 152 GPa:
Implications for Earth’s inner core. Science, 288, 1626–1629.
Merkel, S., Jephcoat, A.P. et al. (2002a) Equation of state, elasticity, and shear strength of pyrite under pressure.
Phys. Chem. Miner., 29, 1 –9.
Merkel, S., Wenk, H.R. et al. (2002b) Deformation of polycrystalline MgO at pressures of low temperatures of the
lower mantle. J. Geophys. Res., 107(B11), 2271–2287.
Merrill, R.T., McElhinny, M.W. and McFadden, P.L. (1996) The Magnetic Field of the Earth: Paleomagnetism,
the Core, and the Deep Mantle. San Diego, CA: Academic Press.
References 1163

Mignel, M.C., Vespignami, A., Zapperi, S., Weiss, J. and Grasso, J.R. (2001) Intermittent dislocation flow in
visco-plastic deformation. Nature, 410, 667–670.
Militzer, B., Magro, W. and Caperley, D. (1998) In: Kalman, G.J., Blagoev, K.B. and Rommel, J.M. (eds.),
Strongly Coupled Coulomb Systems. New York: Plenum Press.
Millard, R.L., Peterson, R.C. and Hunter, B.K. (1992) Temperature dependence of ion disorder in MgAl2O4 spinel
using 27Al and 17O magic-angle spinning. Am. Mineral., 77, 44–52.
Miller, G.H., Rossman, G.R. and Harlow, G.E. (1987) The natural occurrence of hydroxide in olivine. Phys.
Chem. Miner., 14, 461– 472.
Miller, G.H., Stolper, E.M. and Ahrens, T.J. (1991a) The equation of state of a molten komatiite. 1. Shock wave
compression to 36 GPa. J. Geophys. Res., 96, 11831–11848.
Miller, G.H., Stolper, E.M. and Ahrens, T.J. (1991b) The equation of state of a molten komatiite. 2. Application to
komatiite petrogenesis and the hadean mantle. J. Geophys. Res., 96, 11849– 11864.
Miller, D.M., Goldstein, S.L. and Langmuir, C.H. (1994) Nature, 368, 514–520.
Millis, A.J. (1996) Phys. Rev. B, 53, 8434.
Millis, A.J. (1998) Lattice effects in magneto-resistive manganese perovskites. Nature, 392, 147–148.
Millis, A.J., Littlewood, P.B. and Shraiman, B.I. (1995) Phys. Rev. Lett., 75, 5144.
Miniti, M.E. and Rutherford, M.J. (1998) Assessment of the shock effects on hornblende: water contents and
isotopic compositions. Proc. Lunar Planet. Sci. Conf., 29, 1435–1436.
Mirwald, P.W. and Massonne, H.J. (1980) The low high quartz and quartz–coesite transition to 40 kbar between
6008C and 16008C and some reconnaissance data on the effect of NaAlO2 component on the low quartz–
coesite transition. J. Geophys. Res., 85, 6983–6990.
Misener, D.J. (1973) Cationic diffusion in olivine to 14008C and 35 kb. Geochem. React. Kinet., Carnegie Inst.
Washington Publ., 634, 117–129.
Misener, D.J. (1974) Cationic diffusion in olivine — I: cobalt and magnesium. Geochim. Cosmochim. Acta,
44, 759 –762.
Mishima, O. (1994) Reversible first-order transition between the two H2O amorphous at 0.2 GPa and 135 K.
J. Chem. Phys., 100, 5910–5912.
Mishima, O. (1996) Nature, 384, 546.
Mishima, O. and Stanley, H.E. (1998a) Decompression-induced melting of ice IV and the liquid–liquid transition
in water. Nature, 392, 164–168.
Mishima, O. and Stanley, H.E. (1998b) Rev. High Press. Sci. Technol., 7, 1103.
Mishima, O. and Stanley, H.E. (1998c) The relationship between liquid, supercooled and glassy water. Nature,
396, 329 –335.
Mishima, O., Calvert, L.D. and Whalley, E. (1984) Melting ice Ih at 77 K and 10 kbar: a new method of making
amorphous solids. Nature, 310, 393–395.
Mishima, O., Calvert, L.D. and Whalley, E. (1985) An apparently first-order transition between two amorphous
phases of ice induced by pressure. Nature, 314, 76– 78.
Mitra, S. (1989) Fundamentals of Optical, Spectroscopic and X-Ray Mineralogy. New York: Wiley,
236 pp.
Mitra, S. (1992) Applied Mössbauer Spectroscopy — Theory and Practice for Geochemists and Archeologists.
New York: Pergamon Press, 400 pp.
Mitra, S. and Bhattacharya, D. (1982) Defects and dislocations in the upper mantle (asthenosphere) and
attenuation of shear waves. Gerlands Beitr. Geophys., 91(6), 518–524.
Mitra, S., Pal, T. and Pal, T.N. (1991a) Electron localization at B-site: a concomitant process for oxidation of
Cr-spinels to a partly inverse form. Solid State Commun., 77, 297–301.
Mitra, S., Pal, T. and Pal, T.N. (1991b) Petrogenetic implication of the Mössbauer hyperfine parameters of
Fe3þ-chromites from Sukinda (India) ultramafites. Mineral. Mag., 55, 535–540.
Mitrovica, J. (1996) Haskell 1935 revisited, J. Geophys. Res., 101(B1), 555–570.
Mizoguchi, T. and Tanaka, M. (1963) The nuclear quadrupole interactions of 57Fe in spinel-type oxides. J. Phys.
Soc. Jpn, 18, 1301–1306.
Mocala, K., Navratosky, A. and Sherman, D.M. (1992) High temperature heat capacity of Co3O4 spinel: thermally
induced spin unpairing transition. Phys. Chem. Miner., 19, 88 –95.
1164 References

Mocquet, A., Vacher, P., Grasset, O. and Sotin, C. (1996) Theoretical seismic models of Mars: the importance of
the iron content of the mantle. Planet. Space Sci., 44, 1251–1268.
Mohn, P. (1999) A century of zero expansion. Nature, 400, 18– 19.
Mojzsis, S.I., Harrison, T.M. and Pidgeon, R.T. (2001) Nature, 409, 178–181.
Molin, G.M., Saxena, S.K. and Brizi, E. (1991) Iron–magnesium order–disorder in an orthopyroxene crystal
from the Johnstown meteorite. Earth Planet. Sci. Lett., 105, 260–265.
Molnar, P., England, P. Martinod, J.. (1993) Mantle dynamics, uplift of the Tibetan Plateau, and the Indian
monsoon. Rev. Geophys., 31(4), 357–396.
Molnar, P. and Lyon-Caen, H. (1988) Geol. Soc. Am. Spec. Pap., 218, 179.
Molinari, A., Canova, G.R. and Ahzi, S.A. (1987) A self-consistent approach of the large deformation polycrystal
viscoplasticity. Acta Metall., 35, 2983–2994.
Montagner, J.-P. and Gulliot, L. (2000) In: Boschi, E., Ekstrom, G. and Morelli, A. (eds.), Problems in Geophysics
for the New Millennium. Rome: Editrice Compostiori, 217 pp.
Mook, H.A., Pencheng, D., Hayden, S.M., Aeppll, G., Perring, T.G. and Dogan, F. (1998) Spin fluctuations in
YBa2Cu3O6.6. Nature, 395, 580–582.
Moore, P.B. and Louisnathan, J. (1967) Fresnoite: unusal titanium coordination. Science, 156, 1361– 1362.
Morbidelli, A. et al. (2000) Meteor. Planet. Sci., 35, 1309.
Moreira, M., Kunz, J. and Allègre, C.J. (1998) Rare gas systematics popping rock: isotopic and elemental
compositions in the upper mantle. Science, 279, 1178–1181.
Morelli, A. and Dziewonski, A.M. (1993) Body-wave travel times and a spherically symmetric P- and S-wave
velocity model. Geophys. J. Int., 112, 178 –194.
Morelli, A., Dziewonski, A.M. and Woodhouse, J.H. (1986) Geophys. Res. Lett., 13, 1549.
Morgan, J.W. and Anders, E. (1979) Geochim. Cosmochim. Acta, 43, 1601.
Mori, S., Chen, C.H. and Cheong, S.W. (1998) Pairing of charge-ordered stripes in (La,Ca)MnO3. Nature, 392,
473–476.
Morin, F.J. (1950) Phys. Rev., 78, 819.
Morin, F.J. (1954) Phys. Rev., 93, 1195.
Moritomo, Y., Asamitsu, A. and Tokura, Y. (1995) Pressure effect on the double-exchange ferromagnet
La12xSrxMnO3(0.15 # x # 0.5). Phys. Rev., 51(22), 50–53.
Moriya, T. (1960) Phys. Rev., 117, 635.
Moroni, E.G., Grimvall, G. and Jarborg, T. (1996) Free energy contributions to the hcp–bcc tranformation in
transition metals. Phys. Rev. Lett., 76, 2758–2761.
Morris, S. (1992) Proc. R. Acad. Lond., 436, 203.
Morris, E.R. and Williams, Q. (1997) Electrical resistivity of Fe3O4 to 48 GPa: compression-induced changes in
electron hopping at mantle pressure. J. Geophys. Res., 102, 18139–18148.
Morris, M.C., Mcmurdie, H.F., Evans, E.H., Paretzkin, B., Parker, H.S., Panagiotopoulous, N.C. and Hubbard,
C.R. (1981) Standard X-Ray Diffraction Powder Patterns, National Bureau of Standards Monograph
25 — Section, Vol. 18, 37 pp.
Mott, N.F. (1979) Adv. Phys., 21, 785.
Mott, N. (1990) Metal–Insulator Transitions, 2nd ed. London: Taylor & Francis.
Mott, N.F. and Davies, E.A. (1972) Electronic Processes in Noncrystalline Materials. Oxford: Clarendon Press.
Muhlhausen, C. and Gordon, R.G. (1981) Electron-gas theory of ionic crystals, including manybody effects. Phys.
Rev., 23, 900 –923.
Müller, H.J. and Raab, S. (1997) Elastic wave velocities of granite at experimental simulated partial melting
conditions. Phys. Chem. Earth, 22, 93–96.
Muller, O. and Roy, R. (1974) The Major Ternary Structural Families. Berlin: Springer.
Murad, E. and Wagner, F.E. (1987) The Mössbauer spectrum of almandine. Phys. Chem. Miner., 14, 264–269.
Murakam, T., Utsinomiya, T., Simazu, Y. and Prasad, N. (2001) Earth Planet. Sci. Lett., 184,
525–528.
Murnaghan, F.D. (1944) The compressibility of media under extreme pressures. Proc. Natl Acad. Sci. USA, 3,
224–247.
References 1165

Murthy, V.R. (1991) Earth differentiation of the Earth and the problem of mantle siderophile elements: a new
approach. Science, 253, 303– 306.
Musgrave, M.J.P. (1970) Crystal acoustic: introduction to the elastic waves and vibrations in crystals, Holden-Day
Mag Physics. San Francisco, CA: Holden-Day, 288 pp.
Musselwhite, D. and Lunine, J. (1995) Alteration of volatile inventories by polar clathrate formation on Mars,
J. Geophys. Res., 100(E11), 23301–23306.
Myers, D.J., Urdahl, R.S., Cherayil, B.J. and Fayer, M.D. (1997) J. Chem. Phys., 107, 9741.
Mysen, B.O. (1990) Effect of pressure, temperature, and bulk composition on the structure and species
distribution in depolymerized alkali aluminosilicate melts and quenched melts. J. Geophys. Res.,
95(B10), 15733–15744.
Mysen, B.O. (1998) Interaction between aqueous fluid and silicate melt in the pressure and temperature regime of
the Earth’s crust and upper mantle. N. Jb. Miner. Mh., 172, 227–244.
Mysen, B.O. and Boettcher, A.L. (1975) Melting of a hydrous mantle II. Geochemistry of crystals and liquids
formed by anatexis of mantle peridotite at high pressures and high temperatures as a function of
controlled activities of water, hydrogen and carbon dioxide. J. Petrol., 16, 549–590.
Mysen, B. and Virgo, D. (1978) Influence of pressure, temperature, and bulk composition on melt structures in the
system NaAlSi2O6 –NaFe3þSi2O6. Am. J. Sci., 278, 1306–1322.
Mysen, B. and Virgo, D. (1985) Iron-bearing silicate melts: relations between pressure and redox equilibria. Phys.
Chem. Miner., 12, 191– 200.
Mysen, B.O., Virgo, D. and Seifert, F. (1982) The structure of silicate melts: implications for chemical and
physical properties of natural magma. Rev. Geophys., 20, 353–383.
Mysen, B.O., Ulmer, P., Konzett, J. and Schmidt, M.W. (1998a) The upper mantle near convergent plate
boundaries: ultrahigh-pressure mineralogy. In: Hemley, R.J. (ed.), Physics and Chemistry of the Earth’s
Deep Interior, Reviews in Mineralogy, Vol. 37, pp. 97–132.
Mysen, B.O., Virgo, D., Popp, R.K. and Bertka, C.M. (1998b) The role of H2O in Martian magmatic systems.
Am. Mineral., 83, 942 –946.
Nagai, T., Ito, T., Hattori, T. and Yamanaka, T. (2000) Compression mechanism and amorphization of portlandite,
Ca(OH)2: structural refinement under pressure. Phys. Chem. Miner., 27, 462–466.
Naka, S., Ito, S. and Inagaki, M. (1974) Kinetic studies of transitions from amorphous silica and quartz to coesite.
J. Am. Ceram. Soc., 57, 217– 219.
Nakagiri, N., Manghnani, M.H., Ming, L.C. and Kumara, S. (1986) Crystal structure of magnetite under pressure.
Phys. Chem. Miner., 13, 238 –244.
Nakai, S. and Takabe, H. (1996) Rep. Prog. Phys., 59, 1071.
Nakajima, Y. and Ogasawara, Y. (1997) Garnet lherzolite with olivine-Ti mineral intergrowth from Rongcheng in
the Sulu UHP terrane, eastern China. EOS Trans. Am. Geophys. Union, 78, F787.
Nakamoto, K., Margoshes, M. and Rundle, R.E. (1955) Stretching frequencies as a function of distances in
hydrogen bonds. J. Am. Chem. Soc., 77, 6480–6488.
Nakamura, A. and Schmalzreid, H. (1983) On the nonstoichiometry and point defects of olivine. Phys. Chem.
Miner., 10, 27–37.
Nakatsuka, A., Yoshiasa, A., Yamanaka, T., Ohtaka, O., Katsura, T. and Ito, E. (1999) Symmetry change of
majorite solid-solution in the system Mg3Al2Si3O12 – MgSiO3. Am. Mineral., 84, 1135–1143.
Narayan, C., Luo, H., Orloff, J. and Rouff, A.L. (1998) Solid hydrogen at 342 GPa: no evidence for an alkali
metal. Nature, 393, 46–49.
Nasu, S., Kurimoto, K., Nagatomo, S., Endo, S. and Fujita, F. (1986) Iron 57 Mössbauer study under high
pressure: e-iron and iron oxide (Fe2O3), Hyperfine Interact., 29, 1583.
Nasu, S., Abe, T., Yamamoto, K. and Endo, S. (1991) 57Fe Mössbauer study of SrFeO3 under ultra-high pressure.
Hyperfine Interact., 67, 529–532.
Nasu, S., Abe, T., Yamamoto, K., Endo, S., Takano, M. and Takeda, Y. (1992) Mössbauer study of CaFeO3 under
external high-pressure. Hyperfine Interact., 70, 1063–1066.
Natoli, V.M., Martin, R.M. and Ceperley, D.C. (1993) Crystal structure of atomic hydrogen. Phys. Rev. Lett.,
70, 1952–1955.
1166 References

Navrotsky, A. (1989) Thermochemistry of perovskites. In: Navrotsky, A. and Weidner, D.J. (eds.), Perovskite: A
Structure of Great Interest to Geophysics and Materials Science, Geophysical Monograph, Vol. 45.
Washington, DC: AGU, pp. 67–80.
Navrotsky, A., Weidner, D.J., Lieberman, R.C. and Prewitt, C.T. (1992) Materials science of the Earth’s deep
interior, MRS bulletin. May, 30– 57.
Navrotsky, A. (1999) A lesson from ceramics. Science, 284, 1788–1789.
Navrotsky, A., Weidner, D.J. (ed.) (1989) Perovskite: A structure of Great Interest Geophysics and Material
Science, Geophysical Monograph, Vol. 45. Washington, DC: AGU.
Navrotsky, A., Geisinger, K.L., McMillan, P. and Gibbs, G.V. (1985) The tetrahedral framework in glasses and
melts — inferences from molecular orbital calculations and implications for structure, thermodynamics,
and physical. Phys. Chem. Miner., 11, 284–298.
Neaton, J.B. and Ashcroft, N.W. (1999) Pairing in dense lithium. Nature, 400, 141–144.
Nell, J. and Wood, B.J. (1991) High temperature electrical measurements and thermodynamic properties of
Fe3O4 –FeCr2O4 –MgCr2O4 – FeAl2O4 spinels. Am. Mineral., 76, 405–426.
Nell, J., Wood, B.J. and Mason, T.O. (1989) High-temperature cation distribution Fe3O4 –MgFe2O4 –FeAl2O4 –
MgAl2O4 spinels from thermopower and conduction measurements. Am. Mineral., 74, 339–351.
Nellis, W.J. and Mitchell, A.C. (1998) Bull. Am. Phys. Soc., 43, 408.
Nellis, W.J. et al. (1988) Science, 240, 779.
Nellis, W.J. et al. (1991) J. Chem. Phys., 94, 2244.
Nellis, W.J., Ross, M. and Holmes, N.C. (1995) Temperature measurements of shock-compressed liquid hydrogen
implications for the interior of Jupiter. Science, 269, 1249.
Nellis, W.J., Holmes, N.C., Mitchell, A.C., Hamilton, D.C. and Nicol, M. (1997) J. Chem. Phys., 107, 9096.
Neumeier, J.J., Cornelius, A.L. and Schilling, J.S. (1994) Physica B, 198, 324.
Neumeier, J.J., Hundley, M.F., Thompson, J.D. and Heffneer, R.H. (1995) Substantial pressure effects on the
electrical resistivity and ferromagnetic transition temperature of La12xCaxMnO3. Phys. Rev., 52,
7006– 7009.
Newsom, H.E. (1986) In: Hartmann, W.K., Phillips, R.J. and Taylor, G. (eds.), Origin of the Moon. Houston, TX:
Lunar Planetary Institute, pp. 203–229.
Newsom, H.E. and Sims, K.W.W. (1991) Science, 252, 926 –933.
Newton, R.C., Charlu, T.V. and Kleppa, O.J. (1977) Thermochemistry of high pressure garnet and clinopyroxenes
in the system CaO–MgO– Al2O3 –SiO2. Geochim. Cosmochim. Acta, 41, 369– 377.
Nguyen, J. and Holmes, N.C. (1998) Iron sound velocities in shock wave experiments up to 400 GPa. EOS Trans.
Am. Geophys. Union, 79, F846.
Nicholas, J.B., Winans, R.E., Harrison, R.J., Iton, L.E., Curtiss, L.A. and Hopfinger, A.J. (1992) Ab initio
molecular orbital study of the effects of basis set size on the calculated structure and acidity of hydroxyl
groups in framework molecular sieves. J. Phys. Chem., 96, 10247–10257.
Nickel, K.F. and Green, D.H. (1985) Empirical geothermobarometry for garnet peridotites and implications for the
nature of the lithosphere, kimberlites and diamonds. Earth Planet. Sci. Lett., 73, 158–170.
Nielsen, O.H. (1986) Phys. Rev. B, 34, 5808–5819.
Niesler, H. and Jackson, I. (1989) Pressure derivatives of elastic wave velocities from ultrasonic interferometric
measurements on jacketed polycrystals. J. Acoust. Soc. Am., 86, 1573–1585.
Nininger, R.C. Jr. and Schroeer, D. (1978) J. Phys. Chem. Solids, 39, 137.
Nishizawa, O. and Akimoto, S. (1973) Partition of magnesium and iron between olivine and spinel and between
pyroxene and spinel. Contrib. Mineral. Petrol., 41, 217–230.
Nolas, G., Slack, G., Morelli, D.T., Tritt, T.M. and Ehrlich, A.C. (1996) J. Appl. Phys., 79, 4002.
Nolet, G. and Zierws, A. (1994) J. Geophys. Res., 99, 15813–15820.
Nolet, G., Grand, S.P. and Kennett, B.L.N. (1994) J. Geophys. Res., 99, 23753–23766.
Novak, A. (1974) Hydrogen bonding in solids. Correlation of spectroscopic and crystallographic data. Struct.
Bond., 18, 177– 216.
Nyblade, A.A. (2001) Hard-cored continents. Nature, 411, 38–39.
Nye, J.F. (1975) Physical Properties of Crystals. Oxford: Clarendon Press.
References 1167

Nye, J.F. (1985) Physical Properties of Crystals: Their Representation by Tensors and Matrices. Oxford: Oxford
University Press.
Nyman, H. and O’keeffe (1978) Sodium titanium silicate, Na2TiSiO5. Acta Crystallogr., 34, 905–906.
O’Connell, R.J. (1977) On the scale of mantle convection. Tectonophysics, 38, 119–136.
Ogasawara, Y., Liou, J.G. and Zhnag, R.Y. (1997) Thermochemical calculation of log f O2 –T– P stability
relations of diamond-bearing assemblages in the model system CaO – MgO – SiO2 – CO2 – H2O.
Russ. Geol. Geophys., 38, 587– 598.
Ohashi, Y. (1982) A program to calculate the strain tensor from two sets of unit-cell parameters. In: Hazen, R.W.
and Finger, L.W. (eds.), Comparative Crystal Chemistry. New York: Wiley, pp. 92 –102.
Ohtaka, O., Tobe, H. and Yamanaka, T. (1997) Phase equilibria for the Fe2SiO4 –Fe3O4 system under high
pressure. Phys. Chem. Miner., 24, 555–560.
Ohtani, E. and Kamaya, N. (1992) Geophys. Res. Lett., 19, 2239.
Ohtani, E. and Kumazawa, M. (1981) Melting of forsterite Mg2SiO4 up to 15 GPa. Phys. Earth Planet. Interiors,
27, 32 –38.
Ohtani, E., Kagawa, N. and Fujino, K. (1991a) Stability of majorite (Mg,Fe)SiO3 at high pressures and 18008C.
Earth Planet. Sci. Lett., 102, 158 –166.
Ohtani, E., Kato, T. and Ito, E. (1991b) Transition metal partitioning between lower mantle and core materials at
27 GPa. Geophys. Res. Lett., 18, 85– 88.
Ohtani, E., Nagata, Y., Suzuki, A. and Kato, T. (1995a) Melting relations of peridotite and the density cross-over
in planetary mantle. Chem. Geol., 120, 207 –221.
Ohtani, E., Shibata, T. and Kudo, T. (1995b) Stability of hydrous phases in the transition zone and the uppermost
part of the lower mantle. Geophys. Res. Lett., 22, 2553–2556.
Ohtani, E., Yurimoto, H. and Seto, S. (1997) Element partitioning between metallic liquid, silicate liquid and
lower-mantle minerals: implications for core formation of the Earth. Phys. Earth Planet. Interiors,
100, 97–114.
Ohtani, E., Mizobata, H. and Yurimoto, H. (2000) Stability of dense hydrous magnesium silicate phases in the
system Mg2SiO4 –H2O and MgSiO3 –H2O at pressures up to 27 GPa. Phys. Chem. Miner., 27,
533–544.
Okada, Y., Arima, T., Tokura, Y., Maruyama, C. and Mori, N. (1993) Phys. Rev. B, 48, 9677.
Okal, E.A. and Cansi, Y. (1998) Earth Planet. Sci. Lett., 164, 23.
Okamoto, K. and Maruyama, S. (1999) The high pressure synthesis of lawsonite in the MORB þ H2O system.
Am. Mineral., 84, 362 –373.
Okay, A.I., Sengör, A.M.C. and Satir, M. (1993) Tectonics of an ultrahigh-pressure metamorphic terrane: the
Dabie Shan/Tongbai Shan Orogen, China. Tectonics, 12, 1320–1334.
O’Keefe, M., Hyde, B.G. and Bovin, J.O. (1979) Contribution to the crystal chemistry of orthorhombic
perovskites: MgSiO3 and NaMgF3. Phys. Chem. Miner., 4, 299–305.
Okhulkov, A.V., Demianets, Y.N. and Gorbaty, Y.E. (1992) X-ray scattering in liquid water at pressure of up to
7.7 kbar. Test of fluctuation model. J. Chem. Phys., 100, 1578–1588.
Okimoto, Y., Katsufuji, T., Ishikawa, T., Urushibara, A., Arima, T. and Tokura, Y. (1995) Phys. Rev. Lett., 75, 109.
Okuchi, T. (1997) Hydrogen partitioning into molten iron at high pressure implications for Earth’s core. Science,
278, 1781–1784.
Okuchi, T. (1998) The melting temperature of iron hydride at high pressures and its implications of the Earth’s
core. J. Phys. Condens. Matter, 10, 11595– 11598.
Olafsson, M. and Eggler, D.H. (1983) Phase relations of amphibole, amphibole –carbonate, and phlogopite–
carbonate peridotite: petrologic constraints on the asthenosphere. Earth Planet. Sci. Lett., 64, 305–315.
Olijnyk, H. and Jephcoat, A.P. (1998) Phys. Rev. B, 57, 879.
Olijnyk, H., Paris, E. and Geiger, C.A. (1991) Compressional study of katoite [Ca3Al2(O4H4)3] and grossular
garnet. J. Geophys. Res., 96, 14313–14318.
Olsen, J.S., Gerward, L. and Jiang, J.Z. (1999) J. Phys. Chem. Solids, 60, 229.
Omori, S. et al. (2002) Bull. Earthq. Res. Inst. Univ. Tokyo, 76, 455.
O’Neill, B. and Jeanloz, R. (1993) Petrology of the lower mantle. A view through the diamond cell (abstract). EOS
Trans. AGU, Fall Meeting Suppl., 74, 551.
1168 References

O’Neill, B. and Jeanloz, R. (1994) MgSiO3 –FeSiO3 –Al2O3 in the Earth’s lower mantle: perovskite and garnet at
1200 km depth. J. Geophys. Res., 99, 19901–19915.
O’Neill, H.St.C. and Wood, B.J. (1979) An experimental study of Fe –Mg partitioning between garnet and olivine
and its calibration as a geothermometer. Contrib. Mineral. Petrol., 70, 59–70.
O’Neill, B., Bass, J.D., Smyth, J.R. and Vaughan, M.T. (1989) Elasticity of a grossular-pyrope almandine garnet.
J. Geophys. Res., 94, 17819–17824.
O’Neill, B., Bass, J.D., Rossman, G.R., Geiger, C.A. and Langer, K. (1991a) Elastic properties of pyrope. Phys.
Chem. Miner., 17, 617– 621.
O’Neill, B., Brown, D. and Jeanloz, R. (1991b) Effect of simultaneous Fe and Al substitution on the volume of
magnesium-silicate perovskite. EOS Trans. Am. Geophys. Union, 72(44).
O’Neill, H.St. (1991c) The origin of the Moon and the early history of the Earth — a chemical model, Part I.
Geochim. Cosmochim. Acta, 55, 1135–1157.
O’Neill, H.St. (1991d) The origin of the Moon and the early history of the Earth. A chemical model. Part 2.
Geochim. Cosmochim. Acta, 55, 1159–1172.
O’Neill, H.St.C. and Navrotsky, A. (1983) Simple spinels: crystallographic parameters, ionic radii, lattice
energies and cation distribution. Am. Miner., 68, 181 –185.
O’Neill, H.St. and Navrotsky, A. (1984) Cation distributions and thermodynamic properties of binary spinel solid
solutions. Am. Mineral., 69, 733 –753.
O’Neill, H.St.C. and Wall, V.J. (1987) The olivine– orthopyroxene–spinel oxygeobarometer, the nickel
precipitation curve, and the oxygen fugacity of the Earth upper mantle. J. Petrol., 28, 1169–1191.
O’Neill, H.St., Annersten, H. and Virgo, D. (1992) The temperature dependence the cation distribution in
magnesioferrite (MgFe2O) from powder XRD structure refinements and Mössbauer spectroscopy. Am.
Mineral., 77, 725 –740.
O’Neill, H.St.C., Rubie, D.C., Canil, D., Geiger, C.A., RossII, C.R., Seifert, F. and Woodland, A.B. (1993)
Ferric iron in the upper mantle in the transition zone assemblages: implications for relative oxygen
fugacities in the mantle. In: Evolution of the Earth and Planets, Geophysical Mono, 75, IUGG 14,
73–88.
O’Neill, H.St.C., McCammon, C.A., Canil, D., Rubie, D.C., Ross, C.R. and Seifert, F. (1993a) Mössbauer
spectroscopy of mantle transition zone phases and determination of minimum Fe3þ content.
Am. Mineral., 78, 456 –460.
O’Neill, B., Bass, J.D. and Rossman, G.R. (1993b) Elastic properties of hydrogrossular garnet and implications
for water in the upper mantle. J. Geophys. Res., 98, 20031–20037.
O’Nions, R.K. and Tolstikhin, I.N. (1994) Earth Planet. Sci. Lett., 124, 131–138.
Onuma, N., Higuchi, H., Wakita, H. and Nagasawa, H. (1968) Trace element partition between two pyroxenes and
the host lava. Earth Planet. Sci. Lett., 5, 47–51.
Orenstein, J. and Millis, A.J. (2000) Advances in physics of high-temperature superconductivity. Science, 288,
468–474.
Osako, M. and Ito, E. (1991) Thermal diffusivity of MgSiO3 perovskite. Geophys. Res. Lett., 18, 239–242.
Osako, M. and Kobayashi, Y. (1979) Phys. Earth Planet. Interiors, 55, 292.
Osborne, M.D., Fleet, M.E. and Bancroft, G.M. (1981) Fe2þ –Fe3þ ordering in chromite and Cr-bearing spinels.
Contrib. Mineral. Petrol., 77, 251 –255.
Osborne, M.D., Fleet, M.E. and Bancroft, G.M. (1984) Next-nearest-neighbour effects in the Mössbauer spectra of
(Cr, Al) spinel. J. Solid State Chem., 53, 174 –183.
Ostapenko, G.T. (1971) Thermodynamics of first-order phase transitions under non-hydrostatic stress. Geochem.
Int., 771–778, translated from Geokhimiya.
Ostapenko, G.T. (1973) Thermodynamics of second-order phase transitions under non-hydrostatic stress.
Geochem. Int., 148–155, translated from Geokhimiya.
Otter, M.L. and Gurney, J.J. (1989) Kimberlites and related rocks 2: their mantle/crust setting. In: Ross, J. et al.
(eds.), Diamonds and Diamond Exploration. Victoria: Blackwell Scientific, pp. 1042–1053.
Ottonello, G., Princivalle, F. and Della Giusta, A. (1990) Temperature composition and f O2 effects on
intersite distribution of Mg and Fe2þ in olivines: experimental evidence and theoretical interpretation.
Phys. Chem. Miner., 17, 301 –312.
References 1169

Ozawa, S. (1991) Trivalent cations in olivine and their implication to the upper mantle tectonics as inferred from
the high pressure experiments. Doctoral Thesis, Univ. of Tokyo.
Ozima, M. (1986) Nature, 321, 813 –814.
Ozima, M. (1998) Noble gases under pressure in the mantle. Nature, 393, 303–304.
Pacalo, R.E.G. and Gasparik, T. (1990) Reversals of the orthoenstatite–clinoenstatite transition at high pressures
and high temperatures. J. Geophys. Res., 95, 15853– 15858.
Pacalo, R.E.G. and Parise, J.B. (1992) Crystal structure of superhydrous B a hydrous magnesium silicate
synthesized at 14008C and 20 GPa. Am. Mineral., 77, 681–684.
Pacalo, R.E.G. and Weidner, D.J. (1995) Elasticity of superhydrous B. Phys. Chem. Miner., 23, 520–525.
Pacalo, R.E.G. and Weidner, D.J. (1997) Elasticity of majorite, MgSiO3 tetragonal garnet. Phys. Earth Planet.
Interiors, 99, 145 –154.
Pal, T., Moon, H.S. and Mitra, S. (1994) Distribution of cations in natural chromites at different stages of
oxidation — a 57Fe Mössbauer investigation. J. Geol. Soc. India, 44, 53 –64.
Palmer, D.C. and Finger, L.W. (1994) Pressure-induced phase transition in cristobalite: an X-ray powder
diffraction study to 4.4 GPa. Am. Mineral., 79, 1–8.
Palmer, D.C., Downs, R.T. and Hemley, R.J. (1991) High Pressure Phase Transitions in Cristobalite, Condensed
Matter and Materials Physics (CMMP 91), 17– 19 December, 1991. Birmingham: The Institute of
Physics, Abstracts with Program.
Palmer, D.C., Hemley, R.J. and Prewitt, C.T. (1994) Raman spectroscopic study of high-pressure phase transitions
in cristobalite. Phys. Chem. Miner., 21, 481–488.
Pal’yanov, Y.N. et al. (1997) Russ. Geol. Geophys., 4, 920 –945.
Pal’yanov, Yu.N., Sokol, A.G., Borzdov, Yu.M., Khokhryakov, A.F. and Sobolev, N.V. (1999) Diamond
formation from mantle carbonate fluids. Nature, 400, 417– 418.
Papike, R.L. and Cameron, M. (1976) Crystal chemistry of silicate minerals of geophysical interest. Rev. Geophys.
Space Phys., 14, 37–80.
Pappalardo, R.T., Head, J.W., Greeley, R. et al. (1998) Geological evidence for solid-state convection in Europa’s
ice shell. Nature, 391, 365–367.
Paris, E., Ross, C.R. II and Olijnyk, H. (1992) Mn3O4 at high pressure: a diamond anvil cell study and a structural
modelling. Eur. J. Mineral., 4, 87–93.
Paris, J.B., Leinenweber, K., Weidner, D.J., Tan, K. and Von Dreck, R.B. (1994) Pressure-induced H bonding:
neutron diffraction study of brucite, Mg(OH)2 to 9.3 GPa. Am. Miner., 79, 193–196.
Parise, J.B., Wang, Y., Yeganeh-Haeri, A., Cox, D.E. and Fei, Y. (1990) Crystal structure and thermal expansion
of (Mg,Fe)SiO3 perovskite. Geophys. Res. Lett., 17, 2089.
Parise, J.B., Leinenweber, K., Weidner, D.J., Tan, K. and Von Dreck, R.B. (1994) Pressure-induced H bonding :
neutron diffraction study of brucite, Mg(OH)2, to 9.3 GPa. Am. Miner., 79, 193–196.
Parise, J.B., Wang, Y., Gwanmesia, G.D., Zhang, J., Sinelnikov, Y., Chmielowski, J., Weidner, D.J. and
Liebermann, R.C. (1996) The symmetry of garnet on the pyrope (Mg3Al2Si3O12) –majorite (MgSiO3)
join. Geophys. Res. Lett., 23, 3799– 3802.
Parise, J.B., Therous, B., Li, R., Loveday, J.S., Marshall, W.G. and Klotz, S. (1998a) Pressure dependence
of hydrogen bonding in metal deuteroxides: a neutron powder diffraction study of Mn(OD)2 and
b-Co(OD)2. Phys. Chem. Miner., 25, 130–137.
Parise, J.B., Loveday, J.S., Nelmes, R.J. and Kagi, H. (1999) Hydrogen repulsion “transition” in Co(OD)2 at high
pressure? Phys. Rev. Lett., 83, 328– 331.
Park, J.H., Vescovo, E., Kim, H.J., Kwon, C., Ramesh, R. and Venkatesan, T. (1998) Direct evidence for a half-
metallic ferromagnet. Nature, 392, 794.
Parkhomenko, E.J. (1982) Electrical resistivity of minerals and rocks at high temperature and pressure.
Rev. Geophys. Space Phys., 20, 193–218.
Parkinson, I.J., Hawkesworth, C.J., Cohen, A.S. (1998). Ancient mantle in a modern arc: Osmanium isotopes in
Izu-Bonin-Mariana fore arc peridotites, Science, 281, 2011–2013.
Parman, S.W., Dann, J.C., Grove, T.L. and de Wit, M.J. (1997) Emplacement conditions of komatiite magmas
from the 3.49 Ga Komati Formation, Barberton Greenstone Belt, South Africa. Earth Planet. Sci. Lett.,
150, 303 –323.
1170 References

Parmentier, E.M. and Zuber, M.T. (2002) Lunar Planet. Sci., 33, 1737.
Parrinello, M. and Rahman, A. (1980) MD simulation. Phys. Rev. Lett., 45, 1196.
Paschek, D. and Geiger, A. (1999) Simulation study on the diffusion motion in deeply supercooled water. J. Phys.
Chem. B, 103, 4139–4146.
Passaglia, E. and Rinaldi, R. (1984) Katoite, a new member of the Ca3Al2(SiO4)3 –Ca3Al2(OH)12 series and a
new nomenclature for the hydrogrossular group of minerals. Bull. Mineral., 107, 605– 618.
Pasternak, M., Taylor, R.D., Dillon, T. and Jeanloz, R. (1993) EOS Trans. Am. Geophys. Union, Fall Meeting
Suppl., 74, 416.
Pasternak, M.P., Nasu, S., Wada, K. and Endo, S. (1994) High-pressure of magnetite. Phys. Rev., 50,
6446– 6449.
Pasternak, M.P., Taylor, R.D., Jeanloz, R., Li, X., Nguyen, J.H. and McCammon, C.A. (1997) Phys. Rev. Lett., 79,
5046.
Pasternak, M.P., Rozenberg, G.K., Machavariani, G.Y., Naaman, O., Taylor, R.D. and Jeanoz, R. (1999)
Breakdown of the Mott–Hubbard state in Fe2O3: a first-order insulator–metal transition with collapse of
magnetism at 50 GPa. Phys. Rev. Lett., 82, 4663–4666.
Paterson, M.S. (1978) Experimental Rock Deformation — The Brittle Field. New York: Springer, 30 pp.
Paterson, M.S. (1982) The determination of hydroxyl by infrared absorption in quartz, silicate glasses and similar
materials. Bull. Mineral., 105, 20–29.
Paul, D.Mc., Balakrishnan, G., Bernhoeft, N.R., David, W.I.F. and Harrison, W.T.A. (1987) Anomalous
structural behaviour of the superconducting compound La1.85 –Ba0.15CuO4. Phys. Rev. Lett., 58,
1976– 1978.
Pauling, L. (1935) J. Am. Chem. Soc., 57, 2680– 2684.
Pauling, L. (1961) The Nature of the Chemical Bond and the Structure of Molecules and Crystals, 4th ed. Ithaca,
NY: Cornell University Press, 400 pp.
Pavese, A., Catti, M., Ferraris, G. and Hull, S. (1997) P – V equation of state of portlandite, Ca(OH)2, from powder
neutron diffraction data. Phys. Chem. Miner., 24, 85 –89.
Pavese, A., Artioli, G. and Hull, S. (1999a) In situ powder neutron diffraction of cation partitioning vs. pressure in
Mg0.94Al2.04O4 synthetic spinel. Am. Mineral., 84, 905–912.
Pavese, A., Artioli, G., Russo, U. and Hoser, A. (1999b) Cation partitioning versus temperature in
(Mg0.70Fe0.23)Al1.97O4 synthetic spinel, by in situ neutron powder diffraction. Phys. Chem. Miner.,
26(3), 242–250.
Pawley, A.R. (1994) The pressure and temperature stability limits of lawsonite: implications for H2O recycling in
subduction zones. Contrib. Mineral. Petrol., 118, 99–108.
Pawley, A.R. and Wood, B.J. (1995) The high-pressure stability of talc and 10 Å-phase: potential storage for H2O
in subduction zones. Am. Mineral., 80, 998–1003.
Pawley, A.R. and Wood, B.J. (1996) The low-pressure stability of phase. A Mg7Si2O8(OH)6. Contrib. Mineral.
Petrol., 124, 90–97.
Pawley, A.R., McMillan, P.F. and Holloway, J.R. (1993) Hydrogen in stishovite with implications for mantle
water content. Science, 261, 1024– 1026.
Pawley, A.R., Redfern, S.A.T. and Wood, B.J. (1995) Thermal expansivities and compressibilities of hydrous
phases in the system MgO– SiO2 –H2O: talc, phase A and 10 Å phase. Contrib. Mineral. Petrol.,
122, 301–307.
Pawley, A.R., Redfern, S.A.T. and Holland, T.J.B. (1996) Volume behaviour of hydrous minerals at high pressure
and temperature. In: Thermal expansion of lawsonite, zoisite, clinozoisite, and diaspore. Am. Mineral.,
81, 335 –340.
Pawley, A.R., Chinnery, N.J. and Clark, S.M. (1998) Volume measurements of zoisite at simultaneously elevated
pressure and temperature. Am. Mineral., 83, 1030–1036.
Payne, M.C., Teter, M.P., Allen, D.C., Arias, T.A. and Joannopouls, J.D. (1992) Interactive minimisation
techniques for ab initio total-energy calculations: molecular dynamics and conjugate gradient. Rev. Mod.
Phys., 64, 1045–1097.
Peacock, S.M. (1990) Fluid processes in subduction zones. Science, 248, 229–345.
Peacock, S.M. (1995) Island Arc, 4, 376.
References 1171

Peacock, S.M. and Wang, K. (1999) Seismic consequences of warm versus cool subduction metamorphism:
examples from southwest and northeast Japan. Science, 286, 937– 939.
Peacock, S.M., Rushmer, T. and Thompson, A.B. (1994) Partial melting of subducting oceanic crust. Earth
Planet. Sci. Lett., 121, 227–244.
Pearson, D.G., Davies, G.R. and Nixon, P.H. (1993) Geochemical constrains on the petrogenesis of diamond
facies. In: Coleman, R.G. and Wang, X. (eds.), Ultrahigh Pressure Metamorphism. Cambridge:
Cambridge University Press, pp. 456–510.
Pearson, D.G. et al. (1995) Geochim. Cosmochim. Acta, 59, 959–977.
Pechar, F., Fuess, H. and Joswig, W. (1989) Refinement of the crystal structure of kaersutite (Vlcibra, Bohemia)
from neutron diffraction. N. Jb. Miner. Mh., 89, 137–143.
Pengcheng, Dai1, Mook1, H.A. and Dogan2, F. (1998) Incommensurate Magnetic Fluctuations in YBa2Cu3O6.6.
Phys. Rev. Lett., 80, 1738–1741.
Pentinghaus, H. (1978) Crystal chemistry of hollandites AxM8(O,OH)16 (x # 2). Phys. Chem. Miner., 3,
85–86.
Pepin, R.O. (1998) Nature, 394, 664– 667.
Percival, M.L.J. (1990) Optical absorption spectroscopy of doped materials: the P21 3 – P21 21 21 phase transition
in K2(Cd0.98Co0.02)2(SO4)3. Mineral. Mag., 54, 525 –535.
Perdew, J.P., Burke, K. and Ernzerhof, M. (1996) Generalized gradient approximation made simple. Phys. Rev.
Lett., 77, 3856–3868.
Perkins, D., Edgar, W.F. and Essene, E.J. (1980) The thermodynamic properties and phase relations of
some minerals of the system CaO–Al2O3 –SiO2 –H2O. Geochim. Cosmochim. Acta, 44, 61– 84.
Perottoni, C.A. and da Jornada, J.A.H. (1998) Pressure-induced amorphization and negative thermal expansion in
ZrW2O8. Science, 280, 886–889.
Peselnick, L. and Robie, R.A. (1963) Elastic constants of calcite. J. Appl. Phys., 34, 2494–2495.
Peterson, R.C., Lager, G.A. and Hitterman, R.L. (1991) A time-of-flight powder on fraction study of MgAl2O4 at
temperature up to 1273 K. Am. Mineral., 76, 1455–1458.
Peterson, E., Horz, F. and Chang, S. (1997) Modification of amino acids at shock pressures of 3.5 to 32 GPa.
Science, 61(18), 3937– 3950.
Petersons, H.F. and Constable, S. (1996) Global mapping of the electrically conductive lower mantle. Geophys.
Res. Lett., 23(12), 1461–1464.
Peyonneau, J. and Poirier, J.P. (1991) J. Geophys. Res., 96, 6113.
Pfleiderer, C., Uhlarz, M., Hayden, S.M., Vollmer, R., Löhneysen, H.v., Bernhoeft, N.R., Lonzarich, G.G., (2001)
Coexistence of superconductivity and ferromagnetism in the d-band metal ZrZn2. Nature, 412, 58 –61.
Phillips, R. and Hansen (1998) Nature, 279, 1492–1494.
Phillips, B.L. and Kirkpatrick, R.J. (1995) High-temperature 29Si MAS-NMR spectroscopy of anorthite
(CaAl2Si2O8) and its P1 – I 1 structural phase transition. Phys. Chem. Miner., 22, 267–275.
Phillips, B.L., Kirkpatrick, R.J. and Carpenter, M.A. (1992) Investigation of short-range Al, Si order in synthetic
anorthite by 29Si MAS-NMR spectroscopy. Am. Mineral., 77, 484–494.
Phillips, B.L., Thompson, I.G., Xiao, Y. and Kirkpatrick, R.J. (1993) Constrains on the structure and dynamics of
the b-cristobalite polymorphs of SiO2 and AlPO4 from 31P, 27Al and 29Si NMR spectroscopy to 770 K.
Phys. Chem. Miner., 20, 341 –352.
Phillips, B.L., Burnley, P.C., Worminghaus, K. and Navtrosky, A. (1997) 29Si and 1H NMR spectroscopy of
high-pressure hydrous magnesium silicates. Phys. Chem. Miner., 24, 179–190.
Pickett, W.E., Cohen, R.E. and Krakauer, H. (1991) Lattice instabilities, isotope effect, and high-Tc
superconductivity in La22xBaxCuO4. Res. Rev. Lett., 67, 228–231.
Pierazzo, E., Vickery, A.M. and Melosh, H.J. (1997) Icarus, 127, 408.
Piermarini, G.J., Block, S. and Barnett, J.D. (1973) Hydrostatic limits in liquids and solids to 100 kbar. J. Appl.
Phys., 44, 5377–5382.
Piermarini, G.J., Block, S., Barnett, J.D. and Forman, R.A. (1975) Calibration of the pressure dependence of the R,
ruby fluorescence line to 195 kbar. J. Appl. Phys., 46, 2774–2778.
Piskoti, C., Yarger, J. and Zettl, A. (1998) C36, a new carbon solid. Nature, 393, 771– 774.
1172 References

Plymate, T.G. and Stout, J.H. (1994) Pressure–volume–temperature behaviour of g-Fe2SiO4(spinel) based on
static compression measurements at 4008C. Phys. Chem. Miner., 21(6), 413–420.
Poe, B. and Rubie, D. (2000) Transport properties of silicate melts at high pressure. In: Aoki, H., Syono, Y.
and Hemley, R.J. (eds.), Physics Meets, Mineralogy, Cambridge University Press,
340–353.
Poe, B., McMillan, P.F., Cote, B., Massiot, D. and Coutures, J.P. (1992) J. Phys. Chem., 96, 8220.
Poe, B.T., Xu, Y. and Rubie, D.C. (1998) Electrical conductivity of mantle minerals: in situ high pressure high
temperature complex impedance spectroscopy. In: Nakahara, M. (ed.), Review of High Pressure Science
and Technology, 7, 22–24, Japan Society High Pressure Science and Technology, Kyoto.
Poirer, J.P. (1986) Dislocation-mediated melting of iron and the temperature of the Earth’s core. Geophys. J. R.
Astron. Soc., 85, 315–328.
Poirier, J.P. (1981) Martinsitic olivine–spinel transformation and plasticity of the mantle transition zone.
In: Paterson, M.S. and Nicolas, A. (eds.), Anelasticity in the Earth, Geodynamics Series 4.
Washington, DC: AGU, pp. 113 –117.
Poirier, J.P. (1985) Creep of Crystals. Cambridge: Cambridge University Press.
Poirier, J.P. (1991) Introduction to the Physics of the Earth’s Interior. Cambridge: Cambridge University Press.
Poirier, J.P. (1994) Light elements in the Earth’s outer core: a critical review. Phys. Earth Planet. Interiors,
85, 319 –337.
Poirier, J.P. (1982) On transformation plasticity. J. Geophys. Res., 87, 6791–6797.
Poirier, J.P. and Duba, A.G. (1997) On power-law kinetics of transport phenomena in minerals. Phys. Chem.
Miner., 24, 495–499.
Poirier, J.P. and Peyronneau, J. (1992) In: Syono, Y. and Manghnani, M.H. (eds.), High-Pressure Research:
Application to Earth and Planetary Sciences. Tokyo/Washington, DC: Terra Scientific/AGU,
pp. 77 –87.
Poirier, J.-P. and Price, G.D. (1999) Primary slip system of e-iron and anisotropy of the Earth’s inner core. Phys.
Earth Planet. Interiors, 110, 147 –156.
Poirier, J.P. and Tarantola, A. (1998) A logarithmic equation of state. Phys. Earth Planet. Interiors,
109, 1 –8.
Poirier, J.P., Peyronneau, J., Gesland, J.Y. and Brebec, G. (1983) Viscosity and conductivity of the lower
mantle, an experimental study on a MgSiO3 perovskite analog KZnFe3. Phys. Earth Planet. Interiors,
32, 273 –287.
Poirier, J.P., Goddat, A. and Peyronneau, J. (1996) Ferric iron dependence of the electrical conductivity of the
Earth’s lower mantle material. Phil. Trans. R. Soc. Lond. A, 354, 1361–1369.
Poli, S. and Schmidt, M.W. (1995) H2O transport and release in subduction zones: experimental constraints on
basaltic and andesitic systems. J. Geophys. Res., 100, 22299–22314.
Poli, S. and Schmidt, M.W. (1998) The high pressure stability of zoisite and phase relationships of zoisite-bearing
assemblages. Contrib. Mineral. Petrol., 130, 162–175.
Polian, A., Grimsditch, M. and Philippot, E. (1993) Phys. Rev. Lett., 71, 3143.
Poole, P.H., Sciortino, E., Grande, T., Stanley, H.E. and Angell, C.A. (1994) Phys. Rev. Lett., 73,
1632– 1635.
Popp, R.K., Virgo, D. and Phillips, M.W. (1995a) H deficiency in kaersutitic amphiboles: experimental
verification. Am. Mineral., 80, 1347–1350.
Popp, R.K., Virgo, D., Yoder, H.S., Hoering, T.C. and Phillips, M.C. (1995b) An experimental study of phase
equilibria and Fe oxy-component in kaersutitic amphibole: implications for the fH2 and aH2O in the upper
mantle. Am. Mineral., 80, 534–548.
Porcelli, D.R. and Wasserburg, G.J. (1995) Geochim. Cosmochim. Acta, 59, 4921–4937.
Prawer, S., Smith, T.F. and Finlayson, T.R. (1985) Aus. J. Phys., 38, 63.
Prewitt, C.T. and Downs, R.T. (1998) High-pressure crystal chemistry. In: Hemley, R.J. (ed.), Ultrahigh Pressure
Mineralogy, Reviews in Mineralogy. Washington, DC: Mineralogical Society of America, pp. 283–317,
Chapter 9.
References 1173

Prewitt, C.T. and Finger, L.W. (1992) Crystal chemistry of high pressure hydrous magnesium silicates. In:
Syono, Y. and Manghmami, M.H. (eds.), High-Pressure Research: Application to Earth and
Planetary Sciences. Washington, DC: AGU, pp. 269–274.
Prewitt, C.T. and Parise, J.B. (2000) Hydrous phases and hydrogen bonding at high pressure. In: Hazen, R.M.
(ed.), Crystal Chemistry at High Pressures and Temperatures, Reviews in Mineralogy, Vol. 41.
Washington, DC: Mineralogical Society of America, pp. 309–334.
Prewitt, C.T. and Sleight, A.W. (1969) Garnet-like structures of high-pressure cadmium germanate and calcium
germanate. Science, 163, 386–387.
Prewitt, C.T., Gramsch, S.A. and Fei, Y. (2002) High-pressure crystal chemistry of nickel sulphides. J. Phys.
Condens. Matter, 14, 11411–11415.
Pruzzan, P.J. (1994) Mol. Struct., 322, 279–286.
Putirka, K.J. (1999) Melting depths and mantle heterogeneity beneath Hawaii and the East Pacific Rise:
Constraints from Na/Ti and rare earth elements ratios. J. Geophys. Res., 104(B2), 2817–2830.
Raffo, P.L. and Less, J. (1969) Common Met., 17, 133.
Ralph, R.L., Finger, L.W., Hazen, R.M. and Ghose, S. (1984) Compressibility and crystal structure of andalusite
at high pressure. Am. Mineral., 69, 513 –519.
Ramamurthy, V. (1991) Science, 253, 303.
Ramasesha, S.K., Mohan, M., Singh, A.K. and Rao, C.N.R. (1994) High pressure study of Fe3O4 through the
verwey transition. Phys. Rev., 50, 13789–13791.
Ramdas, A.K., Rodriguez, S., Grimsditch, M., Anthony, T.R. and Banholzer, W.F. (1993) Effect of isotopic
constitution of diamond on its elastic constants 14C diamond, the hardest known material. Phys. Rev.
Lett., 71(1), 189 –192.
Ramirez, A.P. (1997) J. Phys. Condens. Matter, 9, 8171–8199.
Ramirez, A.P. (1999) Cohabitation in the cuprates. Nature, 399, 527–528.
Ramirez, A.P., Schiffer, P., Cheong, S.W., Chen, C.H., Bao, W., Palstra, T.T.M., Gammel, P.L., Bishop, D.J. and
Zegarski, B. (1996) Phys. Rev. Lett., 76, 3188.
Ramirez, A.P., Hayashi, A., Cava, R.I., Siddharthan, R. and Shastry, B.S. (1999) Nature, 399, 333–335.
Ranalli, G. (1995) Rheology of the Earth, 2nd ed. London: Chapman & Hall.
Ransom, B. and Helseson, H. (1995) A chemical and thermodynamic model of dioctahedral 2:1 layer clay
minerals in diagenetic processes: dehydration of dioctahedral aluminous smectite as a function of
temperature and depth in sedimentary basins. Am. J. Sci., 295, 245–281.
Rao, C.N.R. and Gopalakrishnan, J. (1986) New Directions in Solid State Chemistry. Cambridge: Cambridge
University Press, 515 pp.
Rapp, R.P., Watson, E.B. and Miller, C.F. (1991) Partial melting of amphibolite/eclogite and the origin of
Archean trodhjemites and tonalites. Precambr. Res, 51, 1–125.
Rateb, M., Abu, A. and Burns, R.G. (1976) The effect of pressure on the degree of covalency of the cation–
oxygen bond in minerals. Am. Mineral., 61, 391– 397.
Raterron, P., Béjina, F., Doukhan, J.C., Jaoul, O. and Libermann, R.C. (1998) Olivine/Fe-metal equilibrium under
high pressure: an ATEM investigation. Phys. Chem. Miner., 25, 485–493.
Ray, P.C. (1923) Proc. R. Soc. Lond. Ser. A, 102, 640.
Reddy, K.P.R., Oh, S.M. et al. (1980) Oxygen diffusion in forsterite. J. Geophys. Res., 85, 322–326.
Redfern, S.A.T. (1992) The effect of Al/Si disorder on the I 1 – P1 co-elastic phase transition in Ca-rich
plagioclase. Phys. Chem. Miner., 19, 246 –255.
Redfern, S.A.T. (1996) Length-scale dependence of high-pressure amorphization: the static amorphization of
anorthite. Mineral. Mag., 60(3), 493– 498.
Redfern, S.A.T. and Angel, R.J. (1999) High-pressure behaviour and equation of state of calcite, CaCO3. Contrib.
Mineral. Petrol., 134, 102–106.
Redfern, S.A.T. and Salje, E. (1987) Thermodynamics of plagioclase temperature evolution of the spontaneous
strain at the I 1 – P1 phase transition in anorthite. Phys. Chem. Miner., 14, 189–192.
Redfern, S.A.T. and Salje, E. (1992) Microscopic dynamic and macroscopic thermodynamic character of the
I 1 – P1 phase transition in anorthite. Phys. Chem. Miner., 18, 526–533.
1174 References

Redfern, S.A.T. and Wood, B.J. (1992) Thermal expansion of brucite, Mg(OH)2. Am. Mineral., 77,
1129– 1132.
Redfern, S.A.T., Salje, E. and Navortsky, A. (1989) High-temperature enthalpy at the orientational order–
disorder transition in calcite: implications for the calcite/aragonite phase equilibrium. Contrib. Mineral.
Petrol., 101, 479–484.
Redfern, S.A.T., Clark, S.M. and Henderson, C.M.B. (1993a) High-pressure phase transition in gillespite:
new evidence from energy-dispersive diffraction. Mater. Res. Forum, 133– 136, 615– 620.
Redfern, S.A.T., Wood, B.J. and Henderson, C.M.B. (1993b) Static compressibility of magnesite to 20 GPa:
implications for MgCO3 in the lower mantle. Geophys. Res. Lett., 20, 2099–2120.
Redfern, S.A.T., Wood, B.J. and Henderson, C.M.B. (1993c) Static compression of magnesite to 20 GPa:
implications for MgCO3 ilmenite. Am. Mineral., 81, 45– 50.
Reichlin, R., McMahan, A.K., Ross, M., Martin, S., Hu, J., Hemley, R.J., Mao, H.K. and Wu, Y. (1994) Phys. Rev.
B, 49, 3725–3733.
Reid, A.F. and Ringwood, A.E. (1969a) Earth Planet. Sci. Lett., 6, 205– 208.
Reid, A.F. and Ringwood, A.E. (1969b) Six-coordinate silicon: high pressure strontium and barium
aluminosilicates with the hollandite structure. J. Solid State Chem., 1, 6–9.
Reinecke, T., Vander Klauw, S. and Stockhert, B. (1992) Ultrahigh-pressure metamorphism of oceanic crust in
the western Alps. 29th Int. Geol. Cong., Kyoto Japan Abstr., 2, 599.
Reisberg, L.C., Allegre, C.J. and Luck, J.M. (1991) The Re–Os systematics of the Ronda ultramafic complex of
southern Spain. Earth Planet. Sci. Lett., 105, 196–213.
Reisberg, L. and Lorand, J.P. (1995) Nature, 376, 159–162.
Renner, B. and Lehmann, G. (1986) Correlation of angular and bond length distortions in TO4 units in crystals.
Z. Kristallogr., 175, 43–59.
Revenaugh, J.S. and Meyer, R. (1997) Seismic evidence of partial melt within a possibly ubiquitous low velocity
layer at the base of the mantle. Science, 277, 670–673.
Reynard, B. and Guyot, F. (1994) High-temperature properties of geikielite (MgTiO3 – ilmenite) from high-
temperature high-pressure Raman spectroscopy some implications for MgSiO3 –ilmenite. Phys. Chem.
Miner., 21, 441–450.
Reynard, B. and Rubie, D.C. (1996) High-pressure, high-temperature Raman spectroscopic study of ilmenite-type
MgSiO3. Am. Mineral., 81, 1092–1096.
Reynard, B., Petit, P., Guyot, F. and Gillet, P. (1994) Pressure-induced structural modifications in Mg2GeO4 –
olivine: a Raman spectroscopic study. Phys. Chem. Miner., 20(8), 556–562.
Reynard, B., Fiquet, G., Itie, J. and Rubie, D.C. (1996a) High-pressure X-ray diffraction study and equation of the
state of MgSiO3 ilmenite. Am. Mineral., 81(1), 45 –50.
Reynard, B., Takir, F., Guyot, F., Gwanmesia, G.D., Liebermann, R.C. and Gillet, P. (1996b) High-temperature
Raman spectroscopic and X-ray diffraction study of b-Mg2SiO4: insights into its high-temperature
thermodynamic properties and the b- to a-phase transformation mechanism and kinetics. Am. Mineral.,
80, 585 –594.
Ribbe, P.H. (1980) The humite series and Mn-analogs. In: Ribbe, P.H. (ed.), Orthosilicates, Reviews in
Mineralogy, Vol. 5. Washington, DC: Mineralogical Society of America, pp. 231–272.
Rice, S.B., Benimoff, A.L. and Sclar, C.B. (1989) “3.65 Å Phase” in system MgO–SiO2 –H2O at pressure greater
than 90 kbar: crystallo-chemical implications for mantle phase (abstract). Abstr. Int. Geol. Cong.,
Washington, DC, Meet, 694–695.
Richardson, C.F. and Ashcroft, N.W. (1997) High temperature superconductivity in metallic hydrogen: Electron-
electron enhancements. Phys. Rev. Lett., 78, 118.
Richet, P. (1988) Superheating, melting and vitrification of high pressure minerals. Nature, 331, 56–58.
Richet, P. and Gillet, P. (1997) Eur. J. Mineral., 9, 907.
Richet, P., Mao, H.K. and Bell, P.M. (1989) Static compression and equation of state of CaO to 1.35 Mbar.
J. Geophys. Res., 93, 15279–15288.
Richet, P., Gillet, P., Ali Bouhfid, M., Daniel, I. and Fiquet, G. (1993) A versatile heating stage for measurements
up to 2700 K with applications to phase relationships determinations, Raman spectroscopy and X-ray
diffraction. J. Appl. Phys., 73, 5441–5451.
References 1175

Richter, F.M. and McKenzie, D.P. (1981) On some consequences and possible causes of layered mantle
convection. J. Geophys. Res., 86, 6133–6142.
Ridgen, S.M., Ahrens, T.J. and Stolper, E.M. (1989) High-pressure equation of state of molten anorthite and
diopside. J. Geophys. Res., 94, 9508–9522.
Riedel, M.R. and Karato, S. (1997) Grain-size evolution is subducted oceanic lithosphere associated with
the olivine-spinel transformation and its effects on rheology. Earth Planet. Sci. Lett., 148,
27–44.
Rigden, S.M., Stolper, E.M. and Ahrens, T.J. (1988) Shock compression of molten silicates: results for model
basaltic composition. J. Phys. Res., 93, 367– 382.
Rigden, S.M., Gwanmesia, G.D., FitzGerald, J.D., Jackson, I. and Liebermann, R.C. (1991) Spinel elasticity and
structure of the transition zone of the mantle. Nature, 354, 143–145.
Rigden, S.M., Gwanmesia, G.D., Jackson, I. and Libermann, R.C. (1992) In: Syono, Y. and Maghnani, M. (eds.),
High Pressure Research: Applications to Earth and Planetary Sciences. Washington, DC: AGU,
pp. 167 –182.
Rigden, S.M., Gwanmesia, G.D. and Liebermann, R.C. (1994) Elastic wave velocities of a pyrope–majorite
garnet to 3 GPa. Phys. Earth Planet. Interiors, 86, 35–44.
Righter, K., Drake, M. and Yaxley, G. (1997) Phys. Earth Planet. Interiors, 100, 115–134.
Rimstidt, J.D., Chermak, J.A. and Gagen, P.A. (1994) Rates of reaction of galena, sphalerite, chalcopyrite, and
arsenopyrite with Fe3þ in acidic solutions. In: Alpers, C.N. and Blowes, D.W. (eds.), Environmental
Geochemistry of Sulfide Oxidation. Washington, DC: American Chemical Society, 681 pp.
Ringwood, A.E. (1975) Composition and Petrology of the Earth’s Mantle. New York: McGraw-Hill.
Ringwood, A.E. (1977) Composition of the core and implications for origin of the earth. Geochem. J.,
11, 111–135.
Ringwood, A.E. (1979) Origin of the Earth and Moon. New York: Springer.
Ringwood, A.E. (1984) The Earth core: its composition, formation and bearing upon the origin of the Earth. Proc.
R. Soc. Lond. A, 395, 1– 46.
Ringwood, A.E. (1991) Phase transformations and their bearing on the constitution and dynamics of the mantle.
Geochim. Cosmochim. Acta, 55, 2083–2110.
Ringwood, A.E. (1992) Volatile and siderophile element geochemisry of the Moon: a reappraisal. Earth Planet.
Sci. Lett., 11, 537 –555.
Ringwood, A.E. (1994) Role of the transition zone and 600 km-discontinuity in mantle dynamics. Phys. Earth
Planet. Interiors, 86, 524.
Ringwood, A.E. and Hibberson, W. (1990) The system Fe –FeO revisited. Phys. Chem. Miner., 17, 313–319.
Ringwood, A.E. and Hibberson, W. (1991) Solubilities of mantle oxides in molten iron at high pressures and
temperatures: implications for the composition and formation of Earth’s core. Earth Planet. Sci. Lett.,
102, 235 –251.
Ringwood, A.E. and Irifune, T. (1988) Nature of the 650-km discontinuity: implications for mantle dynamics and
differentiation. Nature, 331, 131–136.
Ringwood, M. and Major, A. (1967) High-pressure reconnaissance investigations in the system Mg2SiO4 –MgO–
H2O. Earth Planet. Sci. Lett., 2, 130–133.
Ringwood, A.E. and Major, A. (1970) The system Mg2SiO4 –Fe2SiO4 at high pressures and temperatures. Phys.
Earth Planet. Interiors, 3, 89 –108.
Ringwood, A.E. and Major, A. (1971) Synthesis of majorite and other high pressure garnets and perovskites.
Earth Planet. Sci. Lett., 12, 411 –418.
Ringwood, A.E. and Reid, A.E. (1969) High pressure transformations of spinels (I). Earth Planet. Sci. Lett.,
5, 245 –250.
Ringwood, A.E., Reid, A.F. and Wadsley, A.D. (1967) High-pressure KAlSi3O8, an aluminosilicate with sixfold
coordination. Acta Crystallogr., 23, 1093–1095.
Ringwood, A.E., Kesson, S.E., Ware, N.G., Hibberson, W. and Major, A. (1979) Immobilization of high level
nuclear reactor wastes in Synroc. Nature, 278, 219– 223.
Ringwood, A.E., Kato, T., Hibberson, W. and Ware, N. (1990) High pressure geochemistry of Cr, V, and Mn and
implications for origin of the Moon. Nature, 347, 72–76.
1176 References

Robbins, M., Wertheim, G.K., Sherwood, R.C. and Buchanan, D.N.E. (1971) Magnetic properties and site
distributions in the system FeCr2O4 – Fe3O4 – (Fe2þCr(22x)Fe3þ x O4). J. Phys. Chem. Solids, 32,
717–729.
Robert, F. (2001) The origin of water on Earth. Science, 293, 1056–1059.
Robert, J.J. and Tyburczy, J.A. (1993) Impedance spectroscopy of single and polycrystalline olivine: evidence for
grain boundary transport. Phys. Chem. Miner., 20, 19–26.
Robert, J.L., Hardy, M. and Sanz, J. (1995) Excess protons in synthetic micas with tetrahedrally coordinated
divalent cations. Eur. J. Mineral., 7, 457–461.
Robertson, G. and Woodhouse, J. (1995) Geophys. J. Int., 123, 85.
Robie, R.A., Hemingway, B.S. and Fisher, J.R. (1978) Thermodynamic properties of minerals and related
substances at 298.15 K and 1 bar (105 pascals) pressure and at higher temperatures. US Geol. Surv. Bull.,
1452, 456.
Robinson, K., Gibbs, G.V. and Ribbe, P.H. (1971) Quadratic elongation: a quantitative measure of distortion in
coordination polyhedra. Science, 172, 567– 570.
Roeloffs, E. (1999) Radon and rock deformation. Nature, 399, 104–105.
Romanowicz, B., Li, X.D. and Durek, J. (1996) Anisotropy in the inner core: could it be due to low-order
convection? Science, 274, 963– 996.
Ross, N.L. (1997) The equation of state and high-pressure behaviour of magnesite. Am. Mineral., 82,
682–688.
Ross, N.L. and Hazen, R.M. (1989) Phys. Chem. Miner., 16, 415.
Ross, N.L. and Hazen, R.M. (1990) High-pressure crystal chemistry of MgSiO3 perovskite. Phys. Chem. Miner.,
17, 228 –237.
Ross, N.L. and Reeder, R.J. (1992) High-pressure structural study of dolomite and ankerite. Am. Mineral.,
77, 412 –421.
Ross, R.G., Anderson, P., Sundqvist, B. and Bäckström, G. (1984) Rep. Prog. Phys., 47, 1347.
Ross, N.L., Shu, J.F. and Hazen, R.M. (1990) High-pressure crystal chemistry of stishovite. Am. Miner.,
75, 739–747.
Rossman, G.R. (1988) Optical Spectroscopy, Reviews in Mineralogy, Vol. 18. Washington, DC: Mineralogical
Society of America, pp. 207–254.
Rossman, G.R. (1996) Studies of OH in nominally anhydrous minerals. Phys. Chem. Miner., 23,
299–304.
Rotter, C.A. and Smith, C.S. (1966) Ultrasonic equation of state of iron at low pressure, room temperature. J. Phys.
Chem. Solids, 27, 267– 276.
Rozenberg, G.K., Hearne, G.R., Pasternak, M.P., Metcalf, P.A. and Honig, J.M. (1996) Nature of the Verwey
transition in magnetite (Fe3O4) to pressures of 16 GPa. Phys. Rev., 53, 6482– 6487.
Rozenberg, G.K., Pasternak, M.K., Hearne, G.R. and McCammon, C.A. (1997) High pressure and electronic-
magnetic properties of hexagonal cubanite (CuFe2S3). Phys. Chem. Miner., 24(7), 569–573.
Rubie, D.C. (1984) The olivine–spinel transformation and the rheology of the subducting lithosphere. Nature,
308, 305.
Rubie, D.C. (1990) In: Berber, D.J. and Meredith, P.G. (eds.), Deformation Process in Minerals, Ceramics, and
Rocks. London: Unwin Hyman, 262 pp.
Rubie, D.C. and Brearley, A.J. (1994) Phase transitions between b and g (Mg,Fe)2SiO4 in the Earth’s mantle:
mechanisms and rheological implications. Science, 264, 1445–1448.
Rubie, D.C. and Ross, C.R. II (1994) Kinetics of the olivine–spinel transformation in subducting lithosphere:
experimental constraints and implications for deep slab processes. Phys. Earth Planet. Interiors,
86, 223–241.
Rubie, D.C. and Thompson, A.B. (1985) Metamorphic reactions: Kinetics, texture and deformation. In:
Thompson, A.B. and Rubie, D.C. (eds.), Advances in Physical Geochemistry, Springer-Verlag, 27 –79.
Rubo, R. and Ohata, A. (1972) A quantum theory of double exchange. Phys. Soc. Jpn, 33, 21–32.
Rueff, J.P., Kao, C.C. et al. (1999) Pressure-induced high-spin to low-spin transition in FeS evidenced by X-ray
emission spectroscopy. Phys. Rev. Lett., 82(16), 3284–3287.
References 1177

Rumble, D. (1998) Stable isotope geochemistry of ultrahigh-pressure rocks. In: Hacker, B.R. and Liou, J.G. (eds.),
Geodynamics and Geochemistry of Ultrahigh-Pressure Rocks. Dordrecht: Kluwer, pp. 241–259.
Rumble, D. and Yui, T.F. (1998) The Qinglongshan oxygen and hydrogen isotope anomaly near Donghai in
Jiangsu Province, China. Geochim. Cosmochim. Acta, 62, 3307–3321.
Runcorn, S.K. (1967) Flow in the mantle inferred from the low degree harmonics of the geopotential. Geophys. J.
R. Astron. Soc., 14, 375.
Ruoff, A.L. (1965) Mass transfer problem in ionic crystals with charge neutrality. J. Appl. Phys., 36,
2903– 2907.
Ruoff, A.L., Lincoln, R.C. and Chen, Y.C. (1973) A new method of absolute high pressure determination. J. Phys.
D: Appl. Phys., 6, 1295–1306.
Ruoff, A.L., Xia, H., Luo, H. and Vohra, Y.K. (1990) Rev. Sci. Instrum., 61, 3830.
Ruoff, A.L., Luo, H. and Vohra, Y.K. (1991) J. Appl. Phys., 69, 6413– 6416.
Ruoff, A.L., Xia, H. and Xia, Q. (1992) Rev. Sci. Instrum., 63, 4342– 4348.
Russell, C.T. (1993) J. Geophys. Res., 98, 18681.
Russell, J.H. and Mao, H.K. (1997) Solids, static high-pressure effects. Encyclopedia Appl. Phys., 18,
555–570.
Russell, J., Hemley, R.J. and Cohen, R.E. (1992) Silicate perovskite. Annu. Rev. Earth Planet. Sci., 20, 553–600.
Russell, S.A., Lay, T. and Garnero, E.J. (1998) Nature, 396, 255–258.
Russo, M. and Silver, G. (1996) Cordillera formation, mantle dynamics, and the Wilson cycle. Geology, 24(6),
511–514.
Sacerdoti, M. and Passaglia, E. (1985) The crystal structure of katoite and implications within the hydrogrossular
group of minerals. Bull. Mineral., 108, 1–8.
Sack, R.O. and Ghiorso, M.S. (1991) Chromian spinels as petrogenetic indicators: thermodynamics and
petrological applications. Am. Mineral., 76, 827 –847.
Safronov, V.S. and Vitjazev, A.V. (1986) The origin and evolution of the terrestrial planets. In: Saxena, S.K. (ed.),
Chemistry and Physics of the Terrestrial Planets. New York: Springer, pp. 1–29.
Saitoh, T., Bocquet, A., Mizokawa, T. and Fujimori, A. (1999) Systematic variation of the electronics structures of
3rd transition metal oxides. Phys. Rev. B, 52, 7934–7938.
Sales, B.C., Mandrus, D. and Williams, R.K. (1997) Thermoleastic materials New Approaches and Advances,
Mater. Res. Soc. Proc. 47B.
Salje, E.K.H. (1989) Characteristics of perovskite-related materials. Phil. Trans. R. Soc. Lond. A, 328,
409–416.
Salje, E.K.H. (1990) Phase Transitions in Ferroelastic and Co-elastic Crystals. Cambridge: Cambridge
University Press, 366 pp.
Salje, E.K.H. (1992a) Application of Landau theory for the analysis of phase transitions in minerals. Phys. Rep.,
215, 49 –99.
Salje, E.K.H., Wrock, B., Graeme-Barber, A. and Carpenter, M.A. (1993) Experimental test of rate equations:
time evolution of Al, Si ordering in anorthite CaAl 2Si2O8. J. Phys. Condens. Matter, 5,
2961– 2968.
Samara, G.A. (1969) Effects of pressure on the magnetic properties of magnetite. Bull. Am. Phys. Soc., 14, 308.
Samara, G.A. and Giardini, A.A. (1969) Effect of pressure on the Néel temperature of magnetite. Phys. Rev.,
186, 577–580.
Sankaran, A.V. (1997) Subducting sea floors bridge the mantle divide. Curr. Sci., 73(11), 901–903.
Sanloup, C., Guyot, F., Gillet, P. et al. (2000) Structural changes in liquid Fe at high pressures and high
temperatures from synchrotron X-ray diffraction. Europhys. Lett., 52(2), 151–157.
Sanloup, C., Mao, H.K. and Hemley, R.J. (2002) High-pressure transformations in xenon hydrates. PNAS, 99(1),
25–28.
Sarda, P., Staudacher, T. and Allègre, C.J. (1998) Earth Planet. Sci. Lett., 91, 73.
Sasaki, S., Prewitt, C.T., Sato, Y. and Ito, E. (1982a) Single-crystal X-ray study of g-Mg2SiO4. J. Geophys. Res.,
87, 7829–7832.
Sasaki, S., Takeuchi, Y., Fujino, K. and Akimoto, S. (1982b) Electron-density distributions of three
orthopyroxenes, Mg2Si2O6, Co2Si2O6, and Fe2Si2O6. Z. Kristallogr., 158, 279–297.
1178 References

Sato, Y., Akaogi, M. and Akimoto, S. (1978) Hydrostatic compression of the synthetic garnets pyrope and
almandine. J. Geophys. Res., 83, 335–338.
Sat-Sorensen, Y. (1983) Phase transitions and equations of state for the sodium halides: NaF, NaBr, and NaI.
J. Geophys. Res., 88, 3543–3548.
Sato, R.H., McMillan, P.F., Dennison, P. and Dupree, R. (1991) J. Phys. Chem., 95, 4483.
Saul, A. and Wagner, W. (1989) J. Phys. Chem. Ref. Data, 18, 1537.
Saumon, D. and Chabrier, G. (1992) Fluid hydrogen at high density: Pressure ionization. Phys. Rev. A, 46,
2084– 2100.
Saumon, D. et al. (1996) Astrophys. J., 460, 993.
Sautter, V., Haggerty, S.E. and Field, S. (1991) Ultradeep (.300 km) ultramafic xenoliths: petrological evidence
from the transition zone. Science, 252, 827–830.
Sawamoto, H. (1977) Orthorhombic perovskite (Mg1Fe)SiO3 and constitution of the lower mantle. In: Manghnani,
M.H. and Akimoto, S. (eds.), High-Pressure Research Applications in Geophysics. New York: Academic
Press, pp. 219–244.
Sawamoto, H. (1987) Phase diagram of MgSiO3 at pressures up to 24 GPa and temperatures up to 22008C: phase
stability and properties of tetragonal garnet. In: Manghnani, M.H. and Syono, Y. (eds.), High-Pressures
in Mineral Physics. Washington, DC: AGU, pp. 209–219.
Sawamoto, H. and Horiuchi, H. (1990) b-(Mg0.9Fe0.1)2SiO4: single crystal structure, cation distribution, and
properties of coordination polyhedra. Phys. Chem. Miner., 17, 293–300.
Sawamoto, H., Weidner, D.J., Sasaki, S. and Kumazawa, M. (1984) Science, 224, 749–751.
Saxena, S.K. and Dal Negro, A. (1983) Petrogenetic application of Mg–Fe2þ order–disorder in orthopyroxene to
the cooling history of rocks. Bull. Mineral., 106, 443 –449.
Saxena, S.K. and Dubrovinsky, L.S. (2000) Iron phases at high pressures and temperatures: phase transition and
melting. Am. Mineral., 85, 372 –375.
Saxena, S.S. and Littlewood, P.B. (2001) Iron cast in exotic role. Nature, 412, 290–291.
Saxena, S.K. and Shen, G. (1992) Assessed data on heat capacity, thermal expansion and compressibility for some
oxides and silicates. J. Geophys. Res., 97, 19813–19826.
Saxena, S.K., Chatterjee, N., Fei, Y. and Shen, G. (1993a) Thermodynamic Data on Oxides and Silicates.
New York: Springer, 428 pp.
Saxena, S.K., Shen, G. and Lazor, P. (1993b) Experimental evidence for a new iron phase and implications for
Earth’s core. Science, 260, 1312–1314.
Saxena, S.K., Dubrovinsky, L.S., Häggkvist, P., Cerenius, Y., Shen, G. and Mao, H.K. (1995) Synchrotron X-ray
study of iron at high pressure and temperature. Science, 269, 1703–1704.
Saxena, S.K., Dubrovinsky, L.S., Lazor, P., Cerenius, Y., Häggkvist, P., Hanfland, M., Hu, J. (1996a) Stability of
Perovskite (MgSiO3) in the earth’s mantle. Science, 274, 1357–1359.
Saxena, S.K., Dubrovinsky, L.S. and Häggqvist, P. (1996b) X-ray evidence for the new phase b-iron at high
temperature and high pressure. Geophys. Res. Lett., 23, 2441–2444.
Saxena, S.K., Dubrovinsky, L.S., Lazor, P. and Hu, J. (1998) In situ X-ray study of perovskite (MgSiO3): phase
transition and dissociation at mantle conditions. Eur. J. Mineral., 10, 1275–1281.
Saxena, S.K., Dubrovinsky, L.S., Tutti, F. and Bihan, T.L. (1999a) Equation of state of perovskite (MgSiO3) based
on experimentally measured data. Am. Mineral., 48, 226–232.
Saxena, S.K., Dubrovinsky, L.S., Tutti, F. and Bihan, T.L. (1999b) Equation of state of perovskite (MgSiO3)
based on experimentally measured data. Am. Mineral., 84, 226–232.
Scambelluri, M., Multener, O., Hermann, J., Piccardo, G.B. and Trommsdorff, V. (1995) Subduction of water into
mantle: history of an alpine peridotite. Geology, 23, 459–462.
Scarfe, C.M., Mysen, B.O. and Rai, C.S. (1979) Invariant melting behaviour of mantle material: partial melting
of two lherzolite nodules. Carnegie Institute Washington Yearbook, Vol. 78, pp. 498–501.
Schäfer, H. et al. (1983) Phys. Chem. Miner., 9, 248–252.
Scharmli, G.H. (1982) In: Schereyer, W. (ed.), High Pressure Research in Geosciences. Stuttgart:
Schwizerbartsche, pp. 349– 373.
Schenk and Bulmer (1998) Nature, 279, 1514–1515.
References 1179

Schiano, P., Clocchiatti, N., Shimizu, N., Maury, R.C., Jochun, K.P. and Hofmann, A.W. (1995) Hydrous,
silica-rich melts in the sub-arc mantle and their relationship with erupted arc lavas. Nature, 377,
595–600.
Schloessin, H.H. and Dvorak (1972) Geophys. J. R. Astron. Soc., 27, 499.
Schloessin, H.H. and Timco, G.W. (1977) The significance of ferroelectric phase transitions for the Earth and
planetary interiors. Phys. Earth Planet. Interiors, 14, 6–12.
Schmalzried, H. (1983) Thermodynamics of compounds with narrow range of stoichiometry. Ber. Bunsenges
Phys. Chem., 87, 726– 733.
Schmidt, M.W. (1995) Lawsonite: upper pressure stability and formation of higher density hydrous phase. Am.
Mineral., 80, 1286–1292.
Schmidt, M.W. (1996) Experimental constraints on recycling of potassium from subducted oceanic crust. Science,
272, 1927–1930.
Schmidt, M.W. and Poli, S. (1994) The stability of lawsonite and joisite at high pressure: experiments in CASH to
92 kbar and implications for the presence of hydrous phases in subducted lithosphere. Earth Planet. Sci.
Lett., 124, 105–118.
Schmidt, M.W. and Poli, S. (1998) Experimental based water budget for dehydration slabs and consequences for
arc magma generation. Earth Planet. Sci. Lett., 163, 361–379.
Schmidt, S.C., Shaner, J.W., Samara, G.A. and Ross, M. (1994) High-Pressure Science and Technology — 1993.
New York: American Institute of Physics.
Schmidt, M.W., Poli, S., Comodi, P. and Zanazzi, P.F. (1997) The high pressure behaviour of kyanite:
decomposition of kyanite into stishovite and corundum. Am. Mineral., 82, 460–466.
Schmidt, M.W., Finger, L.W., Angel, R.J. and Dinnebier, R.E. (1998) Synthesis crystal structure and phase
relations of AlSi3OH a high-pressure hydrous phase. Am. Mineral., 83, 881–888.
Schmidtbauer, E. (1987a) 57Fe Mössbauer spectroscopy and magnetization of cation deficient Fe2TiO4 and
FeCr2O4 Pt. I: 57Fe Mössbauer spectroscopy. Phys. Chem. Miner., 14, 533– 541.
Schmidtbauer, E. (1987b) Fe2TiO4 and FeCr2O4 Pt. II: Magnetization data. Phys. Chem. Miner., 15, 201–207.
Schmidtbauer, E. and Amthauer, G. (1998) Study of the electrical charge transport in ilvaite using impedance
spectroscopy and thermopower data. Phys. Chem. Miner., 25, 522–533.
Schmidtbauer, E., Kunzmann, Tm., Fehr, Th. and Hochleitner, R. (1996) Electrical conductivity, thermopower and
57
Fe Mössbauer spectroscopy of an Fe-rich amphibole, arfvedsonite. Phys. Chem. Miner., 23, 99–106.
Schofield, P.F., Charnock, J.M., Cressey, G. and Henderson, C.M.B. (1994) An EXAFS study of cation site
distortions through the P2=c – P1 transition in the synthetic cuproscheelite–sanmartinite solid solution.
Mineral. Mag., 58, 185–199.
Schopf, W.J. (1993) Science, 260, 640– 646.
Schrauder, M. and Navon, O. (1993) Solid carbon dioxide in a natural diamond. Nature, 365, 42–44.
Schrauder, M. and Navon, O. (1994) Geochim. Cosmochim. Acta, 58, 761–771.
Schreiber, E. and Anderson, O.L. (1966) Pressure derivatives of the sound velocity of polycrystalline alumina.
J. Am. Ceram. Soc., 49, 184– 190.
Schreiber, H.D. and Haskin, L.A. (1976) Chromium in basalts: experimental determination of redox states
and partitioning among synthetic silicate phases, Proceedings of the 7th Lunar Science Conference.
New York: Pergamon Press, pp. 1221– 1259.
Schreyer, W. (1998a) Experimental studies on metamorphism of crystal rocks under mantle pressures. Mineral.
Mag., 51, 1–26.
Schreyer, W. and Baller, T.H. (1981) Calderite, Mn3Fe2Si3O12, a high-pressure garnet. Proc. XII M.A. Meeting,
Novosibrisk. Exp. Mineral., 68– 77.
Schreyer, W., Massone, H. and Chopin, C. (1987) Continental crust subducted to depths near 100 km:
implications for magma and fluid genesis in collision zones. In: Mysen, B. (ed.), Magmatic Processes:
Physiochemical Principles, Spec. Publ. Geochem. Soc., Vol. 1, pp. 155–163.
Schulze, D.J. and Helmstaedt, H. (1998) Coesite–sanidine eclogites from kimberlite: products of mantle
fractionation or subduction. J. Geol., 96, 435 –443.
Schulze, F., Behrens, H., Holtz, F., Roux, J. and Johannes, W. (1996a) The influence of water on the viscosity of a
haplogranitic melt. Am. Mineral., 81, 1155–1165.
1180 References

Schwartz, K.B., Nolet, D.A. and Burns, R.G. (1980) Mössbauer spectroscopy and crystal chemistry of natural Fe–
Ti garnets. Am. Mineral., 65, 142 –153.
Sclar, A.B. (1990) The system MgO–SiO2 –H2O at high pressures and high temperatures (abstract). VM
Goldschmidt Conference; Program and Abstract, 80.
Sclar, C.B. and Monxenti, S.P. (1971) High-pressure synthesis and geophysical significance of a new hydrous
phase in the system MgO–SiO2 –H2O (abstract). Geol. Soc. Am. Abstr. Program, 3, 698.
Sclar, C.B., Carrison, L.C. and Schwatz, C.M. (1965) High pressure synthesis and stability of a new hydronium
bearing layer silicate in the system MgO–SiO2 –H2O (abstract). EOS, 46, 184.
Sclar, C.B., Carrison, L.C. and Stewart, O.M. (1967) High pressure synthesis of a new hydroxylated pyroxene in
the system MgO–SiO2 –H2O (abstract). EOS, 48, 226.
Scott, H.P. and Williams, Q. (1999) An infrared spectroscopic study of lawsonite to 20 GPa. Phys. Chem. Miner.,
26, 437 –445.
Seal, M. (1984) High Temp. High Press., 16, 573–579.
Seifert, F. (1983) Mössbauer line broadening in aluminous orthopyroxenes: evidence for next nearest neighbours
interactions and short-range order. N. Jb. Miner. Abh., 148, 141– 162.
Seifert, F. and Röthlisberger, F. (1993) Macroscopic and structural changes at the incommensurate– normal phase
transition melilites. Mineral. Petrol., 48, 179–192.
Seifert, F., Czank, M., Simons, B. and Schmahl, W. (1987) A commensurate–incommensurate phase
transition in iron-bearing transition in iron-bearing akemanites. Phys. Chem. Miner., 14,
26–35.
Seki, K. and Toyoshima, M. (1998) Preserving tradigrades under pressure. Nature, 395, 853–854.
Sempolinski, D.R. and Kingery, W.D. (1980) Ionic conductivity and magnesium vacancy mobility in magnesium
oxide. J. Am. Ceram. Soc., 63, 664–669.
Serghiou, G.C. and Hammack, W.S. (1993) Pressure-induced amorphization of wollastonite (CaSiO3) at room
temperature. J. Phys. Chem., 98, 9830–9834.
Serghiou, G., Zerr, A. and Boehler, R. (1998) (Mg,Fe)SiO3 –perovskite stability under lower mantle conditions.
Science, 280, 2093.
Shankland, T.J. and Chung, D.H. (1974) General relations among sound speeds. 2. Theory and discussion. Phys.
Earth Planet. Interiors, 8, 121 –129.
Shankland, T.J. and Peyronneau, J.P. (1993) Electrical conductivity of the Earth’s lower mantle. Nature,
366, 453–455.
Shankland, T.J., Duba, A.G. and Woronow, A. (1974) Pressure shifts of optical absorption bands in iron-bearing
garnet, spinel, olivine, and pyroxene. J. Geophys. Res., 79, 3273–3282.
Shankland, T.J., Nitsan, U. and Duba, A.G. (1979) Optical absorption and radiative heat transport in olivine at
high temperature. J. Geophys. Res., 84, 1603–1610.
Shannon, R.D. (1976) Revised C in shock ionic radii and systematic studies of interatomic distances in halides and
chalcogenides. Acta Crystallogr. A, 32, 751–757.
Shanon, R.D. and Prewitt, C.T. (1969) Acta Crystallogr. B, 25, 925–946.
Shapkov, V.P., Tse, J.S., Belosludov, V.R. and Beloslodov, R.V. (1997) Elastic moduli and instability in
molecular crystals. J. Phys. Condens. Matter, 9, 5853–5865.
Sharma, S.K. (1990) Applications of Raman spectroscopy in Earth and planetary sciences. In: Gopalan, K., Gaur,
V.K. and Macdougall, J.D. (eds.), From Mantle to Meteorites. Bangalore: Indian Academy of Sciences,
pp. 263 –308.
Sharma, S. and Matson, D. (1984) Raman spectra and structure of sodium aluminium germanate glasses. J. Non-
Cryst. Solids, 69, 81–96.
Sharma, S. and Sikka, S.W. (1996) Pressure induced amorphization of materials. Prog. Mater. Sci., 40, 1–77.
Sharma, S.K., Mammone, J.F. and Nicol, M.F. (1981) Raman investigation of ring configurations in vitreous
silica. Nature, 292, 140–141.
Sharma, S., Matson, D. et al. (1984) Raman study of the structure of glasses along the join SiO2 –GeO2. J. Non-
Cryst. Solids, 68, 99–114.
Sharp, T.G. and Rubie, D.C. (1995) Catalysis of the olivine to spinel transformation by high clinoenstatite.
Science, 269, 1095–1098.
References 1181

Sharp, T.G., El Goresy, A., Wopenko, B. and Chen, M. (1999) A post-stishovite SiO2 polymorph in the meteorite
Shergotty: implications for impact events. Science, 284, 1511–1513.
Sharp, Z.D., Essene, E.J. and Hunzider, J.C. (1993) Stable isotope geochemistry and phase equilibria of
coesite-bearing white schists, Dora Maira Massif, Western Alps. Contrib. Mineral. Petrol., 114,
1–12.
Shearer, P.M. (1990) Seismic imaging of upper-mantle structure with new evidence for a 520-km discontinuity.
Nature, 344, 121–126.
Shearer, P.M. and Masters, T.G. (1992) Nature, 355, 791.
Shearer, P.M., Toy, K.M. and Orcutt, J.A. (1988) Axi-symmetric Earth models and inner-core anisotropy. Nature,
333, 228 –232.
Shen, A.H., Bassett, W.A. and Chou, I.M. (1992) Hydrothermal studies in a diamond anvil cell: pressure
determination using the equation of state of H2O. In: Syono, Y. and Manghnani, M.H. (eds.), High
Pressure Research: Application to Earth and Planetary Sciences. Tokyo/Washington, DC: Terra
Scientific/AGU, pp. 61–68.
Shen, Y. and Forsyth, D.W. (1995) Geochemical constraints on initial and final depth of melting beneath
mid-oceanic-ridges. J. Geol. Phys., 100, 2211–2237.
Shen, G. and Heinz, D.L. (1998) High-pressure melting of deep mantle and core materials. In: Hemley, R.J. (ed.),
Ultrahigh-Pressure Mineralogy: Physics and Chemistry of the Earth’s Deep Interior, Reviews in
Mineralogy, Mineralogical Society of America, Vol. 37, pp. 369–392.
Shen, A. and Keppler, H. (1995) Infrared spectroscopy of hydrous silicate melts to 10008C and 10 kbar: direct
observation of H2O speciation in a diamond-anvil cell. Am. Mineral., 80, 1335–1338.
Shen, G. and Lazor, P. (1999) Measurement of melting temperatures of some minerals under lower mantle
pressures. J. Geophys. Res., 100, 17699–17715.
Shen, G., Fei, Y., Halenius, U. and Wang, Y. (1994) Optical absorption spectra of (Mg, Fe)SiO3 silicate
perovskites. Phys. Chem. Miner., 20, 478 –482.
Shen, G., Mao, H.K. and Hemley, R.J. (1998a) Melting and crystal structure of iron at high pressures and
temperatures. Geophys. Res. Lett., 25(3), 373–376.
Shen, T., Solomon, S.C., Bjarnason, I.Th. and Wolfe, C.J. (1998b) Nature, 395, 62–65.
Sherghiou, G., Zerr, A., Chopelas, A. and Boehler, R. (1998) The transition of pyrope to perovskite. Phys. Chem.
Miner., 25, 193–196.
Sherman, D.M. (1991) Hartree–Fock band structure, equation of state, and pressure-induced hydrogen bonding in
brucite, Mg(OH)2. Am. Mineral., 76, 1769– 1772.
Sherman, D.M. (1991) J. Geophys. Res., 96, 14299.
Sherman, D.M. (1995) Stability of possible Fe– FeS and Fe–FeO alloy phases at high pressure, and the
composition of the Earth’s core. Earth Planet. Sci. Lett., 132, 87–98.
Sherman, D.M. (1988) High-spin to low-spin transition of iron (II) oxides at high pressures: possible effects on the
physics and chemistry of the lower mantle. In: Ghosh, S., Coey, J.M.D. and Salje, E. (eds.), Structural
and Magnetic Phase Transitions in Minerals, Advances in Physical Geochemistry, Vol. 7. New York:
Springer, pp. 113–128.
Shichiri, T., Kishimoto, Y. and Maruyama, M. (1994) Morphological change of ice crystals under pressure and
by mixing of isotope. J. Jpn. Assoc. Cryst. Growth, 21, 61– 67.
Shieh, S.R., Mao, H.K., Hemley, R.J. and Ming, L.C. (1998) The decomposition of phase D in the lower
mantle and the fate of dense hydrous silicates in subducting slabs. Earth Planet. Sci. Lett., 159,
12–23.
Shieh, S.R., Mao, H.K., Konzett, J. and Hemley, R.J. (2000) In situ high pressure X-ray diffraction of phase E to
15 GPa. Am. Mineral., 85, 765–769.
Shim, S. and Duffy, T. (2000) Constraints on the P – V – T equation of state of MgSiO3 perovskite. Am. Mineral.,
85, 354 –363.
Shim, S., Duffy, T.S. and Shen, G. (1998) P – V – T equation of state of MgSiO3 and CaSiO3 perovskites to 60 GPa
and 2000 K. EOS Trans., AGU, F861.
Shimakawa, Y., Kubo, Y. and Manako, T. (1996) Giant magnetoresistance in Tl2Mn2O7 with the pyrochlore
structure. Nature, 379, 53–54.
1182 References

Shimizu, H., Ohnishi, M., Sasaki, S. and Ishibashi, Y. (1995) Phys. Rev. Lett., 74, 2820.
Shimizu, K., Suhara, K., Ikumo, M., Eremets, M.I., Amaya, K. (1998) Superconductivity in oxygen. Nature, 393,
767–769.
Shimizu, K., Kimura, T., Furomoto, S., Takeda, K., Kontani, K., Onuki, Y., Amaya, K. (2001) Superconductivity
in the non-magnetic state of iron under pressure. Nature, 412, 316–318.
Shimobayashi, N. and Kitmura, M. (1991) Phase transition in Ca-poor clinopyroxenes: a high-temperature
transmission electron microscopic study. Phys. Chem. Miner., 18, 153–160.
Shiratani, E. and Sasai, M. (1998) J. Chem. Phys., 108, 3264– 3276.
Shu, J., Mao, H.K., Hu, J., Fei, Y. and Hemley, R.J. (1998) Single-crystal X-ray diffraction of wüstite to 30 GPa
hydrostatic pressure. N. Jb. Miner. Abh., 172, 309–323.
Sikka, S.K. (1992) In: Singh, A.K. (ed.), Recent Trends in Pressure Research. New Delhi: IBH, pp. 254.
Sikka, S.K. and Sharma, M.S. (1995) In: Banerjee, S. and Ramanujan, R.V. (eds.), Advances in Physical
Metallurgy. New York: Gordon and Breach.
Silver, P.G. and Wakita, H. (1996) A search for earthquake precursors. Science, 273, 77–78.
Sikka, S.K., Sharma, S.M. and Chidambaram, R. (1994) In: Schmidt, S.C. et al. (eds.), High Pressure Science &
Technology — 1993. New York: AIP Press, 291 pp.
Silvera, L.F. (1980) The solid molecular hydrogen in the condensed phase: fundamentals and static properties.
Rev. Mod. Phys., 32, 393–452.
Simmons, G. and Wang, H. (1971) Single Crystal Elastic Constants and Calculated Aggregate Properties.
Cambridge, MA: MIT Press.
Simon, A.T., Redfern, S.A.T. and Angel, R.J. (1999) High-pressure behaviour and equation of state of calcite,
CaCO3. Contrib. Mineral. Petrol., 134, 102–106.
Simpson, D.W. and Negmatullaev, S.K. (1981) Bull. Seismol. Soc. Am., 71, 1561–1586.
Sinelnikov, Y.D., Chen, G., Neuville, D.R., Vaughan, M.T. and Liebermann, R.C. (1998) Ultrasonic shear wave
velocities of MgSiO3 perovskite at 8 GPa and 800 K and lower mantle composition. Science, 281,
677–679.
Singh, A.K. (1993a) The lattice strain in a specimen (cubic system) compressed under non-hydrostatic pressure.
J. Appl. Phys., 73, 4278–4286.
Singh, A.K. (1993b) Analysis of C60 fullerite compression under non-hydrostatic pressure. Phil. Mag. Lett., 67,
379–384.
Singh, D. (1995) Local density and generalized gradient approximation studies of KNbO3 and BaTiO3.
Ferroelectrics, 164, 143.
Singh, A.K. and Balasingh, C. (1994) The lattice strains in a specimen (hexagonal system) compressed
nonhydrostatically in an opposed anvil high pressure set up. J. Appl. Phys., 75, 4956–4962.
Singh, A.K. and Balasingh, C. (1996) Bull. Mater. Sci., 19, 601.
Singh, A.K. and Kennedy, G.C. (1974) Compression of calcite to 40 kbar. J. Geophys. Res., 79, 2615–2622.
Singh, A.K., Balasingh, C., Mao, H.K., Hemley, R.J. and Duffy, T.S. (1998a) Measurement and analysis of
lattice strains in a polycrystalline sample under ultra-high pressure. In: Sengupta, S.P. (ed.), Proceedings
of the International School of Powder Diffraction. India: Allied Publishers, pp. 64–77.
Singh, A.K., Balasingh, C., Mao, H.K., Shu, J. and Hemley, R.J. (1998b) Analysis of lattice strains measured
under non-hydrostatic pressure. J. Appl. Phys., 83, 7567–7575.
Singh, A.K., Mao, H.K., Shu, J. and Hemely, R.J. (1998c) Estimation of single-crystal elastic moduli
from polycrystalline X-ray diffraction at high pressure. Applications to FeO and iron. Phys. Rev. Lett.,
80, 2157–2160.
Singh, S.C., Taylor, M.A.J. and Montagner, J.P. (2000) On the presence of liquid in Earth’s inner core. Science,
287, 2471–2474.
Sinogeikin, S.V. and Bass, J.D. (1995) Elasticity of Mg4Si4O12 –Mg3Al2Si3O12 garnet and end-member tetragonal
Mg4Si4O12 garnet by Brillouin scattering. EOS Trans. Am. Geophys. Union, 76, S276.
Sinogeikin, S.V. and Bass, J.D. (1999) Single-crystal elastic properties of chondrodite: implications for water in
the upper mantle. Phys. Chem. Miner., 26, 297– 303.
Sinogeikin, S.V., Bass, J.D., O’Neill, B.O. and Gasparik, T. (1997) Elasticity of tetragonal end-member
majorite and solid solutions in the system Mg4Si4O12 –Mg3Al2Si3O12. Phys. Chem. Miner., 24, 115–121.
References 1183

Sinogeikin, S.V., Katsura, T. and Bass, J.D. (1998) Sound velocities and elastic properties of Fe-bearing
wadsleyite and ringwoodite. J. Geophys. Res., 103, 20819–20825.
Sisson, T.W. and Grove, T.L. (1993) Temperature and H2O contents of low-MgO high-alumina basalts. Contrib.
Mineral. Petrol., 113, 167–184.
Skinner, B.J. (1966) Thermal expansion. In: Clark, S.P. (ed.), Handbook of Physical Constants, Vol. 97. Boulder,
CO: Geological Society of America, pp. 75– 95.
Skogby, H. (1992) Order–disorder kinetics in orthopyroxenes of ophiolite origin. Contrib. Mineral. Petrol.,
109, 471–478.
Skriver, H.L. (1984) The LMTO Method Muffin–Tin Orbitals and Electronic Structure. Berlin: Springer.
Slack, G.A. and Ross, R.G. (1985) J. Phys. C, 18, 3957.
Slade, R.C.T., Southern, J.C. and Thompson, I.M. (1990) J. Mater. Chem., 1, 563.
Slater, J.C. (1951) A simplifications of the Hartree–Fock method. Phys. Rev., 18, 385–390.
Slater, J.C. (1974) Quantum Theory of Molecules and Solids, Vol. 4. New York: McGraw-Hill.
Sleep, N.H. (1990) Hotspots and mantle plumes: some phenomenology. J. Geophys. Res., 95, 6715–6736.
Sleep, N.H. (1994) Martian plate tectonics. J. Geophys. Res., 99, 5639–5655.
Smith, J.V. (1981) The first 800 million years of Earth’s history. Phil. Trans. R. Soc. Lond. A, 301, 401–422.
Smith, D.C. (1988) Eclogites and Eclogite-Facies Rocks. Amsterdam: Elsevier, 524 pp.
Smith, J.V. and Brown, W.L. (1988) Feldspar Minerals: Crystal Structures, Physical, Chemical and
Microtextural Properties, Vol. 1. New York: Springer, 2nd revised and extended ed., 828 pp.
Smith, P. and Buseck, P.R. (1981) Science, 212, 322–324.
Smith, J.V. and Dawson, J.B. (1981) Storage of F and Cl in the upper mantle: geochemical implications. Lithos,
14, 133 –147.
Smith, K.L. and Lumpkin, G.R. (1993) Structural features of zirconolite, hollandite and perovskite, the major
waste-bearing phase in Synroc. In: Boland, J.N. and Fitz Gerald, J.D. (eds.), Defects and Processes in
Solid State: Geoscience Applications. Amsterdam: Elsevier, 470 pp.
Smith, J.V. and Mason, B. (1970) Pyroxene–garnet transformation in Coorara meteorite. Science, 168, 832–833.
Smith, G.S. and Snyder, R.L. (1979) FN: a criterion for rating powder diffraction patterns and evaluating the
reliability of powder-pattern indexing. J. Appl. Crystallogr., 12, 60–65.
Smith, J.V., Artioli, G. and Kvick, A. (1986) Low albite, NaAlSi3O8: neutron diffraction study of crystal structure
at 13 K. Am. Mineral., 71, 727– 733.
Smith, G.P., Douglas, A.W., Karen, M.F., Leroy, M.D., Spahr, C.W., and John, A.H. (2001a). A complex pattern
of mantle flow in the Lau Backarc. Science, 292, 713– 716.
Smith, G.P. (2001b) J. Geophys Res., 106(E10), 23689–23722.
Smylie, D.E. (1999) Viscosity near Earth’s solid inner core. Science, 284, 461–463.
Smyth, J.R. (1990) b-Mg2SiO4: a potential host for water in the mantle. Am. Mineral., 72, 1051–1055.
Smyth, J.R. (1994) A crystallographic model for hydrous wadsleyite (b-Mg2SiO4): an ocean in the Earth’s
interior? Am. Mineral., 79, 1021–1024.
Smyth, J.R., Swope, R.J. and Pawley, A.R. (1995) H in rutile-type compounds: II. Crystal chemistry of Al
substitution in H-bearing stishovite. Am. Mineral., 80, 454–456.
Smyth, J.R., Kawamoto, T., Jacobsen, S.D., Swope, R.J., Hervig, R.L. and Holloway, J.R. (1997) Crystal structure
of monoclinic hydrous wadsleyite [b-(Mg, Fe)2SiO4]. Am. Mineral., 82, 270–275.
Snelnikov, Y.D., Chen, G., Neuville, D.R., Vaughan, M.T. and Liebermann, R.C. (1998) Ultrasonic shear wave
velocities of MgSiO3 perovskite at 8 GPa and 800 K and lower mantle composition. Science, 281, 677.
Sobolev, A. and Chaussidon, M. (1996) H2O concentrations in primary melts from supra-subduction zones
and mid-ocean ridges: implications for H2O storage and recycling in the mantle. Earth Planet. Sci. Lett.,
137, 45 –55.
Sobolev, N.V. and Shatsky, V. (1990) Diamond inclusions in garnets form metamorphic rocks: a new
environment for diamond formation. Nature, 343, 742–746.
Söderlind, P., Moriarty, J.A. and Wills, J.M. (1996) First-principal theory of iron up to earth-core pressure:
structural, vibrational, and elastic properties. Phys. Rev., 53, 14063–14072.
Solheim, L.P. and Peltier, W.R. (1994) Phase boundary detections at 660-km depth and episodically layered
isochemical convection in the mantle. J. Geophys. Res., 99.
1184 References

Solomatov, V.S. and Stevenson, D.J. (1994) Earth Planet. Sci. Lett., 125, 267.
Solomon, S.C. (2002) An older face for Mars. Nature, 418, 27–28.
Solomon, S.C. et al. (2001) The MESSENGER mission to Mercury: scientific objectives and implementation.
Planet. Space Sci., 49, 1445–1465.
Somayazulu, M.S., Sharma, S.M. and Sikka, S.K. (1994a) Phys. Rev. Lett., 78, 98.
Somayazulu, M.S., Sharma, S.M., Sikka, S.K., Gard, N. and Chaplot, S.L. (1994b) In: Schmidt, S.C., Shaner,
J.W., Samara, G.A. and Ross, M. (eds.), High Pressure Science and Technology — 1993. New York: AIP
Press, 815 pp.
Somayazulu, M.S., Finger, L.W., Hemley, R.J. and Mao, H.K. (1996) New high-pressure compounds in
methane–hydrogen mixtures. Science, 271, 1400–1402.
Sonder, L. and England, P. (1986) Earth Planet. Sci. Lett., 77, 81–90.
Song, X. (1996) Anisotropy in central part of inner core, J. Geophys. Res., 101(B7), 16089–16098.
Song, X. (1997) Anisotropy of the Earth’s inner core. Rev. Geophys., 35, 297–313.
Song, X. and Ahrens, T.J. (1994) Pressure– temperature range of reactions between liquid iron in the outer core
and mantle silicates. Geophys. Res. Lett., 21, 153 –156.
Song, X. and Helmberger, D.V. (1993) Anisotropy of Earth’s inner core. Geophys. Res. Lett., 20, 2591–2594.
Song, X. and Hemlberger, D.V. (1995) Depth dependence of anisotropy of Earth’s inner core. J. Geophys. Res.,
100, 9805–9816.
Song, X. and Helmberger, D.V. (1998) Seismic evidence for an inner core transition zone. Science, 282, 924–927.
Song, X. and Richards, P.G. (1996) Seismological evidence for differential rotation of the Earth’s inner core.
Nature, 382, 221–224.
Souriau, A. (1998) Is the rotation real? Science, 281, 56.
Souriau, A. and Romanowicz, B. (1996) Geophys. Res. Lett., 23, 1.
Sowa, H. and Asbahs, H. (1996) Amorphization of quartz-type structure and a new high-pressure phase of
AlAsO4. Z. Kristallogr., 211(2), 96–100.
Sowa, H., Macavei, J. and Schulz, H. (1990) The crystal structure of berlinite AlPO4 at high pressure.
Z. Kristallogr., 192, 119–136.
Spearing, D.R., Farnan, I. and Stebbins, J.F. (1992) Dynamics of the a–b phase transitions in quartz and cristobalite
as observed by in situ high temperature 29Si and 17O NMR. Phys. Chem. Miner., 19, 307–321.
Spearing, D.R., Stebbins, J.F. and Farnan, I. (1994) Diffusion and the dynamics of displacive phase transition in
cryolite (Na3AlF6) and chiolite (Na5Al3F14): multinuclear NMR studies. Phys. Chem. Miner., 21,
373–386.
Speedy, R.J. (1996) J. Phys. Condens. Matter, 8, 10907.
Spetzler, H. (1970) Equation of state of polycrystalline and single-crystal MgO to 8 kilobars and 800 K.
J. Geophys. Res., 75, 2073–2087.
Spetzler, H., Shen, A., Chen, G., Hermannsdorfer, G., Schulze, H. and Weigel, R. (1996) Ultrasonic
measurements in a diamond-anvil cell. Phys. Earth Planet. Interiors, 98, 93–99.
Spitzig, W.A. and Leslie, W.C. (1971) Acta Metall., 19, 1143.
Srikanth, V., Roy, R., Graham, E.K. and Voigt, D.E. (1991) J. Am. Ceram. Soc., 74, 3145.
Stacey, F.D. (1969) Physics of the Earth. New York: Wiley, 324 pp.
Stacey, F.D. (1992) Physics of the Earth, 3rd ed. Brisbane: Brookfield Press.
Stacey, F.D. (1995) Theory of thermal and elastic properties of the lower mantle and core. Phys. Earth Planet.
Interiors, 89, 219 –245.
Stacey, F.D. (1996) Thermoelasticity of (Mg, Fe)SiO3 perovskite and a comparison with the lower mantle. Phys.
Earth Planet. Interiors, 98, 65 –77.
Stacey, F.D. (1997) Bullen’s seismological homogeneity, h, applied to mixture of minerals: the case of the lower
mantle parameter. Phys. Earth Planet. Interiors, 99, 189–193.
Stacey, F.D. (1998) Thermoelasticity of a mineral composite and reconsideration of lower mantle properties.
Phys. Earth Planet. Interiors, 106, 219–236.
Stacey, F.D. and Isaak, D.G. (2000) Extrapolation of lower mantle properties to zero pressure: constraints on
composition and temperature. Am. Mineral., 85, 345 –353.
Stacey, F.D. and Loper, D.E. (1983) Phys. Earth Planet. Interiors, 33, 45.
References 1185

Stassis, C. (1994) Inelastic neutron scattering of g-iron, and the determination of the elastic constants by lattice
dynamics. In: Schmidt, S.C., Shaner, J.W., Samara, G.A. and Ross, M. (eds.), High-Pressure Science and
Technology. Woodbury, NY: American Institute of Physics, 1922 pp.
Stau-Olsen, J., Cousins, C.S.G., Gerward, L., Jhans, H. and Sheldon, B.J. (1991) A study of the crystal structure
of Fe2O3 in the pressure range up to 65 GPa using synchrotron radiation. Phys. Scripta, 43, 327–330.
Stein, S. and Stein, C.A. (1996) Thermo-mechanical evolution of oceanic lithosphere: implications for the
subduction process and deep earthquakes. In: Bebout, G.E. et al. (eds.), Subduction: Top to Bottom,
Geophysical Monograph Series 96. Washington, DC: AGU, pp. 1–17.
Steinle-Neumann, G., Stixrude, L. and Cohen, R.E. (1998) First-principles elastic constants for the hcp transition
metals Fe, Co, and Re at high pressure. Phys. Rev. B, 60, 791–799.
Steinle-Neumann, G., Stixrude, L., Cohen, R.E. and Gulseren, O. (2001) Elasticity of iron at the temperature of
the Earth’s inner core. Nature, 413, 57–60.
Stevenson, D.J. (1999) The world between crust and core. Nature, 400, 52.
Stevenson, D.J. (1987) In: Schmidt, S.C. and Holmes, N.C. (eds.), Shock Waves in Condensed Matter. New York:
Elsevier, 1988.
Stevenson, D.J. (1994) Weakening under stress. Nature, 372, 129.
Stevenson, D.J. (1995) Light from tungsten on core construction. Nature, 378, 763–764.
Stewart, D.N., Basu, E.H., Walker, K.A. and Gubbins, D. (1995) Geomagnetism, Earth rotation and the electrical
conductivity of the lower mantle. Phys. Earth Planet. Interiors, 92, 199–214.
Stishov, S.M. and Popova, S.V. (1961) A new dense modification of silica. Geochemistry, 10, 923–926.
Stixrude, L. and Brown, J.M. (1998) The Earth’s core. In: Hemley, R.J. (ed.), Ultrahigh Pressure Mineralogy:
Physics and Chemistry of the Earth’s Deep Interior, Reviews in Mineralogy, Mineralogical Society of
America, Vol. 37, pp. 261 –280.
Stixrude, L. and Bukowinski, M.S.T. (1990) Fundamental thermodynamic relations and silicate melting with
implications for the constitution of D00 . J. Geophys. Res., 95, 19311– 19325.
Stixrude, L. and Cohen, R.E. (1993) Stability of orthorhombic MgSiO3 perovskite in the Earth’s lower mantle.
Nature, 364, 613–616.
Stixrude, L. and Cohen, R.E. (1995a) High pressure elasticity of iron and anisotropy of Earth’s inner core.
Science, 267, 1972–1975.
Stixrude, L. and Cohen, R.E. (1995b) Constraints on the crystalline structure of the inner core: mechanical
instability of fcc iron at high pressure. Geophys. Res. Lett., 22(2), 125–128.
Stixrude, L., Hemley, R.J., Fei, Y. and Mao, H.K. (1992) Science, 257, 1099.
Stixrude, L., Cohen, R.E. and Singh, D.J. (1994) Iron at high pressure: linearized augmented plane wave
calculations in the generalised gradient approximation. Phys. Rev. B, 50, 6442–6445.
Stixrude, L., Cohen, R.E. and Hemley, R.J. (1998a) Theory of minerals at high pressure. In: Hemley, R.J. (ed.),
Ultrahigh-pressure Mineralogy: Physics and Chemistry of the Earth’s Deep Interior, Reviews in
Mineralogy, Mineralogical Society of America, Vol. 17, pp. 639–671.
Stixrude, L., Wasserman, E. and Cohen, R.E. (1998b) First-principles investigations of solid iron at high pressure
and implications for the Earth’s inner core. In: Manghnani, M.H. and Yagi, T. (eds.), Properties of Earth
and Planetary Materials at High Pressure and Temperature, Geophysical Monograph, Vol. 101.
Washington, DC: AGU, pp. 159–171.
Stöffler, D. (1966) Contrib. Mineral. Petrol., 12, 15.
Stöffler, D. (1984) J. Non-Cryst. Solids, 67, 465.
Stöffler, D.R. and Hornemann, U. (1972) Meteoritics, 7, 371.
Stoffler, D., Bischoff, A., Buchwald, V. and Rubin, A.E. (1988) Shock effects in meteorites. In: Kerridge, J.F. and
Matthews, M.S. (eds.), Meteorites and the Early Solar System. Tucson, AZ: University of Arizona Press,
pp. 165 –202.
Stokes, H.T. and Hatch, D.M. (1988) Isotropy Subgroups of the 230 Crystallographic Space Groups. Singapore:
World Scientific, 580 pp.
Stølen, S. and Grønvold, F. (1996) Calculation of the phase boundaries of wüstite at high pressure. J. Geophys.
Res., 101B, 11531–11540.
1186 References

Stølen, S., Glöckner, R., Grønvold, F., Atake, T. and Izumisawa, S. (1996) Heat capacity and thermo-
dynamic properties of nearly stoichiometric wüstite from 13 to 450 K. Am. Mineral., 81,
973–981.
Stolper, E.M. and Ahrens, T.J. (1987) Geophys. Res. Lett., 14, 1231.
Stompor, R. et al. (2001) Astrophys. J., 561, 17–110.
Stopler, E. (1982a) The speciation of water in silicate melts. Geochim. Cosmochim. Acta, 46,
2609– 2620.
Storey, B.C. (1995) The role of mantle plumes in continental breakup — case histories from Gondwanaland.
Nature, 377, 301–308.
Straub, J.K., Holian, B.L. and Swanson, R.E. (1980) Bull. Am. Phys. Soc., 25, 549.
Strixrude, L. (1995) Constraints on the crystalline structure the inner core: mechanical instability of BCC iron at
high pressure. Geophys. Res. Lett., 22(2), 125–128.
Strong, H.M., Tuft, R.E. and Hanneman, R.E. (1973) The iron fusion curve and g– d-liquid triple point. Metall.
Trans., 4, 2657–2661.
Struzhkin, V.V., Goncharov, A.F., Hemley, R.J. and Mao, H.K. (1997a) Cascading Fermi resonances and the soft
mode in dense ice. Phys. Rev. Lett., 78, 4446–4449.
Struzhkin, V.V., Timofeev, Y.A., Hemley, R.J. and Mao, H.K. (1997b) Superconduction TC and electron– phonon
coupling in Nb to 132 GPa: magnetic susceptibility at megabar pressures. Phys. Rev. Lett., 79,
4262– 4265.
Strzhemechny, M.A. and Hemley, R.J. (2000) New ortho–para mechanism in dense solid hydrogen. Phys. Rev.
Lett., 85(26), 5595–5598.
Strzhemechny, M.A., Hemley, R.J. et al. (2002) Ortho– para conversion of hydrogen at high pressures. Phys. Rev.
B, 66(18), 014103.
Su, W.J. and Dziewonski, A.M. (1991) Predominance of long-wavelength heterogeneity in the mantle. Nature,
352, 121 –126.
Su, W.J. and Dziewonski, A.M. (1994) Seismol. Res. Lett., 65, 23.
Su, W.J., Woodward, R.L. and Dziewonski, A.M. (1994) Degree 12 model of shear velocity heterogeneity in the
mantle. J. Geophys. Res., 99, 6945–6981.
Su, W., Dziewonski, A.M. and Jeanloz, R. (1996) Planet within a planet: of the inner core of Earth. Science,
274, 1883–1887.
Sudo, A. and Tatsumi, Y. (1990) Phlogopite and K-amphibole in the upper mantle: implications for magma
genesis in subduction zones. Geophys. Res. Lett., 17, 29– 32.
Sueno, S., Kimata, M. and Prewitt, C.W. (1984) The crystal structure of high clinoferrosilite. Am. Mineral., 69,
264–269.
Sugano, S. et al. (1970) Multiplets of Transition Metal Ions in Crystals. New York: Academic Press, Chapter 10.
Sugiyama, M., Endo, S. and Koto, K. (1987) The crystal structure of stishovite under pressure up to 6 GPa.
Mineral. J., 13, 455 –466.
Sumino, Y. and Anderson, O.L. (1989) Elastic constant of minerals. In: Carmichael, R.S. (ed.), Handbook of
Physical Properties of Rocks, Vol. III. Boca Raton, FL: CRC Press, pp. 139–280.
Sumino, Y. and Anderson, O.L. (1989) In: Carmichael, R.S. (ed.), Handbook of Physical Properties of Rocks, Vol.
III. Boca Raton, FL: CRC Press, pp. 39–137.
Sumita, T. and Inoue, T. (1996) Melting experiments and thermodynamic analyses on silicate H2O systems up to
12 GPa. Phys. Earth Planet. Interiors, 96, 187– 200.
Sun, T., Herbst, C., King, H.E. Jr. and Kruger, M. (1994) Bull. Am. Phys. Soc., 302, 668.
Sung, C.M. (1976) New modification of the diamond anvil press: a versatile apparatus for research at high
pressure and temperature. Rev. Sci. Instrum., 47, 1133–1136.
Sung, C.M. and Sung, M. (1996) Carbon nitride and other speculative superhard materials. Mater. Chem. Phys.,
43, 1 –18.
Sung, C.M., Goetz, C. and Mao, H.K. (1977) Pressure distribution in the diamond anvil press and the shear
strength of fayalite. Rev. Sci. Instrum., 48, 1386–1391.
Surkov, Y.A., Moskalyeava, L.P., Khrynkova, V.P., Dulin, A.D., Smirnov, S.Y. and Zaitseva, S.Y. (1986) Venus
rock composition and the Vega 2 landing site. J. Geophys. Res., 91, 215– 218.
References 1187

Sutton, S.R., Jones, K.W., Gordon, B., Rivers, M.L., Bajt, S. and Smith, J. (1993) Reduced chromium in olivine
grains from lunar basalt 15555: X-ray absorption near edge structure (XANES). Geochim. Cosmochim.
Acta, 57, 461– 468.
Suzuki, I. (1975) Thermal expansion of periclase and olivine and their anharmonic properties. J. Phys. Earth,
23, 145 –159.
Suzuki, T. and Akaogi, M. (1995) Element partitioning between olivine and silicate melt under high pressure.
Phys. Chem. Miner., 22, 411 –418.
Suzuki, I. and Anderson, O. (1983) Elasticity and thermal expansion of a natural garnet up to 1000 K. J. Phys.
Earth, 31, 125 –138.
Suzuki, I., Ohtani, E. and Kumazawa, M. (1979) Thermal expansion of g-Mg2SiO4. J. Phys. Earth, 27,
53–61.
Suzuki, I., Ohtani, E. and Kumazawa, M. (1980) Thermal expansion of modified spinel, b-Mg2SiO4. J. Phys.
Earth, 28, 273 –280.
Suzuki, T., Yagi, T., Akimoto, S., Ito, A., Morimoto, S. and Syono, Y. (1985) X-ray diffraction and Mössbauer
spectrum on the high pressure phase of Fe2O3. In: Minomura, S. (ed.), Solid State Physics Under
Pressure. Recent Advance with Anvil Devices. Tokyo: Terra Scientific, pp. 149– 154.
Swainson, I.P. and Dove, M.T. (1995) Molecular dynamics simulation of a and b-cristobalite. J. Phys. Condens.
Matter, 7, 1771–1788.
Sweeney, R.J., Prozesky, W. and Przybylowicz, W. (1995) Selected trace and minor element partitioning between
peridotite minerals and carbonatite melts at 18 –46 kb pressure. Geochim. Cosmochim. Acta, 59(18),
3671– 3683.
Swope, R.J., Smyth, J.R. and Larson, A.C. (1995) H in rutile-type compounds: I. Single-crystal neutron and X-ray
diffraction study of H in rutile. Am. Mineral., 80, 448–453.
Syono, Y. (1984) Shock-induced phase transition in oxides and silicates. In: Sunagawa (ed.), Materials Science of
the Earth’s Interior. Tokyo: Terra Scientific, pp. 395–414.
Syono, Y. and Manghnani, M.H. (1992a) High Pressure Research: Applications to Earth and Planetary Science.
Tokyo/Washington, DC: Terra Scientific/AGU.
Syono, Y. and Manghnani, M.H. (1992b) First-principles predictions of elasticity and phase transitions in high
pressure SiO2 and geophysical implications. Am. Geophys. Union Washington, 425–431.
Syono, Y., Akimoto, S. and Endo, Y. (1971a) J. Phys. Chem. Solids, 32, 243.
Syono, Y., Tokonami, M. and Matsui, Y. (1971b) Crystal field effects on the olivine–spinel transformation. Phys.
Earth Planet. Interiors, 4, 347 –352.
Syono, Y., Goto, T. and Nakagawa, Y. (1977) High Pressure Research: Applications in Geophysics. New York:
Academic Press, 463 pp.
Syono, Y., Yamauchi, H., Ito, A., Someya, Y., Ito, E., Matsui, Y. and Akimoto, S. (1980) Magnetic properties of
the disordered ilmenite FeTiO3 II synthesized at very high pressure. FERRITES: Proceedings of the
International Conference, Japan, 1980.
Syuno, Y., Takei, H., Goto, T., Ito, A. (1981) Single-crystal x-ray and Mössbauer study of shocked ilmenite to
80 GPa. Phys. Chem. Miner., 7, 82–87.
Syono, Y., Ito, A., Morimoto, S., Suzuki, T., Yagi, T. and Akimoto, S. (1984) Mössbauer study on the high
pressure phase of Fe2O3. Solid State Commun., 50, 97.
Syono, Y., Noguchi, Y., Fukuoka, K., Kusaba, K. and Atou, T. (1998) Shock-induced phase transition of
MnO and several other transition metal oxides. In: Schmidt, S.C. et al. (eds.), Shock Compression
of Condensed Matter — 1997. New York: American Institute of Physics, pp. 151–154.
Tackley, P.J. (2000) Mantle convection and plate tectonics: Toward an integrated physical and chemical theory.
Science, 288, 2002–2006.
Tackley, P.J., Stevenson, D.J., Glatzmaier, G.A. and Schubert, G. (1993) Effects of an endothermic phase
transition at 670 km depth in a spherical model of convection in the Earth’s mantle. Nature, 361,
699–704.
Takahashi, E. (1986a) Melting of a dry peridotite KLB-1 up to 14 GPa: implications on the origin of peridotitic
upper mantle. J. Geophys. Res., 91, 9367–9382.
1188 References

Takahashi, E. (1986b) Genesis of calc-alkali andesite magma in a hydrous mantle-crust boundary: petrology of
lherzolite xenoliths from the Ichinomegata crater, Oga peninsula, Northeast Japan, Part II. J. Volcanol.
Geotherm. Res., 29, 355–395.
Takahashi, E. (1990) Speculations on the Archean mantle: missing link between komatiite and depleted garnet
peridotite. J. Geoprocesses EOS, 71, 644.
Takahashi, E. and Ito, E. (1987) Mineralogy of mantle peridotite along a model geotherm up to 700 km depth.
In: Manghnani, M. and Syono, Y. (eds.), High Pressure Research in Mineral Physics, Geophysical
Monograph, Vol. 39. Washington, DC: AGU, pp. 427 –437.
Takahashi, E. and Kushiro, I. (1993) Melting of a dry peridotite at high pressures and basalt magma genesis.
Am. Mineral., 68, 859 –879.
Takahashi, T. and Liu, L. (1970) Compression of ferromagnesian garnets and effect of solid solutions on the bulk
modulus. J. Geophys. Res., 75, 5757– 5766.
Takahashi, T., Bassett, W.A. and Mao, H.K. (1968) Isothermal compression of the alloys of iron up to 300
kilobars at room temperature: iron–nickel alloys. J. Geophys. Res., 73, 4717–4725.
Takano, M. et al. (1977) Mater. Res. Bull., 12, 923.
Takano, M., Nasu, S., Abe, T., Yamamoto, K., Endo, S., Takeda, Y. and Goodenough, J.B. (1991) Pressure-
induced high-spin to low-spin transition in CaFeO3. Phys. Rev. Lett., 67(23), 3267–3270.
Takeda, H. (1990) Diffusion of ions in mantle minerals at high pressure and high temperature. In: Marumi, F.
(ed.), Dynamic Processes of Material Transport and Transformation in the Earth’s Interior. Tokyo:
Terra Scientific, pp. 217 –238.
Takeda, T. and Komura, S. (1980) Proc. Int. Conf., 385.
Takeda, T. et al. (1972) J. Phys. Soc. Jpn, 33, 967.
Takeda, Y. et al. (1978) Mater. Res. Bull., 13, 61.
Takeda, Y., Kanno, K., Takeda, T., Yamamoto, O., Tankano, M., Nakayama, N. and Bando, Y. (1986)
Phase relation in the oxygen nonstoichiometric system, SrFeOx(2.5 # x $ 3.0). J. Solid State Chem.,
63, 237.
Takele, S. and Hearne, G.R. (1999) Electrical transport magnetism and spin-state configuration of high pressure
phases of FeS. Phys. Rev. B, 60, 4401–4403.
Takley, P.J. (2000) Mantle convection and plate tectonics: toward an integrated physical and chemical theory.
Science, 288, 2002–2006.
Tamai, H. and Yagi, T. (1989) High-pressure and high-temperature phase relations in CaSiO3 and CaMgSi2O6 and
elasticity of perovskite-type CaSiO3. Phys. Earth Planet. Interiors, 54, 370–377.
Tanabe, T. and Sugano, S. (1954) On the absorption spectra of complex ions. J. Phys. Soc. Jpn, 9,
766–779.
Tanaka, H. (2000) Simple physical model of liquid water. J. Chem. Phys., 112, 799–809.
Tanaka, S. and Hamaguchi, H. (1997) Degree one heterogeneity and hemispherical variation of anisotropy in the
inner core from PKP(BC)–PKP(DF) times. J. Geophys. Res., 102, 2925–2938.
Tanimoto, T. and Anderson, D.L. (1984) Mapping convection in the mantle. Geophys. Res. Lett., 11,
287–290.
Taran, M.N., Langer, K. and Platonov, A.N. (1996) Pressure and temperature effects on exchange-coupled-pair
bands in electronic-spectra of some oxygen-based iron-bearing minerals. Phys. Chem. Miner., 23,
230–236.
Taura, H., Yurimoto, H., Kurita, K. and Sueno, S. (1998) Pressure dependence on partition coefficients
for trace elements between olivine and the coexisting melts. Phys. Chem. Miner., 25, 469–484.
Taylor, W.H. (1982) The structure of silimanite and mullite. Z. Kristallogr., 71, 205–218.
Taylor, S.R. and McLennan, S.M. (1985) The Continental Crust: Its Composition and Evolution. Oxford:
Blackwell Scientific, pp. 1– 312.
Taylor, H.P. Jr. and Sheppard, S.M.F. (1986) Igneous rocks. I. Processes of isotopic fractionation and isotope
systematics. Rev. Mineral., 16, 227–271.
Taylor, S., Pastenak, R.M. and Jeanloz, R. (1991) Hysteresis in the high pressure transformation of bcc- to hcp-
iron. J. Appl. Phys., 69, 6126–6128.
References 1189

Tayor, M. and Brown, G.E. (1979) Structure of silicate mineral glasses. T. The feldspar glasses, NaAlSiO3,
KAlSi2O8, CaAl2Si2O8. Geochim. Cosmochim. Acta, 43, 61–77.
Telxelra, J. (1998) The double identity of ice X. Nature, 392, 232–233.
Tera, F., Brown, L., Morris, J. and Sacks, I.S. (1986) Sediment incorporation in island-arc magmas: inferences
from 10Be. Geochim. Cosmochim. Acta, 50, 535–550.
Terhume, R.W. et al. (1985) Phys. Rev. B, 32, 8416.
Terry, M.P., Robinson, P., Carswell, D.A. and Gasparik, T. (1999) EOS, 80, 5359.
Teslic, S., Fgami, T. and Vichland, D. (1997) Local atomic structure of PZT and PLZT studied by pulsed neutron
scattering. J. Phys. Chem. Solids, 57, 1537–1543.
Teter, D.M. (1998) Mater. Res. Sci. Bull., 22–27.
Teter, D.M. and Hemley, R.J. (1996) Low-compressibility carbon nitrides. Science, 271, 53 –55.
Teter, D.M., Gibbs, G.V., Boisen, M.B., Allan, D.C. and Teter, M.P. (1995) First principles study of several
hypothetical silica framework structures. Phys. Rev., 52, 8064–8073.
Teter, D.M., Hemley, R.J., Kresse, G. and Hafner, J. (1998) High pressure polymorphism in silica. Phys. Rev.
Lett., 80, 2145–2148.
Thibault, Y. and Walter, M.J. (1995) The influence of pressure and temperature on the metal–silicate partition
coefficients of nickel and cobalt in a model C1 chondrite and implications for metal segregation in a deep
magma ocean. Geochim. Cosmochim. Acta, 59, 991 –1002.
Thiel, M. and van Ree, F.H. (1989) Int. J. Thermophys., 10, 227.
Thomas, S.D. and Hemley, R.J. (1995) Some like it hot: the temperature structure of the Earth. Am. Geophys.
Union, 5–9.
Thompson, A.B. (1992) Water in the Earth’s upper mantle. Nature, 358, 295–302.
Thompson, P. and Grimes, N.W. (1978) Observation of low energy phonons in spinels in spinel. Solid State
Commun., 25, 609–611.
Thong, N. and Schwarzenbach, D. (1979) The use of electric field gradient calculation in charge density
refinements. II. Charge density refinement of the low-quartz structure of aluminium phosphate. Acta
Crystallogr., 35, 658–664.
Thyaparam, S., Heine, V., Dove, M.T. and Hammonds, K.D. (1996) A computational study of Al/Si ordering in
cordierite. Phys. Chem. Miner., 23, 127–139.
Tilley, R.J.D. (1987) Defect Crystal Chemistry and its Applications. Glasgow: Blackie, 236 pp.
Timco, G.W. and Schloessin, H.H. (1974) High Temp. High Press., 6, 541–544.
Timco, G.W. and Schloessin, H.H. (1976a) High Temp. High Press., 8, 73–82.
Timco, G.W. and Schloessin, H.H. (1976b) Ferroelectrics, 11, 409–412.
Tingle, T.N., Tyburezy, J.A., Ahrens, T.J. and Becker, C.H. (1992) The fate of organic matter during planetary
accretion: preliminary studies of the organic chemistry of experimentally shocked Murchison meteorite.
Origins Life EVol. Biosphere, 21, 385 –397.
Tingle, T.N., Green, H.W., Scholz, C.H. and Korzynski, T.A. (1993) The theology of faults triggered by the
olivine– spinel transformation in Mg2GeO4 and its implications for the mechanisms for deep focus
earthquakes. J. Struct. Geol., 15, 1249–1256.
Tinkham, M. (1996) Introduction to Superconductivity. New York: McGraw-Hill.
Tokura, Y., Taguchi, T., Moritomo, Y., Kumagai, K., Suzuki, T. and Iye, Y. (1993) Phys. Rev. B, 48, 14063.
Tolentino, H., Dartyge, E., Fontaine, A. and Tourillon, G. (1988) J. Appl. Crystallogr., 21, 15.
Tomioka, Y. et al. (1996) Phys. Rev., 53, 1689–1692.
Tonks, W.B. and Melosh, H.J. (1990) In: Newsom, H.E. and Jones, J.H. (eds.), Origin of the Earth. New York:
Oxford University Press, pp. 151 –174.
Toriumi, M., Karato, S. and Fujii, T. (1984) Transient and steady state creep of olivine. In: Sunagawa, I. (ed.),
Materials Science of the Earth’s Interior. Tokyo: Terra Scientific, pp. 281–300.
Tossel, J.A. (1976) Electron structure of iron-bearing oxidic minerals at high pressure. Am. Mineral., 61,
130–144.
Tossel, J.A. (1977) A comparison of silicon-oxygen bonding in quartz and magnesium olivine from X-ray spectra
and molecular orbital calculations. Am. Mineral., 62, 136–141.
1190 References

Tranquada, J.M., Gehring, P.M., Shirane, G., Shamoto, S. and Sato, M. (1992) Neutron-scattering study of the
dynamical spin susceptibility in YB2Cu3O6.6. Phys. Rev. B, 46, 5561– 5575.
Tranquada, J.M. et al. (1995) Nature, 375, 561–563.
Treiman, A.H. (1985) Amphibole and hercynite spinel in Shergotty and Zagami: magmatic water, depth of
crystallization, and metasomatism. Meteoritics, 20, 229–243.
Tribandino, M. and Talarico, F. (1992) Orthopyroxenes from granulite rocks of the Wilson Terrane (Victoria
Land, Antarctica): Crystal chemistry and cooling history. Eur. J. Mineral., 4, 453–463.
Trieloff, M., Kunz, J., Clague, D.A., Harrison, D. and Allègre, C. (2000) Science, 288, 1036.
Tritt, T.M. (1999) Holey and unholey semiconductors. Science, 283, 804–805.
Trojer, F.J. (1969) The crystal structure of a high-pressure polymorph of CaSiO3. Z. Kristallogr., 130, 185–206.
Tromp, J. (1993) Support for anisotropy of the Earth’s inner core from free oscillations. Nature, 366, 678–681.
Tromp, J. and Dziewonski, A.M. (1998) Two views of the deep mantle. Science, 281, 655–656.
Trønnes, R.G. (1990) Low-Al, high-K amphiboles in subducted lithosphere from 200 to 400 km depth:
experimental evidence (abstr.). EOS, 71, 1587.
Trønnes, R.G., Canil, D. and Wei, K. (1992) Element partitioning better silicate minerals and coexisting melts at
pressures of 1 –17 GPa: implications for mantle evolution. Earth Planet. Sci. Lett., 111, 241–255.
Troullier, N. and Martians, J.L. (1990) A straightforward method for generating soft transferable
pseudopotentials. Solid State Commun., 74, 613– 616.
Troullier, N. and Martians, J.L. (1991a) Efficient pseudopotentials for plane-wave calculation I. Phys. Rev.,
43, 1993–2006.
Troullier, N. and Martians, J.L. (1991b) Efficient pseudopotentials for plane-wave calculations II. Operators for
fast iterative diagonalization. Phys. Rev., 43, 8861–8869.
Trunin, R.F., Simakov, G.V., Podurets, M.A., Moiseyev, L. and Popov, L.V. (1970) Izv. Earth Phys., 1, 13.
Truskett, T.M., Torquato, S. and Debenedetti, P.G. (2000b) Towards a quantification of disorder in materials.
Distinguishing equilibrium and glassy sphere packing. Phys. Rev. E, 62, 993–1001.
Tsai, C.H. and Liou, J.G. (2000) Eclogite-facies and inferred ultrahigh-pressure metamorphism in the North Dabie
Complex central-eastern China. Am. Mineral., 85, 1–8.
Tschauner, O., Zerr, A., Specht, S., Rocholl, A., Boehler, R. and Palme, H. (1999) Partitioning of nickel and cobalt
between silicate perovskite and metal at pressures up to 80 GPa. Nature, 398, 604.
Tse, J.S. (1992) Mechanical instability in ice Ih: a mechanism for pressure-induced amorphization. J. Chem. Phys.,
96, 5482–5487.
Tse, J.S. and Klug, D.D. (1991) Mechanical instability of a-quartz: a molecular dynamics study. Phys. Rev. Lett.,
67, 3559–3562.
Tse, J.S. and Klug, D.D. (1992) Structural memory in pressure-amorphized AlPO4. Science, 255, 1559–1561.
Tse, J.S. and Klug, D.D. (1993) Anisotropy in the structure of pressure-induced disordered solids. Phys. Rev. Lett.,
70, 174 –177.
Tse, J.S., Klug, D.D., Tulk, C.A., Swainson, I., Svensson, E.C., Loong, C.K., Shpakov, V., Belosludov, V.R.,
Belosludov, R.V. and Kawazoe, Y. (1999) The mechanisms for pressure induced amorphization of ice Ih.
Nature, 400, 647–649.
Tsuchiya, T., Yamanaka, T. and Matsui, M. (2000) Molecular dynamics study of pressure-induced transformation
of quartz-type GeO2. Phys. Chem. Miner., 27(3), 149 –155.
Tsukada, M., Satoko, C. and Adachi, H. (1980) J. Phys. Soc. Jpn, 48, 200.
Tsuneyuki, S., Matsui, Y., Aoki, H. and Tsukuda, M. (1989) Nature, 339, 209–211.
Turcott, D.L. and Schubert, G. (1982) Geodynamics: Applications of Continuum Physics to Geological Problems.
New York: Wiley.
Turcotte, D.L. et al. (2001) J. Geophys. Res., 106, 4265.
Turner, S. and Hawkesworth, C. (1995) The nature of the subcontinental mantle: constraints from the major-
element composition of continental flood basalts. Chem. Geol., 120, 295–314.
Tyburezy, J.A., Frisch, B. and Ahrens, T.J. (1986) Shock-induced volatile loss from a carbonaceous chondrite:
implications for planetary accretion. Earth Planet. Sci. Lett., 80, 201–207.
Uchida, T., Funamori, N. and Yagi, T. (1996) Lattice strains in crystal under uniaxial stress field. J. Appl. Phys.,
80, 739 –746.
References 1191

Ueda, K., Tabata, H. and Kawali, T. (1998) Ferromagnetism in LaFeO3 –LaCrO3 superlattices. Science, 280, 1064.
Uehara, M., Mori, S., Chen, C.H. and Cheong, S.W. (1999) Nature, 399, 560–563.
Ulmer, P. and Trommsdorff, V. (1995) Serpentine stability to mantle depths and subduction-related magmatism.
Science, 268, 858 –861.
Urakawa, S., Kondo, T., Igawa, N., Shimomura, O. and Ohno, O. (1994) Synchrotron radiation study on the high-
pressure and high-temperature phase relations of KAlSi3O8. Phys. Chem. Miner., 21(6), 387–391.
Urakawa, S., Igawa, N., Shimomura, O. and Ohno, H. (2000) Geophys. Monogr., 101, 241.
Urbain, G., Bottinga, Y. and Richet, P. (1982a) Viscosity of liquid silica, silicates, and aluminosilicates. Geochim.
Cosmochim. Acta, 60, 75–86.
Urbain, G., Bottinga, Y. and Richet, P. (1982b) Viscosity of liquid silica, silicates and aluminosilicates. Geochim.
Cosmochim. Acta, 46, 1061–1071.
Usselman, T.M. (1975) Am. J. Sci., 275, 278.
Utsumi, W. and Yagi, T. (1991) Light-transparent phase from room temperature compression of graphite. Science,
252, 1542–1544.
Utsumi, W., Funamori, N., Yagi, T., Ito, E., Kikegawa, T. and Shimomura, O. (1995) Thermal expansivity of
MgSiO3 perovskite under high pressures up to 20 GPa. Geophys. Res. Lett., 22, 1005–1008.
Utsumi, W. et al. (1997) Rev. High Press. Sci. Technol., 6, 512, Spec. Issue.
Utsumi, W., Weidner, D.J. and Liebermann, R.C. (1998) Volume measurements of MgO at high pressure and high
temperatures. In: Syono, I. and Manghnani, M.H. (eds.), Properties of Earth and Planetary Materials at
High Pressure and Temperature. Washington, DC: AGU, pp. 327–333.
Vaidya, S.N., Bailey, S., Pasternack, T. and Kennedy, G.C. (1973) Compressibility of fifteen minerals to
45 kilobars. J. Geophys. Res., 78, 6893–6898.
Van den Berg, A.P. and Yuen, D.A. (1996) Is the lower-mantle rheology Newtonian today? Geophys. Res. Lett.,
23, 2033–2036.
Vanderbilt, D. (1990) Soft self-consistent pseudopotentials in a generalized eigenvalue formalism. Phys. Rev. B,
41, 7892–7895.
Van der Hilst, R.D. and Karason, H. (1999) Compositional heterogeneity in the bottom 1000 kilometers of Earth’s
mantle: toward a hybrid convection model. Science, 283, 1885–1888.
Van der Hilst, R., Widiyantoro, S. and Engdahl, E.R. (1997) Evidence for deep mantle circulation from global
tomography. Nature, 386, 578–584.
Van Horn, H.M. (1991) Science, 252, 384.
Van Keken, P.E., Hauri, E.H. and Ballentine, C.J. (2002) Mantle mixing: the generation preservation and
destruction of chemical heterogeneity. Rev. Earth Planet. Sci., 30, 493–525.
Van Schilfgaarde, M., Abrikosov, I.A. and Johansson, B. (1999) Nature, 400, 46–49.
Van Tendeloo, G., Ghose, S. and Amelinckx, S. (1989) A dynamical model for the P1 – I 1 phase
transition in anorthite, CaAl2Si2O8 I. Evidence from electron microscopy. Phys. Chem. Miner., 16,
311–319.
Van Thiel, M. and Ree, F.H. (1989) Int. J. Thermophys., 10, 227.
Varotsos, P., Alexopoulos, K. et al. (1993) Tectonophysics, 224, 269–288.
Vassilious, M.S. and Ahrens, T.J. (1982) The equation of state of Mg0.6Fe0.4O to 200 GPa. Geophys. Res. Lett.,
9, 127 –130.
Vaughan, P.J. and Kohlstedt, D.L. (1981) Cation stacking faults in magnesium germanate spinel. Phys. Chem.
Miner., 7, 241–245.
Vaughan, M.T. and Weidner, D.J. (1978) The relationships of elasticity and crystal structure in andalusite and
silimanite. Phys. Chem. Miner., 3, 133–144.
Veblen, D.R. (1989) Am. Mineral., 74, 509.
Velde, B. (1980) Ordering in synthetic aluminous serpentines; infrared spectra and cell dimensions. Phys. Chem.
Miner., 6, 209–220.
Venturini, E.L., Morosin, B. and Graham, R.A. (1985) J. Appl. Phys., 57, 3814.
Verhoogen, J. (1980) Energetics of the Earth. Washington, DC: National Academy of Science.
Verwey, E.J.W. (1939) Electronic conduction of magnetite (Fe3O4) and its transition point at low temperatures.
Nature, 144, 327–328.
1192 References

Vidale, J.E. and Benz, H.M. (1992) Upper mantle seismic discontinuities and the thermal structure of subduction
zones. Nature, 356, 678–683.
Vidale, J.F. and Hedlin, M.A.H. (1998) Evidence for partial melt at the core– mantle boundary north of Tonga
from the strong scattering of seismic waves. Nature, 391, 682–685.
Vinet, P., Ferrante, J., Rose, J.H. and Smith, J.R. (1987) Compressibility of solids. J. Geophys. Res., 92,
9319– 9325.
Vinet, P., Ross, J.H., Ferrante, J. and Smith, J.R. (1989) Universal features of the equation of state of solids.
J. Phys. Condens. Matter, 1, 1941.
Vinnik, L., Farra, V. and Romanowicz, B. (1989) Observational evidence for diffracted SV in the shadow of the
Earth’s core. Geophys. Res. Lett., 16, 519 –522.
Vinnik, L., Chevrot, S. and Montagner, J.P. (1997) Evidence for a stagnant plume in the transition zone? Geophys.
Res. Lett., 24, 1007–1011.
Vinnik, L., Breger, L. and Romanowicz, B. (1998a) Anistropic structures at the base of the Earth’s mantle. Nature,
393, 564.
Vinnik, L., Breger, L. and Romanowicz, B. (1998b) On the inversion of Sd particle motion for anisotropy in D00 .
Geophys. Res. Lett., 25, 679–682.
Virgo, D. and Hafner, S. (1969) Fe2þ, Mg order-disorder in heated orthopyroxenes. Mineral. Soc. Am. Spl Paper,
2, 67 –81.
Virgo, D. and Mysen, B. (1985) The structural state of iron in oxidized vs. reduced glasses at 1 atm: a 57Fe
Mössbauer study. Phys. Chem. Miner., 12, 65 –76.
Virgo, D., Fei, Y., Mysen, B.O., Wang, Y. and Mao, H.K. (1994) Temperature-dependent electron delocalization
in (Mg, Fe) SiO3 perovskite. Am. Mineral., 79, 826–837.
Vizgirda, J. and Ahrens, T.J. (1982) Shock compression of aragonite and implications for the equation of state of
carbonates. J. Geophys. Res., B87, 4747–4758.
Vocadlo, L. and Price, G.D. (1996) Phys. Chem. Miner., 23, 42.
Vocadlo, L., Poirer, J.P. and Price, G.D. (2000a) Grüneisen parameters and isothermal equation of state. Am.
Mineral., 85, 390 –395.
Vocaldo, L., Brodholt, J., Alfè, D., Gillan, M.J. and Price, G.D. (2000b) Ab initio calculations on the polymorphs
of iron at core conditions. Phys. Earth Planet. Interiors, 117, 123.
Vogel, N. (1995) Foothills hazard: Homes built on mine tailings. San Jose Mercury News, 1.
Von Bargon, N. and Boehler, R. (1990) High Press. Res., 6, 133.
Von Huene, R. and Scholl, D. (1991) Observations at convergent margins concerning sediment subduction,
subduction erosion, and the growth of continenal crust. Rev. Geophys., 29, 279–316.
Vos, W.L., Finger, L.W., Hemley, R.J., Hu, J.Z., Mao, H.K. and Shouten, J.A. (1992) High-pressure van der
Waals compound in solid nitrogen–helium mixtures. Nature, 358, 46 –48.
Vos, W.L., Finger, L.W., Hemley, R.J. and Mao, H.K. (1993) Novel H2-H2O clathrates at high pressures.
Phys. Rev. Lett., 71, 3150–3153.
Vos, W.L., Finger, L.W., Hemley, R.J., Mao, H.K. and Yoder, H.S. (1994) In: Schmidt, S.C. et al. (eds.), High-
Pressure Science and Technology. New York: AIP, 857 pp.
Wadati, K. (1935) On the activity of deep-focus earthquakes in the Japanese islands and neighbourhoods.
Geophys. Mag., 1, 162 –202.
Wadhawan, V.K. (1982) Ferroelasticity and related properties of crystals. Phase Trans., 3, 3–103.
Wagner, M. (1968) Investigation of lattice dynamics by means of vibronic spectra. Z. Phys., 214,
79–107.
Wagner, W., Saul, A. and Pruss, A. (1994) J. Phys. Chem. Ref. Data, 23, 515.
Walcott, R.I. (1970) Flexural rigidity, thickness, and viscosity of the lithosphere. J. Geophys. Res., 75,
3941– 3954.
Walck, M.C. (1984) The P-wave UM structures beneath an active spreading centre of the Gulf of California.
Geophys. J. R. Astron. Soc., 76, 697–723.
Walter, M.J. (1997) One step to Earth. Nature, 387, 453.
Walter, M.J. (1999) Screws tighten on the core. Nature, 398, 563–565.
Wanamaker, B.J. (1994) Geophys. Res. Lett., 21, 21.
References 1193

Wanamaker, B.J. and Duba, A.G. (1993) Electrical conductivity of Carlos olivine along [100] under oxygen-
and pyrope buffer conditions and implications for defect equilibria. J. Geophys. Res., 98,
489–500.
Wang, C.Y. (1968) Constitution of the lower mantle as evidenced from shock wave data for some rocks.
J. Geophys. Res., 73, 6459–6476.
Wang, X. and Liou, J.G. (1993) Ultrahigh-pressure metamorphism of carbonate rocks in the Dabie Mountains,
central China. J. Meta. Geol., 11, 575 –588.
Wang, H. and Simmon, G. (1974) Elasticity of some mantle crystal structures 3. Spessartite– almandine garnet.
J. Geophys. Res., 79, 2607–2613.
Wang, Y. and Weidner, D.J. (1994) Thermoelasticity of CaSiO3 perovskite and implications for the lowe mantle.
Geophys. Res. Lett., 21(10), 895 –898.
Wang, Y. and Weidner, D.J. (1996) Pure and Applied Geophysics, Special Issue 146(3/4), 533–549.
Wang, Y., Weidner, D.J., Liebermann, R.C., Liu, X., Ko, J., Vaughan, M., Zhao, T., Yeganeh, Y., Haeri, A. and
Pacalo, J.B. (1991a) Science, 251, 410 –413.
Wang, Y., Zhao, Y., Weidner, D.J., Liebermann, R.C., Parise, J.B. and Cox, D.E. (1991b) High temperature
behavior of (Mg, Fe)SiO3 perovskite at 1bar. EOS Trans. Am. Geophys. Union, 72(44), 464, Abstracts.
Wang, Y., Gasparik, T. and Liebermann, R.C. (1993) Modulus microstructure in synthetic majorite.
Am. Mineral., 78, 1165–1173.
Wang, Y., Weidner, D.J., Liebermann, R.C. and Zhao, Y. (1994a) P – V – T equation of state of (Mg,Fe)SiO3
perovskite: constraints on composition of the lower mantle. Phys. Earth Planet. Interiors, 83, 13– 40.
Wang, A., Meyer, H.O.A. and Dele-Dubois, M.L. (1994b) Magnesite: as inclusion in natural diamond. GSA/MSA
Annual Meeting Seattle, A417.
Wang, Y., Weidner, D.J. and Guyot, F. (1996) Thermal equation of state of CaSiO3 perovskite. J. Geophys. Res.,
101, 661 –672.
Wang, W. and Sueno, S. (1997a) Min. J., 18, 9.
Wang, Y., Martinez, I., Guyot, F. and Liebermann, R.C. (1997b) The breakdown of olivne to perovskite and
magnesiowüstite. Science, 275, 510 –513.
Wänke, H. (1981) Constitution of terrestrial planets. Phil. Trans. R. Soc. London A, 303, 287–302.
Wänke, H. and Dreibus, G. (1988) Chemical composition and accretion history of terrestrial planets. Phil. Trans.
R. Soc. London Ser. A, 325, 545–557.
Wänke, H., Dreibus, G. and Jagoutz, E. (1984) Mantle chemistry and accretion history of the Earth. In: Kroner, A.
et al. (eds.), Archaean Geochemistry. Berlin: Springer, pp. 1–24.
Warde (1992) Science, 258, 770.
Wark, J.S., Whitlock, R.R., Hauer, A., Swain, P.J. and Solone (1988) In: Schmidt, S.C. and Holmes, N.C. (eds.),
Shock Waves in Condensed Matter. Amsterdam: Elsevier, 781 pp.
Wasilewski, P. (1977) Adv. Earth Planet. Sci., 1, 123.
Wasserburg, G.E. et al. (1994) Astrophys. J., 424, 412–428.
Watanabe, H. (1987) In: Manghnani, M.H. and Syono, Y. (eds.), High-Pressure Research in Mineral Physics.
Washington, DC: AGU, pp. 275–278.
Watson, R.E. (1958) Analytic Hartree– Fock solutions for O¼. Phys. Rev., 111, 1108–1110.
Watson, E.B. (1979) Calcium diffusion in a simple silicate melt to 30 kbar. Geochim. Cosmochim. Acta,
43, 313–322.
Watson, L.L., Hutcheon, I.D., Epstein, S. and Stolper, E.M. (1994) Water on Mars: clues from deuterium/
hydrogen and water contents of hydrous phases in SNC meteorites. Science, 265, 86 –90.
Watt, J.P., Davies, G.F. and O’Connell, R. (1976a) The elastic properties of composite materials. J. Rev. Geophys.
Space Phys., 14, 541–563.
Watt, J.P., Davies, G.F. and O’Connell, R.J. (1976b) The elastic properties of composite materials. Geophs. J. R.
Astron. Soc., 54, 601–630.
Watt, J.P., Davies, G.F. and O’Connell, R.J. (1976c) The elastic properties of composite materials. Rev. Geophys.
Space Phys., 14, 541–563.
Waychunas, G.A. (1986) Performance and use of Mössbauer goodness-of-fit parameters: response to spectra of
varying signal noise ratio and possible misinterpretations. Am. Mineral., 71, 1261– 1265.
1194 References

Wayte, G.J., Worden, R.H., Rubie, D.C. and Droop, G.T.R. (1989) Contrib. Mineral. Petrol., 101, 426.
Weathers, M.S. and Bassett, W.A. (1987) Melting of carbon at 50 to 300 kbar. Phys. Chem. Miner., 15, 105.
Webb, S.L. (1989) The elasticity of the upper mantle orthosilicates olivine and garnet to 3 GPa. Phys. Chem.
Miner., 16, 684–692.
Webb, S.L. and Dingwell, D.B. (1990) Non-newtonain rheology of igneous melts at high stresses and strain
rates: experimental results for rhyolite, andesite, basalt, and nephelinite. J. Geophys. Res., 95(B10),
15695–15701, 10.1029/90JB00247.
Webb, S. and Courtial, P. (1996) Compressibility of melts in the CaO–Al2O3 –SiO2 system. Geochim.
Cosmochim. Acta, 60(1), 75 –86.
Webb, S.L. and Dingwell, D.B. (1990) J. Geophys. Res., 95, 15695– 15701.
Webb, S.L. and Dingwell, D.B. (1995) Viscoelasticity. Rev. Miner., 32, 95– 118.
Webb, S.L., Ross, C.R. and Liebertz, J. (1992) Thermal expansion and spontaneous strain associated with the
normal incommensurate phase transition in melilites. Phys. Chem. Miner., 18, 522–525.
Weber, T., Herbig, J., Mark, M. et al. (2003) Bose-Einstein condensation of cesium. Science, 299, 232–235.
Wechsler, B.A. and Prewitt, C.T. (1984) Crystal structure of ilmenite (FeTiO3), at high temperature and at high
pressure. Am. Mineral., 69, 176–185.
Weidner, D.J. (1986) Mantle models based on measured physical properties of minerals. In: Saxena, S.K. (ed.),
Chemistry and Physics of Terrestrial Planets. Berlin: Springer, pp. 251–274.
Weidner, D.J. and Ito, E. (1985) Elasticity of MgSiO3 in the ilmenite phase. Phys. Earth Planet. Interiors,
40, 65 –70.
Weidner, D.J., Wang, H. and Ito, J. (1978) Elasticity of the orthoenstatite. Phys. Earth Planet. Interiors, 17,
7–13.
Weidner, D.J., Bass, J.D., Ringwood, A.E. and Sinclair, W. (1982a) The single-crystal elastic moduli of stishovite.
J. Geophys. Res., 87, 4740–4746.
Weidner, D.J., Bass, J.D. and Vaughan, M.T. (1982b) The effect of crystal structure and composition on elastic
properties of silicates. In: Akimoto, S. and Manghnani, M. (eds.), High-pressure Research in Geophysics.
Japan: Center for Academic Publications, pp. 125 –133.
Weidner, D.J., Sawamoto, H., Sasaki, S. and Kumazawa, M. (1984) Single crystal elastic properties of the spinel
phase of Mg2SiO4. J. Geophys. Res., 89, 7852–7860.
Weidner, D.J., Wang, Y. and Vaughan, M.T. (1994) Science, 266, 419–422.
Weinstein, S.A. (1992) Induced compositional layering in a convecting fluid layer by an endothermic phase
transition. Earth. Planet. Sci. Lett., 113, 23– 39.
Weir, S.T., Mitchel, A.C. and Nellis, W.J. (1996) Metallization of fluid molecular hydrogen fat 140 GPa
(1.4 Mbar). Phys. Rev. Lett., 76, 1860–1863.
Weisz, R.S. (1951) Interatomic distances and ferromagnetism in spinels. Phys. Rev., 84, 379.
Wells, P.R.A. (1977) Pyroxene thermometry in simple and complex systems. Contrib. Mineral. Petrol.,
62, 129–139.
Wells, A.F. (ed.) (1984) Structural Inorganic Chemistry, 5th ed. Oxford, England: Clarendon Press, 1095 pp.
Wen, L. and Helmberger, D.V. (1998) Ultra-low velocity zones near the core-mantle boundary from broadband
PKP precursors. Science, 279, 1701.
Weng, K., Mao, H.K. and Bell, P.M. (1982) Lattice parameters of the perovskite phases in the system MgSiO3 –
CaSiO3 –Al2O3. Carnegie Inst. Washington Yearb., 81, 273–277.
Weng, K., Xu, J., Mao, H.K. and Bell, P.M. (1983) Preliminary Fourier-transform infrared spectral data on the
SiO82
6 octahedral group in silicate perovskite. Carnegie Inst. Washington Yearb., 82, 355– 356.
Wenk, H.R. (1978) Geology, 6, 507.
Wenk, H.R. (1999) A voyage through the deformed earth with the self-consistent model. Model Simul. Mater. Sci.
Eng., 7, 699–722.
Wenk, H.R. and Kroll, H. (1984) Analysis of P1I  1 and C1 plagioclase structures. Bull. Mineral., 107,
467–487.
Wenk, H.R., Takeshita, T. and Jeanloz, R. (1988) Development of texture and elastic anisotropy during
deformation of hcp metals. Geophys. Res. Lett., 15, 76–79.
References 1195

Wenk, H.-R., Matthies, S., Hemley, R.J., Mao, H.-K. and Shu, J. (2000a) The plastic deformation of iron at
pressures of the Earth’s inner core. Nature, 405, 1044–1046.
Wenk, H.-R., Baumgardener, J.R., Lebensohn, R.A. and Tome, C.N. (2000b) A convection model to explain
anisotropy of the inner core. J. Geophys. Res., 105, 5663– 5667.
Wentrof, R.H. and DeVries, R.C. (1987) High-pressure synthesis (chemistry). Encyclopedia Phys. Sci. Technol.,
6, 492 –506.
Wentzcovitch, R.M. and Price, G.D. (1996) High pressure studies of mantle minerals by ab initio variable cell
shape molecular dynamics. In: Silvi, B. and Darco, P. (eds.), Molecular Engineering, Kluwer: Dordrecht
p. 31 –61.
WentCovitch, R.M. and Stixrude, L. (1997) Crystal chemistry of forsterite: a first principle study. Am. Miner., 82,
663–671.
Wentzcovitch, R.M., Martins, J.L. and Price, G.D. (1993) Ab initio molecular dynamics with variable, cell shape;
Application to MgSiO3 perovskite. Phys. Rev. Lett., 70, 3947–3950.
Wentzcovitch, R.M., Hugh-Jones, D.A., Angel, R.J. and Price, G.D. (1995a) Ab initio study of MgSiO3 C2=c
enstatite. Phys. Chem. Miner., 22, 453–460.
Wentzcovitch, R.M., Ross, N.L. and Price, G.D. (1995b) Ab initio study of MgSiO3 and CaSiO3 perovskite at
lower mantle pressure. Phys. Earth Planet. Interiors, 90, 101–112.
Wentzcovitch, R.M., Silva, C., Chelikowsky, J.R. and Binggeli, N. (1998) Phys. Rev. Lett., 80, 2149.
Wetherill, G.W. (1986) Accumulation of the terrestrial planets and implications concerning lunar origin.
In: Hartman, W.K. et al. (eds.), Origin of the Moon. Houston, TX: Lunar Planet Institute, pp. 519–550.
Wetherill, G.W. (1990) Formation of the Earth. Annu. Rev. Earth Planet. Sci., 18, 205–256.
Whaler, K. and Holme, R. (1996) Catching the inner core in a spin. Nature, 382, 205–206.
White, W.M. (1996) Best friend hides deep secret. Nature, 379, 117–118.
White, R.M. and Geballe, T.H. (1979) Long Range Order in Solids. New York: Academic Press, p. 142, and
references therein.
White, N. and Lovell, B. (1997) Measuring the pulse of a plume with the sedimentary record. Nature,
387, 888–890.
Widmer, R., Masters, F. and Gillbert, F. (1991) J. Int., 104, 541.
Widom, E. and Shirey, S.B. (1996) Os isotope systematics in the Azores: implications for mantle plume sources.
Earth Planet. Sci. Lett., 142(3–4), 451 –465.
Wigner, E. and Huntington, H.B. (1935) On the possibility of a metallic modification of hydrogen. J. Chem. Phys.,
3, 764 –770.
Wijs, G.A., Kresse, G. et al. (1998) The viscosity of liquid iron at the physical conditions of the Earth’s core.
Nature, 392, 805–807.
Wilburn, D.R., Bassett, W.A., Sato, Y. and Akimoto, S. (1978) X-ray diffraction compression studies of hematite
under hydrostatic, isothermal conditions. J. Geophys. Res., 83, 3509–3512.
Wilde, S.A., Valley, J.W., Peck, W.H. and Graham, C.M. (2001) Nature, 409, 175–178.
Will, G., Hoffbauer, W., Hinze, E. and Lauterjung, J. (1986) The compressibility of forsterite up to 300 kb with
synchrotron radiation. Physica, B139/140, 193–197.
William, A.B., Wu, T.C., Chou, I.M., Haselton, H.T. Jr., Frantz, B.O., Mysen, W.L., Huang, Sharma, S.K. and
Schifert, D. (1996) The hydrothermal diamond anvil cell (HDAC) and its applications. In: Dyar, M.D.,
McCammon, C.A. and Schaefer, M.W. (eds.), Mineral Spectroscopy: A Tribute to Roger G. Burns,
The Geochemical Society Special Publication, 5, pp. 261–272.
Williams, R.J.P. (1959) Deposition of trace elements in a basic magma. Nature, 184, 44.
Williams, Q. (1998) The temperature contrast across D00 . In: Gurnis, M., Wysession, M.E., Knittle, E. and
Buffett, B. (eds.), The Core Mantle Boundary Region. Washington, DC: American Geophysical Union,
pp. 73 –81.
Williams, Q. (1992) A vibrational spectroscopic study of hydrogen in high pressure mineral assemblages. In:
Syono, Y. and Manghnani, M.H. (eds.), High-Pressure Research: Application to Earth and Planetary
Sciences. Tokyo/Washington DC: Terra Scienific/American Geophysical Union, pp. 289–296.
Williams, Q. and Garnero, E. (1996) Seismic evidence for partial melt at the base of Earth’s mantle. Science,
273, 1528–1530.
1196 References

Williams, Q. and Jeanloz, R. (1986) Phys. Rev. Lett., 56, 163.

Williams, Q. and Jeanloz, R. (1987) Phys. Rev. Lett., 59, 1132.
Williams, Q. and Jeanloz, R. (1988) Spectroscopic evidence for pressure induced coordination change in silicate
glasses and melts. Science, 239, 902 –905.
Williams, Q. and Jeanloz, R. (1989) Static amorphization of anorthite at 300 K and comparison with diaplectic
glass. Nature, 338, 413 –415.
Williams, Q., Jeanloz, R. and McMillan, P. (1987) Vibrational spectrum of MgSiO3 perovskite: zero-pressure
Raman and mid-infrared spectra to 27 GPa. J. Geophys. Res., 92, 8116–8128.
Williams, Q., Knittle, E., Reichlin, R., Martin, S. and Jeanloz, R. (1990) Structural and electronic properties of
Fe2SiO4-fayalite at ultrahigh pressure: amorphization and gap enclosure. J. Geophys. Res., 95,
21549–21563.
Williams, Q., Hemley, R.J., Kruger, M.B. and Jeanloz, R. (1993) High-pressure infrared spectra of a-quartz,
coesite, stishovite and silica glass. J. Geophys. Res., 98, 22157–22170.
Williams, Q., Revenaugh, E. and Garnero, E. (1998) A correlation between ultralow basal velocities in the mantle
and hot spot. Science, 280, 546 –549.
Wilson, J.A. (1972) Systematics of the breakdown of Mott insulation in binary transition metal compounds.
Adv. Phys., 21, 143–198.
Wilson, C.C. (1994) Structural studies of schultenite in the temperature range 125–324 K by pulsed single-crystal
neutron diffraction: hydrogen ordering and structural distortions. Mineral. Mag., 58, 629–634.
Windley, B.F. and Smith, J.V. (1976) Archean high grade complexes and modern continental margins. Nature,
260, 671 –675.
Winkler, B. and Dove, M.T. (1992) Thermodynamic properties of MgSiO3 perovskite derived from large scale
molecular dynamics simulations. Phys. Chem. Miner., 18, 407–415.
Winkler, B., Langer, K. and Johannsen, P.G. (1989) The influence of pressure on the OH valence vibration of
zoisite: an infrared spectroscopic study. Phys. Chem. Miner., 16, 668–671.
Winter, J.K. and Ghose, S. (1979) Thermal expansion and high-temperature crystal chemistry of the Al2SiO5
polymorphs. Am. Mineral., 64, 573–586.
Winters, R.R., Garg, A. and Hammack, W.S. (1992) Phys. Rev. Lett., 69, 3751.
Wittlinge, R.J., Werner, S. and Schulz, H. (1997) On the amorphisation of ZnCr2S4 spinel under high pressure:
X-ray diffraction studies. Phys. Chem. Miner., 24(7), 597–600.
Wolf, G.H. and Bukowinski, M.S.T. (1985) Ab initio structural and thermoelastic properties of orthorhombic
MgSiO3 perovskite. Geophys. Res. Lett., 12, 809– 812.
Wolf, G.H. and Bukowinski, S.T.M. (1987) Theoretical study of the structural properties and equations of state of
MgSiO3 and CaSiO3 perovskites: implications for lower mantle composition. In: Manghnani, M.H. and
Syono, Y. (eds.), High-Pressure Research in Mineral Physics. Tokyo/Washington: Terrs Scientific/
American Geophysical Union, pp. 313–331.
Wolf, G.H. and Jeanloz, R. (1985) Lattice dynamics and structural distortions of CaSiO3 and MgSiO3 perovskites.
Geophys. Res. Lett., 12, 413–416.
Wolf, G.H., Durben, D.J. and McMillan, P.F. (1990a) J. Chem. Phys., 93, 2280.
Wolf, G.H., Durben, D.J. and McMillan, P.F. (1990b) Raman spectroscopic study of the vibrational properties and
reversion of CaGeO3 and MgSiO3 perovskites as a function of temperature. EOS Trans. Am. Geophys.
Union, 71, 1667, Abstract.
Wolfe, C.J. (1998) Prospecting for hotspot roots. Nature, 396, 212–213.
Wolfe, J.P. and Hauser, M.R. (1995) Acoustic wavefront imaging. Ann. Physik, 4, 99 –126.
Wolfe, C.J., Bjarnason, I.Th., VanDecar, J.C. and Solomon, S.C. (1997) Seismic structure of the Iceland mantle
plume. Nature, 385, 245–247.
Wolfenstine, J. and Kohlstedt, D.L. (1994) High-temperature creep and kinetic decomposition of Ni2SiO4. Phys.
Chem. Miner., 21(4), 234–243.
Wood, B.J. (1990a) An experimental test of the spinel peridotite oxygen barometer. J. Geophys. Res. Solid Earth
Planets, 95, 15845–15851.
References 1197

Wood, B.J. (1990b) Postspinel transformations and the width of the 670-km discontinuity: a comment on
“Postspinel transformations in the system Mg2SiO4 –Fe2SiO4 and some geophysical implications” by
E. Ito and E. Takahashi. J. Geophys. Res., 95, 12681–12685.
Wood, B.J. (1993) Carbon in the core. Earth Planet. Sci. Lett., 117, 593–607.
Wood, B.J. (1995) The effect of H2O on the 410-kilometer seismic discontinuity. Science, 268, 74 –76.
Wood, B.J. (1997) Hydrogen: an important constituent of the core? Science, 278, 1727.
Wood, B.J. (2000) Phase transformations and partitioning relations in peridotite under lower mantle conditions.
Earth Planet. Sci. Lett., 174, 341 –354.
Wood, B.J. and Henderson, C.M.B. (1978) Compositions and unit-cell parameters of synthetic non-stoichometric
tschermakitic clinopyroxene. Am. Mineral., 42, 109–124.
Wood, B.J. and Nell, J. (1991) High-temperature electrical conductivity of the lower mantle phase (Mg, Fe)O.
Nature, 351, 309–311.
Wood, B.J. and Rubie, D.C. (1996) The effect of alumina on phase transformations at the 660-kilometer
discontinuity from Fe–Mg partitioning experiments. Science, 273, 1522.
Wood, B.J. and Virgo, D. (1989) Upper mantle oxidation state: ferric iron contents of lherzolite spinels by
57
Fe Mössbauer spectroscopy and resultant oxygen fugacities. Geochim. Cosmochim. Acta, 53,
1277– 1291.
Wood, B.J., Holland, T.J.B., Newton, R.C. and Kleppa, O.J. (1980) Thermochemistry of jadeite–diopside
pyroxenes. Geochim. Cosmochim. Acta, 44, 1363–1371.
Wood, T.E., Siedle, A.R., Hill, J.R., Skarjune, R.P. and Goodbrake, C.J. (1990) Better Ceramics through
Chemistry (IV). Mater. Res. Soc. Symp. Proc., 180, 97.
Wood, B.J., Pawley, A. and Frost, D.R. (1994) Water and carbon in the Earth’s mantle. Phil. Trans. R. Soc.
London Ser. A, 354, 1495– 1511.
Woodcock, L.V. (1997) Entropy difference between the face-centred cubic and hexagonal close-packed crystal
structures. Nature, 385, 141–143.
Woodhead, J.D., Greenwood, P., Harmon, R.S. and Stoffers, P. (1993) Oxygen isotope evidence for recycled crust
in the source of FM type ocean island basalt. Nature, 362, 809–813.
Woodland, A.B. and Angel, R.J. (1997) Reversal of the orthoferrosilicate High-P clinoferrosilite transition,
a phase diagram for FeSiO3 and implications for the mineralogy of the Earth’s upper mantle.
Eur. J. Mineral., 9, 245–254.
Woodland, A.B. and Angel, R.J. (1998) Crystals structure of a new spinelloid with the wadsleyite structure in the
system Fe3O4 –Fe2SiO4 and implications for the Earth’s mantle. Am. Mineral., 83, 404–408.
Woodland, A.B. and O’Neill, H.St.C. (1995) Phase relation between Ca3Fe3þ 2þ 3þ
2 Si3O12 – Fe3 –Fe2 Si3O12 garnet
and CaFeSi2O6 – Fe2Si2O6 pyroxene solid solutions. Contrib. Mineral. Petrol., 121(1), 87 –98.
Woodland, A.B., McCammon, C. and Angel, R.J. (1997) Intersite partitioning of Mg and Fe in Ca-free high-
pressure C2=c clinopyroxene. Am. Mineral., 82, 923–930.
Woutersen, S., Emmerichs, U., Han-Kwang, N. and Bakker, H.J. (1998) Anomalous temperature dependence of
vibrational lifetimes in water and ice. Phys. Rev. Lett, 81, 1106–1109.
Wren, M.C. and Ackland, G.J. (1996) Ab initio studies of structural instabilities in magnesium silicate perovskite.
Phys. Chem. Miner., 23, 107 –118.
Wright, K. and Catlow, C.R.A. (1994) A computer simulation study of OH defects in olivine. Phys. Chem. Miner.,
20, 515 –518.
Wright, K. and Catlow, C.R.A. (1996) Calculations on the energetics of water dissolution in wadsleyite. Phys.
Chem. Miner., 23, 38– 41.
Wright, K. and Price, G.D. (1993) Computer simulation of defects and diffusion in perovskite. J. Geophys. Res.,
98, 22245–22253.
Wright, K., Freer, R. and Catlow, C.R.A. (1994) The energetics and structure of the hydrogarnet defect in
grossular: a computer simulation study. Phys. Chem. Miner., 20, 500– 503.
Wu, C.C. and Mason, T.O. (1981) Thermopower measurement of cation distribution in magnetite. J. Am. Ceram.
Soc., 64, 520–522.
1198 References

Wu, T.C., Bassett, W.A., Huang, W.L., Guggenheim, S., Koster, A.F. and Van Groos (1997) Montmorrillonite
under high H2O pressures: stability of hydrate phases, rehydration hysteresis, and the effect of interlayer
cations. Am. Miner., 82, 69–78.
Wunder, B. (1996) Gleichgewichts expeiment in system MgO–SiO2 –H2O: volaufige stabilities felder von der
klinohumit (Mg9Si4O16(OH)2), chondrodite (Mg5Si2O8(OH)2) and phase A (Mg7Si2O8(OH)6. Berichte
der deutchen Mineralogischen Gesellschaft, Beiheft I zum. Eur. J. Mineral., 8–321.
Wunder, B. (1998) Equilibrium experiments in the system MgO–SiO2 – H2O (MSH): stability fields of clinohumite-
OH [Mg9Si4O16(OH)2], chondrodite– OH [Mg5Si2O8(OH)2] and phase A (Mg7Si2O8(OH)6). Contrib.
Mineral. Petrol., 132, 111 –120.
Wunder, B. and Schreyer, W. (1992) Metastability of the 10 Å-phase in the system MgO–SiO2 –H2O (MSH):
what about MSH phases in subduction zones? J. Petrol., 334, 877–889.
Wunder, B. and Schreyer, W. (1997) Antigorite: high-pressure stability in the system MgO–SiO2 –H2O(MSH).
Lithos, 41, 213– 227.
Wunder, B., Medenbach, O., Krause, W. and Schreyer, W. (1993a) Synthesis, properties and stability of
Al3Si2O7(OH)3 (phase Pi), a hydrous high-pressure phase in the system Al2O3 – SiO2 –H2 (ASH). Eur.
J. Mineral., 5, 637 –649.
Wunder, S., Rubie, D.C., Ross, C.R., Medenbach, O., Seifert, F. and Schreyer, W. (1993b) Synthesis, stability and
properties of Al2SiO4(OH)2: a fully hydrated analogue of topaz. Am. Mineral., 78, 285–297.
Wunder, B., Medenbach, O., Daniels, P. and Schreyer, W. (1995) First synthesis of the hydroxyl end member of
humite, Mg7 – Si3 –O12(OH)2. Am. Mineral., 80, 638–640.
Wyckoff, R.W.G. (1964) Crystal Structures, 2nd edn. Vol. 1. New York: Interscience.
Wyllie, P.J. and Huang, W.L. (1976) Carbonation and melting reactions in the system CaO– MgO–SiO2 –CO2
at mantle pressures with geophysical and petrological applications. Contrib. Mineral. Petrol., 54,
79–107.
Xiao, Y. and Lasaga, A.C. (1996) Geochim. Cosmochim. Acta, 57, 2283.
Xiao, Y.H., Kirkpatrick, R.J., Hay, R.L., Kim, Y.J. and Phillips, B.L. (1995) Investigation of Al, Si order in
K-feldspars using Al-27 and Si-29 MAS NMR. Mineral. Mag., 59, 47–61.
Xie, X., Minitti, M., Chen, M., Mao, H.K., Wang, D., Shu, J. and Fei, Y. (2002) Natural high-pressure polymorph
of merrillite in the shock melt veins of the Suizhou L6 chondrite. Geochim. Cosmochim. Acta, 66,
2439– 2444.
Xu, S. (1992) Science, 256, 80.
Xu, S., Okay, A.I., Ji, S., Sengor, A.M.C., Su, W. et al. (1992) Diamond from the Dabie Shan metamorphic rocks
and its implication for tectonic setting. Science, 256, 80–82.
Xu, J., Hu, J., Ming, L., Huang, E. and Xie, H. (1994) The compression of diaspore. AlO(OH) at room temperature
up to 27 GPa. Geophys. Res. Lett., 21, 161 –164.
Xu, Y., McCammon, C.A. and Poe, B.T. (1998) The effect of alumina on the electrical conductivity of silicate
pervoskite. Science, 282, 922.
Xu, J., Mao, H.K., Hemley, R.J. and Hines, E. (2002) The moissanite anvil cell: a new tool for high-pressure
research. J. Phys: Condens. Matter., 14, 11543–11548.
Xue, X., Stebbins, J.F., Kanzaki, M., McMillan, P.F. and Poe, B. (1991) Pressure-induced silicon coordination and
tetrahedral structural changes in alkali oxide-silica melts up to 12 GPa: NMR Raman and infrared
spectroscopy. Am. Mineral., 76, 8–26.
Yaganeh-Haeri, A., Weidner, D.J. and Ito, E. (1989) Elasticity of MgSiO3 in the perovskite structure. Science,
243, 787 –789.
Yagi, T. (1978) Experimental determination of thermal expansivity of several alkali halides at high pressures.
J. Phys. Chem. Solid, 39, 563 –571.
Yagi, T. (1994) Feldspars at high pressure. In: Parsons, I. (ed.), Feldspars and Their Reactions, NATO ASI series
C, Mathematical and Physical Sciences. Dordrecht: Kluwer, pp. 271– 321.
Yagi, A. and Akaogi, M. (1991) Phase transition of KAlSi3O8. Programme and abstracts of papers of the 32nd
High Pressure Conference, Japan, 1991, 70– 71.
Yagi, T. and Hishinuma, T. (1995) Iron hydride formed by the reaction of iron, silicate, and water implications for
the light-element of the Earth’s core. Geophys. Res. Lett., 22, 1933–1936.
References 1199

Yagi, T. and Mao, H.K. (1977) Crystal field spectra of the spinel polymorph of Ni2SiO4 at high pressure. Ann. Rep.
Geophys. Lab. Yearb., 76, 505–508.
Yagi, T., Mao, H.K. and Bell, P.M. (1978) Structure and crystal chemistry of perovskite-type MgSiO3. Phys.
Chem. Miner., 3, 97–110.
Yagi, T., Bell, P.M. and Mao, H.K. (1979) Phase relations in the system MgO–FeO–SiO2 between 150 and
700 kbar at 10008C. Carnergie Inst. Washington Yearb., 78, 614–618.
Yagi, T., Suzuki, T. and Akimoto, S. (1985) Static compression of wüstite (Fe0.98O) to 120 GPa. J. Geophys. Res.,
90, 8784–8788.
Yagi, T., Akaogi, M., Shimomura, O., Tamai, H. and Akimoto, S. (1987) High pressure and high temperature
equations of state of majorite. In: Manghnani, M.H. and Syono, Y. (eds.), High-Pressure Research in
Mineral Physics, Geophysical Monograph Series, Vol. 39. AGU, pp. 141–147.
Yagi, T., Uchiyama, Y., Akaogi, M. and Ito, E. (1992) Isothermal compression curve if MgSiO3 tetragonal garnet.
Phys. Earth Planet. Interiors, 74, 1–7.
Yagi, T., Funamori, N. and Utsumi, W. (1993) Stability and thermal expansion of orthorhombic
perovskite type MgSiO3 under lower mantle condition (abstract). EOS Trans. AGU, Fall. Mtg. Suppl.,
74, 571.
Yagi, A., Suzuki, T. and Akaogi, M. (1994a) High pressure transitions in the system KAISi3O8 –NaAlSi3O8. Phys.
Chem. Miner., 21(2), 12–17.
Yagi, T. et al. (1994b) In: Schmidt, S.C. et al. (eds.), High-Pressure Science and Technology. New York:
AIP, 943 pp.
Yagi, T., Fukuoka, K., Takei, H. and Syono, Y. (1998a) Geophys. Res. Lett., 15, 816.
Yagi, T., Kondo, T. and Syono, Y. (1998b) High pressure in situ X-ray diffraction study of MnO to 137 GPa and
comparison with shock compression experiment. In: Schmidt, S.C. et al. (eds.), Shock Compression of
Condensed Matter-1997. New York: American Institute of Physics, pp. 159–162.
Yamada, H., Matsui, Y. and Ito, E. (1983) Mineral. Mag., 47, 177–181.
Yamaguchi, Y., Akai, J. and Tomita, K. (1978) Contrib. Mineral. Petrol., 66, 263.
Yamamoto, K. and Akimoto, S.i. (1974) High pressure and high temperature investigations in the system
MgO–SiO2 – H2O. J. Solid State Chem., 9, 187–195.
Yamamoto, K. and Akimoto, S. (1977) The system MgO–SiO2 –H2O at high pressures and temperatures, Stability
field of hydroxyl, chondrodite, hydroxyl-clinohumite and 10 Å-phase. Am. J. Sci., 277, 288–312.
Yamanaka, T. and Takeuchi, Y. (1993) Order– disorder transition in MgAl2O4 spinel at high temperatures up to
17008. Z. Kristall., 165, 65–78.
Yamanaka, T., Takeuchi, Y. and Tokonami, M. (1984) Anharmonic thermal vibrations of atoms in MgAl2O4
spinel at temperature up to 1,933 K. Acta Crystallogr., B40, 96–102.
Yan, H. (1992). Dislocation recovery in olivine. MSc Thesis. University of Minnesota.
Yang, J. and Epstein, S. (1983) Interstellar organic matter in meteorites. Geochim. Cosmochim. Acta, 47,
2199– 2216.
Yang, M.H. and Flynn, C.P. (1994) Instrinsic diffusion properties of an oxide MgO. Phys. Rev. Lett., 73,
1809– 1812.
Yang, H. and Ghose, S. (1994) Thermal expansion, Debye temperature and Grüneisen parameter of synthetic
(Fe,Mg)SiO3 orthopyroxenes. Phys. Chem. Miner., 20, 575– 586.
Yang, H. and Hazen, R.M. (1998) Crystal chemistry of cation order–disorder in pseudobrookite -type MgTi2O5.
J. Solid State Chem., 138, 238– 244.
Yang, H. and Hazen, R.M. (1999) Comparative high-pressure crystal chemistry of karooite, MgTi2O5, with
different ordering states. Am. Mineral., 84, 130–137.
Yang, H. and Konzett, J. (2000) High-pressure synthesis of Na2Mg6Si6O18(OH)2—a new hydrous silicate phase
isostructural with aenigmatite. Am. Miner., 85, 259–262.
Yang, H. and Prewitt, C.T. (1999a) Crystal structure and compressibility of a tw-layer polytype of
pseudowollastonite (CaSiO3). Am. Miner., 84, 1902–1905.
Yang, H. and Prewitt, C.T. (1999b) On the crystal structure of pseudowollastonite. Am. Mineral., 84, 929–932.
Yang, H., Downs, R.T., Finger, L.W., Hazen, R.M. and Prewitt, C.T. (1997a) Compressibility and crystal
structure of kyanite, Al2SiO5 at high pressure. Am. Mineral., 82, 467–474.
1200 References

Yang, H., Hazen, R.M., Downs, R.T. and Finger, L.W. (1997b) Structural change associated with the
incommensurate normal phase transition in akermanite Ca2MgSi2O7 at high pressure. Phys. Chem.
Miner., 24, 510–519.
Yang, H., Hazen, R.M., Finger, L.W., Prewitt, C.T. and Downs, R.T. (1997c) Compressibility and crystal
structure of sillimanite, Al2SiO5, at high pressure. Phys. Chem. Miner., 25, 39–47.
Yang, H., Prewitt, C.T. and Frost, D.J. (1997d) Crystal structure of the dense hydrous magnesium silicate, phase
D. Am. Mineral., 82, 651 –654.
Yang, H., Hazen, R.M., Prewitt, C.T., Finger, L.W., Lu, R. and Hemley, R.J. (1998) High-pressure single crystal
x-ray diffraction and infrared spectroscopic studies of the C2=m – P21 =m phase transition in
cummingtonite. Am. Mineral., 83, 288 –299.
Yang, H., Finger, L.W., Conrad, G., Prewitt, C.T. and Hazen, R.M. (1999) A new pyroxene structure at high
pressure: single crystal X-ray and Raman study of the Pbcm–P21cn phase transition in protopyroxene.
Am. Mineral., 84, 245 –256.
Yarger, J.L., Smith, K.H., Nieman, R.A., Diefenbacher, J., Wolf, G.H., Poe, B.T. and McMillan, P.F. (1995) Al
coordination changes in high-pressure aluminosilicate liquids. Science, 270, 19641–19965.
Yeganeh-Haeri, A. (1994) Synthesis and reinvestigation of the elastic properties of single-crystal magnesium-
silicate perovskite. Phys. Earth Planet. Interiors, 87, 111–121.
Yeganeh-Haeri, A., Weidner, D.J. and Ito, E. (1989) Single-crystal elastic moduli of magnesium metasilicate
perovskite. In: Navrotsky, A. and Weidner, D.J. (eds.), Perovskite: A Structure of Great Interest to
Geophysics and Materials Science, Geophysical Monograph Series, 45. Washington, DC: AGU,
pp. 13 –25.
Yeganeh-Haeri, A., Weidner, D.J. and Ito, E. (1990a) Elastic properties of the pyrope-majorite solid solution
series. Geophys. Res. Lett., 17, 2453–2456.
Yeganeh-Haeri, A., Weidner, D.J. and Ito, E. (1990b) Elasticity of MgSiO3 in the perovskite structure. Science,
248, 787.
Yesinowski, J.P., Eckert, H. and Rossman, G.R. (1988) Characterization of hydrous species in minerals by high-
speed 1H MAS-NMR. J. Am. Chem. Soc., 110, 1367–1375.
Yin, M.T. and Cohen, M.L. (1983) Phys. Rev. Lett., 50, 2006.
Yogurtcu, Y.K., Miller, A.J. and Saunders, G.A. (1990) Elastic behavior of YAG under pressure. J. Phys. C: Sol.
St. Phys., 13, 6585–6597.
Yoneda, A. (1990) Pressure derivative of elastic constants of single crystal MgO and MgAl2O4. J. Phys. Earth, 38,
19–55.
Yoneda, A. and Morioka, M. (1992) Pressure derivatives of elastic constants of single crystal forsterite,
high-pressure research: application to earth and planetary science. Geophys. Monogr. Ser., 67,
207–214.
Yoo, C.S. and Nellis, W.J. (1991) Phase transformations in carbon fullerences at high shock pressure. Science,
254, 1489–1491.
Yoo, C.S., Holmes, N.C. and See, E. (1992) Shock-induced optical changes in Al2O3 at 200 GPa:
implications for shock temperature measurements in metals. In: Schmidt, S., Dick, R.D., Forbes,
J.W. and Pasker, D.G. (eds.), Shock Compression of Condensed Matter. Amsterdam: North-Holland,
pp. 733 –736.
Yoo, C.S., Holmes, N.C. and Ross, M. (1993) Shock temperatures and melting of iron at earth core conditions.
Phys. Rev. Lett., 70, 3931–3934.
Yoo, C.S., Akellaa, J., Campbell, A.J., Mao, H.K. and Hemley, R.J. (1995) Phase diagram of Iron by in situ X-ray
diffraction: implications for Earth’s core. Science, 270, 1473–1475.
Yoo, C.S., Söderlind, P. and Campbell, A.J. (1996) DHCP as possible new 10 phase of iron at high pressure and
temperature. Phys. Lett., A15, 321– 328.
Yossida, S., Sumita, I. and Kumazawa, M. (1996) Growth model of the inner core coupled with outer core
dynamics and the resultant elastic anisotropy. J. Geophys. Res., 101, 28085–28103.
Young, D.A. and Grover, R. (1988) In: Shock Waves in Condensed Matter (eds. Schimidt and Holmes), North-
Holland, Amsterdam, 131 pp.
References 1201

Young, T.E., Green, H.W. II, Hofmeister, A.M. and Walker, D. (1993) Infrared spectroscopic investigation of
hydroxyl in b-(Mg, Fe)2SiO4 and coexisting olivine: implications for mantle evolution and dynamics.
Phys. Chem. Miner., 19, 409 –422.
Young, T.E., Green, H.W., II, Hofmeister, A.M. and Walker, D. (1994) J. Geophys. Res., 99, 15821.
Yu, R. and Krakauer, H. (1994) Linear-response calculations within the LAPW method. Phys. Rev. B 49,
4467– 4477.
Yu, R. and Krakauer, H. (1995) First principles determination of chain-structure instability in KNbO3. Phys. Res.
Lett., 74, 4067–4070.
Yuen, D.A., Cadek, O., Chopelas, A. and Matyaska, C. (1993) Geophysical evidences for thermal chemical
plumes in the lower mantle. Geophys. Res. Lett., 20, 899–903.
Yui, T.F., Rumble, D. III, and Lo, C.H. (1995) Unusually low d18O ultra-high pressure metamorphic rocks from
the Sulu Terrain, eastern China, Geochim et Cosmochim Acta, 59(13) 2859– 2864.
Yund, R.A., Blanpied, M.L., Tullis, T.E. and Weeks, J.D. (1990) J. Geophys. Res., 95, 15589.
Yurimoto, H. and Ohtani, E. (1992) Element partitioning between majorite and liquid: a secondary ion mass
spectrometric study. Geophys. Res. Lett., 19, 17 –20.
Yusa, H. and Inoue, T. (1997) Compressibility of hydrous wadsleyite (b-phase) in Mg2SiO4 by high pressure
X-ray diffraction. Geophys. Res. Lett., 24, 1831–1834.
Yutatake, H. and Shimada, M. (1978) Phys. Earth Planet. Interiors, 17, 193.
Zaho, Y. and Anderson, D.L. (1994) Mineral physics constraints on the chemical composition of the Earth’s lower
mantle. Phys. Earth Planet. Interiors, 85, 273–292.
Zang, J., Bishop, A.R. and Roder, H. (1996) Double degeneracy and Jahn–Teller effects in collossal
magnetoresistance perovskites. Am. Phys. Soc., 53, 8840–8843.
Zatman, S. and Bloxham, J. (1997) Nature, 388, 760.
Zaug, J., Abramson, E.H., Brown, J.M. and Slutsky, L.J. (1993) Sound velocities in olivine at Earth mantle
pressure. Science, 260, 1487– 1489.
Zener, C. (1951) Phys. Rev., 82, 403– 405.
Zerr, A. and Boehler, R. (1993) Mg, Fe, SiO3-perovskite to 625 kilobars: indication of a high melting temperature
in the lower mantle. Science, 262, 553–555.
Zerr, A., Serghiou, G. and Boehler, R. (1997) Melting of CaSiO3 perovskite of 430 kbar and first in-situ
measurements of lower mantle eutectic temperatures. Geophys. Res. Lett., 24(8), 909–912.
Zerr, A., Diegeler, A. and Boehler, R. (1998) Solidus of earth’s deep mantle. Science, 281,
243–246.
Zha, C.S., Duffy, T.S., Mao, H.K. and Hemley, R.J. (1993a) Elasticity of hydrogen to 24 GPa from single-crystal
Brillouin scattering and synchrotron x-ray diffraction. Phys. Rev. B, 48, 9246–9255.
Zha, J.M., Abramson, E.H., Brown, J.M. and Slutsky, L.J. (1993b) Sound velocities in olivine at earth mantle
pressures. Science, 260, 1487– 1489.
Zha, C.S., Duffy, T.S., Downs, R.T., Mao, H.K. and Hemley, R.J. (1996) Sound velocities and elasticity of single-
crystal forsterite to 16 GPa. J. Geophys. Res., 101, 17535–17545.
Zha, C.S., Hemley, R.J., Mao, H.K. and Duffy, T.S. (1997) Elasticity measurement and equation of state from
MgO to 60 GPa. EOS Trans. Am. Geophys. Union, 78, F752.
Zha, C.S., Duffy, T.S., Downs, R.T., Mao, H. and Hemley, R.J. (1998a) Brillouin scattering and x-ray
diffraction of San Carlos olivine: direct pressure determination to 32 GPa. Earth Planet. Sci. Lett., 159,
25–33.
Zha, C.S., Mao, K., Hemley, R.J. and Duffy, T.S. (1998b) Recent progress in high-pressure brillouin
scattering: olivine and ice. In: Rev. High Pressure Sci. Tech., 7(4), pp. 739–741. Kyoto: Japan
Society of High-Pressure Science and Technology.
Zha, C.S., Duffy, T.S., Downs, R.T., Mao, H.K., Hemley, R.J. and Weidner, D.J. (1998c) Single-crystal elasticity
of the a and b of Mg2SiO4 polymorphs at high pressure. Geophysical Monograph, 101, 9– 16, American
Geophysical Union.
Zhang, S.C. (1997) Science, 275, 1089– 1096.
Zhang, L. (1998) Single crystal hydrostatic compression of (Mg, Mn, Fe, Co)2SiO4 olivines. Phys. Chem. Miner.,
25, 308 –312.
1202 References

Zhang, J. (1999) Effect of pressure on the thermal expansion of MgO up to 8.2 GPa. Phys. Chem. Miner.,
27, 145 –148.
Zhang, H. and Bukowinski, M.S.T. (1991) Modified potential-induced-breathing model of potentials for close-
shell ions. Phys. Rev. B, 44, 2495–2503.
Zhang, J. and Guyot, F. (1999) Thermal equation of state of iron and Fe0.91Si0.09. Phys. Chem. Miner., 26,
206–211.
Zhang, L. and Hafner, S.S. (1992) High-pressure 57Fe g-resonance and compressibility of Ca(Fe, Mg)Si2O6
clinopyroxenes. Am. Mineral., 77, 462 –473.
Zhang, J. and Herzberg, C. (1994) Melting experiments on anhydrous peridotite KLB-1 from 5.0 to 22.5 GPa.
J. Geophys. Res., 99, 17729–17742.
Zhang, J. and Herzberg, C. (1996) Melting experiments on anhydrous peridotite, KLB-1: compositions of magmas
in the upper mantle and transitions zone. J. Geophys. Res., 101, 8271–8295.
Zhang, R.Y. and Liou, J.G. (1994) Significance of coesite inclusions in dolimite from eclogite in the southern
Dabie mountains, China. Am. Mineral., 80, 181 –186.
Zhang, R.Y. and Liou, J.G. (1996) Partial transformation of gabbro to coesite-bearing eclogites from the Sulu
terrane, eastern China. J. Meta. Geol., 15, 183–202.
Zhang, R.Y. and Liou, J.G. (1997) Partial transformation of gabbro to coesite-bearing eclogites from Yongkou,
the Sulu terrane, Eastern China. J. Meta. Geol., 15, 183–202.
Zhang, R.Y. and Liou, J.G. (1998) Ultrahigh-pressure metamorphism of the Sulu terrane, Eastern China: a
prospective view. Cont. Dyn.
Zhang, J. and Weidner, D.J. (1998) EOS, 709, F861.
Zhang, J., Liebermann, R.C., Gasparik, T., Herzberg, C.T. and Fei, Y. (1993) J. Geophys. Res., 98, 19785.
Zhang, H., Daniels, W.B. and Cohen, R.E. (1994a) Exciton energy and its pressure dependence in alkali halides.
Phys. Rev. B, 30(1), 70–74.
Zhang, R.Y., Liou, J.G. and Cong, B. (1994b) J. Metamorph. Geol., 12, 169.
Zhang, M., Salje, E.K.H., Bismayer, U., Unruh, H.G., Wruck, B. and Schmidt, C. (1995a) Phase transitions in
titanite CaTiSiO5: an infrared spectroscopic, dielectric response and heat capacity study. Phys. Chem.
Miner., 22, 41–49.
Zhang, R.Y., Liou, J.G. and Cong, B.L. (1995b) Ultrahigh-pressure metamorphism talc-, magnesite- and
Ti-clinohumite-bearing mafic-ultramafic complex, Dabie Mountains, east central China. J. Petrol.,
36, 1011–1038.
Zhang, R.Y., Liou, J.G. and Ernst, W.G. (1995c) Ultrahigh-pressure metamorphism and decompressional
P – T paths of eclogites and country rocks from Weihei, eastern China. Island Arch., 4, 293– 309.
Zhang, J., Li, B., Utsumi, W. and Liebermann, R.C. (1996a) In situ X-ray observations of the coesite–stishovite
transition: reversed phase boundary and kinetics. Phys. Chem. Miner., 23, 1– 10.
Zhang, J., Martiinez, I., Guyot, F., Gillet, P. and Saxena, S.K. (1996b) X-ray diffraction study of magnesite at high
pressure and high temperature. Phys. Chem. Miner., 24, 122–130.
Zhang, L., Ahsbahs, H., Kutoglu, A. and Hafner, S.S. (1997) Single crystal compression and crystal structure of
clinopyroxene up to 10 GPa. Am. Mineral., 82, 245–258.
Zhang, I., Ahsbahs, H. and Kutoglu, A. (1998a) Hydrostatic compression and crystal structure of pyrope to
33 GPa. Phys. Chem. Miner., 25, 301–307.
Zhang, Y.F., Fu, B., Li, Y.L., Xiao, Y. and Li, G.G. (1998b) Oxygen and hydrogen isotope geochemistry of
ultrahigh pressure eclogites from the Dabie Mountains and the Sulu terrane. Earth Planet. Sci. Lett.,
155, 113–129.
Zhang, L., Stanek, J., Hafner, S.S., Ahsbash, H., Grünsterudel, H.F., Metge, J. and Rüffer, R. (1999) 57Fe nuclear
forward scattering of synchrotron radiation in hedenbergite CaFeSi2O6 at hydrostatic pressure up to
68 GPa. Am. Mineral., 84, 447–453.
Zhao, Y. and Anderson, D.L. (1994) Mineral physics constraints on the chemical composition of the Earth’s lower
mantle. Phys. Earth Planet. Interiors, 85, 273–292.
Zhao, L.S. and Helmberger, D.V. (1993) Upper mantle compressional velocity structure beneath the North-West
Atlantic Ocean. J. Geophys. Res., 98, 14185–14196.
References 1203

Zhao, Y., Parise, J.B., Wang, Y., et al. (1994) High-pressure crystal chemistry of neighborite, NaMgF3: an angle-
dispersive diffraction study using monochromatic synchrotron x-radiation. Am. Miner., 79,
615–621.
Zhao, Y., Schiferl, D. and Shankland, T.J. (1995) A high P –T single crystal x-ray diffraction study of
thermoelastic of MgSiO3 orthoenstatite. Phys. Chem. Miner., 22, 393–398.
Zharikov, V.A., Ishbulatov, R.A. and Chundinovskikh, L.T. (1984) High-pressure clinopyroxene and the eclogite
barrier. Sov. Geol. Geophys., 25, 53– 61.
Zharkov, V.N. and Gudkova, T.V. (1992) In: Syono, Y. and Manghnani, M.H. (eds.), High-Pressure Research:
Application to Earth and Planetary Sciences. Tokyo: Terra Scientific, pp. 393– 401.
Zheng, H., Xie, H., Xu, Y., Song, M., Zhang, Y., Wang, M. and Wu, H. (1995) Experimental study of the influence
of hydrous minerals on the melting [behaviour] of rocks at high temperatures and pressures. Acta Geol.
Sinica, 69(4), 326– 336.
Zheng, Y.F., Fu, B., Li, Y.L., Xiao, Y. and Li, S.G. (1998) Oxygen and hydrogen isotope, geochemistry of
ultrahigh-pressure eclogites from the Dabie Mountains and the Sulu terrane. Earth Planet. Sci. Lett.,
155, 113–129.
Zhong, W., Vanderbilt, D. and Rabe, K.M. (1994) Phase transitions in BaTiO3 from first principles. Phys. Rev.
Lett., 73, 1861–1864.
Zhou, H. (1996) A high-resolution P-wave model for the top 1200 km of the mantle. J. Geophys. Res.,
101, 27791–27810.
Zimmer, C., Khurana, K.K. and Kivelson, M.G. (2000) Icarus, 147, 329– 347.
Zindler, A. and Hart, S. (1986) Annu. Rev. Earth Planet. Sci., 14, 493–571.
Zinner, F. and Whetten, R. (1992) Acc. Chem. Res., 25, 119–126.
Zinner, E., Amari, S., Wopenka, B. and Lewis, R.S. (1995) Meteoritics, 30, 209.
Zou, G., Mao, H.K., Bell, P.M. and Virgo, D. (1980) High pressure experiments on the iron oxide wüstite
(Fe12xO). Carnegie Inst. Washington Yearb., 79, 374–376.
Zou, G., Bell, P.M. and Mao, H.K. (1981) Application of the solid helium pressure medium in a study of the a– 1
Fe transition under hydrostatic pressure. Annu. Rep. Geophys. Lab., 80, 272–274.
Zuber, M.T. (2001) The crust and mantle of the Mars. Nature, 412, 220–227.
This page is intentionally left blank
 1205

Glossary

Adiabatic temperature gradient Temperature variation with depth due only to

the compression of the material. In a convecting system like the Earth’s mantle,
assuming constant viscosity and without internal heating, the temperature – depth
profile lies along an adiabat.
Anharmonic vibrations In solids the vibrations of particles loose harmonicity due to
differential interactive forces and nonuniformity in the lattice.
Atomic form factor Fourier transform of the atomic scattering potential.
Basalt A rock made by 5– 20% melting of peridotite. It is produced under mid-ocean
ridges (mid-ocean ridge basalt, MORB) by upwelling of hot mantle. Ocean island
basalts (OIBs) are products of similar processes under ocean islands.
Basis In a crystal the positions of the atoms in a unit cell written as fractions of
translation vectors constitute the basis vectors or basis. The crystal structure is defined
by specifying the lattice and the basis vectors, which then identify how each atom in
the crystal is positioned.
Bcc lattice Body-centered-cubic lattice. Lattice sites are at the vertices and the center of
a simple cubic unit cell.
Bernal model Random close packed sphere model for a liquid or an amorphous solid.
Body waves In homogeneous, isotropic, elastic solid only P and S waves can travel with
the velocities given by:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffi
l þ 2m m
a¼ ; b¼
r r

In a perfect liquid or gas, where m ¼ 0; only P-waves can exist. P-type waves are
called acoustic waves; in contrast S-type waves behave like electromagnetic waves.
Generally, heavier materials have higher a and b than lighter ones, as in most cases l
and m increase faster than r: The above relations also indicate that a and b are
independent of the frequency or wave number. Hence P-and S-waves do not change
shape as they travel.
p Employing known Poisson’s relation one obtains, s ¼ 0:25; and
l ¼ m; a ¼ b 3: In the Earth, a and b generally decrease downwards, because of
densification. a and b are lowest in sedimentary rocks, higher in igneous and
metamorphic rocks, and highest in basic igneous rocks.
Boltzman’s constant kB , is defined as

KB ¼ R/NA ¼ 1:38 £ 10223 J/deg ¼ 1:38 £ 10216 erg/deg:

where NA the Avogadro’s number.
1206 Glossary

Bonds in solids
Ionic bonds are characterized by their ionic (homopolar) bonds between alternating
positive and negative ions at lattice points, e.g., NaCl, LiF, metal oxides, sulfides,
carbides selenides etc. The potential energy of the bond ranges in the order of
hundred of kcal/mole (e.g., 180 kcal/mole for NaCl, 240 kcal/mole for LiF) or
105 joules/mole. Ionic crystals show high m.p., strong absorption in the IR region,
low thermal and electrical conductivity’s at low temperatures.
Metallic bonds are developed by interaction between positive ions and electron gas
(formed collectivization of valence electrons and uniform distribution of electron
density throughout the lattice). The potential energy of interaction of metallic
bonds is in the order of dozens of kcal/mole e.g., 26 kcal for Na and 94 kcal
for Fe.
Bonding orbital Singlet configuration of electrons shared between two atoms or
molecules giving rise to an attractive interaction.
Bose –Einstein statistics describe symmetric wave funcitons, for systems of particles
having integral spin called bosons (e.g., Photons and certain nuclei), with unlimited
number of particles (‘condensate’) in a given cell (of 6-dimensional phase space).
Bragg’s law Relationship between the scattering angle u; wavelength l; and periodic
layer spacing d; leading to Bragg scattering: 2d sinu ¼ nl:
Brillouin peaks Peaks in the density – density correlation function, Snn ðq; vÞ; in fluids at
v ¼ ^cq; where c is the velocity of sound. These contrast with Rayleigh peaks.
Brillouin scattering The scattering of light by phonons in solids and liquids.
Brillouin zone A Wigner-Seitz cell of a reciprocal lattice, hence the cell of smallest
volume enclosed by the planes that are perpendicular bisectors of reciprocal lattice
vectors.
Broken symmetry Term associated with states or thermodynamic phases with a lower
symmetry than that of the interaction Hamiltonian. For example, the Ising Hamiltonian
is invariant under reversal of all spins, but the all-spins-up or all-spins-down ground
states are not; and hence are broken-symmetry states.
Bulk modulus The second derivative (K) of the elastic free energy with respect to
isotropic strain. It is reciprocal of the compressibility b: For isotropic solids it is related
to the Lamé coefficients by K ¼ l þ 2m=d; where d is the spatial dimension. This
constant measures the resistance to a change in volume under pressure.
K ¼ 2dP=du
(The minus sign is necessary for K to be . 0, because when u , 0; P . 0).
With reference to l and m; its relation is K ¼ l þ 2=3 mA. perfectly incompressible
ðK ¼ 1Þ is not necessarily perfectly rigid: m – 1; l ¼ 1:
Burgers vector Vector specifying the strength of dislocations in periodic and quasi-
periodic structures including crystalline solids and smectic liquid crystals. A loop
which encloses a dislocation line will contain an extra step (the Burgers vector)
corresponding to a direct lattice vector. If the Burgers vector is parallel to the
dislocation line, we have a screw dislocation. If it is perpendicular, we have an edge
dislocation.
Glossary 1207

Chalcogens The chalcogen elements, members of Group VIa, generate a variety of

phases under pressure.
Chalcophile Elements having strong affinity for sulfur.
Chiral Molecule A molecule that does not have a mirror plane. A molecule with a
central atom and four different atoms at the vertices of a tetrahedron is a simple
example of a chiral molecule.
Chondritic meteorites A class of meteorites that record the early history of the Solar
System. They are thought to represent primitive planetary materials.
Clapeyron slope The slope in pressure –temperature space ðdP=dTÞ of an equilibrium
phase boundary separating the regions in which different phases are thermodynami-
cally stable. The Clapeyron slope is defined thermodynamically by the ratio of the
differences in entropy and volume between the product and reactant phases at
equilibrium. Transformations of the a-, b-, g- polymorphs of (Mg,Fe)2SiO4 have
positive Clapeyron slopes (negative volume change and negative entropy change),
whereas the spinel ! magnesiowűstite þ perovskite reaction has a negative
Clapeyron slope (negative volume change but positive entropy change), as do most
reactions producing perovskite at equilibrium.
Climb Motion of dislocation in a crystal perpendicular to the Burgers vector, b,
requiring motion of an entire plane of atoms.
Coherent and incoherent scattering If the scattering cross-section, bi ; of the particles
in a sample varies, there is coherent scattering proportional to lkbll2 and incoherent
scattering proportional to kb2i l 2 lkbll2 : Coherent scattering provides information
about inter-particle correlations, whereas incoherent scattering provides information
about the motion of individual particles. For neutrons, the scattering cross-section can
change for different isotopes or even for the same isotope in the different spin states.
This is responsible for the large incoherent cross section of hydrogen as compared to
that of deuterium.
Commensurate lattice Lattice that can be divided into two or more sub-lattices, each
of whose basis vectors is a rational multiple of the basis vector of the other sub-
lattices.
Compressibility Modulus K measuring change of volume V in response to change
in pressure P: KT ¼ 2V 21 ð›V=dPÞT ¼ n2 ð›n=›mÞT ; where n is the number density
and m is the chemical potential. Inverse of the bulk modulus K:
Constitutive equation Following Hooke’s law the stresses in a body are considered as
linear combinations of the strains, that is the relation.
Pxy ¼ aexx þ beyy þ cezz þ dexy þ feyz þ gezx

holds good, for other relations as well. When the principal axes of stress are considered
identical to the principal axes of strain, the relations become: Pxy ¼ 2mexy ; Pyz ¼
2meyz ; etc., where m is the rigidity (also called modulus of rigidity or shear modulus).
Thus, the shear stresses are proportional to the shear strain. On the other hand, the
relations for the normal stresses are
Pxx ¼ lu þ 2mexx
1208 Glossary

Pyy ¼ lu þ 2meyy ; etc., or generalizing, Pij ¼ ludij þ 2meij ; where dij (the Kronecker
delta) is 1 if i ¼ j; and 0 otherwise, and l is a constant (first Lame elastic constant),
and
 
1 dui duj
eij ¼ þ
2 dXj dXi

The equation is called the constitutive equation for isotropic elasticity, in which each
normal stress depends on the dilation and on the corresponding normal strain.
Conventional unit cell Unit cell whose shape most directly reflects the symmetry of the
lattice. Thus, the conventional unit cell of all cubic lattices, including bcc and fcc, is a
cube. The primitive cells for fcc and bcc lattices are not cubes.
Creep Time-dependent evolution of the strain in a solid subjected to a constant stress.
An ideal solid responds elastically to stress (with a time-independent strain). Creep
results from the motion of defects, i.e., vacancies, interstitials, dislocations, and grain
boundaries.
Critical density Density at the liquid – gas critical point.
Critical point Point in a phase diagram characterized by singularities in derivatives of
the free energy and related thermodynamics quantities.
Crystallographic point group Point group compatible with the symmetry of any
periodic crystal lattice composed of regularly repeated identical unit cells. Point
groups with five-fold symmetry operations (such as the icosahedral group in three
dimensions) are among those that are not crystallographic point groups.P
Cubic anisotropy Anisotropy of a cubic lattice; it leads to a term v i f4i in the
Hamiltonian of an On field theory.
Curie spin susceptibility Response of noninteracting isolated spins to an external
magnet field, x ¼ m2 =kB T; where m is the spin’s magnetic moment. In Curie law any
linear response is proportional to 1=T:
Defect Imperfection in an ordered structure. There are local defects, such as missing
atoms (vacancies) or extra atoms at points other than lattice sites (interstitials) in a
crystal, and there are topological defects such as dislocations characterized by some
nonvanishing quantized line (or surface) integral on a loop (surface) enclosing the
defect core.
Defects Three groups of defects: (i) zero dimensional or point defects, (ii) one-
dimensional or line defects, and (iii) three dimensional or planar defects. Point defects
include vacancies (atomic), interstitial atoms, clusters of atoms, and combination of
vacancies with atoms or electrons. These defects may contribute to colour and
electrical conductivities. Ex. OH-defects: In silicate structure OH-ions can convert the
Si –O –Si bridge to Si – OH: HO – Si by introducing weak OH –HO bond. This cause
ease in recrystallization and lessening mechanical strength. Line defects such as
dislocation may lead to fracture (brittle).
Deformation type When a material deforms with a break in cohesion it is called brittle
deformation. But when the deformation occurs by distributing the strain throughout the
mass of the body in a smoothly varying manner it is called ductile deformation. Above
Glossary 1209

a critical value of stress an ideally plastic material continuously deforms in a

permanent way.
Depletion (geochemical) It refers to the melt extraction of basaltic components (Si, Fe,
and Ca) from geochemically fertile mantle lherzolite. The residue that remains in the
mantle has low Si and high Mg/Fe ratios; the rocks are dunite (mostly olivine) and
harzburgite (olivine þ Mg-pyroxene). Melt extraction in early Earth also included
olivine-and pyroxene-rich komatiite (volcanic peridotite).
Deep earthquakes Defined conventionally as earthquakes with source depths greater
than 300 km. The depth distribution of earthquakes shows a minimum near 300 –
325 km and an increase at deeper depths, suggesting that the deeper earthquakes are a
distinct population caused by a different failure process than the shallower
intermediate-depth earthquakes. In many Wadati-Benioff zones, clusters of inter-
mediate-depth earthquakes extend to depths as great as 350 km. One thus considers the
deeper population as deep earthquakes, such that the upper limit of deep seismicity
can vary between arcs. The average global onset of deep earthquakes is at about
325– 350 km.
d isotopes (e.g., d 18O) The composition of heavy, H (e.g., 18O), and light, L (e.g., 16O),
isotopic elements (E) are expressed as:
H 
H Esamp L Esamp
d E¼ H 2 1 £ 1; 000
Estd L Estd
where samp ¼ sample and std ¼ standard. The ratio is expressed as per mil. The ratio
is expressed as L E=H E or simply as d:
Diffusion Term applies to processes controlled by the diffusion equation: df=dt ¼
D72 f; where f is a scalar field and D is the diffusion constant. The diffusion
current is j ¼ 2D7f: The field, f, can, for example, be the temperature or relative
concentration of two species. For ideal noninteracting particles in suspension,
Brownian motion controls both the mean square displacement of a labeled particle
and the relaxation of concentration fluctuations. For more complex situations, the
self- (or labeled) diffusion constant, Ds ; is defined by kR2 l ¼ 2dDs t; while the
cooperative or gradient diffusion, Dc ; is defined by j ¼ 2Dc 7n; where j is
the particle current and n is the density. As repulsive interactions are increased, Ds
decreases and Dc increases from the noninteracting value D0 :
Dilatation Just as in stresses, the sum of the normal strains across any three orthogonal
planes at a point is constant; it is called dilatation (or dilation), designated by D;
which is

D ¼ exx þ eyy þ ezz ¼ du=dx þ dv=dy þ dw=dz

Dislocation Topological defect in periodic and quasi-periodic solids in which the phase
of a mass density wave changes by 2p in one circuit around a core. Alternatively, a
defect terminates an extra plane of atoms in the crystal.
Domain wall Defect separating two distinct but energetically equivalent states in
systems with a broken discrete symmetry.
1210 Glossary

Earthquake source Abrupt releases of the potential elastic energy stored in rocks (over
a period a few years to thousands of years); laws of elasticity do not apply at focus, but
operates away from it.
Eclogite The basaltic melt from partial melting of peridotite, when trapped from the
crystalline garnet and Ca –Na pyroxene (omphacite) to make the bimineralic rock,
ecologite. Subducted ocean floor basalts are transformed to amphibolite and at higher
pressures to eclogite.
Einstein relation Equation relating the diffusion constant, D, of a particle to its
mobility. 1=a: D ¼ kB T=a: See also Stokes’ law.
Electron-band theory A generic term for the collection of methods used in solving the
quantum-mechanical (Schrödinger) equations for electrons in a lattice.
Emissivity Ratio of the emissive powers of a real (gray) body and an ideal (black) body.
EM-1, EM-2 basalts OIBs having isotopic similarities to recycled continental
materials.
Enstatite (orthoenstatite and clinoenstatite) A member of the pyroxene group of
minerals having composition (MgxFe12x)3 with x $ 0:9: This mineral typically has
orthohombic structure at pressures below 6– 10 GPa (orthoenstatite) and monoclinic
structure at higher pressure (clinoenstatite). Orthoenstatite is a primary constituent of
most peridotites in the upper mantle.
Fcc lattice Face-centered-cubic lattice; lattice in which lattice sites are at the vertices
and face centers of a regular cube. Being at once a close-packed structure and of the
highest crystalline symmetry (cubic), the fcc lattice is the second most common solid
structure for the elements at room temperature, with 18 elements assuming this
structure.
Ferroelastics In these a cooperative crystal distortion induces a spontaneous strain
below Tc : The domain boundaries of a ferroelastic can be controlled by applied
stress.
Ferroelectric crystals Contain aligned electric dipoles arranged in randomly oriented
domains (no overall polarity). In a strong electric field (n £ 102 volts/mm) the domains
realign parallel to the field and the crystal develops a bulk polarity. Hitting or
squeezing such a ferroelectric crystal would produce a surface charge (piezoelectric
effect used in gas lighter and microphones).
Ferroelectrics These are characterized by different orientational states below Tc ; but the
spontaneous polarization is induced by cooperative crystallographic distortion. In
these the domain boundaries can be controlled by an external field.
Ferromagnets In these, long-range magnetic order below the Curie temperature
induces a spontaneous magnetization, Ms resulting in many magnetic domains,
each with a Ms vector oriented in a direction different from that in adjacent
domains.
Friction loss, Q Since rocks or most materials are not perfectly elastic, some elastic
energy is dissipated by transformation to heat or other means. Engineering materials
showing least absorption of energy per wavelength are marked as of high quality
factor, Q: Therefore Q21 is proportional to the amplitude loss per cycle (or per wave-
length). A large Q means energy loss is small. Since the energy, E, is proportional to
Glossary 1211

the square of the amplitude, A:

   
2p 2p
A ¼ E0 exp 2 t ¼ E0 exp 2 x
QT QL

where L is the wavelength. Most of seismic wave energy is lost in shear; dilatation
causes only a little loss. Hence, for shear waves, Qb is smaller than that of P-waves, Qa
ðQa < 2Qb Þ: In the Earth’s mantle Qb is of the order of 300; and in the sedimentary
layers and in low velocity zone (LVZ) it may be as low as 100 or less. Subducting slabs
usually have high Q values.
Frustration The inability of a model system to satisfy all of its bonds, usually because
of topological constraints. A “fully frustrated” system is frustrated on every
elementary unit. The “glass” transition in many disordered systems is regarded as
frustration (especially in spin glasses).
Gibbs free energy This is a thermodynamic potential GðT; P; NÞ; which is a natural
function of temperature T, pressure P, and particle number N.
GL Geophysical Laboratory (Washington).
Glass It is a phase of matter with no long-range order but with a nonzero shear rigidity.
Usually a disordered material which has a restoring force against shear (or its
generalization). Conventionally, any disordered material with a viscosity greater than
1013 poise, i.e., 1012 N s m22 is called a glass.
Grain boundary The boundary between different microcrystallites. Often, crystallite
boundaries consist of a low-energy periodic arrangement of dislocations, which orients
crystallites at small angles with respect to each other.
Grüneisen Constant Ratio of the logarithmic derivative of the Debye temperature to
that of the volume. It is used to calculate the thermal properties of a solid.
Gruneisen parameter Frequently used in geophysics, shock physics, and theoretical
high-temperature equations of state. Parameter is related to the adiabatic gradient by
g ¼ KS =TðdT=dPÞS ¼ aKT =cpr ¼ aKT =cVr ; where KS and KT are the adiabatic and
isothermal bulk moduli KS;T ¼ 2VðdP=dVÞST ; a is the thermal expansion coefficient
a ¼ 1=VðdV=dTÞV ; CP and CV are the specific heats at constant pressure and volume,
respectively; and r is the density.
Hard spheres Random close packing of hard spheres fills space to 63% and represents a
reasonable approximation to the structure of glasses and dense liquids. Periodic close
packing leads to fcc and hcp structures with a filling fraction of 0.7404. Dense
systems of hard spheres are liquids for f , 0:49; have coexisting liquid ðf ¼ 0:49Þ
and fcc solid ðf ¼ 0:54Þ phases for 0:49 , f , 0:54; and are fcc solids for f . 0:54;
where f is the volume fraction.
Helmholts free energy This is a thermodynamic potential FðT; V; NÞ; which is a natural
function of the temperature T; the volume V; and the number of particles N:
Hexagonal close-packed (hcp) Hard spheres at maximum packing fraction (74%) form
into hexagonal layers which then stack in an hcp or fcc structure. In an hcp lattice,
alternate layers are the same. Each particle has 12 nearest neighbors. (This is not a
Bravais lattice, requiring a unit cell with two particles, i.e., a basis of 2.) The layer
1212 Glossary

spacing is 2/3 times the distance between sphere centers. This is the most common
structure of the elements, with 24 elements assuming this structure at room
temperature.
HIMU basalts OIBs having isotopic similarities to recycled oceanic crust.
Hookean Elasticity Where the relationship between stress and strain is linear the
behaviour is called Hookean elasticity. The relationship between stress and strain can
be presented as.

s ¼ E1

where s is the stress, 1 is the strain and E is the Young’s modulus (a constant).
HP High pressure.
Hugoniot Pressure – volume – temperature relationship during shock compression,
material specific (usually pressure –volume plot). The Hugoniot is the locus of
density, pressure, and energy states in a material after passage of a single shock. It is a
well-defined curve on the EOS surface. The compression (ratio of shocked to
unshocked density) generally increases with pressure and approaches a value of 4 for
an ideal gas in the high-pressure limit. Thus, Hugoniot is a locus of pressure –volume
points that can be reached by a shock compression experiment starting from a given
initial condition.
Hund’s rule Electrons on an atom in an unclosed shell occupy the orbitals in such a way
as to maximize the total spin. This configuration allows Pauli exclusion to keep like
spins apart, and hence reduce their Coulomb repulsion.
HT High temperature.
Hydrogen bond forms when Hydrogen atom having a single covalent (atomic) bond is
bonded to two atoms, by giving up an electron to the molecule and the remaining
proton (Hþ) can provide bond up to two atoms, e.g., in H2O molecules with a dipole
moment. The potential binding energy of hydrogen bond is , 5 kcal/mole.
Hydrophilic and hydrophobic Liking or disliking water and, therefore, soluble or not
soluble in water, respectively. Hydrophilic interactions are complicated and
controversial. Water consists of dipolar molecules, and hence the charged or polar
molecules can gain some polarization energy in water. Organic molecules on the other
hand, with low polarizability and no dipoles will disrupt the preferred structure of
nearby water molecules and increase their energy. Oils, therefore, separate from water
and will not dissolve.
Hysteresis History-dependent properties of a system. The thermodynamic properties of
a system are completely determined by the temperature, pressure, etc., and are
independent of how the system reached these conditions. Therefore, a hysteretic state
must be out of equilibrium, or metastable. Hysteresis is often associated with first-
order phase transitions in which there is finite barrier to nucleation of a new phase and
a latent heat to be dissipated. It is also associated with glassy systems frozen into
nonequilibrium states.
Icosahedral symmetry The symmetry of an icosahedron, particularly noteworthy for
five-fold rotation axes. It is the point group of highest symmetry, and one which is
Glossary 1213

crystallographically forbidden, i. e., periodic crystals with this symmetry cannot be

formed (see quasi-crystals). However, the dense packing and high symmetry of
icosahedra have led people to look for short-range icosahedral order in liquids
and glasses.
Ilmenite As employed in the geophysical literature, the compound with composition
(Mg,Fe)SiO3 when it has the structure of the mineral ilmenite, an iron– titanium oxide.
With increasing pressure at low temperatures, metastable clinoenstatite ((Mg,Fe)SiO3)
transforms directly to the ilmenite structure.
Incommensurate crystal A structure with long-range periodic order but with two or
more periodicities with an irrational ratio. Common examples of incommensurate
crystals are magnetic systems in which the magnetic period (e.g., helical order) is
irrational with respect to the atomic lattice and systems, with density wave instabilities
at wave vector qDW related to the Fermi wave vector kF (in one dimension, qDW ¼ 2kF ),
which is often unrelated (hence usually irrationally related) to the underlying lattice.
Incompatible/compatible When rock starts to melt, some elements (incompatible) are
concentrated in the liquid silicate while others (compatible) remain in the solid
minerals.
Inverse Problem In seismology one is required to determine the variation of velocity,
say with depth when the travel-time curve is secured. This inference of the model of
the earth from the data constitutes the inverse problem.
Inviscid Flow Fluid flow with no viscosity and hence no dissipation. For incompressible
fluids, such flow obeys Euler’s equation.
Irreducible representation In group theory, a representation of a symmetry operation
that cannot be expressed in terms of lower dimensional representations.
Isochore A line which represents the relationship between the P and T, keeping V
constant.
Isotope data Isotope data are reported as parts per thousand differences from a reference
standard. The d notation is defined as
dX ¼ 1; 000 £ ðRX 2 RSTD Þ=RSTD
where RX and RSTD are the isotope ratios (e.g., 18O/16O or D/H (2H/1H)) of the sample
(X) and standard (STD), respectively. Data for O and H are reported relative to VSMOW
(Vienna Standard Mean Ocean Water) and for carbon relative to VPDB (Vienna Peedee
Belemnite).
IVCT Intervalence Charge transfer.
KCR The prevalent taxonomy of KCR is 50– 90% olivine and variable contents of mica,
amphibole, calcite, mineral oxides, and diamond.
Kinetic hindrance A condition under which chemical reactions take place sufficiently
slowly so that metastability occurs. For example, because reaction rates decrease with
decreasing temperature, a phase can persist metastably outside its stability field if
temperatures are sufficiently low.
Kink A point defect in a one-dimensional model with discrete symmetry, e.g., a spin-flip
in a one-dimensional Ising model with a spin-up chain on one side and a spin-down
chain on the other side of the kink.
1214 Glossary

Lamé coefficient For an isotropic solid in d dimensions, there are two elastic constants,
known as Lamé coefficients, l and m; related to the bulk and shear moduli by: K ¼
l þ ð2m=dÞ and G ¼ m:
Lattice A periodic array of points defined by the linear combination with integer
coefficients, li ; of a set of primitive translations vectors a1 ; ………; ad : The points are
located at R1 ¼ l1 a1 þ l2 a2 þ · · · þ ld ad :
Lattice Dynamics Refers to the vibrations of atoms in a solid about their equilibrium
positions and the effect of these vibrations on the properties of the solid.
Lattice wave and sound wave Oscillator can be a progressive displacement of wave in a
discrete lattice of atoms, called lattice wave, or a displacement wave in an elastic
continuum, called sound wave.
Lherzolite Ultramafic rock containing garnet or spinel plus olivine and Ca- and Mg-
pyroxene.
Lindemann melting relation Not a melting theory but relates melting to the amplitude
of atomic vibration (see Poirier (1991) for a review).
Lithophile Elements having strong affinity for oxygen, and are concentrated in the
silicate Earth.
Local Density Approximation The simplification made in atomic and molecular
physics for replacing the complicated electron exchange and correlation energies by
simple free-electron-gas expressions.
Magnetic scattering Neutrons have a magnetic moment and are, therefore, scattered by
magnetic fields. Thus, they are useful probes of magnetic structure such as occurs in
paramagnets, ferromagnets, antiferromagnets, etc., or in flux phases of superconduc-
tors. With the advent of high photon fluxes in synchrotrons, it is now possible to use the
weak scattering of X-rays from magnetic fields (a relativistic effect) to study magnetic
structures, but the technique is nowhere near as popular as magnetic neutron
scattering.
Magnetization The magnetic dipole moment per unit volume. It is the order parameter
for a magnetic transition and is the derivative of the energy with respect to magnetic
field.
Magnon A quantized spin wave. In ferromagnets the long wavelength dispersion is
quadratic, v , q2 ; while for antiferromagnets it is linear, v , lql:
Majorite It is the high-P (. 8 GPa) polymorph of pyroxene. Because majorite has
the garnet structure, it forms a solid solution with garnet. Majorite garnet is
recognized by having . 3 Si atoms per 12 O atoms and excess Si (beyond that
accommodated in the low-P garnet structure) is in octahedral rather than tetrahedral
coordination. Unmixed majoritic garnet with crystallographically controlled
lamellae of pyroxene is present in mantle xenoliths, and is indicative of depths
of origin of . 300 km but with equilibrium at lower (2 – 3 GPa) pressures (Haggerty
and Sautter, 1990).
Martensitic transformation An isochemical trasformation in which the product phase
(the martensite or martensitic polymorph) is a result of transformation of the host
phase by shear stress, but maintains a structurally continuous, nearly strain-free
interface with its host. Shear stress favors the stability of the martensitic polymorph.
Glossary 1215

Mineralogical examples include the orthoenstatite ! clinoenstatite transformation

and, under high stress, the olivine ! spinel transformation.
Metasomites These rocks are enriched in large-ion lithosphile elements (LIL) such
as K, Na, Sr, and Ba, typical of the crust; and in high-field-strength elements such as
Ti, Zr, Nb and P. These show unusual mineralogies of ferriphlogopite (mica),
K-richterite (amphibole), and exotic alkali titanites in substrates of previously
depleted harzburgite.
Metastability A condition where a material is not in its most stable (lowest energy)
form. Under near-equilibrium conditions at high temperatures, one phase cannot
exist significantly into the pressure –temperature stability field of another polymorph,
so reactions are closely reversible across the phase boundaries. However because
reaction rates decrease with decreasing temperature, a phase can persist metastably
for an indefinite time well outside its stability field if temperatures are sufficiently low.
For example, diamonds created at very high pressures deep in the mantle can persist
indefinitely in a metastable state at Earth’s surface conditions even though they are
grossly outside their stability field.
Meteorites These are extraterrestrial bodies and are classed as Stony-irons, siderolites,
pallasites: composed entirely of metal and silicates. Stones, acrolites: composed
entirely of metal silicate materials.
Mie-Grüneisen Equation An equation of state of the solid that employs the Grüneisen
constant to calculate the thermal contribution to the pressure.
Molecular bonds are formed by van der Waals bonding through close packing of
molecules such as, Ar, CH4, paraffin and some organic compounds. The potential
energy of these bonds is of the order of several kcal/mole e.g., for Ar: 18 kcal/mole;
CH4: 2.4 kcal/mole. When inert gases are solidified they acquire a structure of cubic
close packing.
Molecular Dynamics Computer simulation of molecular properties using numerical
integration methods to solve Newton’s equations for N interacting particles.
Molecular dynamics simulation Computer simulation of a large number of interacting
particles. This yields structural thermodynamic and transport properties.
Molecular orbital approximation The electronic states of a many particle system are
often approximated by ignoring the Coulomb interaction among electrons and finding
the single particle states in the potential of the bare ions. The states are then occupied
by electrons at the lowest energy consistent with the Pauli exclusion principle. The
opposite approximation, Heitler-London, assumes strong correlations from the
electron repulsions, and hence allows no double occupancy of a site.
Monte Carlo Method Computer simulation of molecular properties.
Mott insulators Magnetic crystals that are insulators by virtue of local magnetic
moments are known as Mott insulators.
Multiple scattering It occurs in any system in which the mean-free path, l; is not much
greater than the dimensions of the system, L: Almost all scattering applies to the limit
l q L; where there is only single scattering in the sample. The sample is essentially
transparent to the probe by particle or wave. A scattering probability approaching unity
is already much too high to allow meaningful interpretation of scattering data.
1216 Glossary

Navier – Stokes equations These are hydro-dynamical equations for an incompressible

fluid: Pðdv=dtÞ þ rðv·7Þv ¼ 27P þ h72 v; with 7·v ¼ 0; where r is the mass density,
v is the velocity, P is the pressure, and h is the viscosity.
Nucleation A first-order transition from one phase to another is characterized by a
discontinuous jump in the order parameter, f, and by an energy barrier between the
two phases (or values of f). Because of the barrier, there is a surface tension, s;
associated with an interface between the two phases. A droplet (nucleus) of the new
equilibrium phase gains bulk free energy but costs surface energy. For the droplet to
grow, its radius, R, must exceed a critical radius, Rc ¼ 2s=Df ; where Df is the gain in
bulk free energy density. The critical nucleus may form thermodynamic fluctuations
(homogeneous nucleation) or on a surface or dust particle (heterogeneous nucleation).
Order parameter parameter distinguishing an ordered from a disordered phase. For
example, the order parameter for a ferromagnet is the average magnetization kml: kml
is zero in the high-temperature paramagnetic phase where spins are randomly
oriented and nonzero in the ferromagnetic phase where spins align on average along
a common direction. More generally, the order parameter is the average, kfl; of an
operator, f; which is a function of the dynamical variables in the system Hamiltonian
kfl is zero in the disordered phase and nonzero in the ordered phase. It is chosen to
have values in different equivalent ordered phases reflecting the symmetry of the
Hamiltonian H (i.e., to transform under an irreducible representation of the symmetry
group of h).
Order-parameter exponent for a second-order transition, the order parameter ðfÞ;
goes to zero as T ! Tc with an exponent b: kfl , ðTc 2 TÞb : In mean-field theory,
b ¼ 1=2; in the three-dimensional Ising model, b < 1=3:
Paramagnetic having a positive magnetic susceptibility so that the system energy is
decreased upon application of an external field.
Paramagnetic phase Disordered high-temperature phase of a magnet.
Peridotites Rocks made up predominantly of olivine (Mg,Fe)2SiO4, with lesser amounts
of orthopyroxene (Mg,Fe)SiO3 and clinopyroxene Ca(MG,Fe)Si2O6. They appear to
be the dominant rock type of the mantle.
Phase Diagram A convenient method of representing the stability regions of solid,
liquid, gas, under the various conditions of temperature and pressure.
Phase transition Transition between two equilibrium phases of matter whose signature
is a singularity or discontinuity in some observable quantity. First-order transitions are
characterized by a discontinuity in a first derivative of a thermodynamic potential. In
particular, entropy S, which is the temperature derivative of a free energy ðS ¼
2dF=dTÞ has a discontinuity in DS leading to a latent heat L ¼ TDS: For second-order
transitions, first derivatives of thermodynamic potentials are continuous. See order
parameter and hysteresis.
Phonons Phonons are quanta of energy of the normal modes of vibration of a crystal
lattice or of an elastic continuum. A photon in electrodynamics is considered as
quantization of the energy of a single wave. Energy can be added or removed from the
oscillator only in integral numbers of phonons.
Plasma An electrically conductive gas of ionized particles.
Glossary 1217

Poisson ratio The Poisson ratio, s; is the negative ratio of change in width to change in
length when a material is pulled along its length: uxx ¼ 2suzz : Normally, the width
will decrease as the length increases. In an isotropic three-dimensional solid with bulk
modulus K and shear modulus G, thermodynamic stability allows for values of s ¼
1=2ð3K 2 2GÞ=ð3K þ GÞ between 2 2 and 1/2. If volume is conserved, then s ¼ 1=2:
It is very unusual to find a three-dimensional material with a negative s, but some have
been artificially made. Two-dimensional fluctuating membranes embedded in three
dimensions have negative values of s:Its relation with l and m is:
s ¼ l=2ðl þ mÞ:

In the Earth, Poisson’s ratio ranges from as low as 0.1 to as high as 0.38 near the
surface (Birch, 1961).The relations between elastic parameters involving s are

Es E
l¼ m¼
ð1 þ sÞð1 2 2sÞ 2ð1 þ sÞ

E l 2s
k¼ ¼
3ð1 2 2sÞ m 1 2 2s
p
Employing known Poisson’s relation one obtains s ¼ 0:25; and l ¼ m; a ¼ b 3:
However, generally the value of s is assumed as 0.25, but if the body is known to be
incompressible the value is taken as 0.50.
Powder diffraction X-ray diffraction method using polycrystalline samples; yields
lattice constants.
Pressure The pressure, P, is defined as
P ¼ 21=3ðPxx þ Pyy þ Pzz Þ:
In liquid (at rest) Pxx ¼ Pyy ¼ Pzz ¼ 2P (; hydrostatic pressure). The minus sign is
used because the pressure is conventionally taken as positive for compression, whereas
the normal stresses are positive for a tension.Compressive normal stresses are
generally taken as positive. Compressive normal stress is normally considered as
“pressure”. For compression P is negative, whereas in tension normal stresses are
positive. Pressure in the Earth is lithostatic and can be measured by
P ¼ rgh
where r is density, g is gravitational acceleration and h is depth.
Pressure Calibration Discontinuous fixed point calibrants In these pressure-induced
phase transitions are observed as discontinuous change in electrical conductivity. In
the continuous calibrant as ruby, a continuous pressure-shift of ruby fluorescence
wavelength is observed. Ruby scale has been calibrated up to 180 GPa. A new
pair of variables has been chosen to improve the accuracy of primary calibration for
anvil devices. They are the density (r) measured with X-ray diffraction, and
acoustic velocity ðVw Þ measured with ultrasonic method or Brillouin scattering on the
same sample under the same compression (Mao and Hemley, 1996). Pressure is
1218 Glossary

derived directly from

ð
P ¼ Vw d r

The resultant P – r relation is a primary pressure standard.

Pressure Units Primary units are force/area or dyn/cm. Literature values are commonly
quoted in terms of MPa (105 dyn), bar (106 dyn), kbar (103 bar), GPa (1011 dyn or
10 kbar). The SI unit of pressure is the pascal, Pa, defined as one newton per sq. meter,
which is equivalent to one joule per cubic meter. 1 kb ¼ 1021 GPa. 1 GPa ¼ 9,872
atm ¼ 145,000 lb/in2.
Pyrolite Fictitious model rock representing the mantle composition composed mainly
of olivine and orthopyroxene, some calcic clinopyroxene, and an aluminous phase
(e.g., garnet).
Pyrophyllite A hydrated sheet silicate which is a good thermal and electrical insulator.
In a gasket of pyrophyllite it is possible to use resistance heaters to generate
temperatures ,1,5008C, while keeping the stressed cylinder walls at low temperatures.
Quasicrystal An ordered structure that exhibits (i) long-range incommensurate
crystal translation order, and (ii) long-range orientational order with a crystal-
lographically disallowed point group symmetry. The first condition is often
referred to as quasiperiodicity. It implies the existence of a diffraction pattern with
Bragg peaks on a dense set of points in reciprocal space. Crystallographically
disallowed point groups are those that are incompatible with tiling of space with a
single kind of tile. In two dimensions, all point group symmetries with order n not
equal to 2, 3, 4 or 6 are disallowed. In three dimensions, icosahedral symmetry
is disallowed. Penrose tilings provide an example of a two-dimensional
quasicrystal with 5-fold symmetry. Physical examples of quasicrystals include
AlMn alloys with icosahedral symmetry and AlCuCo alloys with decagonal
symmetry.
Raman spectroscopy Scattering of monochromatic laser light, which reduces a
spectrum based on the change in polarizability during atomic vibrations. This yields
information about the vibrational frequencies in a solid.
RT Room temperature.
Rayleigh number In a medium of thickness h, limited by a cold upper boundary and a
hot lower boundary, heat is transferred by convection or conduction, depending on the
value of the Rayleigh number (Ra) defined as.

rgah3 DT
Ra ¼
nk
DT is the difference of temperature between the upper and lower boundary; and r; g; a;
n; and k are the density, the gravity, the thermal expansion coefficient, the viscosity,
and the thermal diffusivity of the medium respectively. A critical value of the Rayleigh
number, RaCr, marks the onset of convection. If Ra , RaCr ; conduction occurs; and
convection will start when RaCr , Ra:
Glossary 1219

Refractory elements These condense from gas to solid at high temperature.

Shock waves These are transient waves, typically in the times scale of 10,6 s. The shock
pressure can be as high as 500 GPa, and the accompanying temperature can be , 2,000 –
15,000 K. The shock wave data are based on the Rankie –Hugoniot equations that
follow from conservation of momentum, mass, and energy in the response of the system
to the impact. As a result of a high-velocity impact shock wave front passing through the
medium can generate extremely high pressure. The peak pressure is controlled by the
intensity of the shock but not limited by the strength of the materials. Pressures as high as
50 TGPa (1 tetrapascal ¼ 1012 Newton m22). In such dynamic compression tempera-
ture also increases steeply with pressure. Thus a condition of simultaneous high P – T is
obtained, which can hardly be simulated by experiments other than acoustic
measurements and vibrational spectroscopy. In shock-wave measurements the paired
variables are particle velocity (VP) and shock velocity (Vs), from which Hugoniot
equations of state, the pressure – density relations are calculated.
Siderites These are composed entirely of metal (Ni – Fe).
Siderophile Elements that dissolve readily in metallic iron under moderately reducing
conditions are siderophile, and are concentrated in the core.
Solidus temperature Temperature at which the first melt is produced in a multi-
component system.
Sp.Gr./S.G/SG Space group.
Spinodal decomposition Process of decay towards equilibrium in an unstable region of
a phase diagram. Typically seen in quenches of binary mixtures.
Stefan – Boltzmann law states that the total radiant energy (ET ) of a black body is
proportional to the fourth power of its absolute temperature, T,

ET ¼ s T 4

where s is the universal Stefan –Boltzmann constant, which is equal to 5.67 £

1028 Watt/m20k4.
Stoke’s law It says that the drag force ( f ) on a spherical particle of radius a in a fluid of
viscosity h moving at velocity V is f ¼ 6phaV: Equivalently, the mobility is a21 ¼
ð6phaÞ21 :
Strain Relative distortion of a solid. Application of stress to a solid produces strain;
shear strain is displacement perpendicular to the gradient. It is a measure of the
relative change in the length or shape of an object

ðFinal 2 initialÞlength in x-direction

1xx ¼ ¼ du=dx
Initial length in x-direction

where du; dv and dw are the increment in length of the small component dx; dy and
dz: Similarly, 1yy ¼ dv=dy: 1zz ¼ dw=dz: In the elastic wave passing through the
Earth’s interior the attending displacement is small and the resultant strains are
much smaller than 1023 so the infinitesimal strain theory works very well. But
when the strains are larger than 1021 we need finite strain theory (Birch, 1952).
Finite strain theory is employed in viscous flow, such as folding. Here the
1220 Glossary

displacements are replaced by velocities, and the strains are replaced by the rate of
strain. The latter remain infinitesimal over the time interval. Strain is measured by
the strain tensor, uij ¼ ð1=2Þð7i uj þ 7j ui Þ; where u is the displacement of the solid
relative to a reference solid (in Lagrangian coordinates). Longitudinal strain has
displacement along the gradient. In a solid, gij can be regarded as the strain per unit
time or strain-rate tensor. In a fluid phase, application of a stress results in a
continuous deformation of the system with time, and stress is linearly related to
strain rate by the viscosity (tensor). The stress s is related to strain rate 1_ in a linear
manner.
s ¼ h1:
_
When the body is ideal viscous or Newtonian, the constant h is the viscosity.
Stress tensor The stress tensor at a point P is an array of numbers (matrix).
Pxx Pxy Pxz
Pyx Pyy Pyz
Pzx Pzy Pzz
Stress tensor is symmetrical, because
Pxz ¼ Pzx ; Pyx ; Pxy and ¼ Pzy ¼ Pyz

P ¼ 21=3ðPxx þ Pyy þ Pzz Þ – 0

Stress Force per unit area acting on an element of matter through its surfaces. Normal
forces (to the surface) are associated with compression, while transverse forces are
associated with shear. The force density is related to the gradient of the (symmetric)
stress tensor: fi ¼ 7j sij : At any given point the sum of the normal stresses on any three
orthogonal planes is constant (i.e., invariant with respect to rotation of the coordinate
axes).
Supercooling For a first-order transition, there is a region between the phase boundary
and the spinodal line where the system is in metastable equilibrium. Here the free
energy prefers the low-temperature phase but there is a barrier to overcome to form a
critical nucleus. If there are no inhomogeneities to aid nucleation and the activation
energy is sufficiently high, then the high-temperature phase can remain until the
sample is supercooled to near the spinodal line. For pure water, the freezing transition
can be suppressed to 2 408C.
Synchrotron radiation Electromagnetic radiation emitted by electrons or
positrons orbiting in a storage ring. This radiation results from the centripetal
acceleration of the electrons by the magnetic fields in the ring. The spectrum
contains high-energy X rays with high intensity, useful for X-ray diffraction on very
small samples.
Tensor rank A zero-rank tensor is a vector, which has both magnitude and direction. A
second-rank tensor is a vector associated with direction. (Note: Forces within a body
are not vectors.)
Glossary 1221

Thermal boundary layer Layer in which the thermal gradient deviates from an adiabat
due to thermal conduction. Heat transfer through the boundary layer controls the
cooling of the Earth.
Thermal expansion coefficient (a) a ¼ 1=VðdV=dTÞp : In the Earth, the a values range
from 1 to 5 £ 1025 K21.
Thermal pressure Pressure increase due to heating at constant volume PTH ¼ aKT DT:
Transformational faulting A form of faulting found in some mineral systems that
exhibit polymorphism. If the low-pressure polymorph is pressurized at low
enough temperature, it persists metastably and is subsequently deformed, however,
it may fault suddenly (without essential loss of cohesion) by localization of
transformation to its denser polymorph (Kirby et al., 1991). “Anticrack” faulting
is another term occasionally used for transformational faulting (Green II, 1994).
This term emphasizes the nucleation of “crack-like” inclusions of the high-
pressure phase under stress. The term “transformational faulting” is preferred
because it emphasizes the connection of the faulting process with mineralogical
changes of state.
Ultra-low-velocity-zone A thin (5 –40 km thick) layer above the core-mantle boundary
(at a depth of 2,900 km), where a reduction of the seismic velocities of 10% and more
has been observed.
Vacancy A point-like defect in a periodic solid consisting of a missing particle at a
crystal site. Point defects in three-dimensional systems have integrable strain fields,
and therefore do not destroy long-range order the way that dislocations can.
Van der Waals attraction 1=r 6 attractive interaction between neutral atoms.
Viscosity The coefficient relating the shear stress to the shear strain rate. It measures the
rate of momentum transfer across a transverse velocity gradient.
Wadati-Benioff zone An inclined zone of seismicity at a convergent plate boundary
marking the presence of cold subducting lithosphere. This structure was first noted in
Japan by Wadati in 1935. Benioff (1949) proposed that the inclined seismic zone was
due to large-scale thrust faulting. With the discovery of plate tectonics, it was
recognized that although many shallow earthquakes in subduction zones result
from interplate thrusting, others, especially at depths exceeding about 50 km,
result from deformation within the slab. Frohlich (1987) used the term “Wadati-
Benioff Zone”.
Waves
L-wave Low-frequency, long-wavelength transverse vibrations, which develop in the
immediate neighborhood of the epicenter and are responsible for most of the
destructive force of earthquakes. They are confined to the outer skin of the Earth.
S-waves high-frequency, short wavelength transverse waves, which propagated in all
directions from the focus and travel at varying velocities (proportional to density)
through the solid parts of the Earth’s crust, mantle and core.
P-waves high-frequency, short-wavelength longitudinal waves, which have many
of the same characteristics as S-waves — the major difference being that
P waves travel not only through the solid part of the Earth but also the liquid part
of the core.
1222 Glossary

Wigner-Seitz cell The unit cell of a lattice that is the interior envelope of all planes that
are perpendicular bisectors of bonds connecting a lattice site to all other lattice sites.
Yield strength Stress above which a material deforms plastically (elastic limit).
Young’s modulus l The ratio of uniaxial stress to uniaxial strain in a crystal (with
orthogonal stresses equal to zero).Young’s modulus measures the resistance to
extension, and is defined as
 
Pxx ð3l þ 2mÞm
1¼ ¼
exx lþm

In terms of Lamé coefficients, l ¼ 9K=ð3K þ GÞ in three dimensions.

 1223

Subject Index

a-PbO2 phase, 298, 301 ALH 84001 meteorite, 23, 94, 272, 280
b-phase in various study, 989 Alkali metals, 63, 95–97, 977
e –g transition, 979, 992–993 Alkaline volcanoes, 242
g– e– l triple point for iron, 995 Allende meteorite, 110, 118, 120, 275
k and convective power of core, 1077 Alpe Arami Massif, 309, 547–548, 596, 599
k at D00 zone, 1074 Amino acid racemization, 126
k at mantle depths, 1072 Ammonia in Uranus and Neptune, 70
k under shock pressure, 1077 Ammonia, 32–33, 47, 70–74
ns (Si– O) and , Si–O–Si . , 846 Amorphous ice polymorphism, 89
p-polarized spectra, 335 Amorphous ice, 86, 88–89, 91, 507
s-polarized spectra, 335 Anderson–Grüneisen parameter, 209, 418, 471, 586,
z-O2 phase, 47 749, 932
10 Å phase, 909 Anisotropic intrinsic moments, 306, 720
129
Xe anomaly, 30 Antiferro distortive transition, 482, 1063
18
O isotope in UHP rocks, 151 Antiparallel Cooper spin pairs, 1052
182
W fractionation, 173 Antisymmetric many body wave function, 402
187
Os/188Os ratios, 149, 170, 221, 224–225, 245–246 Anomalous low-velocity zone, 40, 75, 194, 232
27
Al NMR spectra, 637 Ar –H2 system, 75
29
Si MAS-NMR spectroscopy study, 636–637 Arc magmatism, 98, 140, 160, 242
29
Si NMR peaks of phase B, 913 Arrhenian approximation, 798–799
3.65 Å phase, 909 Atom–atom pair potential, 393
3
He/4He reservoirs, 154 Axial and volume compressibility
40
K in the core, 147, 175 of clinopyroxenes, 595
5
Eg-5T2g triplets, 746
87
Sr/86Sr ratio, 139, 148, 159 B1-B2 phase transition in MgO, 862
9-coordinated Ti, 298 B8 and anti-B8 super-lattice structure, 868
Baddeleyite, 298–301, 408, 839
Ab initio approximation, 388 Band assignment of infrared spectra of lawsonite, 944
Ab initio simulation, 71, 590 Band structure, 48, 375, 393, 408, 410, 435, 496, 498,
Acoustic emission, 217, 332, 890, 906–907, 909, 1087 756, 782, 854–855, 858, 860
Acoustic phonon, 367, 995, 1072 Bandwidth of a polaron, 1061
Acoustic velocity surfaces, 370 Barophilic organisms, 34
Activation enthalpy, 386, 461–463, 495, 510, 532, Barotolerant microbes, 34
749, 1069 Basalt sources, 227
Activation volume, 386, 461 –463, 495, 510–511, BC8 metallic phase, 106
541–542, 701, 750, 895, 1044, 1049, 1093 bcc –fcc phase boundary, 985
Adiabatic bulk modulus, 296, 421, 586, 668, 737, 880 Berlinite/Scheelite structure, 309
African plate, 219 Bernal–Fowler ‘Ice rules’, 79
Age of the Universe, 19 Big Bang, 19
Al2O3 –SiO2 –H2O system, 9, 954 Black hydrogen, 61
Al3+ partitioning, 549 Bloch states, 395, 410
1224 Subject Index

Bloch’s theorem, 395 Cauchy relation, 415, 858

Blue-shift of polariton energies, 288 Cauchy violation, 415
Bond length, bond valences in hydrous Cell distortion of perovskite, 724
wadsleyite, 557 CFSE in spinels, 679
Bore hole strain meter, 215 CH4 –H2 system, 40
Born– Durand premelting, 428, 998 Charge (and mass) density, 36
Born– Mayer type pair potential, 41 Charged carbon clasters, 118
Born’s stability condition, 416 Charged coupled device, 364, 994
Bose– Einstein condensation, 40, 1193 Chemical vapour deposition (CVD), 27, 109
Bragg lines and cell parameter of SiO2 and Chijiadian lherzolite, 551, 815
CaCl2-type structure, 832 Chile earthquake, 160, 167, 215
Breakdown of selection rule, 118, 366 Chirality retention, 126
Breathing distortion of O6 octahedron, 1062 Chloride fluids, 228
Brillouin zone, 322, 372, 406, 429, 571, 587, 687, 723, Chondritic meteorite, 24 –27, 100, 110, 134,
756, 837, 877, 970, 995, 1204 167, 173, 184
Broad-band luminescence, 116 Chondrule ages, 30
Broken symmetry, 51, 404 –405 Christoffel’s equation, 369
Bulk moduli data for olivines, 535 Clapeyron equation, 106
Bulk moduli for unit cell parameter of Anhy-B and Clathrate cages, 39
Shy-B, 915–916 Clathrate hydrate, 38–40, 47, 74
Bulk moduli of clinopyroxences, 204, 356, 432, 454, Clino-humite and chondrodite, 541, 925–926
616, 636, 639, 668, 682–683, 741 CMR perovskite, 398
Bulk modulus and EOS, 737 CMSCHO system, 114
Bulk modulus and pressure derivative for MgSiO3 CO2-V quartz-like, 99
perovskite, 738 Co-seismic strain, 212
Bulk modulus of calcite, 878, 880 Collision induced absorption in H2-, 48, 63
Bulk modulus of garmets, 656 Colossal magnetoresistance, 791, 1059–1061
Bulk modulus of magnetite, 686, 691–693 Colsest-packing distortion parameter in kyanite, 645
Bulk modulus of MgCO3, 880–881 Columbite structure, 257
Burgers vectors, 566, 1081 Cometary water on Mars, 69
Communal entropy, 462
(CMAS) system, 646 Composition of continental crust and pelagic
C in stellar and interstellar materials, 102 sediments, 808
Ca translation, 760–761 Composition of Vulcan’s Throne kaersutite, 938
Ca-Al-inclusions (CaAIs), 30 Compositional models, 134, 177, 186
Ca-ferrite, 253–255, 625, 654 –655, 697 –698 Compressibilities and bulk modulus of lawsonite, 943
Ca-phases in mantle discontinuities, 250 Compressibility and bulk modulus
Calc-alkaline magmatism, 228 of clinopyroxenes, 595
Calcium content: thermobarometer, 651 Compton scattering, 377, 850
Cambridge Edinburgh Total Energy Package, 390 Conductance of water, 215
Cambridge Serial Total Energy Package, 390 Configurational entropy, 345, 689, 782, 794–795,
CaO–MgO (FeO)–SiO2 system, 337 800–801, 838
Carbon hybrid orbital, 102 Continental flood basalts, 140, 225
Carbon in space, 100 Conversion energy channels, 65
Carbon isotopes, 114 Conversion table for units of stress, 323
Carbon nanotubes, 102, 119, 122 Cooperative Jahn–Teller effect, 348, 397
Carbon polymorphs, 104 Core viscosity, 191
Carbonado, 111, 113 Cosmochemistry, 23, 967
Caribbean and Pacific anomalies, 264, 266, 269 Cotunnite, 298, 300
Carr–Parrinello techniques, 392 Coulombic attraction, 36–38, 401
Cation distribution in spinel, 316, 481, 491, 570, 575, Covalancy, 76, 346
677, 680, 688, 690, 705 Cr-dimerization 353
Subject Index
Cr2+ in olivine, 548
Cr2SiO4: Cr2+-orthosilicates, 576
Cr3+-bearing minerals, 74, 271, 349–350, 355–356,
454, 863, 893, 897, 1041
Cr3+ and Al3+ in wadsleyite, 547
Craton keels, 222
Cratonic microdiamond, 110
Cratonization, 137–138, 187–188
Creep rate, 577, 894 –895, 1079, 1086– 1087
Critical point of water, 67
Cross polarization magic angle, 384
Crust and mantle of Mars, 271
Crustal fractionation, 137
Crustal recycling, 154, 244, 1100
Cryogenic mathod, 338
Crystal field and charge transfer, 745
Crystal field splitting, 344
Crystal filed stabilization energy, 157, 353, 355, 743
Crystalization age of achondrite, 29, 272
Crystalline structure of iron phases, 969–970,
977, 1003
CsI metallization, 291
Cubic perovskite, 254, 262, 307, 315, 713, 788
Cyanobacteria, 115
D/H ratio, 68–69, 75, 95–96, 139
D/H ratios in minerals, 95
d-wave symmetry, 1050
Debye temperature, 382, 418, 453, 455, 466, 471, 474,
607, 837, 970, 1048
Deep mantle melting, 251
Deep-ocean hydrothermal system, 32
Defects in oxide perovskites, 717
Deformation equivalence, 446–448
Deformational twinning, 427
Dehydration melting of metabasalt, 817
Dehydration temperature, 954–956
Delta-function like states, 36, 401
Dense hydrous magnetio-silicates (DHMS), 69
Dense phases of hydrogen, 903
Densification, 456, 635–636, 795, 826, 852–853,
867, 1071
Density changes and buoyancy, 1000
Density functional theory calculation, 118
Density functional theory, 389 –390, 567, 968
Deutarium at high pressure, 93
Deuterium in Mars, 94
Deviatoric stress, 190, 230, 334, 379, 424, 443, 635,
771, 854, 872, 943, 962, 990, 1012, 1082
Diamond melts, 106
Diamond window, 59, 61–62, 339
Diamond elastic moduli, 116
1225
Dielectric properties, 219, 341, 369, 407, 788
Diffraction patterns for AlSiO3OH, 944
Diffuse interstellar bands, 118
Diffusion coefficient of oxygen, 73
Diffusive anomalies, 73
Diopsite– jadeite join, 136, 245, 599–601, 925
Dipole-forbidden internal stretching mode, 48
Dirac equation, 398
Discrete variational-Xa calculation, 346
Dislocation (power-law)creep-diffusion creep, 1087
Dislocation climb, 1082, 1088, 1090, 1098
Dislocation creep, 231, 569, 854, 891, 1081–1090
Dislocation glide and climb, 1082
Dislocation recovery, 542, 560, 1081–1083, 1087
Displacive transition, 429, 477, 482, 484, 487, 635,
751, 840, 842, 1049
Distortional disorder, 492
Dolomite stability and depths, 882
Double exchange model, 791
Double hexagonal stacking, 68
Double-exchange ferromagnets, 396–397, 433–434,
791, 1054, 1058
Doubled well potential, 488– 489
Doubly-seismic slab, 235
Early crust, 25, 136, 271
Early sun, 23, 27
Earth Resource Satellite-1 (ERS-1), 220
Earth Scope, 213
Earth’s magnetism and orbital obliquity, 270
Earthquake in California, 220
Earthquake in Kobe, 214
East African Rift System, 219
ECP-band intensity, 355
Edge dislocation, 1081, 1087– 1089
Effusive rocks, 227
Elastic constant, 84 –88, 181–257, 348–397,
410–509, 555–588, 639–699, 721,
734–736, 827–877, 976–1004
Elastic constants of MgO and MgAl2O4, 699
Elastic moduli of wadsleyite, 555– 556
Elastic properties of (Mg0.9,Fe0.1)2SiO4, 533
Elastic properties of humites and olivine, 927
Elastic properties of Mg2SiO4 polymorphs, 534
Elastic property change, 355
Elastic softening, 667–668
Elasticity of hcp Fe, 997
Electric quadrupole –quadrupole
interaction, 64, 381
Electrical conductivity and Fe3+, 109, 342, 703, 793,
839, 894, 1041, 1044, 1069, 1074
Electrical conductivity of olivine, 540, 895
1226
Electrical conductivity, 47, 71, 143, 180, 259, 335,
353, 498, 540, 575, 678, 691, 705, 727,
749, 793, 839, 871, 886, 895, 1009, 1020,
1041, 1099
Electrical resistivity measurements, 691
Electrical resistivity of polycrystalline materials, 110
Electron density changes, 383
Electron energy loss and auger spectroscopy, 375
Electron excitation, 406
Electron paramagnetic resonance, 332
Electron phonon coupling, 397, 1048, 1060,
1063, 1185
Electronic character of iron, 967
Element partitioning in mineral, 158
Emplacement of garnet peridotites, 821
Energitics of isosymmetric transition, 495
Energy dispersive diffraction study, 376
Enstatite–diopsite join, 599–603, 646–648
Enstatite–jadeite join, 600
Entropy of ice, 79
EOS data for silica in SiO6 octahedra, 836
EOS of magnesite, 880
Equation of state parameter of Brucite, 881, 962
Equation of state parameters for phases, 465, 981
Equlibrium in the system, CaO –MgO–Al2O3 –SiO2
and the corresponding parameters,
649–650
Escherechia coli, 32, 34, 291
Eukaryotes, 32
Eulerian strain, 205 –206, 450, 923–924
Europa, 34, 67, 81, 83, 140
Europian Synchrotron Radiation, 49, 77, 832
Evolutionary history of the Solar system, 31
Exciton in alkali halides, 286–287
Exsolutions in garnets, 815
F-point Brillouin zone, 815, 821, 667
Fabry–Perot tandem interferometer, 368, 533
Faulting instability, 211, 213
Fcc vs hcp: c44/c66 ratio, 1003
Fcc-dhcp (or hcp)-melt triple point, 993
Fe in wadsleyite, 562
Fe on elastic anisotropy, 858
Fe–FeS eutectic, 26
Fe–S –O ternary system, 599, 611, 727, 794, 887, 897
Fe2+ in a crystalline field, 499, 744
Fe2+ in perovskite, 743, 745, 1068
Fe3+ in protonation, 563
Fe3+ in glass, 804
Fe3+ in perovskite, 745, 1067
Fe3O4 and Fe2O3, 705
Fe/Mg in velocity relation, 544
Subject Index
Fe-rich Martian mantle, 277
FeO in D00 zone 857
Ferrodistortive transition, 482, 1063
Ferroelastic transition, 486, 667, 830, 848– 849
Ferroelectric alignment, 79
Ferroelectric depolarization pulses, 1078
Ferroelectric–antiferroelectric transition, 1078
Ferroelectricity, 407, 730, 763, 782, 784–785,
788–789, 1049–1050
Ferromagnetic manganese perovskite, 1065
Ferropericlase, 111– 112, 528
Fertile mantle, 250
FeS III, monoclinic, 1019
FeS IV, hexagonal, 1020
FeSiO3 solubility, 610
First mode Grüneisen parameter, 365, 469, 537
First order Birch–Murnaghan EOS, 348, 451
First principles calculations, 209, 301, 306, 388–389,
393, 443, 567, 826, 839, 968–1017, 1112
First principles linearized Muffin Tin orbital, 389
Fischer–Tropsch-type reactions, 6, 33
Five-fold symmetry, 404
Fluid transport, 228–229
Fluids in the lower crust, 141
Fluorescence spectra of Cr3+ -doped compounds,
372–373, 658
Formation of the moon, 30, 174
Forsterite–jadeite join, 611, 1130
Fossil slab, 236, 261
Fractal dimensions, 1079
Free protons, 67
Frustrating metallic behaviour, 60
Fullerene road, 123–124
Fullerenes, 102, 104, 107– 108, 118–125, 490
Galapagos island, 223
Gaskets, 116, 321, 330, 333–334, 348
Gehlenite (Ca2Al2SiO7), 27, 760
Generalized gradient approximation, 970–971
Geomagnetic field propagation, 1010
Geomagnetic reversals, 270
Geothermics, 137
Ghost magnetism, 1052
Giant magnetoresistance, 1059, 1181
Gibb’s free energy of reaction, 532, 958
Gibb’s free energy, 495, 972
Global free oscilations of Jupiter, 45
Global positioning system (GPS), 216
Gold standard, 426
Grain-boundary diffusion, 511, 577, 1085–1087, 1092
Granulites, 141– 142, 208, 583, 629, 817
Graphite and amorphous carbon, 110
Subject Index
Graphite–diamond conversion, 109
Gravitational perturbation, 24
Green–Kubo theory, 394
Greenstone belts, 138
Grenvillian orogeny, 138
Grüneisen parameter for liquid iron, 980, 1012
Grüneisen parameter of lawsonite, 469
H diffusion and conductivity, 1090
H2 – CH4 system, 38, 47
H2 – H2O system, 74
H2 – He mixture, 71
Hadean history of the Earth, 134
Halogens in DHMS phases, 902
Hardness of polycrystalline materials, 300
Harmonics of the geopotential, 218–219
Hartree–Fock program, 590, 725
Hartree–Fock wave function, 388
He –Ne mixture, 75
Heat sources at core, 29, 199, 1078, 1001
Heavy fermion semiconductors, 1051–1052
HED parent body, 24, 29
HED suite, 24
Helioseismology, 44
Helmholtz free energy, 459, 974
Hexagonal and cubic diamond, 105
Hexagonal and rhombohedral graphite, 105
High-pressure VI Si compounds, 255
High pressure absorption spectra, 345
High sanidine, 617 –618, 622
High temperature superconductors, 48, 121, 433, 711,
1112, 1135, 1052, 1054
High-degree harmonic, 218
Higher isomorphs of ice, 86
HMS Challenger, 34
Hollandite, 145–146, 253 –255, 399, 505, 620 –627,
630, 698, 809, 830– 831
Holy Grail of physics, 96
Hook’s law, 204– 205, 410, 1205
Hot line in the upper mantle, 219
Hotspot in Hawaii, La Reunion, 223
HS–LS transition in FeS, 501
HS–LS transition, 1026
Hudsonian orogeny, 138
Hugoniot pressure and volume, 422
Hugoniot sound velocity, 217
Hugoniot temperature for e iron, 202, 261, 873,
979, 1012
Hund coupling J, 1063, 1047
Hydrogen dimer molecules, 60
Hydrodynamics, 78, 404, 421, 955
Hydrogen in Jupiter, 66
1227
Hydrogen stretching mode, 52
Hydrogen-bearing astrophysical objects, 59
Hydrothermal vents, 34–36
Hydrous phases in Earth mantle, 901
Hydrous ringwoodite (g-Mg2SiO4), 573
Hydrous wadsleyite, 555, 557, 559– 560, 562– 563,
573, 899
Hydroxyl concentration: Bi and z Values, 541
Ice Ih stability boundary, 84
Ice morphology, 83
Ice VI in diamond, 84
Ice VII as pressure medium, 87
Iceland hotspot, 231, 803
Iceland mantle plumes, 243, 1195
Icosahedra, 119, 121, 404–405
Improper ferroelastic transition, 312
Impulsive stimulated scattering, 332, 367
Incommensurate phases, 480–481
Incompatible elements, 45–172, 225–251, 408,
886–896, 1095
Inelastic X-ray scattering, 322
Infrared acoustic, 332
Inner core anisotropy, 200, 428, 1004–1005,
1008, 1180
Insulater to metal transition, 60, 435, 791, 828
Insulating molecular state, 52
Inter-atomic distances of magnetite, 696
Inter atomic potential, 391–393, 729
Interesting metals, 98
Interlayer cation, 956
Intermediate spinels, 316
Intersteller diamond and SiC, 28
Intrinsic anharmonic parameter, 366, 753– 754
Intrinsic anharmonicity, 365, 455, 754
IR-active vibron, 54, 64
Iron hydrite, 96, 1016, 1108, 1167, 1197
Iron in perovskite, 740–741, 1068
Isopropanol, 71
Isosymmetric transition, 494
Isothermal elastic moduli of MgO, 863
Isotopes in MORB and hotspots, 150
Isotopic abundances, 30
Isotropic upper inner core, 1008
Jahn– Teller distortion, 163, 576, 791, 1060
Jahn– Teller effects, 347, 791, 1200
Japanese and Farallon plates, 236
Jupiters moons, 32, 34, 282
K 00 values for spinel group crystals, 699
K-felspar/hollandite, 145
1228 Subject Index

K/T (cretaceous-tertiary), 114 Magaplumes, 154, 244

K2NiF4 structure, 313, 315 Magma fragmentation, 231
Kaersutite amphibole, 936, 939 Magma ocean generation, 135
Kaersutite in SNC meteorites, 938 Magma viscosity, 850
Kimberlite clan rocks, 113 Magnetic ‘spin ice’, 316
KLB-1 natural peridotite, 135 Magnetic behaviour of carbon, 124
Kohn-Sham equation, 390, 855 Magnetic field, heat flow and plate tectonics, 1011
Komatiite, 250–251, 726, 769, 819 –820, 1133, 1138, Magnetic hardening, 720
1162, 1169, 1187, 1207 Magnetic ordering, 433, 486, 695, 713, 1056
Komatiitic melts, 136 Magnetic quasi-particle, 1071
Kondo lattice system, 1063 Magnetite exsolution, 815
Korringa–Kohn–Rostoker method, 398 Magnons, 1071
Koyna reservoir, 215 Majorite and TZ, 316
KSS and Bloch’s theorem, 394 Malaita pipe in Papua, 238
Mantle “wetspots”, 803
Landau first order displacive transition, 841, 845 Mantle flow and viscosity, 662
Landau order parameter, 488 Mantle fluids, 142
Landau theory, 485–486, 843 Mantle geochemistry of Mars, 226
Laporte selection rules, 375 Mantle geochemistry, 275, 1032
Laser heating, 99, 128, 219, 258, 301, 338, 362, 775, Mantle isotopes, 147
888, 917, 986, 989 Mantle phase stability in Mars, 276
Lattice compressibility and KT, 739 Mantle plumes, 136, 241–242, 249, 281, 819
Lattice disorder, 423, 480, 955 Mantle rheology, 223, 230, 890
Latur earthquake, 218 Mantle xenoliths, 68–70, 184, 242, 425, 583, 646,
Laves phases, 75 659, 885, 935
Layered-mantle convection, 185, 714 Many body theory, 405
LDA ice, 91–92 Marguerite (H4), 29
Lead paradox, 228, 809–810 Mars Global Surveyor (MGS), 220, 271
Lehmann discontinuity, 579, 1088, 1146 Martian atmosphere, 23, 69, 94– 95, 271
Lennard–Jones pair potential, 421, 998 Martian meteorites (SNCs), 69
Levy–Shoemaker comet, 44–45, 272 Martian polar caps, 34
Li under pressure, 97 Martian xenon, 40
LIL enrichment, 228 MD simulation, 392–393, 589, 839, 862
Lithofile volatiles 1300-600K, 30 Mean field theory, 405
LiNbO3 structure, 309, 548, 743, 762, 772 Medium range order, 852
Lindemann law of melting, 258, 509 Melt sinking, 251
Line of Mt. Cameroon, 219 Melting point of innert gases, 47
Linear compressibility and interatomic distances, 456, Memory effect, 509
684, 695, 734, 789, 888, 931, 943, 1112 Memory glass, 312– 313, 508–509, 1035, 1108, 1131
Linear preferred orientation (LPO), 246 Metallic hydrogen, 48, 59 –61, 66, 98, 322
Linearized augmented plane wave method, 257, 306, Metallic liquid, 26, 71
389, 435, 784, 786, 859, 1162 Metallic regime, 73
Lithospheric rheology and dynamics, 229 Metallicity of water and ammonia, 71
Local density approximation, 53, 306, 389, 970 Methane clathrate hydrate, 47
Low-degree harmonics, 218 MgO–FeO– SiO2: Thermodynamics Data, 520–522
Lower inner core, 1008 MgO–SiO2 –H2O ternary, 887, 897
Lower mantle Al-phase, 611 MgO–Al2O3 –SiO2 system, 652
Lower mantle phases, 112, 491, 745, 922, 1117 MgSiO3 perovskite density and shear elasticity, 732
Lowest energy crystal structure, 48 MgTiO3-ilmenite, 772, 1173
Micro earthquake, 214
M-O length and T-O-T angles of felspars, 617 Micro-Raman spectrometer, 362–364
Madelung electrostatic force, 408 Microorganism under pressure, 34
Subject Index
Mid-Oceanic Ridge Basalts (MORB), 183, 896
Minor discontinuities, 208
Missing xenon problem, 144–145
Mixed dislocation, 1081
Mode Grüneisen parameter 365, 465, 469, 537,
760–761
Modified electron gas calculation, 584, 721, 837,
854, 1119
Moissanite anvil cell, 333, 1197
Molecular cloud, 21, 23, 88
Molecular dipole hydrogen, 61
Molecular dynamic (MD) simulation, 420
Molecular dynamics simulation, 392, 795, 971
Molecular metallic solid, 52
Molecular regime, 73
Monazite lamellae, 816, 1038
Moon formation, 133
Mott insulators, 326, 408, 410, 857, 1033
Mott transition, 50, 304, 403, 435, 867, 1028,
1069–1070
Muffin-tin potential approximation, 395
Multi-anvil and DAC, 330
Murchison meteorite, 25
n-diamond, 106
Nabarro–Herring creep, 1086–1088, 1096
National Synchrotron Light Source, 329, 731,
850, 920
Nazca plate, 231–232
Ne’el temperature of magnetite, 748
Near infrared bands for Fe2+, 748
Nebular condensation, 23, 26– 27
Negative compressibility, 448
Nephelauxitic series, 349
Neutron star, 19, 21, 44
New metarials, 38, 121
Newtonian flow, 162
NFS spectra, 592
NH3 in the Archaean period, 34
Nitrogen in type Ia diamond, 116
Nitrogen inclusion in diamond, 112
NMR parameter of MgO–SiO2 –H2O phases, 903
No phonon line, 372
Noble gases in mantle melts, 143
Non-ferroic displacive transition, 634
Non-Newtonian high viscosity blobs, 251 –252
Non-metallic ionic regime, 73
Normal and inverse NiAs structures, 857
Normal/inverse spinels, 679
Novel physics, 1053
Nuclear magnetic resonance, 332, 347, 384, 842
Nuclear quadrapole resonance, 307, 332, 1163
1229
Nuclear spin–spin interaction, 64
Nurek reservior, 215
NW Pacific subduction, 235
o-p conversion rate, 55–56, 64 –65
Oceanic island basalts (OIB), 202
Octupole polarization, 289–290
OH2 solubility, 895
Olber’s paradox, 19
Olivine/iron buffer, 1099
Olivine deformation, 1082
Olivine fabric anisotropy, 891
Olivine-spinel phase transition in Mars, 279, 566
One-phonon density of states of diamond, 118
Onuma diagram, 549–550
Ophiolites in Oman, Newfoundland, 222
Ophiuchus cloud complex, 40
Order of transition, 481
Order parameter, 72, 312, 356–358, 404–495, 602,
623–637, 688, 724, 842–843
Ore localization, 218
Oriantational order parameter, 72
Orientational order, 57, 64, 72, 123, 357, 476–477
Orpa kimberlite, 113
Ortho– para conversion, 64–65, 1185
Orthorhombic –tetragonal–cubic transitions, 299,
487, 729
OSPE of transition metal ions, 680
Oxygen level increase, 137
Oxygen sublattice transformation (hcp-fcc), 565
Oxygen vacancies and protonation, 891
P and S velocities and shear modulus, 260, 300, 370,
534, 1215
P and S velocity contrast, 544
p-H2 and o-p mixed crystal, 55 –56
Pair potential, 41–42, 295, 388, 393, 453, 730, 755,
837, 848
Paleo-slabs, 265– 266
Pan-African orogeny, 138
Panspermia, 32
Papua New Guinea, 222, 238
Paramagnetic Meissener effect, 1053, 1112
Parana basin, 232
Partial dislocation, 421, 487, 540, 565, 1095, 1098
Pauli exclusion principle, 38, 41, 97, 326, 388,
408, 436
Pb paradox, 228
Pentagon road, 123– 124
Peridotite mineralogy, 178, 252
Perovskite and mantle convection, 713
Petrogeny’s residua system, 226
1230 Subject Index

Petrotectonics of mantle, 225 Primitive atmosphere, 24, 36

Phase B and super hydrous phase B, 899, 902, 914 Primordial heat, 199, 1000
Phase equilibria in the mantle, 520 Primordial noble-gas, 19, 23, 36, 95–96
Phase transition in anorthite-rich felspars, 630 Principal unit strains of felspars, 619
Phonon gas, 419 Proto-solar nebula, 21
Photo emission method, 375 Proton diffusion in water, 73
Photoluminesence spectroscopy, 374 Proton quantum tunneling, 58
Photon ordering– disordering, 77, 286, 288, 330 Protonium hydride, 60–61
Physical properties of mid-transition zone phases, Pseudo-hcp structure, 74, 772
605, 607 Pseudo potential calculations of bcc and hcp iron,
Piclogite mantle model, 183 740, 855, 975
Piezoelectricity, 407, 711, 717, 731, 761, 763, 849 Pseudo-wollastonite ferroelectric structure, 757– 758
Pinning twin boundaries, 725 Pseudo-potential, 734–735, 740, 786, 855,
Planetary dynamo, 71 862, 971–972
Planetary embryo, 22– 23, 131, 133 Pseudo-Jahn-Teller effect, 788
Planetary interiors, 36, 46, 68, 204, 333, 1080, 1084 Pyrochlore structure, 316, 318
Planetary isentropes, 19, 21–23, 31, 36–37, 45–46, Pyroelectric metarials, 407
71, 74, 203 –204 Pyrolite mantle model, 183, 733
Plastic deformation, 112, 116, 211, 231, 330, 334, 424,
709, 806, 891, 1006, 1080, 1084, 1086, Quadrupole polarizability, 307
1089, 1095, 1098 Quantized phonon spectrum, 119
Plumes and under plating, 244 Quantum condensate, 58
Polariton branch, 288 Quantum molecular solid, 48
Polaron hopping, 1048– 1049 Quantum oscillator, 419
Polarons small and large, 703 –704, 746, 791, Quantum simulation, 48, 51
1043, 1048 Quantum wave functions, 38
Polycyclic aromatic hydrocarbons, 118 Quasi-hydrostatic stress, 336
Positional disorder, 492, 837 Quasi-icosahedral, 119
Post seismic deformation, 213 Quasi-stable phase, 476, 480
Post-Archean crust, 222 Quasicrystal, 1056
Post-seismic strain transients, 213 Quenched water, 91
Post-seismic stress, 213
Potassium mobility in subduction, 809 R25-T-M2 branch, 722–723
Potential-induced breathing model, 394 Racah parameter, 346, 349–351, 499, 501
Powder diffraction data for phase D, 920 Radial X-ray diffraction, 377–378, 996
Power law creep regime dislocation, 261, 1081–1082, Radiative and lattice contribution, 1073
1087, 1097 Radiative heat transfer, 48, 342, 350, 353, 683, 1044
Pr12x CaxMnO3 perovskite, 1067 Radio bursts, 1078
Pre-earthquake strain indicators, 215 Radiogenic heat, 199, 223
Precursors to ScS and PcP, 215–216 Raman-active vibron, 54, 64
Presence of liquid spheroids, 1006 Raman bands of SiO2 polymorphs, 828
Presolar carbonaceous meteorites, 115 Raman spectra: Aluminous perovskite, 732, 734, 759
Presolar diamond, 24 Raman spectroscopy of 13C diamond, 115
Presolar grains, 25 Raman vibron frequencies, 40
Pressure derivatives of the elastic moduli, 374, Range of pressure in the Universe, 21
700, 1029 Rattling-ion model, 787
Pressure effect on viscosity, 127, 548, 659 Rayleigh scattering acoustic wave propagation, 367,
Pressure induced electric dipole, 61 663–664, 1030
Pressure induced mode softening, 47, 53 –54, 60–61, Reaction path blocking, 403
67–69, 89 Red Sea Rift, 219
Pressure measurement techniques, 1144 Red shift of the absorption edge, 118
Pressure units and correspondence, 321 –322, 1216 Redlich– Kister model, 460
Subject Index
Reedmergnerite, 616–617, 620, 626, 637
Refinement of lawsonite X-ray lines, 942
Reflectance spectroscopy, 26
Refractory inclusions, 27
Refractory metals, 27
Regolith, 23, 157
RE manganate perovskite, 1066
Rheological structure, 190
Richter magnitude, 212
Rigid unit mode distortion, 489–490
Rotation angles and distortion, 723
Rotation of inner core, 1001, 1006
Rotational disorder, 492
Roton, 39, 54, 56, 65–66
Ruby fluorescence, 63, 335–336, 338 –339, 371, 534,
536, 754, 851, 994
Ruddlestone-Popper series, 792
Rutile–fluorite-cubic structure, 1070
S, Se and Te, 1018
s-process necleosynthesis, 27
San Andreas fault, 210–211, 214, 425
San Carlos olivine, 531, 534 –537, 1093, 1099
San Francisco earthquake, 211
Satellite-derived gravity data, 218
Saturn, 45–46, 60, 93–94
Saturn’s core, 93
Scattering lengths of some cations and
tetrahedral-oxygen bond length, 678, 689
Screw dislocation, 1081, 1088
Seafloor spreading, 225
Second mode Grüneisen parameter, 469
Second-order Birch–Murnaghan EOS, 655
Seismic anisotropy in ‘D’ layer, 231, 241, 246, 258,
263, 266, 411, 818
Seismic tomography, 147, 181, 195, 201, 208–209,
231, 238, 252, 263, 600
Seismic velocity in TZ, 667
Seismological model, 177, 180, 188–189, 192
Semi-Empirical methods, 388
Shear strains, 205
Shear viscosity of molten iron, 385, 800, 998, 1091
Shear-wave splitting with in the Earth, 1005
Shewanella oneidensis MR-1, 34
Shock produced lonsdaleite and SiC, 113
Shock-Hugoniot data on silicate rocks
and minerals, 778
Shocked transition to glass, 635
Short-and long range order, 29, 1067
Short-lived radio-nuclides, 23, 27, 29
Short-term triggering, 214
Si–O stretching mode, 366, 846
1231
Si–O–Al bond angles in anorthite, 629
Si–O–Si bond angle, 366, 444, 489–490, 841
Si-coordination, 297
Side band Fluorescence ultrasonic, 372, 509
Siderophile elements, 26, 132–133, 158–159,
167–169, 245, 977, 1101
Siderophile volatiles 1300-600K, 30
Silica rods, 605, 816
Silicate activation energies, 520
Silicate melts: O, Si diffusion, 1091
Silicate polymerization, 800
Simple ionic model, 41
Site group analysis for g-Mg2SiO4, 571
Slichter modes, 999
Slip system, 247, 542, 891, 996, 1006, 1081–1083,
1088–1090, 1099
Sm/Nd ratios, 172
SN shock waves, 110
SNCs, 938
Soft modes, 291, 356, 358–359, 365, 407, 482, 484,
723, 751
Solid hydrogen, 49– 50, 52, 54 –55, 59– 66
Solid hydrogen:alkali metal, 63
solid state convection in ice, 83
Solid state diffusion, 1099
Solid state mantle convection, 225
Solid Xe, 38, 40, 144–145, 291
Solidification texturing, 1005–1006
Solidus and lower mantle, 135, 145–147, 258
Sound velocities of MgO and MgAl2O4, 700
sp-hybridized carbon solids, 103
Space geoelectric methods, 211
Space probe, 23
Spectra of grossular garnet, 361
Spectroscopic Grüneisen parameter, 470
Spin fluctuation, 1054
Spin ice, 81, 316
Spin transition depth, 873
Spin-up and spin-down sub-orbitals, 346, 499,
502, 1062
Spin-allowed transition, 354, 744–745
Spin-density waves, 1053
Spin-forbidden transition, 164, 354
Spinel deformation, 1089
Spinel structure, 272, 315, 487, 528–576, 679–706,
815, 1089
Split atoms, 489
Spontaneous electronic polarization, 60
Spontaneous polarization, 49, 54, 407, 485, 763,
784
St Helens eruption, 227, 231
Stability of bcc and fcc phases, 984
1232
Stability of Fe phases, 816
State of stress, 205, 212, 850
Stefans law, 1217
Steric hindrances, 508
Stokes–Einstein diffusivity, 1091
Stokes–Einstein equation, 794, 798
Strain diffusion, 213
Strain hardening, 306
Strain rate, 190–191, 210, 218, 230–231, 325, 505,
778, 798, 801 –802, 891, 1080, 1082, 1089,
1096–1098
Streaming potential, 214 –215
Stress and strain, 204 –205, 1098
Stretch-densification, 456
Strong field dynamo, 1009
Structural anomalies, 72– 73
Structural parameters for FeTiO3, 775
Subducted kerogenous carbon, 111
Subducting andesitic rocks, 935
Substitutional disorder, 492
Subsurface ocean of Europa, 34
Sudbury impact structure, 125
Sulawesi Indonesia subduction zone, 811
Sulfur, 35, 128, 245, 275, 383
Sulu garnet peridotites, 815
Supercooled water, 88– 89
Superfluidity, 40, 1053
Super-ionic conductivity, 838–839, 902
Superionic phase of ammonia, 73 –74
Superionic solid state, 73
Superlattice ordering, 482
Superplasticity, 1089, 1097
Super-plastic regime, 1079, 1097
Superconductivity, 52, 97–98, 128, 291, 326, 401,
403, 405, 978, 1018, 1041, 1043, 1045,
1050–1054
Supernova explosion, 19–20, 110
Supersilicic clinopyroxene, 605, 813, 816
Surface effect, 784–786
Symmetry analysis, 312, 359, 571, 753
Symmetry breaking, 54, 494, 843
Synchro-shear mechanism, 1094, 1098–1099
Synchrotron infrared spectroscopy, 48, 330, 332
Synchrotron radiation, 49, 77, 322, 329, 342, 360,
376–377, 452, 538, 551, 582, 592, 622,
741, 832, 889, 920, 978
Synchrotron source, 329–330, 333, 342, 360, 376,
706, 832, 933
Synchrotron X-ray diffraction, 331 –332
Synthesis of phase D, 920
Synthetic Uranus, 71
Subject Index
T Tauri phase, 21, 23, 25
Talc and phase A, 908, 946
Temporal distribution of aftershock, 213
Tensor quantities, 205
Terrestrial heat-producing radioactive elements, 20,
22, 28, 32–33, 37, 46, 67, 69, 96
Terrigenous and pelagic sediments, 806
Tetrahedral and octahedral bond distances, 164, 567,
677, 679–681, 684, 686, 688, 696, 908
The PREM model, 178, 192, 330, 1001
Thermal anomalies, 200– 201, 232, 266
Thermal expansivity, 186, 209, 341, 366, 437,
446–447, 449, 458, 465, 467–468
Thermobarometry: Alpe Arami, 599
Thermodynamics properties in system MgO-(Al2O3)-
SiO2, 518, 651
Thermodynamic Maxwell relation, 467–468
Thermoelectric materials, 1055
Thermoelectric power, 332
Third-order Birch–Murnaghan EOS, 652
Three-photon spectroscopy, 286–288
Ti4+ in olivine/wadsleyite, 548
Tibetan uplift, 218
Tight binding total energy, 393, 970
Tight-binding model, 43
Tilt angles, bond length distortion, 724, 742
Titius–Bode law, 24
Tolerance factor, 303, 307–308, 715, 728, 765,
1062–1063
Tolerence factor ferromagnetic order, 727
Tonalites, 804, 817
Transformational plasticity, 1098
Tricritical/first order transition, 494
Truncated icosahedral cage, 119
Two branches of PKP travel time, 1004
U-Pb age, 25
UHP iherzolite, 813
Ukrainian craters, 114
Ultra sonic interferometer, 996–997
Ultrasonic interferometry, 217, 332, 370–371,
664, 731
Ultrasonic wave velocities, 217
Ultra sound spectroscopy, 370
Unit cell and atomic positional parameter
of Fe3O4, 694
Upper mantle anisotropy, 246, 267
Values of K0 and K 00 of spinels, 685
Van der Waals forces, 38 –39, 52, 291, 403
Van der walls compounds, 38–40, 52, 291–292,
454, 508
Subject Index
Variationally induced breathing model, 394
Velocity Interferometer System for any Reflector, 399
Very long base line interferometry, 219, 269,
360, 1091
Vibrational entropy of phases, 359, 365, 613, 704, 838
Vibrational excitations in solid, 54–55
Vibrational modes of iron lattices, 669, 732, 991
Vibrational modes of MgSiO3, 572
Vibron, 39 –40, 50– 51, 54–56, 61–64, 75, 355,
473, 788
Vinet EOS, 452, 471–472
Viscoelastic relaxation, 213
Viscosity and conductivity, 1046
Viscosity and strain rate, 190
Vitrinite maturation, 127
Voigt notation, 369, 410
Volatiles and partial melting, 234
Volcanic seamounts, 225
Volume of lower mantle phase, 619
Voyager 2 spacecraft, 71
Wadeite, 145 –146, 254, 580, 620 –623, 625,
807, 809, 830–831
Water in the Earth, 68, 139, 904
Water in the magmatic process, 141
Watson sphere, 388, 394, 721, 1024
1233
Wave velocities in the lower crust, 208
Wave velocities in the lower mantle, 208
Weak field dynamo, 1009
White dwarft stars, 20, 110, 322
Whole mantle convection, 919, 1042
Wohlleben effect, 1053
Wollastonite II, 610
X-ray diffraction data for ilmenite, 78
X-ray diffraction pattern of FeO, 299, 443, 552
X-ray diffraction patterns for dhcp iron, 229, 443, 552
X-ray emission spectra, 304, 305, 866
X-ray fluorescence spectroscopy, 322
Xe-HL, 27
Xe-hydrates, 40
Xenon-clathrates, 40, 47, 144– 145, 291
XES results, 866
Yield strength, 230, 330, 424, 1012
Zero pressure parameters for NaCl, 286–287
Zircon dating, 25
Zoisite and lawsonite, 933
Zone boundary transition, 429, 493– 494
Zone centre vibrational modes, 844