Professional Documents
Culture Documents
High Pressure Geochemistry & Mineral Physics
High Pressure Geochemistry & Mineral Physics
High-Pressure Geochemistry
and Mineral Physics
2. P. Henderson (Editor)
RARE EARTH ELEMENT GEOCHEMISTRY
4. B.O. Mysen
STRUCTURE AND PROPERTIES OF SILICATE MELTS
6. J. Berthelin
DIVERSITY OF ENVIRONMENTAL BIOGEOCHEMISTRY
8. N. Shikazono
GEOCHEMICAL AND TECTONIC EVOLUTION OF
ARC-BACKARC HYDROTHERMAL SYSTEMS
Developments in Geochemistry 9
High-Pressure Geochemistry
and Mineral Physics
Sachinath Mitra
2004
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.
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.
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
Acknowledgements
Sachinath Mitra
Department of Geological Sciences
Jadavpur University
xi
Table of Contents
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
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
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
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 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
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1103
Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1205
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
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
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
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
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
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).
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.
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
1.2. Cosmochemistry
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
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.
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?
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)
(,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).
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.
From the above discussion, the evolutionary history of the Solar System can be
chronologically arranged as below (cf. Halliday, 2001).
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).
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.
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
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
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.
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.
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:
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.
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).
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.
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.
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.
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).
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.
(b) The pressure required for disruption of electronic “shell” structure of atoms to lead to
white dwarf stars can be determined as
(c) To break down nuclei into neutron star the pressure required can be calculated as
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).
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
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).
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.
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 1.8. 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).
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).
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.
Figure 1.11. Principal vibrational excitations of solid H2 (hcp phase) (Hemley et al., 1998, q 1998 Plenum
Publishing Corporation).
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
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.
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).
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.
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.
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
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.
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:
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.
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
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.
(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.
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.
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.
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).
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.
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.
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.
(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).
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.
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”.
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.
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
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
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.
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).
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).
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.
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).
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).
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):
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).
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.
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).
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).
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).
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.
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).
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.
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
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:
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.
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).
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.
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
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).
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).
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)
material closely resembles kerogen. It is generally assumed that the kerogen-like material
formed in the early solar nebula.
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)
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
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.
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).
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.
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
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
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.
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
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.
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.
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):
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).
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.
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.
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.
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).
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).
(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.
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
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.
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).
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.
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.
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
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
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.
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.
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
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.
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).
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
1.21.5.3. 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
1.22.1. K-feldspar/Hollandite
1.22.2. Phlogopite
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).
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
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
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).
(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.
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.
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.
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).
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).
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
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.
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.
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).
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.
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
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
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
(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.
observed by Tschauner et al. (1999) implies drastic chemical disequilibrium between core
and mantle.
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.
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.
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
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)
(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).
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
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):
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
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).
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.
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
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).
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.
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)
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
TABLE 2.3
Representation phase chemistry of pyrolite at 70 and 135 GPa
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.
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).
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.
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 rðzÞ and gðzÞ are, respectively, the density and acceleration due to gravity at depth z:
When we consider the gravitational potential, U,
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).
tij 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)
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
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
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
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].
Figure 2.11. Suggested temperature profile of the earth (modified from Madon, 1992, q 1992 Academic
Press, Inc.).
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).
TABLE 2.5
Temperatures inside the Earth (in Kelvins)
Depth Sources
z (km) W72 V80 A82 SS84 BS85 BM86 P86 W87 IK89
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.
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.
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.
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).
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.
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.
and
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.
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
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
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.
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).
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):
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
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.
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.
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
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.
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).
(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.
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.
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.
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.
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.
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
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
wide spread observation of depleted mantle components of both mid-oceanic ridges and
hotspots.
2.8.2. Petro-tectonics
Africa, produced highly alkaline lavas and carbonate magmas characteristically enriched
in lanthanides, Nb, Zn and Hf.
2.8.3. Xenoliths
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.
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.
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
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:
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
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
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).
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).
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).
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.
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.
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].
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
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).
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.
2.10.4.3. 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.
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.
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].
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).
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).
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.
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.
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).
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
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
TABLE 2.7
[vi]
Predicted and observed high-pressure Si compounds based on the substitution 2[vi]Al ! [vi](Mg þ Si)
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.
2.12.1. Phases
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
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.
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
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.
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
(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).
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.
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
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
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:
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
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).
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
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.
2.16. 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).
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.
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.
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
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.
(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).
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).
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.
2.17. 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).
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).
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).
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
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)
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)
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.
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
(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).
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).
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
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).
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.
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
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.
3.2.1.1. 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).
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
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)
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.
TABLE 3.5
Hardness of polycrystalline materials
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.
(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).
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
TABLE 3.6
Estimated bulk moduli and density for Fe2O3
K0 (GPa) Method
K0 is set equal to 4.
a
K0 ¼ 4:53 determined from the same data.
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).
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
3.4. ABX3
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).
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
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
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
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
12 – 40 GPa
a-berlinite ! Cmcm phase
K0 ¼ 36 GPa very incompressible
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.
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
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.
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
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
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.
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)
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 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
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.
(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.
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
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).
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.
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
TABLE 4.4
High-pressure high-temperature reference phase transition (Wentorf and Devries, 1987)
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.)
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).
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:
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
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
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).
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.
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).
ol=melt
Figure 4.6. ›lnDt =›P of divalent Ni, Co. Fe, Mn and Mg (from Taura et al., 1998).
344 Chapter 4
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
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.
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
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.
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
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.
where K is the bulk modulus of the octahedra. The subscript zero represents the values at
zero pressure and room temperature.
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).
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).
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).
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.
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).
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).
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:
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:
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.
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:
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
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.
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.
Some sources for Raman data of major mantle forming minerals or their analogues
as listed below:
Authors
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.
gi T ¼ KT ðdlnni =dPÞT
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
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.
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.
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
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.
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.
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)
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.
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).
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
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
where C is the angle between the direction of the applied load and the diffraction-plane
normal and QðhklÞ is given by:
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
½2GXR ðhklÞ21 ¼ 1=2½ð2S11 2 S12 2 S13 Þ þ l32 ð25S11 þ S12 þ 5S13 2 S33 þ 3S44 Þ
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
In the above equation, x ¼ hn=kT where h is the Planck’s constant and n the
frequency of harmonic oscillators.
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).
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.
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
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:
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:
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
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
4.5.1. Introduction
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
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.
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.
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.
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.
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.
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).
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
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
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
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).
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.
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.
In atomic computation, most studies use a “Watson sphere” (Watson, 1958), which
is a charged sphere, usually of opposite charge to the ion.
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.
5.1.4. 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:
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).
TABLE 5.2
Number of elastic constants for crystal systems
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
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)
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).
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
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
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
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).
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
dP
KT ðP; TÞ ¼ 2V ð5-14Þ
dV T
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
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
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
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.
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 KS is the adiabatic bulk modulus, G the shear modulus and r the density.
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:
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).
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).
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
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).
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
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;
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
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).
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.
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 relation between two symmetry-breaking strain components (i.e., for no volume
change) is
1sb sb
11 ¼ 2111 :
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
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
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.
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)):
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).
5.7.1. 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.
TABLE 5.5
dTC =dP and dTN =dP some ferromagnetic materials
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 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
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.
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.
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).
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
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
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
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:
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).
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.
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
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).
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 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
aT0 ¼ a þ bT:
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
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 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.
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
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.
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 :
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
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
3KT0 7=3
P¼ ðC 2 C 5=3 Þ ð5-49Þ
2
452 Chapter 5
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 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 :
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
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
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.)
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).
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
(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:
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.
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.
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.
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).
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.
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
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
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.
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.
dT=dP ¼ TDV=DH:
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)
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
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.
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
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
relation is expressed as
ðT
V ¼ V0 exp aðTÞdT ð5-69Þ
T0
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).
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)
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)
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
where
r 2 r0
ap ¼ ;
l
and DE is the binding energy, and r is the length per electron. This gives
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=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.
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.
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)):
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.
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.
477
478
TABLE 5.8 (continued)
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
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.
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)
dP DS DH
dT DV TDV
Superlattice ordering of point defects has been found in metal halides, oxides,
sulfides and other systems.
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).
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
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.
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.
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.
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 :
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.
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):
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
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.
where 1i are the components of the spontaneous strain tensor derived (see Section 5.4).
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.
(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).
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.
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.
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
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
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 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
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.
D ¼ 210 Dq þ 6B þ 5
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
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.
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
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.
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
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
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
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
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)
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)
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).
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
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.)
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).
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.
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
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
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.
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
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.
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.
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:
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
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).
TABLE 6.7
Aggregate elastic properties of Mg2SiO4 polymorphs
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
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
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)
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.
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)
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)
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.
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.
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.
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
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:
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:
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
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
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).
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
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.
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
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.
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
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).
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).
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:
dK=dT ¼ 20:012 ð1Þ Gpa=K; GðRTÞ ¼ 113 ð1Þ GPa; dG=dP ¼ 1:5 ð1Þ;
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)
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).
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.
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).
6.4.2.3. 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).
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).
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).
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).
MgVI IV VI IV
22x Mgx Six Si12x O4 :
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.
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):
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
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.
T ¼ A1g ðRÞ þ Eg ðRÞ þ T1g þ 3T2g ðRÞ þ 2A2u þ 2Eu þ 4T1u ðIRÞ þ 2T2u ð6-20Þ
TABLE 6.15a
Site-group analysis for g-Mg2SiO4 (Chopelas et al., 1994)
TABLE 6.15b
Vibrational modes of g-Mg2SiO4 and assignment Raman modes in cm21
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
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.
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).
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).
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.
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.
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.
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
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
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).
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
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).
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
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.
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)
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
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).
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
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
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.
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).
TABLE 6.16
Volume coefficients of d and DEQ
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).
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)
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).
TABLE 6.18
Bulk moduli (Mbar) of clinopyroxenes
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.
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.
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
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
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.
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.
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
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
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.
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)
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.
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
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:
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
Chapter 7
(K2O, Na2O, CaO)– Al2O3 – SiO2 System
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).
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.
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)
The K0 values observed by Angel (1997) and his co-workers for some felspar
structures are as follows:
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.
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.
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)
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
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).
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
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:
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.
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.
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
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.
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.
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).
and
626 Chapter 7
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.
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.
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.
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.
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.
TABLE 7.3
Parameters of the P1 to I 1 phase transition in anorthite-rich feldspars (Angel, 1994)
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):
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.
TABLE 7.4
Si–O –Al bond angles in anorthite
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
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.
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
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
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)
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:
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:
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
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.
(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
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)
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
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
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.
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).
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.
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
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
TABLE 8.4
Bulk modulus and its pressure derivative for pyrope, Mg3Al2(SiO4)3 (SC single crystal, PC polycrystalline)
(Source: Zhang et al., 1998a)
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)
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.
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
8.2.9. Mg – Cr – garnet
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.).
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).
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
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
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).
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
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
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
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).
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
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:
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
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
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.
TABLE 9.3
Octahedral site preference energies of transition-metal ions in structure
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).
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].
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:
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
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).
TABLE 9.5
Values of K0 and K 00 of spinels
K0 (GPa) K 00 Authors
TABLE 9.6
Bulk modulus of magnetite
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
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
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)
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.
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:
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
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 .
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
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)
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.
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
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
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:
and for u:
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
The pressure dependence of the bond angle (a) of magnetite was calculated to be
6(88) £ 1025 deg/GPa, which is negligibly small.
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
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.
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)
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
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.
TABLE 9.11
Elastic constants of MgO and MgAl2O4
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)
MgO
VS ¼ 6.05(1) þ 0.0381(13)P 2 3.6(4) £ 1024P 2 VS ¼ 6.05(1) þ 6.37(10)(1 2 V/VO)
VP ¼ 9.70(2) þ 0.0704(20)P 2 5.6(6) £ 1024P 2 VP ¼ 9.69(1) þ 12.5(2)(1 2 V/VO)
MgAl2O3
VS ¼ 5.500(5) þ 0.001(1)P VS ¼ 5.49(1) þ 0.1(3)(1 2 V/VO)
VP ¼ 9.785(11) þ 0.047(5)P 2 0.0010(5)P 2 VP ¼ 9.79(1) þ 8.0(4)(1 2 V/VO)
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)
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)
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.
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
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.
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
(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
For further observations on conduction in minerals, readers are referred to Section 15.2.
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):
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.
and other transition and rare-earth elements. These thus play a major role in element
partitioning and geochemical evolution.
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).
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).
10.1.4. 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.
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
3d½B – X 2d cos F
aV ¼ aV0 þ aV F ¼ þ ð10-1Þ
½B – XdT cos F dT
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.
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).
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.
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.)
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
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.
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
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
TABLE 10.1
MgSiO3 perovskite density and shear elasticity
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).
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.
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).
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
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).
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.
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):
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.
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.
TABLE 10.3
Compressibilities and bulk modulus of MgSiO3
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.
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).
TABLE 10.4
Tilt angles (8), bond-length distortions and unit-cell distortions of (Mg12xFez)SiO3 perovskite (from Kudoh
et al., 1990)
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.
Fe2þ in the perovskite structure (Section 10.2.6.3) and the temperature-induced valence
transition for Fe2þ are discussed in later sections.
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
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.
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
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.
(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.
10.3. Transformations
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.
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.
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.
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).
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.
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).
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.)
10.4.2. Pseudo-wollastonite
(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:
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.
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
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
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
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
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.
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).
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.
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
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)
TABLE 10.8
Physical properties used in phase-equilibrium calculations (from Linton et al., 1999)
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.
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)
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
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
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
TABLE 10.10
Structural parameters for FeTiO3
Atom Site x y z
(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)
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)
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:
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.
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.
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
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
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).
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).
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
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).
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:
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).
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
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
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).
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.
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
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
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).
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).)
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:
stands valid over a wide range of viscosities (Dingwell, 1990). Relaxational spectroscopic
data also reveal oxygen-diffusivity change with viscous flow at high viscosities.
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.
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).
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).
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.
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).
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.
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.
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”.
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:
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.
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).
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).
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
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.
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
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.
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.
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).
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.
(< 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
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).
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
Silica rods. The nature of silica rods in clinopyroxenes has already been discussed in the
Section “Supersilicic clinopyroxene” in Chapter 6.
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.
11.6.5. 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).
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).
Chapter 12
Simple Oxides and Carbonates
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).
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).
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).
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).
TABLE 12.1
Raman bands of SiO2 polymorphs (Liu et al., 1997)
12.1.2. Stishovite
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).
TABLE 12.2
Densities of some HP mineral phases
(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
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.
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)
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)
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.
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.
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.
12.1.4. Cristobalite
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.
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)
A1 (R)
A2 (IR) A (R, IR)
B1 (R)
B2 (R) B (R, IR)
E (R, IR)
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.
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.
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.
may be crucial are low fluid availability and the development of transformation stress,
which inhibits nucleation and growth.
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
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).
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.
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.
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.
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.
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).
12.4.1.5. 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.
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:
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).
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
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.
Figure 12.20. The pressure dependence of t for MgO (Duffy et al., 1995a).
862 Chapter 12
TABLE 12.7
Transition pressure, transition volume and total energy difference for the B1–B2 phase transition in MgO
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)
(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)
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.
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).
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.
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
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:
McCammon (1993) concluded that pressure reduces Fe3þ to Fe2þ in FexO wustite
equilibrated with Fe.
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
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 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:
which is ,60% greater than the compressibilities of the corresponding oxide polyhedra as
determined by Hazen and Prewitt (1977).
876 Chapter 12
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).
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.
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
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
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
TABLE 12.12
Bulk modulus of MgCO3 obtained by different workers
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).
TABLE 12.13
Equation of state parameters of magnesite (from Fiquet and Reynard, 1999)
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).
.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.
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).
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
13.1.1. 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.
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).
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.
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:
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.
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
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).
also postulated that the vibration of pinned dislocations in Mg-silicates can attenuate the
seismic waves (Mitra and Bhattacharya, 1982).
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).
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
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.
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).
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
(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
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).
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.
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
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.
(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.
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).
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
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).
13.4.5. Anhy-B
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.
13.4.5.2. 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
with P (whereas areas of the lattice modes remain constant), with H1 being much more
affected.
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)
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.
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
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
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)
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:
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).
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).
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
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.
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:
TABLE 13.9
Calculated volumes of lower-mantle phases (from Frost and Fei, 1999)
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.
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):
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).
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.
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
TABLE 13.10
Single-crystal elastic properties of humites and olivine
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:
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).
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:
[ 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).
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)
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).
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:
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:
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
whereas the upper pressure-stability limit of zoisite is determined to be above 4 GPa from
the relation;
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.
13.8.1. 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
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:
TABLE 13.12
Composition of Vulcan’s Throne kaersutite compared with kaersutite from SNC meteorites (Mysen et al., 1998)
VTa Cb Zc
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.
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)
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.
13.8.3.2. 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):
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
TABLE 13.14
Axial compressibilities and bulk modulus of lawsonite obtained at various temperatures
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.
TABLE 13.15
Band assignments of infrared spectra of lawsonite and the Grüneisen parameter (from Scott and Williams, 1999)
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:
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
Na muscovite GPa ba : bb : bc
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
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.
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).
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.
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):
and
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)
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
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).
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).
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.
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
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 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).
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.
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
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.
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)
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).
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)
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).
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.
Chapter 14
Iron and Siderophile Elements: The Earth’s Core
14.1. Introduction
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 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.
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).
TABLE 14.1
Comparison of pseudo-potential calculations with all-electron calcul ations and experimental values of bcc and
hcp iron
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=8:20 ¼ ½1 þ 5:3P=173ð1=5:3Þ
976 Chapter 14
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
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.
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
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):
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.
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)
981
982 Chapter 14
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.
Okuchi (1997) determined the density reduction of the outer core effected by the
presence of hydrogen, sulphur and carbon as:
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).
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
TABLE 14.4
Fe: bcc ! fcc phase transition
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:
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).
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)
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).
Saxena et al. (1993) Laser heating, phase recognized using power –temperature relation b-Phase inferred over a broad field of P and
T; no structural data
Boehler (1993) Noted a similar behaviour Phase assigned the dhcp structure
Saxena et al. (1995) Laser-heated quenched samples studied with Synchrotron X-ray
Yoo et al. (1995, 1996) Laser-heated quenched samples dhcp phase; partly overlapping the fcc field
Saxena et al. (1996) Laser-heated quenched samples studied under pressure b-phase between P of 40 –60 GPa at T
between 1,450 and 2,150 K
Funamori et al. (1997) 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
Dubrovinsky et al. (1998) 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
Andrault et al. (1998) Laser heated sample with in situ X-ray study Assigned an orthrohombic structure to a new
phase with several common peaks with dhcp
Dubrovinsky et al. (1999) Externally heated sample with in situ X-ray study b-phase appeared P ¼ 301 GPa and
T ¼ 1,350 K
Saxena and Dubrovinsky (2000) 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.
Anderson (1967) found the relation between the Grüneisen parameter and volume as:
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
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
TABLE 14.7
Variation of the thermal Grüneisen parameter of iron with volume as described by equation (14-7)
than in 1-(100) layers. This would also explain a lower specific entropy for the hcp lattice
relative to the high-temperature g-form.
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.
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
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.
14.2.7.1. 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)
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:
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.
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.
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.
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.
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.
14.2.10.2. 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
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).
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
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:
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
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.
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.
At the core P – T condition, the paramagnetic inner core can be responsible for a
variety of geomagnetic phenomena.
TABLE 14.10
Hugoniot temperatures for 1-iron (Boness and Brown, 1990)
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.
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).
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
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.
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.
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
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).
(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.
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.
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).
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).
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).
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.
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
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).
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).
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.
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
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.
14.10. Fe3S2
14.11. Mn –S system
14.11.1. a-MnS
P ¼ 3f ð1 þ 2f Þ5=2 K0 ð1 2 3f =2ð4 2 K 00 Þ
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.
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).)
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.
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).
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).
14.14. Phosphates
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).
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).
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.
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
15.1.1. Introduction
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).
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.
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.
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.
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
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.
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.
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
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.
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).
ZT ¼ ða2 s=kÞT;
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).
15.4.1. 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
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
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.
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
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).
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.
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
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
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
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.
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)
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.
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).
TABLE 15.2
Thermal conductivities from lower mantle to the core
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.
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.
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).
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
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.
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
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
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.
15.13.1. 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.)
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
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).
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
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.
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).
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).
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).
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).
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).
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.
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).
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:
where B is a constant. The mean square distance, kDrðtÞ2 l; travelled by any atom in finite t
becomes linear in long times.
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
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).
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.
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
x2 =D , t; ð15-23Þ
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 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
Although strain rates fluctuate slightly during creep, the empirical creep law holds as
(Karato et al., 1982):
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).
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.
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.
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
ET ¼ s T 4
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
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
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
Cr2+ in olivine, 548 Dielectric properties, 219, 341, 369, 407, 788
Cr2SiO4: Cr2+-orthosilicates, 576 Diffraction patterns for AlSiO3OH, 944
Cr3+-bearing minerals, 74, 271, 349–350, 355–356, Diffuse interstellar bands, 118
454, 863, 893, 897, 1041 Diffusion coefficient of oxygen, 73
Cr3+ and Al3+ in wadsleyite, 547 Diffusive anomalies, 73
Craton keels, 222 Diopsite– jadeite join, 136, 245, 599–601, 925
Cratonic microdiamond, 110 Dipole-forbidden internal stretching mode, 48
Cratonization, 137–138, 187–188 Dirac equation, 398
Creep rate, 577, 894 –895, 1079, 1086– 1087 Discrete variational-Xa calculation, 346
Critical point of water, 67 Dislocation (power-law)creep-diffusion creep, 1087
Cross polarization magic angle, 384 Dislocation climb, 1082, 1088, 1090, 1098
Crust and mantle of Mars, 271 Dislocation creep, 231, 569, 854, 891, 1081–1090
Crustal fractionation, 137 Dislocation glide and climb, 1082
Crustal recycling, 154, 244, 1100 Dislocation recovery, 542, 560, 1081–1083, 1087
Cryogenic mathod, 338 Displacive transition, 429, 477, 482, 484, 487, 635,
Crystal field and charge transfer, 745 751, 840, 842, 1049
Crystal field splitting, 344 Distortional disorder, 492
Crystal filed stabilization energy, 157, 353, 355, 743 Dolomite stability and depths, 882
Crystalization age of achondrite, 29, 272 Double exchange model, 791
Crystalline structure of iron phases, 969–970, Double hexagonal stacking, 68
977, 1003 Double-exchange ferromagnets, 396–397, 433–434,
CsI metallization, 291 791, 1054, 1058
Cubic perovskite, 254, 262, 307, 315, 713, 788 Doubled well potential, 488– 489
Cyanobacteria, 115 Doubly-seismic slab, 235
D/H ratio, 68–69, 75, 95–96, 139 Early crust, 25, 136, 271
D/H ratios in minerals, 95 Early sun, 23, 27
d-wave symmetry, 1050 Earth Resource Satellite-1 (ERS-1), 220
Debye temperature, 382, 418, 453, 455, 466, 471, 474, Earth Scope, 213
607, 837, 970, 1048 Earth’s magnetism and orbital obliquity, 270
Deep mantle melting, 251 Earthquake in California, 220
Deep-ocean hydrothermal system, 32 Earthquake in Kobe, 214
Defects in oxide perovskites, 717 East African Rift System, 219
Deformation equivalence, 446–448 ECP-band intensity, 355
Deformational twinning, 427 Edge dislocation, 1081, 1087– 1089
Dehydration melting of metabasalt, 817 Effusive rocks, 227
Dehydration temperature, 954–956 Elastic constant, 84 –88, 181–257, 348–397,
Delta-function like states, 36, 401 410–509, 555–588, 639–699, 721,
Dense hydrous magnetio-silicates (DHMS), 69 734–736, 827–877, 976–1004
Dense phases of hydrogen, 903 Elastic constants of MgO and MgAl2O4, 699
Densification, 456, 635–636, 795, 826, 852–853, Elastic moduli of wadsleyite, 555– 556
867, 1071 Elastic properties of (Mg0.9,Fe0.1)2SiO4, 533
Density changes and buoyancy, 1000 Elastic properties of humites and olivine, 927
Density functional theory calculation, 118 Elastic properties of Mg2SiO4 polymorphs, 534
Density functional theory, 389 –390, 567, 968 Elastic property change, 355
Deutarium at high pressure, 93 Elastic softening, 667–668
Deuterium in Mars, 94 Elasticity of hcp Fe, 997
Deviatoric stress, 190, 230, 334, 379, 424, 443, 635, Electric quadrupole –quadrupole
771, 854, 872, 943, 962, 990, 1012, 1082 interaction, 64, 381
Diamond melts, 106 Electrical conductivity and Fe3+, 109, 342, 703, 793,
Diamond window, 59, 61–62, 339 839, 894, 1041, 1044, 1069, 1074
Diamond elastic moduli, 116 Electrical conductivity of olivine, 540, 895
1226 Subject Index
Electrical conductivity, 47, 71, 143, 180, 259, 335, Fe-rich Martian mantle, 277
353, 498, 540, 575, 678, 691, 705, 727, FeO in D00 zone 857
749, 793, 839, 871, 886, 895, 1009, 1020, Ferrodistortive transition, 482, 1063
1041, 1099 Ferroelastic transition, 486, 667, 830, 848– 849
Electrical resistivity measurements, 691 Ferroelectric alignment, 79
Electrical resistivity of polycrystalline materials, 110 Ferroelectric depolarization pulses, 1078
Electron density changes, 383 Ferroelectric–antiferroelectric transition, 1078
Electron energy loss and auger spectroscopy, 375 Ferroelectricity, 407, 730, 763, 782, 784–785,
Electron excitation, 406 788–789, 1049–1050
Electron paramagnetic resonance, 332 Ferromagnetic manganese perovskite, 1065
Electron phonon coupling, 397, 1048, 1060, Ferropericlase, 111– 112, 528
1063, 1185 Fertile mantle, 250
Electronic character of iron, 967 FeS III, monoclinic, 1019
Element partitioning in mineral, 158 FeS IV, hexagonal, 1020
Emplacement of garnet peridotites, 821 FeSiO3 solubility, 610
Energitics of isosymmetric transition, 495 First mode Grüneisen parameter, 365, 469, 537
Energy dispersive diffraction study, 376 First order Birch–Murnaghan EOS, 348, 451
Enstatite–diopsite join, 599–603, 646–648 First principles calculations, 209, 301, 306, 388–389,
Enstatite–jadeite join, 600 393, 443, 567, 826, 839, 968–1017, 1112
Entropy of ice, 79 First principles linearized Muffin Tin orbital, 389
EOS data for silica in SiO6 octahedra, 836 Fischer–Tropsch-type reactions, 6, 33
EOS of magnesite, 880 Five-fold symmetry, 404
Equation of state parameter of Brucite, 881, 962 Fluid transport, 228–229
Equation of state parameters for phases, 465, 981 Fluids in the lower crust, 141
Equlibrium in the system, CaO –MgO–Al2O3 –SiO2 Fluorescence spectra of Cr3+ -doped compounds,
and the corresponding parameters, 372–373, 658
649–650 Formation of the moon, 30, 174
Escherechia coli, 32, 34, 291 Forsterite–jadeite join, 611, 1130
Eukaryotes, 32 Fossil slab, 236, 261
Eulerian strain, 205 –206, 450, 923–924 Fractal dimensions, 1079
Europa, 34, 67, 81, 83, 140 Free protons, 67
Europian Synchrotron Radiation, 49, 77, 832 Frustrating metallic behaviour, 60
Evolutionary history of the Solar system, 31 Fullerene road, 123–124
Exciton in alkali halides, 286–287 Fullerenes, 102, 104, 107– 108, 118–125, 490
Exsolutions in garnets, 815
Galapagos island, 223
F-point Brillouin zone, 815, 821, 667 Gaskets, 116, 321, 330, 333–334, 348
Fabry–Perot tandem interferometer, 368, 533 Gehlenite (Ca2Al2SiO7), 27, 760
Faulting instability, 211, 213 Generalized gradient approximation, 970–971
Fcc vs hcp: c44/c66 ratio, 1003 Geomagnetic field propagation, 1010
Fcc-dhcp (or hcp)-melt triple point, 993 Geomagnetic reversals, 270
Fe in wadsleyite, 562 Geothermics, 137
Fe on elastic anisotropy, 858 Ghost magnetism, 1052
Fe–FeS eutectic, 26 Giant magnetoresistance, 1059, 1181
Fe–S –O ternary system, 599, 611, 727, 794, 887, 897 Gibb’s free energy of reaction, 532, 958
Fe2+ in a crystalline field, 499, 744 Gibb’s free energy, 495, 972
Fe2+ in perovskite, 743, 745, 1068 Global free oscilations of Jupiter, 45
Fe3+ in protonation, 563 Global positioning system (GPS), 216
Fe3+ in glass, 804 Gold standard, 426
Fe3+ in perovskite, 745, 1067 Grain-boundary diffusion, 511, 577, 1085–1087, 1092
Fe3O4 and Fe2O3, 705 Granulites, 141– 142, 208, 583, 629, 817
Fe/Mg in velocity relation, 544 Graphite and amorphous carbon, 110
Subject Index 1227
Reedmergnerite, 616–617, 620, 626, 637 Si–O–Al bond angles in anorthite, 629
Refinement of lawsonite X-ray lines, 942 Si–O–Si bond angle, 366, 444, 489–490, 841
Reflectance spectroscopy, 26 Si-coordination, 297
Refractory inclusions, 27 Side band Fluorescence ultrasonic, 372, 509
Refractory metals, 27 Siderophile elements, 26, 132–133, 158–159,
Regolith, 23, 157 167–169, 245, 977, 1101
RE manganate perovskite, 1066 Siderophile volatiles 1300-600K, 30
Rheological structure, 190 Silica rods, 605, 816
Richter magnitude, 212 Silicate activation energies, 520
Rigid unit mode distortion, 489–490 Silicate melts: O, Si diffusion, 1091
Rotation angles and distortion, 723 Silicate polymerization, 800
Rotation of inner core, 1001, 1006 Simple ionic model, 41
Rotational disorder, 492 Site group analysis for g-Mg2SiO4, 571
Roton, 39, 54, 56, 65–66 Slichter modes, 999
Ruby fluorescence, 63, 335–336, 338 –339, 371, 534, Slip system, 247, 542, 891, 996, 1006, 1081–1083,
536, 754, 851, 994 1088–1090, 1099
Ruddlestone-Popper series, 792 Sm/Nd ratios, 172
Rutile–fluorite-cubic structure, 1070 SN shock waves, 110
SNCs, 938
S, Se and Te, 1018 Soft modes, 291, 356, 358–359, 365, 407, 482, 484,
s-process necleosynthesis, 27 723, 751
San Andreas fault, 210–211, 214, 425 Solid hydrogen, 49– 50, 52, 54 –55, 59– 66
San Carlos olivine, 531, 534 –537, 1093, 1099 Solid hydrogen:alkali metal, 63
San Francisco earthquake, 211 solid state convection in ice, 83
Satellite-derived gravity data, 218 Solid state diffusion, 1099
Saturn, 45–46, 60, 93–94 Solid state mantle convection, 225
Saturn’s core, 93 Solid Xe, 38, 40, 144–145, 291
Scattering lengths of some cations and Solidification texturing, 1005–1006
tetrahedral-oxygen bond length, 678, 689 Solidus and lower mantle, 135, 145–147, 258
Screw dislocation, 1081, 1088 Sound velocities of MgO and MgAl2O4, 700
Seafloor spreading, 225 sp-hybridized carbon solids, 103
Second mode Grüneisen parameter, 469 Space geoelectric methods, 211
Second-order Birch–Murnaghan EOS, 655 Space probe, 23
Seismic anisotropy in ‘D’ layer, 231, 241, 246, 258, Spectra of grossular garnet, 361
263, 266, 411, 818 Spectroscopic Grüneisen parameter, 470
Seismic tomography, 147, 181, 195, 201, 208–209, Spin fluctuation, 1054
231, 238, 252, 263, 600 Spin ice, 81, 316
Seismic velocity in TZ, 667 Spin transition depth, 873
Seismological model, 177, 180, 188–189, 192 Spin-up and spin-down sub-orbitals, 346, 499,
Semi-Empirical methods, 388 502, 1062
Shear strains, 205 Spin-allowed transition, 354, 744–745
Shear viscosity of molten iron, 385, 800, 998, 1091 Spin-density waves, 1053
Shear-wave splitting with in the Earth, 1005 Spin-forbidden transition, 164, 354
Shewanella oneidensis MR-1, 34 Spinel deformation, 1089
Shock produced lonsdaleite and SiC, 113 Spinel structure, 272, 315, 487, 528–576, 679–706,
Shock-Hugoniot data on silicate rocks 815, 1089
and minerals, 778 Split atoms, 489
Shocked transition to glass, 635 Spontaneous electronic polarization, 60
Short-and long range order, 29, 1067 Spontaneous polarization, 49, 54, 407, 485, 763,
Short-lived radio-nuclides, 23, 27, 29 784
Short-term triggering, 214 St Helens eruption, 227, 231
Si–O stretching mode, 366, 846 Stability of bcc and fcc phases, 984
1232 Subject Index
Variationally induced breathing model, 394 Wave velocities in the lower crust, 208
Velocity Interferometer System for any Reflector, 399 Wave velocities in the lower mantle, 208
Very long base line interferometry, 219, 269, Weak field dynamo, 1009
360, 1091 White dwarft stars, 20, 110, 322
Vibrational entropy of phases, 359, 365, 613, 704, 838 Whole mantle convection, 919, 1042
Vibrational excitations in solid, 54–55 Wohlleben effect, 1053
Vibrational modes of iron lattices, 669, 732, 991 Wollastonite II, 610
Vibrational modes of MgSiO3, 572
Vibron, 39 –40, 50– 51, 54–56, 61–64, 75, 355, X-ray diffraction data for ilmenite, 78
473, 788 X-ray diffraction pattern of FeO, 299, 443, 552
Vinet EOS, 452, 471–472 X-ray diffraction patterns for dhcp iron, 229, 443, 552
Viscoelastic relaxation, 213 X-ray emission spectra, 304, 305, 866
Viscosity and conductivity, 1046 X-ray fluorescence spectroscopy, 322
Viscosity and strain rate, 190
Xe-HL, 27
Vitrinite maturation, 127
Xe-hydrates, 40
Voigt notation, 369, 410
Xenon-clathrates, 40, 47, 144– 145, 291
Volatiles and partial melting, 234
XES results, 866
Volcanic seamounts, 225
Volume of lower mantle phase, 619
Yield strength, 230, 330, 424, 1012
Voyager 2 spacecraft, 71
Zero pressure parameters for NaCl, 286–287
Wadeite, 145 –146, 254, 580, 620 –623, 625,
807, 809, 830–831 Zircon dating, 25
Water in the Earth, 68, 139, 904 Zoisite and lawsonite, 933
Water in the magmatic process, 141 Zone boundary transition, 429, 493– 494
Watson sphere, 388, 394, 721, 1024 Zone centre vibrational modes, 844