Ebin - Pub Triple Oxygen Isotope Geochemistry 9781501524677 9781946850065

REVIEWS in MINERALOGY
and GEOCHEMISTRY
Volume 86 2021
Triple Oxygen Isotope

Geochemistry
EDITORS
Ilya N. Bindeman
University of Oregon, USA
Andreas Pack
Georg-August-Universität Göttingen, Germany
Series Editor: Ian Swainson

MINERALOGICAL SOCIETY of AMERICA
GEOCHEMICAL SOCIETY
Reviews in Mineralogy and Geochemistry, Volume 86
Triple Oxygen Isotope Geochemistry
ISSN 1529-6466 (print)

ISSN 1943-2666 (online)
ISBN 978-1-946850-06-5
Copyright 2021
The MINERALOGICAL SOCIETY of AMERICA
3635 Concorde Parkway, Suite 500
Chantilly, Virginia, 20151-1125, U.S.A.
www.minsocam.org
The appearance of the code at the bottom of the first page of each chapter in this volume
indicates the copyright owner’s consent that copies of the article can be made for personal
use or internal use or for the personal use or internal use of specific clients, provided
the original publication is cited. The consent is given on the condition, however, that
the copier pay the stated per-copy fee through the Copyright Clearance Center, Inc. for
copying beyond that permitted by Sections 107 or 108 of the U.S. Copyright Law. This
consent does not extend to other types of copying for general distribution, for advertising
or promotional purposes, for creating new collective works, or for resale. For permission
to reprint entire articles in these cases and the like, consult the Administrator of the
Mineralogical Society of America as to the royalty due to the Society.
Geochemistry
86 Reviews in Mineralogy and Geochemistry 86
iii
PREFACE
Since the first volume of the Reviews in Mineralogy was published in 1974, this series
(expanded to Reviews in Mineralogy and Geochemistry in 2000) has grown to an invaluable
library comprising 86 volumes on various topics related to mineralogy and geochemistry.
Often a volume from that series is given by an advisor to a young graduate student or postdoc
as a roadmap when they enter a new research field. The review volumes are, however, also a
steady resource of information and inspiration for more advanced researchers when they need
to get an overview of fields outside their everyday scientific expertise.
The field of stable isotope geochemistry, founded by Harold C. Urey, now looks back
on seven decades of continuing exciting discoveries and developments. Many of these
advancements have already been compiled into five volumes on stable isotope geochemistry
in the Reviews in Mineralogy and Geochemistry series.
Starting in 1986, Volume 16: Stable Isotopes in High Temperature Geological Processes
was the first to review the state of the field and was edited by John W. Valley, Hugh P. Taylor (Jr.),
and James R. O'Neil. This volume was updated in 2001 to emphasize the developments in
laser fluorination approaches. Printed as Volume 43 and entitled Stable Isotope Geochemistry,
it was edited by John W. Valley and David Cole. A rapid accumulation of data and research
in on meteorites and planetary crusts was summarized in 2008 with Volume 68 Oxygen in
the Solar System, edited by Glenn J. MacPherson, David W. Mittlefehldt, John H. Jones, and
Steven B. Simon. Due to new technical developments, most importantly the inductively coupled
multicollector mass spectrometry, the field of stable isotope geochemistry has branched out
to elements other than the traditional H, C, N, O, and S stable isotopes. This development
was summarized in 2004 with Volume 55: Geochemistry of Non-Traditional Stable Isotopes,
edited by Clark M. Johnson, Brian L. Beard, and Francis Albarède, and revisited in 2017
with Volume 82, entitled Non-Traditional Stable Isotopes, edited by Fang-Zhen Teng, James
Watkins, and Nicolas Dauphas.
Why this new volume? During the past two decades, two major and exciting fields
have emerged in the science of stable isotope geochemistry, both related to understanding
tiny variations (down into the lower ppm range) in the minor isotopologue abundances.
The first field is related to the distribution of “clumping” of heavy isotopes in molecules, such
as CO2 or CH4; the second field is the use of small variations in all three oxygen isotopes.
Both approaches allow researchers to resolve absolute temperatures, previously masked
process pathways, and exchange between various reservoirs.
In this new volume of Reviews in Mineralogy and Geochemistry we concentrate on
understanding the latter—understanding of variations among ratios of all three isotopes of
oxygen, with primary emphasis on terrestrial systems. Triple oxygen isotope variations may be
related to large, mass-independent fractionation effects such observed in the Earth atmosphere
or may be small and related to minute variations due to purely mass-dependent processes.
1529-6466/21/0086-0000$00.00 (print) http://dx.doi.org/10.2138/rmg.2021.86.00
1943-2666/21/0086-0000$00.00 (online)
Triple Oxygen Isotope Geochemistry ‒ Preface
Recent advancements in analytical resolution now allow for the identification of processes and
distinct reservoirs that were formerly hidden in the paradigm of a “single terrestrial fractionation
line”. New, high-resolution measurements are accompanied by advances in theoretical
calculations that dovetail with empirical calibrations and applications throughout this volume.
Here, we bring together the leading researchers and asked them to summarize the new
discoveries in their field of expertise. The result is a volume with 14 chapters spanning a wide
range of subjects: from ab-initio theoretical approaches to observation of triple oxygen isotope
variations in the Earth litho-, hydro- and atmosphere.
Triple Oxygen Isotope Geochemistry is a young and rapidly evolving field and some of
the observations reported in this volume will undoubtedly be modified in the future as new
discoveries and improved protocols are developed in the coming years. The triple oxygen
isotope concepts presented in this volume, however, will hopefully prove valuable to new
entrants in this field and lay a foundation for future growth.
As editors, we thank all the authors and reviewers that contributed to this volume. It is
their chapters and their reviews that ensured the high quality of this book. The staff support of
the Mineralogical Society of America, especially Rachel Russell and Ann Benbow, and that of
Ian Swainson, greatly made our editorial work easier and their help is more than appreciated.
October 09, 2020
Ilya Bindeman, University of Oregon
Andreas Pack, Georg-August-Universität Göttingen
v
Geochemistry
86 Reviews in Mineralogy and Geochemistry 86
TABLE OF CONTENTS
1 Why Measure 17O? Historical Perspective,

Triple-Isotope Systematics |and Selected Applications
Martin F. Miller, Andreas Pack
INTRODUCTION.....................................................................................................................1
Non-mass-dependent oxygen triple-isotope distributions..............................................3
Non-mass-dependent oxygen triple-isotope distributions generated by processes
not involving photochemical reactions........................................................................5
OXYGEN TRIPLE-ISOTOPE SYSTEMATICS.......................................................................6
Experimental measurements of ln(17/16α) /ln(18/16α) = θ..................................................8
Defining and quantifying deviations from a reference fractionation relationship..........8
REFERENCE MATERIALS AND STANDARDS.................................................................12
16
O, 17O AND 18O ABUNDANCES AND ISOTOPE RATIO RANGES IN
NATURALLY OCCURRING TERRESTRIAL MATERIALS...........................................15
Oxygen triple-isotope measurements of atmospheric O2............................................................................. 15
Oxygen triple-isotope measurements of terrestrial silicates.........................................16
Oxygen triple-isotope distributions in meteoric waters, snow and ice cores...............18
STANDARDIZING Δ′17O DATA FROM ROCKS AND MINERALS...................................20
SOME EXAMPLES OF THE APPLICATION OF
OXYGEN TRIPLE-ISOTOPE RATIO MEASUREMENTS...............................................21
Corrections to δ13C measurements of CO2...................................................................22
Comparisons of the oxygen triple-isotope compositions of the Earth and Moon........22
Investigating the climate of ‘Snowball Earth’ from hydrothermal rocks.....................24
Insights from Δ′17O measurements on basalts, shales and fluvial sediments...............25
Quantifying gross photosynthetic oxygen production in the oceans............................26
Oxygen triple-isotope ratios in bio-apatite: a proxy for CO2 partial pressure
in paleo atmospheres.................................................................................................26
CONCLUSIONS AND OUTLOOK........................................................................................27
ACKNOWLEDGEMENTS.....................................................................................................28
REFERENCES........................................................................................................................28
vi
Triple Oxygen Isotopes ‒ Table of Contents
2 Discoveries of Mass Independent Isotope Effects in the

Solar System: Past, Present and Future
Mark H. Thiemens, Mang Lin
THE BEGINNING OF ISOTOPES.........................................................................................35

Discovery and chemical physics..................................................................................35
The dawn of stable isotope geochemistry....................................................................37
AN OVERVIEW OF STABLE ISOTOPE GEOCHEMISTRY...............................................38
Mass dependent effects and applications.....................................................................38
Mass independent isotope effects and applications......................................................42
FUNDAMENTAL CHEMICAL PHYSICS OF MASS INDEPENDENT
ISOTOPE EFFECTS: WHAT IS KNOWN AND WHAT NEEDS TO BE KNOWN..........47
Bond formation processes............................................................................................47
The role of symmetry...................................................................................................47
Bond breaking isotope effects......................................................................................52
Physical chemical details of photodissociation general process..................................55
Specific examples of isotope effects in dissociation: carbon dioxide..........................56
PHOTODISSOCIATION IN THE EARLY SOLAR SYSTEM AND
SELF-SHIELDING MODELS.............................................................................................59
Self-shielding models...................................................................................................59
Mass independent isotope effects in CO photodissociation.........................................60
High resolution view of isotopic photodissociation.....................................................64
Self-shielding final test and errors................................................................................68
Nebular fate of water and water ice from photodissociation........................................69
Summary......................................................................................................................75
BOX 1: SELF-SHIELDING PROBLEMS..............................................................................77
SOLID FORMATION IN THE EARLY SOLAR SYSTEM...................................................78
Simultaneous formation of the first solids and their isotopic anomalies......................78
Oxygen isotopic composition of the Solar System......................................................82
A NEW MODEL FOR TRIPLE OXYGEN ISOTOPES, METEORITES, AND
THE ORIGIN OF THE SOLAR SYSTEM..........................................................................84
Revisiting triple oxygen isotopes of CAI.....................................................................84
Material balance of bulk meteorites oxygen isotopes..................................................85
BOX 2: SUMMARY OF CHEMICAL MECHANISM MODEL FOR
PRODUCTION OF METEORITE OXYGEN ISOTOPIC ANOMALIES..........................88
ACKNOWLEDGEMENTS.....................................................................................................89
REFERENCES........................................................................................................................89
vii
3 Climbing to the Top of Mount Fuji: Uniting Theory and

Observations of Oxygen Triple Isotope Systematics
Laurence Y. Yeung, Justin A. Hayles
INTRODUCTION...................................................................................................................97
BASIC CONCEPTS................................................................................................................99
The equations...............................................................................................................99
Quantifying mass-dependent fractionation................................................................102
Natural variability in the triple-oxygen exponents θ..................................................104
A note about anharmonicity.......................................................................................107
ELECTRONIC STRUCTURE CALCULATIONS...............................................................107
Molecular models.......................................................................................................108
Electronic structure methods: different ways of treating chemical interactions........109
Representing electrons using basis sets......................................................................111
Key takeaways............................................................................................................113
A COMPARISON OF THEORETICAL METHODS...........................................................113
Approach....................................................................................................................113
Results........................................................................................................................114
ANALYTICAL CONSIDERATIONS...................................................................................119
Scale distortion and calibration..................................................................................120
Physical and reservoir effects in the real world..........................................................121
CASE STUDY: KINETIC ISOTOPE FRACTIONATION DURING
CARBONATE ACID DIGESTION....................................................................................124
The problems..............................................................................................................124
Evaluation of theory and measurements.....................................................................125
CONCLUDING REMARKS.................................................................................................127
ACKNOWLEDGEMENTS...................................................................................................128
REFERENCES......................................................................................................................128
4 Mass Dependence of Equilibrium Oxygen Isotope Fractionation in

Carbonate, Nitrate, Oxide, Perchlorate, Phosphate, Silicate, and
Sulfate Minerals
Edwin A. Schauble, Edward D. Young
INTRODUCTION.................................................................................................................137
D17O signatures of equilibrium processes...................................................................137
PREDICTING 17O/16O SIGNATURES..................................................................................140
Estimating mass-dependent fractionations in the harmonic approximation..............140
Effects beyond the harmonic approximation..............................................................146
How important are effects beyond the harmonic approximation?..............................150
∆′17O signatures of equilibrium..................................................................................152
SUMMARY...........................................................................................................................171
REFERENCES......................................................................................................................172
viii
5 Standardization for the Triple Oxygen Isotope System:

Waters, Silicates, Carbonates, Air, and Sulfates
Zachary D. Sharp, Jordan A.G. Wostbrock
INTRODUCTION.................................................................................................................179
INTERCALIBRATING WATERS, CARBONATES AND SILICATES FOR d18O..............182
Water...........................................................................................................................182
Carbonates..................................................................................................................183
Silicates......................................................................................................................184
Comparison of all data...............................................................................................184
STANDARDIZATION FOR D17O VALUES OF SELECTED
REFERENCE MATERIALS..............................................................................................185
Method of measurement.............................................................................................185
Water...........................................................................................................................185
Silicates......................................................................................................................187
Air...............................................................................................................................189
Carbonates..................................................................................................................190
Sulfates.......................................................................................................................191
CONCLUSION......................................................................................................................192
REFERENCES......................................................................................................................193
6 Mass-Independent Fractionation of Oxygen Isotopes

in the Atmosphere
Marah Brinjikji, James R. Lyons
INTRODUCTION.................................................................................................................197
PREVIOUS WORK IN THE LOWER ATMOSPHERE.......................................................197
Early measurements...................................................................................................197
Theory and photochemical modeling of O3 MIF signatures......................................198
Applications of oxygen MIF......................................................................................198
MATERIALS AND METHODS...........................................................................................199
VULCAN photochemical model................................................................................199
RESULTS FOR THE LOWER ATMOSPHERE...................................................................200
RESULTS FOR THE UPPER ATMOSPHERE.....................................................................209
DISCUSSION........................................................................................................................212
CONCLUSIONS....................................................................................................................212
REFERENCES......................................................................................................................213
ix
7 Isotopic Traces of Atmospheric O2 in Rocks, Minerals, and Melts

Andreas Pack
INTRODUCTION.................................................................................................................217
THE ISOTOPE COMPOSITION OF THE ATMOSPHERE................................................217
Modern air O2.............................................................................................................217
Why is the reconstruction of the Δ′17O of air O2 from rocks interesting?..................220
AIR–ROCK INTERACTION................................................................................................220
Meteorite fusion crust.................................................................................................220
Cosmic spherules........................................................................................................222
From air to sulfates and beyond.................................................................................224
Deep sea manganese nodules.....................................................................................225
Tektites.......................................................................................................................227
Air inclusions in tektite glasses..................................................................................231
Skeletal apatite and eggshell carbonate......................................................................232
High temperature technical products..........................................................................232
Using Δ′17O as a new tool to distinguish synthetic and natural gems........................234
CONCLUSIONS....................................................................................................................236
REFERENCES......................................................................................................................236
8 Triple Oxygen Isotopes in Evolving Continental Crust, Granites,

and Clastic Sediments
Ilya N. Bindeman
INTRODUCTION.................................................................................................................241
Shales, what do they reflect?......................................................................................242
Shales, granites and continental crust growth............................................................246
PART I: TRIPLE OXYGEN ISOTOPES IN WEATHERING PRODUCTS FROM
MODERN AND RECENT ENVIRONMENTS.................................................................248
Weathering reactions and stable isotopic parameters.................................................248
Using triple oxygen isotopes in weathering products to estimate
T and d′18Ow values..................................................................................................249
Computing the weathering product............................................................................253
Weathering in different climate regions.....................................................................253
Effects of diagenesis on d18O and Δ′17O in shales......................................................255
Effects of metamorphism and anatexis on d18O and Δ′17O.........................................260
PART II: SHALES ACROSS THE GEOLOGIC HISTORY.................................................261
Overview of temporal trend of d18O and ∆′17O and Archean vs.
post Archean shales.................................................................................................261
The Phanerozoic shale record.....................................................................................266
x
PART III: THE TRIPLE OXYGEN ISOTOPE ANALYSIS OF GRANITES IN

COMPARISON WITH SHALES RECORD: INSIGHT INTO THE EVOLUTION
OF THE CONTINENTAL CRUST AND WEATHERING.............................................268
Comparing coeval record of Archean granites and shales..........................................274
Post-Archean granites and shales...............................................................................277
Alternative testable hypotheses of Archean–Proterozoic ∆′17O transition.................278
PART IV. SEDIMENTARY PROXIES SHOWING TEMPORAL D18O INCREASE..........279
SUMMARY: DEFINING THE D′17O CRUSTAL ARRAY...................................................283
REFERENCES......................................................................................................................284
9 Triple Oxygen Isotope Variations in Earth’s Crust

Daniel Herwartz
INTRODUCTION.................................................................................................................291
GUIDING PRINCIPALS.......................................................................................................292
Definitions, reference frames and the significance of TFL′s, MWL′s and q..............293
Mass dependent isotope fractionation in triple oxygen isotope space.......................295
Mixing........................................................................................................................298
QUANTIFYING FRACTIONATION FACTORS.................................................................301
The two directional approach.....................................................................................301
The three-isotope method...........................................................................................302
Isotope thermometry and geospeedometry.................................................................303
WATER–ROCK INTERACTION..........................................................................................304
Constraining input variables for the water–rock mass balance equation...................305
Constraining the isotopic composition of pristine (paleo-) fluids..............................307
Resolving the d18O composition of snowball Earth glaciers......................................307
Resolving the d18O composition of ancient seawater.................................................310
RESOLVING INDIVIDUAL PROCESSES..........................................................................311
Boiling and phase separation observed in well fluids................................................311
Assimilation of low d18O rocks..................................................................................311
Isotopic exchange with CO2 and SO2.........................................................................313
Decarbonation............................................................................................................313
Dehydration................................................................................................................313
Alteration....................................................................................................................314
CONCLUSIONS AND OUTLOOK......................................................................................315
REFERENCES......................................................................................................................316
xi
10 Triple Oxygen Isotope Trend Recorded by Precambrian Cherts:

A Perspective from Combined Bulk and in situ Secondary Ion Probe
Measurements
D.O. Zakharov, J. Marin-Carbonne, J. Alleon, I.N. Bindeman
INTRODUCTION.................................................................................................................323
PART I: PRECAMBRIAN CHERTS AS AN ARCHIVE OF ANCIENT SILICA CYCLE.326
The collection of early Precambrian cherts................................................................328
Silica precipitation and diagenesis, and the effect of oxygen isotopes......................332
Triple oxygen isotopes in silica..................................................................................334
Secondary ion mass spectrometry..............................................................................336
Raman spectroscopy...................................................................................................337
PART II: COMBINED SIMS AND
TRIPLE OXYGEN ISOTOPE ANALYSIS OF PRECAMBRIAN CHERTS....................338
Methods......................................................................................................................339
Results........................................................................................................................341
Discussion: various origins of Precambrian cherts and record of seawater...............343
Triple oxygen isotope evolution of seawater..............................................................352
FUTURE DIRECTION AND CONCLUSIONS...................................................................355
REFERENCES......................................................................................................................357
11 Triple Oxygen Isotopes in Silica–Water and Carbonate–Water

Systems
Jordan A.G. Wostbrock, Zachary D. Sharp
INTRODUCTION.................................................................................................................367
MINERAL–WATER OXYGEN ISOTOPE THERMOMETERS.........................................368
18
O/16O fractionation...................................................................................................368
Triple oxygen isotope fractionation............................................................................370
Equilibrium triple oxygen isotope fractionation—the Δ′17O–δ′18O plot.....................371
KINETIC EFFECTS RESULTING IN OXYGEN ISOTOPE DISEQUILIBRIUM.............372
Disequilibrium effects with dissolved inorganic carbon............................................373
Application to the triple oxygen isotope system........................................................374
SILICA–WATER FRACTIONATION...................................................................................376
Calibration of the triple oxygen isotope silica–water thermometer...........................376
Silica in the terrestrial environment............................................................................377
CARBONATE–WATER FRACTIONATION........................................................................378
Analytical method......................................................................................................378
Triple oxygen isotope fractionation in the calcite (aragonite)–water system.............379
Applications to natural marine samples.....................................................................381
CONSIDERATIONS NECESSARY TO INTERPRET ANCIENT SEDIMENTS...............382
Modelling changing ocean composition in the past...................................................384
Modelling triple oxygen isotope trends during diagenesis.........................................385
Application to carbonate and silica sediments...........................................................388
xii
TRIPLE OXYGEN ISOTOPES OF COEXISTING QUARTZ AND CALCITE..................393

CONCLUSION......................................................................................................................395
REFERENCES......................................................................................................................395
12 Triple Oxygen Isotope Systematics in the Hydrologic Cycle

Jakub Surma, Sergey Assonov, Michael Staubwasser
INTRODUCTION.................................................................................................................401
Analysis......................................................................................................................402
New data presented in this work................................................................................403
TRIPLE OXYGEN ISOTOPES IN WATER.........................................................................404
Triple oxygen isotope fractionation............................................................................404
17
O-excess during the formation of vapor...................................................................404
NATURAL VARIATIONS OF 17O-EXCESS IN WATER.....................................................405
Meaning and purpose of the Global Meteoric Water Line.........................................405
17
O-EXCESS DISTRIBUTION IN ATMOSPHERIC VAPOR..............................................412
FUTURE WORK AND CONCLUSION...............................................................................420
Future work................................................................................................................420
Conclusion..................................................................................................................422
REFERENCES......................................................................................................................423
13 Triple Oxygen Isotopes in Meteoric Waters, Carbonates, and

Biological Apatites: Implications for
Continental Paleoclimate Reconstruction
Benjamin H. Passey and Naomi E. Levin
INTRODUCTION.................................................................................................................429
METEORIC WATERS...........................................................................................................430
Evaporation from the oceans......................................................................................430
Rayleigh distillation and the triple oxygen isotope meteoric water line....................432
Evaporation from isolated bodies of water.................................................................434
Closed-basin and throughflow lakes...........................................................................435
Special considerations and case studies.....................................................................438
ANIMAL AND PLANT WATERS........................................................................................441
Prospects for reconstructing ∆′17O of past atmospheric O2........................................446
Plant waters................................................................................................................446
ANALYSIS OF CARBONATES AND BIOAPATITES........................................................449
Conversion methods...................................................................................................449
Exchange methods......................................................................................................451
Other methods............................................................................................................452
xiii
Interlaboratory reproducibility...................................................................................453
Fractionation exponents for carbonate–water equilibrium, and acid digestion..........454
FUTURE DIRECTIONS AND CONCLUDING REMARKS..............................................455
REFERENCES......................................................................................................................457
14 Small Triple Oxygen Isotope Variations in Sulfate:

Mechanisms and Applications
Xiaobin Cao, Huiming Bao
INTRODUCTION.................................................................................................................463
TRIPLE OXYGEN ISOTOPE SYSTEM..............................................................................465
SULFOXYANIONS–WATER OXYGEN ISOTOPE EXCHANGE.....................................466
Sulfate–water system..................................................................................................466
Sulfite–water system ..................................................................................................466
Thiosulfate–water system ..........................................................................................468
MICROBIAL SULFATE REDUCTION (MSR)...................................................................468
SULFIDE OXIDATION MECHANISMS.............................................................................469
Thiosulfate oxidation on pyrite surface......................................................................469
Sulfite oxidation by O2 in solution.............................................................................471
Sulfite oxidation by Fe3 + in solution..........................................................................472
The role of microbes on oxygen isotope composition of sulfate
derived from sulfur oxidation .................................................................................473
Comments on laboratory experiments........................................................................475
Constraining intrinsic equilibrium and kinetic oxygen isotope effects
during sulfide oxidation...........................................................................................476
APPLICATIONS....................................................................................................................477
Predicted sulfate δ18O and small Δ′17O.......................................................................477
Lake sulfate ...............................................................................................................479
Riverine sulfate...........................................................................................................480
ANALYTICAL METHODS .................................................................................................481
FUTURE OPPORTUNITIES................................................................................................482
REFERENCES......................................................................................................................484
Reviews in Mineralogy & Geochemistry
Vol. 86 pp. 1–34, 2021 1
Copyright © Mineralogical Society of America
Why Measure 17O? Historical Perspective,

Triple-Isotope Systematics and Selected Applications
Martin F. Miller
Planetary and Space Sciences, School of Physical Sciences
The Open University
Walton Hall
Milton Keynes MK7 6AA
UK
m.f.miller@open.ac.uk
Andreas Pack
Georg-August-Universität, Geowissenschaftliches Zentrum
Abteilung Isotopengeologie
Goldschmidtstraße 1
37077 Göttingen
Germany
apack@uni-goettingen.de
INTRODUCTION
For many years, it was considered that measurements of the least abundant stable isotope of
oxygen, 17O, would not provide any information additional to that obtainable from determinations
of the 18O/16O abundance ratio, which, by being a factor of ~5.2 larger than 17O/16O, can be
measured more easily. Here, we summarize significant events in the historical development
of oxygen stable isotope ratio measurements and their application to Earth and planetary
sciences, leading to a consideration of the potential information to be gained from high precision
measurements of the ‘third isotope’. This is followed by a short description of triple oxygen
isotope systematics, together with notation and definitions. In turn, this leads to a discussion of
how improvements in measurement precision, coupled with recent theoretical developments and
empirical findings, have enabled small variations in the 17O/16O isotope abundance ratio relative
to 18O/16O to provide new insights to a remarkable diversity of applications.
The foundations of stable isotope geochemistry can arguably be traced to the year 1947,
when a theoretical framework for calculating equilibrium constants and their temperature
dependence for isotopic exchange reactions was proposed by Urey (1947) and independently by
Bigeleisen and Göppert-Mayer (1947). The partitioning of isotopes at equilibrium results was
considered as a quantum mechanical process; reduced partition function ratios, in conjunction
with the Teller–Redlich approximation (Redlich 1935) formed the basis of the calculations.
The latter approximation allowed the rotational partition function contribution to be eliminated
from the calculations and it was also assumed that isotope substitution has no direct effect
on electronic energies. Thus, equilibrium fractionation factors could be determined from
consideration of the respective molecular vibration frequencies only. From the reported results,
Urey (1947) also suggested that, as a temperature increase from 0 °C to 25 °C should change the
equilibrium 18O abundance in carbonate by a factor of 1.004 relative to that in coexisting water,
accurate determination of the 18O/16O ratio in natural carbonates could be used to determine the

1943-2666/21/0086-0001$05.00 (online)
2 Miller & Pack
temperature at which they formed. Fortunately, these theoretical advances coincided with of the
development by Nier (1947) of a gas source mass spectrometer suitable for routine analyses of
isotope ratios of the lighter elements. Modifications of the Nier mass spectrometer to increase
the precision in relative abundance measurement by an order of magnitude were subsequently
proposed by McKinney et al. (1950). Amongst these, a change-over valve was incorporated, to
permit rapid switching between sample and reference gases being admitted to the ion source.
Furthermore, the delta notation for reporting stable isotope relative abundances was
introduced. Thus, for oxygen:
18
Rsample  18 Rreference (1)
 18 O  18
Rreference
where 18Rsample is the 18O/16O abundance ratio in the sample; 18Rreference is the corresponding
ratio in a reference material. The δ value, being a ratio of dimensionless quantities, is also
dimensionless. Because it is of small magnitude (≪ 1) in natural systems, it is usually reported
as parts per thousand (designated ‘per mil’, or ‘‰’).
The developments in mass spectrometry instrumentation enabled McCrea (1950), Urey
et al. (1951) and Epstein et al. (1951) to investigate the carbonate-water isotopic temperature
exchange scale in detail. It was shown that δ18O measurements of marine carbonates could be
used to determine the ocean water temperatures to ±1 °C on a geologic timescale. Urey et al.
(1951) already noted that the approach requires knowledge of the past seawater composition
and that the δ18O of the carbonate did not change with time. These two factors persist to be
the most important unknowns in paleo-thermometry. Epstein and Mayeda (1953) extended the
measurements to include various natural waters and detail the effects of glaciation. Following
a preliminary investigation by Baertschi (1950), Baertschi and Silverman (1951) extracted
oxygen from silicate crustal rocks by fluorination (with HF and either ClF3 or F2) and reported
a δ18O range of 24‰. They also noted that their measurements (made on O2 as the analyte gas)
could be used to determine the temperature of equilibration and mineral closure. Conversion
of extracted O2 to CO2, by reaction with a heated graphite rod (Clayton 1955) was generally
adopted for δ18O measurements in future studies. This was because CO2 molecular ions occur
in a region of the mass spectrum not coinciding with the major background ions from air O2
and N2. Furthermore, CO2 is less corrosive to the mass spectrometer filament than is O2. Later,
Clayton and Mayeda (1963) used BrF5 as the fluorination reagent for extracting oxygen from
silicate rocks, for improved oxygen yields and reduced systematic errors in the isotope data.
Besides equilibrium exchange, variations of light element stable isotope ratios in
nature may also result from kinetic effects, with the rates of chemical reactions (or physical
processes, such as diffusion) differing for isotopically substituted molecules. Early theoretical
considerations for quantifying the effects of isotopic substitution on the rates of chemical
reactions were published by Bigeleisen (1952) and by Bigeleisen and Wolfsberg (1957).
Despite the existence of the least abundant stable isotope of oxygen, 17O, having been
suggested and confirmed as early as 1924 (Blackett 1925), and its presence identified in Earth’s
atmosphere (Giauque and Johnston 1929) before the discovery of deuterium by H. C. Urey
in 1931, no mention of 17O was made in the early calculations of equilibrium constants and
their temperature dependence for isotope exchange reactions. This was also true for a review
three decades later (Richet et al. 1977). The first application of 17O in the isotope geochemistry
literature was the ‘Craig correction’ (Craig 1957) for quantifying the contribution of 12C16O17O+
isotopologues1 to the mass 45 ion beam when deriving δ13C values from measurements on CO2.
Based on Craig’s (1957) data on CO2 extracted from Peedee Formation calcite by concentrated
H3PO4 at 25.2 °C (McCrea 1950), ~6.3% was attributable to 12C16O17O+. In formulating the
1
Isotopologues (a contraction of isotopic homologues) are molecular entities that differ only in isotopic
composition, e.g. H216O, H217O, H218O.
Historical Perspective, Triple-Isotope Systematics and Selected Applications 3
correction procedure, Craig (1957) postulated that: “For systems in thermodynamic equilibrium,
consideration of the vibrational frequency decrease for the addition of one and of two neutrons
to a nucleus indicates that the fractionation factor for the distribution of 18O between two
compounds should be the square of the fractionation factor for 17O”. The relative abundances of
the three oxygen isotopes should therefore follow a simple power law:
0.5
17
Rsample  18 Rsample  (2)
= 
17
Rreference  18 Rreference 
In delta notation
1   17 O 
(1   18 O)0.5 (3)
Using the Maclaurin series expansion
ln 1  x  x  12 x 2  13 x 3   x (4)
and neglecting higher order terms, this relationship between the three isotopes can thus be
approximated to:
 17 O  0.5 18 O (5)
Isotope ratio modifications that follow this pattern of proportionality are commonly referred to
as ‘mass-dependent’ fractionations. Essentially, they result from the mass difference between
17
O and 16O (1.0042 Da) being approximately half the mass difference between 18O and 16O
(2.0042 Da). Clearly, if all processes that modify δ18O values followed this simple law exactly,
then indeed nothing would be learned from also measuring δ17O. Fortunately, however,
oxygen triple-isotope distributions in natural systems are more variable than is suggested by
this simple analysis.
Non-mass-dependent oxygen triple-isotope distributions
In a study of refractory calcium-aluminum-rich inclusions, fragments, chondrules and
isolated crystals in type-2 and type-3 (C2 and C3) carbonaceous chondritic meteorites,
Clayton et al. (1973) made the surprising discovery that the apparent δ13C of the CO2 used as
the analyte gas for isotope ratio measurements was negatively correlated with the δ18O value,
which varied by nearly 40‰. As the standard (heated graphite rod) method for the conversion
of O2 to CO2 had been used, the authors correctly attributed the apparent change in δ13C to a
change in δ17O. Assuming that the δ13C value was constant, Clayton et al. (1973) found that the
oxygen triple-isotope distributions conformed to δ17O ≈ δ18O, rather than to a mass-dependent
trend with δ17O ≈ 0.5 δ18O. Approximately half of the samples were from the Allende meteorite
(a type CV3). In addition to the data forming an array of slope one on a δ17O versus δ18O plot,
the samples were strongly depleted in both minor isotopes. This led the authors to suggest
that the oxygen isotopic compositions probably resulted from the admixture of a component
of almost pure 16O, of nucleosynthetic origin. That explanation was abandoned almost three
decades later (Clayton 2002), however, in favor of a CO ‘self-shielding’ mechanism. This
is now widely accepted and involves optical shielding of ultraviolet radiation by the most
abundant isotopologue of carbon monoxide, 12C16O. Such a mechanism had previously been
suggested by Thiemens and Heidenreich (1983), although uncertainty about the chemical
speciation of oxygen in the presolar nebula prevented those authors from assigning specifically
which oxygen-bearing gas-phase molecule was most likely to be implicated. Other mechanisms
to explain the δ17O ≈ δ18O relationship observed in many meteoritic components continue to
be discussed, however (e.g., Chakraborty et al. 2013; Thiemens and Lin 2021, this volume).
4 Miller & Pack
Following the discovery of the unusual relationship between δ17O and δ18O in the high-
temperature phases of Allende and other carbonaceous chondrites, further work (with O2 as
the analyte) led to the suggestion by Clayton et al. (1976) and Clayton and Mayeda (1983)
that groups of meteorites whose δ18O and δ17O values plot on a distinct mass-dependent
fractionation line (displaced from one another) originated on separate planetary bodies.
The characterization and classification of meteorites on the basis of oxygen triple-isotope
measurements (nowadays at significantly higher levels of precision) continues to be of
fundamental importance to meteoritics and studies of the solar system.
The next major discovery to stimulate interest in 17O was when Thiemens and Heidenreich
(1983) showed that the generation of ozone from molecular oxygen by electrical discharge
also produces a slope one array on a δ17O versus δ18O plot. This demonstration that mass-
independent (also referred to as ‘non-mass-dependent’) isotopic fractionation of oxygen can
result from a chemical process was of far-reaching significance. At the time, it was suggested
that the effect may result from self-shielding by the major isotopologue, 16O2, although that
was subsequently shown to be improbable (Navon and Wasserburg 1985). Heidenreich and
Thiemens (1986) then suggested that stabilization of the excited transition state during ozone
formation is where the anomalous isotope effect occurs and that symmetry may play a role,
as may isotope-specific reaction rates. The empirical findings have been discussed from a
theoretical perspective in numerous subsequent publications, but details of the mechanism are
still not fully understood. See Thiemens and Lin (2021, this volume) for further information.
The laboratory findings were followed by the discovery that stratospheric ozone is also
characterized by a mass-independent 17O enrichment (Schueler et al. 1990; Krankowsky et
al. 2000; Mauersberger et al. 2001), as are stratospheric CO2 (Thiemens et al. 1991, 1995a,b;
Lämmerzahl et al. 2002) and stratospheric nitrous oxide (Cliff et al. 1999). Lämmerzahl et
al. (2002) showed a tight coupling of the δ17O and δ18O values of stratospheric CO2, with the
ratio being surprisingly high, at 1.70 ± 0.03, for the altitude range sampled (19 to 33 km).
As shown in the same paper, ozone at this altitude range was found to exhibit a δ17O versus
δ18O slope of 0.62 ± 0.06, which was probably controlled largely by the local temperature
(Krankowsky et al. 2000). Tropospheric ozone was also found to be unusually enriched in
17
O (Krankowsky et al. 1995; Johnston and Thiemens 1997), as was tropospheric nitrous
oxide (Cliff and Thiemens 1997). Transfer of 17O enrichment, directly or indirectly, from
stratospheric ozone into most other oxygen-bearing constituents of the atmosphere has since
been documented (and reviewed by Thiemens 2013; Thiemens and Lin 2019). Stratospheric–
tropospheric exchange results in tropospheric O2 being depleted in 17O (Luz et al. 1999)
relative to the mass-dependent composition of photosynthetic O2. Stratospheric CO2 entering
the troposphere (Boering et al. 2004; Thiemens et al. 2014; Liang and Mahata 2015) may
experience successive isotopic exchange with water, especially leaf water, where the process
is promoted by the presence of carbonic anhydrase. The isotopic exchange reduces—and may
eventually eliminate—departure from non-mass-dependent composition.
Few other occurrences of the generation (rather than inheritance) of mass-independent
oxygen isotopic composition have been documented. The fractionation in atmospheric CO
as a result of the CO + ∙OH reaction (Röckmann et al. 1998) is one example. A second is
photodissociation of CO2 by of 185 nm wavelength (ultraviolet) radiation, generating CO
and O2 which are unusually enriched in 17O (Bhattacharya et al. 2000). Carbon monoxide
photodissociation at shorter wavelengths (90–108 nm) in vacuum leads to atomic oxygen
formation, which reacts with CO to produce CO2 enriched in 17O (Chakraborty et al. 2012).
Of considerable significance in a cosmochemical context is the discovery that silica formed
during gas phase oxidation of silicon monoxide by OH is characterized by a slope one array
on the δ17O versus δ18O plot (Chakraborty et al. 2013).
The first examples of rocks or minerals on Earth that are characterized by non-mass-
dependent oxygen isotope composition were discovered by Bao et al. (2000a) in massive
sulfate deposits (gypcretes from the central Namib Desert and sulfate-bearing Miocene
volcanic ash-beds in North America); also in sulfates from Antarctic dry valley soils (Bao
et al. 2000b) and in desert varnish sulfates from Death Valley, USA (Bao et al. 2001).
Substantial non-mass-dependent oxygen isotope distributions were subsequently found in
atmospheric nitrate aerosols, sampled at coastal La Jolla, California (Michalski et al. 2003).
Nitrate minerals (and, to a lesser extent, sulfates) of the Atacama Desert, northern Chile, also
exhibit substantial enrichments of 17O (Michalski et al. 2004), as do nitrates sampled in the dry
valleys of Antarctica (Michalski et al. 2005). Modelling indicated that these unusual isotopic
compositions could be traced to photochemical reactions in the troposphere and stratosphere.
Sulfate in snow and ice sampled at the South Pole (Savarino et al. 2003) was found to
be characterized by strongly non-mass-dependent oxygen isotopic compositions. The authors
suggested that SO2 from explosive volcanic eruptions which ejected substantial quantities
of material into the stratosphere was probably responsible and that reaction with O(3P) in
the stratosphere, rather than with coexisting ∙OH, was the most likely mechanism. Similarly,
volcanic sulfate from volcanic supereruption ash deposits sampled from dry lake beds in
the Tecopa basin, California, contain strong enrichments in 17O relative to mass-dependent
isotopic composition (Martin and Bindeman 2009), indicative of the photolysis and oxidation
of volcanic SO2 by 17O-enriched ozone or ∙OH, in the stratosphere. Eruptions which did not
eject material higher than the troposphere, however, produced sulfate of, or very close to,
mass-dependent oxygen isotopic composition (Martin et al. 2014).
Non-mass-dependent oxygen triple-isotope distributions generated by processes not
involving photochemical reactions
There are three reported examples of non-mass-dependent fractionation of oxygen
isotopes—and not involving photochemistry—having been demonstrated in the laboratory.
The first is that of ozone thermal decomposition (Bhattacharya and Thiemens 1988; Wen and
Thiemens 1990, 1991). Equal 17O and 18O enrichments were found in the product O2 when
the decomposition was performed at 110 °C (Wen and Thiemens 1991). In contrast, photolyic
decomposition by visible light gave a δ17O/δ18O slope of ~0.53 at room temperature; this
increased to ~0.65 if ultraviolet radiation was used.
The second example is that associated with the thermal decomposition of divalent
metal carbonate minerals (calcite, magnesite and dolomite were the examples used), under
conditions that prevent back-reaction and isotopic exchange between the resulting metal oxide
and released CO2 gas. For a collection of carbonates covering a wide range of δ18O values
(and including NBS 18 carbonatite and NBS 19 calcite), it was found that the decomposition
products fitted parallel mass-dependent fractionation arrays, offset from each other by ~ 0.4‰
(Miller et al. 2002). The metal oxides were anomalously depleted in 17O relative to mass-
dependent composition, whereas the CO2 was enriched by a corresponding amount (half the
magnitude of the depletion in the metal oxide). There is still no satisfactory explanation for the
empirical findings. It has been suggested (J. R. Hulston, pers. comm.) that Fermi resonance
might be an influential factor. However, the theoretical analysis required to investigate that
hypothesis has yet to be performed. Fermi resonance occurs when two vibrational energy
levels associated with different diatomic or polyatomic vibrations have nearly the same
energy, that is, may be ‘accidentally degenerate’. It occurs in CO2, where the υ1 vibration
at 1337 cm−1 is close to twice the υ2 vibration at 667 cm−1. A splitting of the 1337 cm−1 line
results, as observed in the Raman spectrum. Essentially, Fermi resonance causes shifting of the
energies and intensities of absorption bands; this has implications for isotope effects. Fermi
resonance might be expected to occur during vibrational excitation of carbonates. In calcite,
for example, υ3 occurs at 1432 cm−1, which is very close to twice the υ4 value of 714 cm−1
6 Miller & Pack
(both bands are Raman active). Depending on whether or not there is a linear progression
in vibrational energy levels associated with substituting 16O by 17O or 18O in the crystalline
carbonate lattice could have significant implications for the isotopic composition of the thermal
decomposition products, if the reaction pathway is influenced by Fermi resonance.
The third example—which does not involve any chemical reaction—is that the diffusion
of molecular O2 gas, at low pressure and in a closed volume, defies mass-dependent isotopic
distributions when a thermal gradient is applied (Sun and Bao 2011a). In contrast to the more
abundant stable isotopes of oxygen, 17O has a non-zero nuclear spin value (5/2). In a follow-up
paper, Sun and Bao (2011b) postulated that a—usually negligible—nuclear spin effect on the
gas diffusion coefficient, amplified by the temperature gradient, may be responsible for the
empirical observations. Such an effect had been predicted, from theoretical considerations,
some 35 years earlier (Zel’dovich and Maksimov 1976).
OXYGEN TRIPLE-ISOTOPE SYSTEMATICS

The equilibrium mass-dependent distribution of the three stable isotopes of oxygen
resulting from isotopic exchange between two chemical entities A and B may be described in
terms of the 17O and 18O abundance ratios relative to 16O as

17
RA  18 RA  (6)
17
 
RB  18 RB 
which, by definition, gives the relationship between the respective fractionation factors as
 
 (7)
17/16
AB   AB
18/16
and as expressed in delta values:


1   17 O A  1   18 O A  (8)
 
1   17 O B  1   18 O B 
The exact magnitude of the exponent θ depends on the identity of A and B, since the
energy differences between the various isotopically-substituted molecules depends not only
on the respective masses of 16O, 17O and 18O, but on the reduced masses in the molecular
vibrations. The first investigation of such variations, and their temperature dependence,
was by Matsuhisa et al. (1978). Calculations were performed according to the procedure
of Urey (1947), i.e. with θ obtained as the ratio of natural logarithms of reduced partition
function ratios, and with vibrational frequencies for 17O-containing species calculated from
published values for the corresponding 16O- and 18O-bearing entities. The harmonic oscillator
approximation for molecular vibrations was assumed throughout. The authors reported that,
for gas phase equilibrium exchange between CO2 and water, θ ranges from 0.5233 at 0 °C to
0.5251 at 727 °C. This variation appears small, compared with the corresponding ln(α18/16)
varying by a factor of approximately twenty, and illustrates that the magnitude of θ is not
particularly sensitive to temperature. For large changes in δ18O, however, small variations in θ
become important and the resulting influence on oxygen triple-isotope distributions is usually
measurable, at current levels of precision.
Matsuhisa et al. (1978) also noted that, at low temperatures, the natural logarithm
of the ratio of partition function ratios for CO2 and water approach limiting values given
by the respective ratios of the zero-point energy differences between the isotopically
substituted molecules. For isotope exchange between water vapor and atomic oxygen, this is
0.53052; the comparable figure for CO2 is 0.52554. At high temperatures, the ratio approaches
a constant value of 0.53053 for all oxygen-bearing molecules, as derived from equation 8 of
Urey (1947) and is in accord with the simple expression
 17Q 
ln   1 1
 16Q  
  m m 16 17 (9)
 18Q  1 1
ln   m m
 16Q  16 18
 
where m16, m17 and m18 are the atomic masses of 16O, 17O and 18O respectively. Matsuhisa et al.
(1978) observed that, for temperatures of interest, ratios of the natural logarithms of partition
function ratios for oxygen-bearing molecules vary from about 0.527 to 0.530. In exchange
reactions, however, the θ values will be somewhat lower and more variable, depending on the
specific reaction considered. The authors reported that a range of 0.520 to 0.528 was obtained
from various representative reactions (unspecified) for which calculations were made. This was
a highly significant finding, which has since been shown experimentally to be largely true. In
the same paper, the authors also reported an experimental determination of 103ln(α17/16) and
103ln(α18/16) for oxygen isotope exchange at equilibrium in the quartz–water system at 250 °C.
The respective values of 4.75 and 9.03 gave θ as 0.526. This number has been confirmed by
more recent theoretical calculations (Cao and Liu 2011) and empirical data (Sharp et al. 2016).
In addition to considering equilibrium processes, Matsuhisa et al. (1978) reported that
kinetic processes involving isotope exchange can be examined using a comparable formalism.
Their calculations for the example of pinhole (Graham’s law) diffusion, however, seem to
have been based on an erroneous algorithm; the reported results are at variance with more
recent discussions of the same example (Young et al. 2002; Dauphas and Schauble 2016). The
relationship as derived in the later works is:
ln 
M16 

diffusion   M17  (10)
 M16 
ln 

 M18 
where Mi refers to the masses of the isotopically substituted molecules. This gives values
ranging from 0.516, for the diffusion of atomic oxygen, to 0.501 in the high molecular mass
limit. As noted by Dauphas and Schauble (2016), the latter value also corresponds to
m16  m17 (11)

diffusion 
m16  m18
Clayton and Mayeda (2009) showed experimentally that the lower limit value of θ for a kinetic
process was approached during thermal dehydration, in vacuum, of Mg–O–H units in the
minerals serpentine and brucite.
The oxygen isotopic compositions of crustal silicates on Earth generally results from
interactions between solids, melts and aqueous fluids, involving many individual processes
and with temperature playing an influential role. Therefore, it is to be expected that, as
suggested by Matsuhisa et al. (1978), collections of silicate rocks and minerals will be
characterized by oxygen triple-isotope distributions that result from the average of the many
individual slopes for specific processes. Measurements on diverse rock and mineral samples
reported during the past decade (and discussed below) indicate that, as expected, equilibrium
processes dominate the slope value and there seems to be little evidence to suggest that the
simple diffusion model is relevant in this context.
8 Miller & Pack
Experimental measurements of ln(17/16α) /ln(18/16α) = θ

Few experimental measurements of the equilibrium exponent θ for oxygen triple-isotope
exchange have been reported to date. Pack and Herwartz (2014) found that, for SiO2 as
chert in equilibration with water at ~50 °C, θ is 0.5235 ± 0.003. At 8 °C, a lower value of
0.5212 ± 0.0008 was determined from opal equilibrated with water. Data from high temperature
rock assemblages (granite and San Carlos lherzolite) as reported in the same paper indicated
that θ was of the order of 0.528 to 0.529, but the associated error bars were significantly larger
because of the very limited δ18O ranges of the samples investigated. Sharp et al. (2016) found
that, for the temperature range ~0 to 50 °C, θ in the SiO2−water system is 0.523−0.524 (see also
Wostbrock and Sharp 2021, this volume). Taken together, these findings are consistent with the
value at 250°C being 0.526, as determined experimentally by Matsuhisa et al. (1978). Oxygen
triple-isotope exchange between biogenic apatite and water under equilibrium conditions has
also been investigated experimentally (Pack et al. 2013), with θ reported to be 0.523 at 37 °C.
Barkan and Luz (2005) measured the fractionation factors 17/16α and 18/16α for water
liquid–vapor exchange at equilibrium and found the ln(17/16α) / ln(18/16α) ratio, θ, to be
0.529 ± 0.001 over the temperature range 11.4 to 41.5°C. The same authors subsequently
conducted evaporation experiments to investigate the relative diffusivities of water vapor
isotopologues in air and derived values for the diffusion fractionation coefficients 17/16αdiffusion
and 18/16αdiffusion (Barkan and Luz 2007). The ratio ln(17/16αdiffusion) / ln(18/16αdiffusion) was found to
be 0.5185 ± 0.0002, in very good agreement with the theoretical value and significantly smaller
than the corresponding ratio reported by the same authors for water liquid–vapor equilibrium.
For CO2–H2O equilibrium exchange, θ was determined by Barkan and Luz (2012) as
0.5229 ± 0.0001 at 25 °C, which is in good agreement with the value of 0.522 ± 0.002 for the
temperature range 2 to 37 °C as reported by Hofmann et al. (2012). Both values, however,
are notably lower than the 0.5246 calculated from theoretical considerations by Cao and Liu
(2011)2 and also somewhat less than 0.5235 as calculated by Matsuhisa et al. (1978).
Defining and quantifying deviations from a reference fractionation relationship
To quantify departures from a specific fractionation relationship between 17O/16O and
16
18
O/ O, Clayton and Mayeda (1988) proposed the term Δ17O, which they defined as:
 17 O 
 17 O – 0.52 18 O (12)
This famous equation has since been used routinely in studies of meteorites; also for quantifying
terrestrial non-mass-dependent isotope effects. It is based on approximating a power law
relationship, as discussed above. Assigning 0.52 as the constant of proportionality was attributed
to measurements made more than a decade earlier (Clayton et al. 1976; Matsuhisa et al. 1978)
of 35 terrestrial samples—including various rocks and waters—which were found to form an
array of slope 0.5164 ± 0.0033 (standard error of the mean, SEM) on a δ17O versus δ18O plot.
Unfortunately, no details of the samples, nor the actual isotope data, were reported. Matsuhisa
et al. (1978) mentioned that, on the basis of those measurements, a slope of 0.520 was chosen
for the quartz–water system as a reasonable compromise between theoretical calculations and
empirical observations. This defined a reference (terrestrial) fractionation line; the extent to
which meteorite samples deviated from this was quantified by the parameter Δ17O. Definitions of
Δ17O using different constants of proportionality have since appeared in the literature, according
to the particular application and associated mass fractionation characteristics. For example, Cliff
and Thiemens (1997) used 0.515 in defining the Δ17O of atmospheric nitrous oxide; Boering et
al. (2004) assigned 0.516 for defining the Δ17O of stratospheric carbon dioxide; Luz et al. (1999)
used 0.521 as determined from the fractionation of O2 during respiration.
2
The θCO2–water value at 25 °C was later recalculated as 0.523 (Y. Liu, pers. comm. in Hofmann et al. 2012).
A consequence of the δ17O versus δ18O relationship not being truly linear is that the
slope given by a collection of data points on such a plot is dependent on the range of δ values
associated with the group of samples and on the isotopic composition of the reference material.
An example of this is shown in Figure 1. For consistency of reporting and comparison of
data sets, it is therefore advantageous to avoid the approximation. The oxygen triple-isotope
distributions in a collection of silicate rock and mineral samples (or waters) of diverse origin
will generally conform to the mass-dependent fractionation relationship

17
Rsample  18 Rsample  (13)
 
17
Rreference  18 Rreference 
as shown experimentally by Meijer and Li (1998) for natural waters. The exponent term in
this case is designated as λ rather than θ, to indicate that the isotope exchange processes are
undefined. Thus, λ is a purely empirical parameter, the magnitude of which results from the
cumulative effects of (unspecified) fractionations associated with the history of the individual
samples. To quantify deviations from a specific mass-dependent fractionation curve, an
additional term is needed:

17
Rsample  Rsample 
18
17
 1  k   18  (14)
Rreference  Rreference 
With δ values rather than absolute ratios, and with k defined to be Δ17O, then, as suggested by
Miller (2002)
1   17 O
 17 O  1 (15)
1   O 
18 
which is of a similar form to the definition of the δ value. In linear format,
ln(1   17 O)   
ln 1   17 O   ln 1   18 O   (16)
It needs to be remembered that the δ17O and δ18O data here are not ‘per mil’ values, as
conventionally reported, but the absolute numbers (≪1). Therefore, it is useful to include
a multiplier of 103 so that the logarithmic terms involving δ17O and δ18O are then of similar
magnitude to the corresponding δ17O and δ18O data reported as ‘per mil’:
 
103 ln(1   17 O)  103 ln 1   17 O  103 ln 1   18 O   (17)
Following similar terminology introduced by Hulston and Thode (1965) in the context of
sulfur multiple isotope ratios, the quantities 103ln(1 + δ17O) and 103ln(1 + δ18O) are sometimes
denoted as δ′17O and δ′18O respectively. As reported by Miller (2002), a plot of 103ln(1 + δ17O)
versus 103ln(1 + δ18O) data obtained from a collection of silicate rocks and minerals (for
example) thus gives a linear array of slope λ, invariant to the range of δ values and to the
isotopic composition of the reference, and with the ordinate deviation of an individual sample
from the mass-dependent reference fractionation line being quantified by 103 ln(1 + Δ17O).
With 103ln(1 + Δ17O) written as Δ′17O, similar to the definitions of δ′17O and δ′18O, we then
obtain the simple, non-approximated relationship:
 17 O   17 O   18 O (18)
This definition is used throughout this volume. What is perhaps not immediately intuitive
is that, instead of assigning the magnitude of λ from a set of measurements, a value may
10 Miller & Pack
alternatively be assigned arbitrarily. For example, reference lines of slope 0.528 (as used
throughout this volume) or 0.5305—and passing through VSMOW—have been defined for
the reporting of Δ′17O in various recent studies. Not accounted for in this Δ′17O definition,
however, is the magnitude of any offset of a reference line from the zero point of the δ scale.
At the time that the definition of Δ′17O was proposed, it was considered that fractionation
arrays formed from silicates (or meteoric waters) were not offset from VSMOW. This was
subsequently found to be incorrect, although quantifying the small magnitude of the offsets
generally requires the highest levels of precision and accuracy. Defining a reference line that is
offset from the zero point of the δ scale by a specified amount may also be useful under some
circumstances. Therefore, an additional term is required in the definition of Δ′17O; it is usually
designated as γ, and with γ′ ≡ 103ln(1 + γ). Thus, we now have:
 17 O   17 O   18 O   (19)
If defined from an empirical data array, the exact magnitude of λ is representative of the
particular group; there is not necessarily an implied relationship between the individual samples.
Different collections of silicate rock and mineral samples generally give slightly different
values of λ (usually between 0.522 and 0.529) and of γ′. Recalibration more accurately to the
VSMOW scale, on the basis of Pack et al. (2016) and Miller et al. (2020), of Open University
measurements of the two silicates arrays reported by Rumble et al. (2007) results in γ′ values
of +8 ± 19 ppm (95% confidence interval) for the hydrothermal quartz and chalk flint array;
–38 ± 5 ppm for the eclogite garnets. These values are 12 ppm lower than given by Hallis et
al. (2010). Similarly, recalibration by Tanaka and Nakamura (2017) of the γ′ value the same
authors reported in 2013, results in a revised γ′ value of –33 ± 5 ppm. This small number of
examples is probably not indicative of the γ′ range limits. Despite the variations of λ and γ′,
individual arrays have frequently been referred to as a ‘Terrestrial Fractionation Line’ (TFL).
20
Data reported relative to the ‘working standard’ O2 :
δ17O = (0.5240 ± 0.0010) δ18O + (0.310 ± 0.006)
15 R2 = 0.99996, n = 46
δ17O (‰)
10
5 Data reported relative to VSMOW:

δ17O = (0.5213 ± 0.0009) δ18O + (–0.018 ± 0.012)
R2 = 0.99996, n = 46
–15 –10 5 10 15 20 25 30 35
–5
δ18O (‰)
–10
Figure 1. Illustrating that a change of reporting reference for δ17O and δ18O measurements (from the labo-
ratory ‘working standard’ O2 to VSMOW, in this example) may be associated with a changed regression
line slope and ordinate offset on a δ17O versus δ18O plot. This is caused by the δ scale being non-linear,
leading to non-linearity of δ17O versus δ18O data arrays. The extent of non-linearity increases with distance
from the zero point of the scale. For this example, the data are from Miller et al. (1999). The δ18O value
of the ‘working standard’ O2 was 10.34‰ relative to VSMOW, with corresponding Δ′17O0.528 = +0.387‰
(as derived from Miller et al. 2015). Converting to the VSMOW scale, from data (shown as open circles)
reported relative to the working standard O2, therefore involves substantial positive shifts of the respective
δ18O and δ17O values. This results in best linear fit to the converted data (filled circles) requiring a different
regression line, as shown. Precision values refer to the 95% confidence interval.
The terminology λRL and γRL has recently been introduced to denote that λRL and γRL
are defined quantities (with RL referring to ‘reference line’) and is gaining acceptance. The
respective values need to be clearly stated, as there is currently no consensus. This is discussed
by Hofmann et al. (2017), who identified six different reference lines used in various studies.
An advantage of writing Δ′17O rather than Δ17O is that it provides a clear distinction from the
original definition introduced by Clayton and Mayeda (1988). For simplicity, the 103 multiplier
term is usually omitted from the definition and γ′RL approximated as γRL, on the basis that | γ |
–4
is usually < 10 or is defined to be zero. Thus,
 17 
O ln 1   17 O   RL ln 1   18 O   RL    (20)
This definition is becoming generally adopted (e.g., Hofmann et al. 2017) and is advocated
for future reporting of Δ′17O data. As a ratio, it may be written (in non-approximated form) as:
1   17 O
 17 O   RL 1 (21)
1   RL  1   18 O 
A graphical representation of the relationship between the various parameters is shown in
Figure 2. Because of the small magnitude of Δ′17O in many studies, it is usually reported as parts
per million (ppm). In some of the earlier literature, the term ‘per meg’ is used instead of ppm.
103 ln(1 + δ17O)
Δʹ17O
(Sample C)
Δʹ17O
(Sample B) C
B
Δʹ17O
(Sample A)
Intercept (γʹ) A
of mass-dependent
fractionation line defined 103 ln(1 + δ18O)
by the sample group
Figure 2. Schematic illustration of the relationship between Δ′17O values for a set of rock, water or gas
samples and: (i) the reference mass-dependent fractionation line characterized by assigned slope λRL
(0.528, for the VSMOW-SLAP scale) and assigned ordinate axis intercept value γ′RL (zero, in this ex-
ample) on the 103 ln(1 + δ17O) versus 103 ln(1 + δ18O) plot; (ii) the mass-dependent fractionation line of
slope λ and ordinate axis intercept value γ′ formed by the collection of samples. Unlike λ, the value of γ′
will depend on the isotopic composition of the reference material (usually VSMOW) relative to which δ17O
and δ18O values are reported; also on the accuracy of the calibration of the ‘working standard’ O2 relative to
that reference. If λRL differs significantly from λ, Δ′17O will vary with increasing δ18O as an artefact of the
increasing divergence (or convergence) of the two mass-dependent fractionation lines.
For completeness, we note that an earlier definition used by Farquhar et al. (1998) also
avoided the approximation of a linear relationship between δ17O and δ18O:
 17 O 1   17 O  (1   18 O)0.52 (22)
12 Miller & Pack
With λRL = 0.52 and γRL = 0, this expression gives very similar results to those obtained from
the recommended Δ′17O definition, for a given triple-isotope data set. We also note that (rather
confusingly) Angert et al. (2003) defined δ′17O − λ δ′18O as 17Δ. The 17Δ notation has not
gained general acceptance, however.
In much of the literature reporting oxygen triple-isotope ratio measurements of waters,
snow and ice cores, Δ′17O is referred to as ‘17O-excess’, a term introduced by Luz and Barkan
(2000) and subsequently defined by the same authors (Barkan Luz 2007) as
17
  
 ln 1   17 O  0.528 ln 1   18 O
O-excess  (23)
This does indicate that 17O-excess is ≥ 0 in all cases, otherwise there is the incongruity of a
‘negative excess’. The definition of 17O-excess is identical to the definition of Δ′17O used
throughout this volume. The expression ‘17O anomaly’ is also sometimes adopted (e.g.,
Thiemens et al. 1995b; Dauphas and Schauble 2016), to describe non-mass-dependent isotope
distributions. For simplicity and consistency, we suggest that it is preferable to use the term
Δ′17O universally and with the values of λRL and γRL clearly stated. For consistency throughout
this volume, the Editors have recommended that Δ′17O be defined using λRL = 0.528 and γRL = 0.
It should be noted that Δ′17O does not behave linearly with regard to mixing calculations,
unlike Δ17O as defined by Clayton and Mayeda (1988). It is therefore necessary to use the
associated δ17O and δ18O data for calculating changes in Δ′17O resulting from the mixing of fluids
characterized by different isotopic compositions. Finally, a detailed discussion of the theoretical
principles and calculations of mass-dependent fractionation processes in the oxygen triple-isotope
system is beyond the scope of this introductory section. For further reading, see Schauble and
Young (2021, this volume); Yeung and Hayles (2021, this volume); Brinjikji and Lyons (2021, this
volume); Thiemens and Lin (2021, this volume). The review by Dauphas and Schauble (2016) is
also recommended. Cao and Liu (2011) and Bao et al. (2016) provide additional insights.
REFERENCE MATERIALS AND STANDARDS

Early measurements of oxygen triple-isotope ratios (e.g., Clayton et al. 1973) were
reported relative to the Standard Mean Ocean Water (SMOW) standard described by Craig
(1961); this practice continued for some considerable time. In 1968, the International Atomic
Energy Agency (IAEA) began distributing two new water reference materials, Vienna Standard
Mean Ocean Water (V-SMOW) and Standard Light Antarctic Precipitation (SLAP). Both were
prepared by H. Craig at the University of California San Diego: V-SMOW by mixing distilled
Pacific Ocean water (collected in July 1967) with small amounts of other waters in order
to make the isotopic composition as close as possible to that of SMOW. V-SMOW is also
referred to as RM 8535 by the US National Institute of Standards and Technology (NIST).
SLAP was prepared from a firn sample collected in 1967 at Plateau Station, Antarctica, by E.
E. Picciotto of the Université Libre de Bruxelles. On the basis of comparative measurements
by 36 different institutions, the δ18O value of SLAP was assigned in 1976 as –55.5‰ exactly,
by the IAEA (Gonfiantini 1978), with V-SMOW (subsequently re-designated as VSMOW)
defined as the zero point of the scale. SLAP is also referred to as RM 8537 by NIST.
For improved consistency of reporting δ18O measurements, it was recommended that data
be normalized to the VSMOW-SLAP scale (Gonfiantini 1978). Thus, δ18O data (relative to
VSMOW) should be scaled by a factor of (–55.5) / (δ18OSLAP/VSMOW) where δ18OSLAP/VSMOW
is the measured value of SLAP relative to VSMOW. It has since been suggested that the
assigned δ18O value of SLAP may be significantly in error, however. On the basis of a
report by Verkouteren and Klinedinst (2004), Kaiser (2008) noted that the ‘true’ δ18O value
seems to be –56.18 ± 0.01‰, illustrating that the VSMOW-SLAP δ18O scale is associated
with significant uncertainties for values far removed from VSMOW. Because measurement
artefacts such as cross-contamination generally lead to a compression of the δ scale, the value
of –56.18 ± 0.01‰ for SLAP is likely to be more accurate. Nevertheless, the value of –55.5‰
has been retained for calibration to the VSMOW-SLAP scale.
With stocks of VSMOW and SLAP becoming low, the IAEA ceased to distribute
those reference waters after November 2006. However, successor materials VSMOW2 and
SLAP2, prepared to have nominally identical isotopic characteristics to VSMOW and SLAP
respectively, were produced at the IAEA Isotope Hydrology Laboratory and made available
from 2007 onwards. Details of the respective preparations are given by Harms and Gröning
(2017). It is important to recognize that the reporting scale for δ18O (and for δ2H) is still
denoted and referred to as the VSMOW–SLAP scale, despite VSMOW and SLAP being no
longer available. The non-availability of VSMOW and SLAP doesn’t prevent calibration (or
normalizing) measurements to VSMOW and SLAP using VSMOW2 and SLAP2, which
provide suitable traceability to VSMOW and SLAP respectively (Dunn et al. 2020).
Oxygen triple-isotope ratio measurements (especially at high precision) are generally
performed using molecular oxygen as the analyte gas and reported relative to the equivalent
ratios in a reference material (usually VSMOW). For investigations involving rocks and minerals,
the use of a water reference material causes significant calibration challenges. The accuracy of
reported δ17O and δ18O data is critically dependent, however, on the accuracy of calibration
(relative to VSMOW) of a laboratory ‘working standard’ O2 against which the isotopic
composition of molecular oxygen extracted from silicates (or waters) is compared. The extent
of instrumental scale compression should also be established, from measurements of molecular
oxygen extracted from SLAP. A few silicates, such as the University of Wisconsin garnet
standard UWG-2 (Valley et al. 1995) and San Carlos olivine, are widely used for calibrating the
δ18O value of individual laboratory’s ‘working standard’ O2, although there remains a lack of
consensus about the exact δ18O values of those silicates relative to VSMOW. Furthermore, there
are currently no standards for δ17O and recent attempts to characterize the δ17O values of UWG-2
and San Carlos olivine have not resulted in consensus (Pack et al. 2016; Sharp et al. 2016; Miller
et al. 2020; Wostbrock et al. 2020). Table 1 lists those recent results. It should be noted that two
distinct variations of San Carlos olivine have been identified (designated as Type I and Type II);
these are characterized by different δ18O values. It is probable that other variations also exist.
Type I is characterized by δ18O = 4.88‰ (Mattey and Macpherson 1993; Thirlwall et al. 2006).
Macpherson et al. (2005) reported a value of 4.84 ± 0.09‰, together with 5.22 ± 0.08‰ for the
more commonly-used Type II. In a more recent investigation, Starkey et al. (2016) compared
δ18O and also δ17O measurements of Types I and II San Carlos olivine. They noted that, despite
the variations in δ18O, the corresponding Δ′17O value seemed to be constant. This latter finding
was subsequently confirmed by Miller et al. (2020). For further discussion on standardizing
oxygen triple-isotope data, see Sharp and Wostbrock (2021, this volume).
Despite atmospheric O2 having a deficit of 17O relative to mass-dependent composition,
a number of studies have used it as a standard, notably in investigations of the triple-isotope
composition of dissolved O2 for estimating global and oceanic biological productivity (e.g.,
Luz et al. 1999; Luz and Barkan 2000). Because δ17O and δ18O measurements of air O2 are
affected by the presence of argon (which was not removed in those investigations), corrections
needed to be applied (Barkan and Luz 2003)3. Converting the resulting isotope data to the
VSMOW-SLAP scale, however, is complicated by the δ17O and δ18O values of air O2 relative
to the VSMOW reference water not having been universally agreed, as discussed below.
3
In more recent studies involving isotopic measurements of air O2 (Yeung et al. 2012; Young et al. 2014;
Pack et al. 2017; Wostbrock et al. 2020), all other atmospheric constituents were removed, including Ar.
14 Miller & Pack
Table 1. Recent high-precision oxygen triple-isotope measurements of widely-used silicate

standards San Carlos olivine and UWG-2 garnet.
Institution n δ18OVSMOW (‰) Δ′17O0.528 (ppm) Reference
San Carlos olivine
Georg-August-Universität 30 5.153 ± 0.161 –36 ± 7 Pack et al. (2016)
Okayama University 5 5.287 ± 0.047 –39 ± 7 Pack et al. (2016)
University of New Mexico 12 5.577 ± 0.095 –54 ± 8 Sharp et al. (2016)
University of New Mexico 18 5.268 ± 0.023 –58 ± 5 Wostbrock et al. (2020)
UWG-2 garnet
University of New Mexico 9 5.696 ± 0.115 –71 ± 5 Wostbrock et al. (2020)
Notes: Measurement precision data are 1σ. All data were calibrated directly to VSMOW and SLAP. With the δ18OVSMOW and Δ′17O0.528 values of
UWG-2 garnet assigned as 5.75‰ and –46 ppm respectively, for calibration purposes, Miller et al. (2020) reported that the Δ′17O0.528 value of San
Carlos olivine is –38 ± 9 ppm (1σ) as measured at Georg-August-Universität Göttingen, with comparable measurements at The Open University
giving –38 ± 8 ppm (i.e., in accord with the inter-laboratory comparison reported by Pack et al. 2016).
For maximum accuracy, linear scaling to VSMOW-SLAP requires that ln(1 + δ18O)
values be adjusted, rather than the corresponding δ18O data, because of the non-linearity of
the δ scale. The ‘true’ scaling factor is therefore ln(1 − 0.0555) / ln(1 + δ18OSLAP/VSMOW), with
δ18OSLAP/VSMOW referring to the measured value of SLAP relative to VSMOW. Fortunately, as
noted by Kaiser (2008), the numerical differences between the conventional and logarithmic
normalization procedures are generally small. Kusakabe and Matsuhisa (2008) did normalize
their ln(1 + δ18O) data to VSMOW-SLAP, with the same scale factor of (–57.10) / (–56.20)
being applied to their ln(1 + δ17O) results. The same approach was adopted by Ahn et al.
(2012). Kaiser (2008) recommended a similar normalization, but in power law format and with
λ assigned to be 0.528, as shown to apply to natural waters (Meijer and Li, 1998):
(1   18 OSLAP/VSMOW, measured )0.528  1 (24)

 17 Osample
VSMOW-SLAP normalized   O VSMOW
17
(1   18 OSLAP/VSMOW, assigned )0.528  1

These scaling procedures preserve the oxygen triple-isotope ratio relationship of the δ17O
and δ18O measurements, whilst ensuring that the measured δ18O values are normalized to
the VSMOW-SLAP scale. To improve inter-laboratory consistency of Δ′17O data from
measurements of water samples, Schoenemann et al. (2013) proposed that δ17O measurements
be normalized to the VSMOW-SLAP scale in the same way as recommended by Gonfiantini
(1978) for normalizing δ18O data, with the δ17O value of SLAP defined to give Δ′17O of
exactly zero, relative to a reference line of λRL = 0.528 and γRL = 0. Schoenemann et al. (2013)
reported that the resulting scale factor for the δ17O measurements is therefore approximately
(–29.6986) / δ17OSLAP/VSMOW. This recommendation involves scaling of the empirical δ17O
results independently of the corresponding δ18O data. Essentially, the experimental data are
adjusted so that the measurements of SLAP fit exactly on a reference line of slope 0.528
and which passes through VSMOW on a ln(1 + δ17O) versus ln(1 + δ18O) plot. The two-
point calibration also addresses other instrument-related effects, as discussed by Yeung et al.
(2018) and by Pack (2021, this volume). Recent measurements of SLAP indicate that it is
characterized by a Δ′17O value slightly less than zero. Schoenemann et al. (2013) obtained
a value of –6 ± 8 ppm (1σ). Wostbrock et al. (2020) reported that averaging their own
measurements (–15 ± 5 ppm) with the –9 ± 7 ppm result reported by Sharp et al. (2016) and
–8 ± 9 ppm obtained at Okayama University (Pack et al. 2016) gives a value of –11 ± 4 ppm.
16
O, 17O AND 18O ABUNDANCES AND ISOTOPE RATIO RANGES IN
NATURALLY OCCURRING TERRESTRIAL MATERIALS
Oxygen is characterized by atomic number 8 and is the third most abundant element in the
solar system, after hydrogen and helium. It is the most abundant element on Earth, with silicate
and oxide minerals of the crust and mantle comprising by far the largest terrestrial reservoir
(~99.5% by mass). Table 2 lists the masses of three stable isotopes of oxygen (Holden et al.
2018), together with their natural abundance ranges (Meija et al. 2016) and abundance ratios
in VSMOW (IAEA reference sheet on VSMOW and SLAP 2006).
Table 2. Mass, natural abundance range, and abundance ratio in the VSMOW reference material, of the three
stable isotopes of oxygen.
Natural abundance Abundance ratio (ppm) in VSMOW,
Isotope Mass (Da)
range (atom%) relative to 16O
16
O 15.994914619 99.738–99.776 106
17
O 16.999131757 0.0367–0.400 379.9 ± 0.8
18
O 17.999159613 0.187–0.222 2005.2 ± 0.45
Although 17O is a factor of ~5.3 less abundant than 18O in seawater, it is nevertheless
approximately twice as abundant as deuterium. Reported δ18O values of terrestrial silicate rocks
and minerals vary from –27.3‰ (Bindeman et al. 2010, 2014), in the most extreme example
of interaction with ‘Snowball Earth’-derived synglacial meteoric waters at depth, to as high
as 60.2‰ (and with Δ17O ranging from 14–21‰) in nitrate deposits from the Atacama desert
(Michalski et al. 2004). However, the largest known 18O enrichments occur in stratospheric ozone
(δ18O up to ~110‰), with only slightly lower values in tropospheric ozone and nitrate aerosols.
Isotopic data for these three species scatter about a slope one line on a δ17O versus δ18O array
(Thiemens 2013). At the other end of the scale, a δ18O value as low as –81.9‰ has been measured
in precipitation collected at Dome Fuji, East Antarctica (Fujita and Abe 2006).
Oxygen triple-isotope measurements of atmospheric O2
M. H. Thiemens (comment reported in Bender et al. 1994) first suggested that there
should be a deficiency of 17O in atmospheric O2, relative to mass-dependent composition. This
was based on a consideration of stratospheric photochemical reactions and mass exchange
between the stratosphere and troposphere (Thiemens et al. 1991). The magnitude of the
deviation from mass dependent composition was subsequently documented from experiments
by Luz et al. (1999). The first high precision δ18O and δ17O measurements of atmospheric
O2 relative to VSMOW directly, however, were by Barkan and Luz (2005), who obtained
δ18O = 23.88 ± 0.02‰, δ17O = 12.08 ± 0.01‰. Measurements on SLAP were reported in the
same paper as –55.11 ± 0.01‰ and –29.48 ± 0.03‰, respectively, giving a Δ′17O value (relative to
λRL = 0.528) of 0.007‰. The air O2 δ18O value was in agreement with 23.79 ± 0.06‰ as obtained
by Horibe et al. (1973) and also with the 23.8 ± 0.14‰ value (Coplen et al. 2002) obtained by
recalibrating the measurements of Kroopnick and Craig (1972). The measurements by Horibe
et al. (1973) and Kroopnick and Craig (1972) were made on CO2 equilibrated with water of air
O2 isotopic composition. In none of those three studies was instrument-related contraction of the
δ scale corrected for, however, which would increase the respective magnitudes of the δ17O and
δ18O data. Relative to a reference line of λRL = 0.528 and passing through VSMOW, Δ′17O for
air O2 as obtained from the Barkan and Luz (2005) data is –0.453‰. Normalizing their data to
the VSMOW-SLAP scale as recommended by Schoenemann et al. (2013) shifts the Δ′17O value
by only 2 ppm, to –0.451‰, despite the corresponding δ18O and δ17O values being increased to
24.05‰ and 12.17‰ respectively. This illustrates the comparative robustness of Δ′17O (resulting
from the correlation of δ17O and δ18O measurement errors) relative to the δ17O and δ18O data.
16 Miller & Pack
Kaiser (2008) considered how ‘true’ linear scaling (i.e., of δ′17O and δ′18O values rather
than δ17O and δ18O) affects the isotope data from Barkan and Luz (2005). If a scaling factor of
ln(1 – 0.0555)) / ln(1 – 0.05511) = 1.00728 is applied to the δ′17O and δ′18O values, then Δ′17O
becomes –0.456‰. Again, the resulting change is very small. Performing the same scaling but
with δ18OSLAP/VSMOW = –56.18‰ as recommended by Kaiser (2008) gives Δ′17O = –0.462‰.
This latter value is in close agreement with subsequent measurements by Kaiser and Abe (2012),
which gave Δ′17O = –0.460‰. The corresponding δ17O and δ18O data were 12.25 ± 0.03‰ and
24.22 ± 0.04‰, corrected for a 0.8% scale contraction (which the authors noted may be typical
for the type of instrument used).
Assigning the Δ′17O value of San Carlos olivine on the VSMOW-SLAP scale as –0.051‰
(based on averaging the data from Pack et al. 2016; Sharp et al. 2016; and Wostbrock et
al. 2020), the Δ′17O value of air O2 relative to VSMOW as reported by Pack et al. (2017)
becomes –0.422‰ (it is –0.409‰ if based on Pack 2016, alone). Applying the same San
Carlos calibration to measurements reported by Young et al. (2014) gives the Δ′17O value of
air O2 as –0.425‰. If the Yeung et al. (2018) Δ′17O value of UWG-2 garnet relative to air O2 is
revised to –0.061‰ (i.e., 10 ppm lower than that of San Carlos olivine), the derived Δ′17O of
air O2 is –0.435‰. Anchoring recent data by Wostbrock et al. (2020) to both San Carlos olivine
at Δ′17O = –0.051‰ and UWG-2 garnet at –0.061‰ results in the Δ′17O value of air O2 as
–0.433‰. Data from different laboratories are thus converging, but more measurements of air
O2, silicates and waters (using similar extraction protocols and the same mass spectrometers)
are needed for consensus values to emerge. At present, is evident that anchoring the Δ′17O
value of air O2 to measurements of silicates results in more positive values than as obtained by
direct measurements of VSMOW relative to air O2. The reason for this is currently unknown.
Oxygen triple-isotope measurements of terrestrial silicates
The first investigation of oxygen triple-isotope mass fractionation relationships
in terrestrial rocks since that reported by Matsuhisa et al. (1978) was by Robert et al.
(1992). Modern cherts were compared with well-preserved examples of age up to 3.5 Ga
(Precambrian), together with mantle-derived rocks (mid-ocean ridge basalts, ocean island
basalts and continental flood basalts). No statistically meaningful distinction was apparent,
on a δ17O versus δ18O plot, between the fractionation line derived from the modern cherts
and that derived from the Precambrian samples. The authors interpreted this as showing that
oxygen isotopic homogeneity between the Precambrian ocean and the Earth’s mantle (through
which the modern ocean has been extensively recycled) was already attained by 3.7 ± 0.1 Ga
ago. A corollary was that the Precambrian sedimentary rocks presented no evidence for the
delivery of extraterrestrial (cometary) water.4 Within measurement error, the mantle-derived
rocks all fitted on the mass-dependent fractionation line defined by the cherts. Meijer and Li
(1998) noted that if the chert data reported by Robert et al. (1992) are re-formulated to obtain
the corresponding λ values, the result is the same (although at significantly lower precision) as
that determined by Meijer and Li (1998) for meteoric waters.
A significant technical advance in oxygen triple-isotope measurements of rocks and
minerals was the demonstration by Sharp (1990) that fluorination of silicates and oxides
by BrF5 vapor could be achieved rapidly and quantitatively using a CO2 laser beam (10.6
μm wavelength) as the heat source, directed onto the sample through a BaF2 window in a
purpose-designed fluorination cell. Far smaller quantities of material were needed than for the
conventional (Clayton and Mayeda 1963) procedure and the attendant blanks were very much
reduced. Furthermore, even the most refractory minerals could be fluorinated, by heating to
incandescence and without heating the surrounding chamber. In that initial report, the resulting
oxygen gas was converted into CO2 by reaction with a hot graphite rod (catalyzed by Pt wire),
4
Rumble (2018) reported that the isotopic homogeneity has since been extended back to 4.3 Ga before pres-
ent, i.e. earlier than the beginning of the Archean Eon.
for δ18O analysis only. Rumble and Hoering (1994) showed that a similar procedure but using
F2 as the fluorinating agent (derived from heating Asprey’s salt, K2NiF6 ∙KF) was also possible
and with the extracted O2 used as the analyte gas for mass spectrometric measurements.
Oxygen triple-isotope data from quartz and spinel samples were reported by those authors.
Laser-assisted fluorination of silicates using focused spot heating by an excimer laser (pulsed,
ultraviolet radiation) was also developed during the 1990s (Wiechert and Hoefs 1995; Rumble
et al. 1997; Young et al. 1998). This technique, however, was limited to spot sizes in the
range ~80–300 µm, was very sensitive with respect to reactive materials in the samples (e.g.,
carbonates, resin, clay minerals), and has thus been largely discontinued.
The development of laser-based silicate fluorination procedures, together with use of O2 as
the analyte for oxygen triple-isotope measurements, coincided with new designs of gas source
mass spectrometers which were better able to resolve the 17O16O ion beam from that of the
major isotopologue, 16O16O, thus facilitating 17O relative abundance measurements at higher
precision. At several laboratories where a laser-based silicate fluorination facility was installed,
an investigation was made of the mass-dependent fractionation characteristics of a collection of
terrestrial rocks and minerals spanning a wide range of δ18O values (Miller et al. 19995; Miller
2002; Rumble et al. 2007; Pack et al. 2007, 2013; Spicuzza et al. 2007; Kusakabe and Matsuhisa
2008; Ahn et al. 2012; Hofmann et al. 2012; Tanaka and Nakamura 2013; Levin et al. 2014; Kim
et al. 2019). In many of these studies, an objective was to define a ‘Terrestrial Fractionation Line’
for use as a Δ′17O reference. Despite the diversity of rock and mineral types in most cases, the
slope (λ) obtained from 103ln(1 + δ17O) versus 103ln(1 + δ18O) data regression in all of the above-
mentioned examples were within the range of 0.5240‒0.5270. Furthermore, in several reports the
associated 95% confidence interval was better than 0.001. Such λ variations are within the range
of theoretical θ values calculated by Matsuhisa et al. (1978) for various oxygen isotope exchange
reactions under equilibrium conditions, thus indicating that kinetic fractionation does not play a
significant role in rock-forming processes.
Rumble et al. (2007), in an inter-laboratory comparison, showed that a mass fractionation
line formed from eclogite facies garnets (high-pressure, medium- to high-temperature
metamorphism) was statistically distinguishable from a comparable array formed from
hydrothermal quartz samples. The respective (best precision) slope values were 0.5262 ± 0.0008
for the garnets and 0.5240 ± 0.0008 (95% confidence interval) for the quartz. This was the first
empirical demonstration that the concept of a single ‘Terrestrial Fractionation Line’ for rocks
and minerals is inaccurate, and that distinct mass fractionation arrays exist. It was later noted
(Hallis et al. 2010) that the two fractionation arrays discussed by Rumble et al. (2007) were
offset from one another. At VSMOW, the origin of the δ scale, the 103ln(1 + δ17O) difference
was 0.046‰. The precision (95% confidence interval) of the intercept value of the eclogite
garnets line was 0.005; that of the hydrothermal quartz line was 0.019. This demonstrated
that mass fractionation arrays from silicate rocks and minerals are offset from VSMOW, by
an amount dependent on the specific fractionation array. This finding, though important, was
overlooked until Tanaka and Nakamura (2013) similarly identified an offset between VSMOW
and a fractionation array formed from a collection of silicate and oxide mineral samples.
Measurements of VSMOW, SLAP and GISP (Greenland Ice Sheet Precipitation) were made,
using the same oxygen purification system and mass spectrometer for the reference waters as
for the mineral and oxide samples. The laboratory ‘working standard’ O2 was calibrated against
O2 obtained from the fluorination of VSMOW and the magnitude of the ordinate offset (γ′) of
5
Miller et al. (1999) reported the slope in δ17O versus δ18O format. Taking the mean of duplicate measurements
(46 samples), the equivalent λ value is 0.5247 ± 0.0010 (95% confidence interval). The same data set was
subsequently extended and discussed by Miller (2002), with all raw data being published in Miller et al.
(2015). A statistically identical slope value of 0.5251 ± 0.0014 (SEM multiplied by Student’s t factor for a
95% confidence limit) was reported by Hofmann et al. (2012) from >700 measurements. This was unchanged
(0.5251 ± 0.0014, 2σ) by increasing the number of samples to 1071 (Pack et al. 2013).
18 Miller & Pack
the silicate array from VSMOW was reported to be –0.070 ± 0.005‰ at the 95% confidence
interval, for the specific collection of samples selected (λ value of 0.5270 ± 0.0005). The
reported γ′ value was later found to be significantly in error (Pack et al. 2016). Tanaka and
Nakamura (2017) subsequently gave a revised figure of –33 ppm.
Further confirmation that distinct mass fractionation lines are associated with specific rock-
and mineral-forming processes was provided by Pack and Herwartz (2014) and, independently,
by Levin et al. (2014). Pack and Herwartz (2014) demonstrated that variations in Δ′17O relate
to the temperature dependence of α and θ for specific fractionations and/or mixing between
reservoirs with different isotopic compositions. They suggested that even a fractionation line
defined by data points from coexisting and well-equilibrated minerals in a rock is conceptually
incorrect, as θ values (by analogy to α values) are specific to two phases that are in equilibrium.
Furthermore, low-temperature (4–50 °C) equilibration between silica and water was shown to
be associated with low (and temperature-dependent) values of θ (0.5211 at ~8 °C), whereas
high-temperature mineral–mineral equilibria in metamorphic and igneous felsic and mafic
rocks are characterized by θ values of the order of 0.528–0.529. In their investigation of oxygen
triple-isotope variations in sedimentary rocks, including Archean and Phanerozoic cherts, Levin
et al. (2014) noted that, although their complete data set produced a fractionation array of
λ value 0.523 ± 0.001, different groups gave distinctive slopes (non-overlapping at the 95%
confidence level). In accord with Pack and Herwartz (2014), Levin et al. (2014) noted that the
slope variations were the result of processes including mass fractionation associated with low-
temperature precipitation during the growth of authigenic minerals; variation in the triple-isotope
composition of the waters from which the sedimentary minerals precipitated; and equilibrium
exchange after initial authigenic formation. The observations collectively provided further
confirmation that no single fractionation line exists for terrestrial materials. Pack and Herwartz
(2014) and Levin et al. (2014) also noted that oxygen triple-isotope studies are particularly
appropriate for investigating (mass-dependent) low-temperature processes. This is because
small variations in θ cause large variations in Δ′17O when the isotope fractionation factors
(18/16α and 17/16α) between water and the mineral of interest are large, as at low temperature.
Comparisons of empirical and predicted triple-isotope arrays can therefore, in principle, be
used to test hypotheses of chert formation and thus inform long-standing debates on the trends
in chert δ18O values, from the Archean through the Phanerozoic.
Oxygen triple-isotope distributions in meteoric waters, snow and ice cores
In much of the literature, the relative abundance of the isotopologues H216O, H217O
and H218O in meteoric precipitation and in the cryosphere has been quantified using the
dimensionless parameter 17O-excess (Angert et al. 2004), which was later defined as
ln(1 + δ17O) − 0.528 ln(1 + δ18O) by Barkan and Luz (2007). The 17O-excess parameter is
therefore identical to the definition of Δ′17O used throughout this volume. The rationale for the
approach introduced by Barkan and Luz (2007) is that, as first shown by Meijer and Li (1998)
and subsequently confirmed by higher precision measurements (Barkan and Luz 2007; Landais
et al. 2008; Luz and Barkan 2010), oxygen triple-isotope distributions in precipitation seemed
to conform closely, on a global scale, to the relationship ln(1 + δ17O) = 0.528 ln(1 + δ18O), with
only small deviations from this trend.
Angert et al. (2004) predicted ppm-scale positive deviations from the array, resulting from
kinetic fractionation during the diffusive transport of water vapor from the (marine) source region
into undersaturated air. This causes the ln(1 + δ17O) versus ln(1 + δ18O) array for natural waters
to be slightly offset from VSMOW. In the same paper it was also suggested that normalized
relative humidity at the vapor source largely controls—and inversely correlates with—the
magnitude of 17O-excess, if turbulence in the marine boundary layer is taken into consideration.
These theoretical predictions could not be tested, however, until a procedure for making oxygen
triple-isotope measurements of waters at the required precision (5 ppm) had been devised.
That was achieved by Barkan and Luz (2005), who developed further a fluorination technique
described by Baker et al. (2002). Small (~2 μL) samples of water samples were injected into a
helium carrier stream which passed through a heated nickel tube containing CoF3 at 370 °C. The
resulting O2 was isolated from the HF also formed, prior to δ17O and δ18O measurements. The
same fluorination method was subsequently established at a small number of other laboratories.
Barkan and Luz (2007) suggested that the difference between the ln(17/16α) / ln(18/16α)
value for water vapor diffusion in air (0.5185 ± 0.0002) and the corresponding ratio (0.529)
for vapor–liquid equilibrium is responsible for the presence of 17O-excess in meteoric waters.
This explanation is in accord with that originally proposed by Angert et al. (2004), although
the relevant fractionation factors were not known accurately in 2004.
Many studies, as reported elsewhere in this volume, have since used 17O-excess
measurements of precipitation as a temperature-insensitive proxy for the humidity at the vapor
source region, above the ocean. Experimental evidence for a non-zero value of 17O-excess
in water vapor of marine air, and a negative correlation between 17O-excess and relative
humidity, was first presented by Uemura et al. (2010). Uechi and Uemura (2019) demonstrated
a similar finding, from a two-year record of 17O-excess in precipitation at the sub-tropical
maritime island of Okinawa, Japan (in the East Asian monsoon region), in conjunction with
the corresponding normalized relative humidity back-calculated from a simple model of
evaporation. Their results were in accord with 17O-excess in the precipitation being determined
largely by diffusional fractionation during evaporation from the surrounding ocean.
From recently published data sets (Li et al. 2015; Tian et al. 2018, 2019), it is evident that
precipitation in the temperate and tropical regions actually conforms to a slightly lower and
more variable proportionality constant than the 0.528 value used in the 17O-excess definition.
Furthermore, it is the inclusion of samples from the polar regions in the earlier data sets of
oxygen triple-isotope ratios in meteoric precipitation that essentially constrained the slope to
be 0.528. Removal of polar data from the sample collection used by Meijer and Li (1998), and
that reported by Luz and Barkan (2010), lowers the slope value significantly (Miller 2018).
Nevertheless, a definition based on a λ value of exactly 0.528 and with the corresponding
Δ′17O of the VSMOW and SLAP reference waters defined to be zero (Schoenemann et al.
2013) provides the basis of a VSMOW-SLAP scale for reporting Δ′17O values to, as well as the
corresponding δ18O data. This is now widely accepted practice for standardizing the reporting
of oxygen triple-isotope data from waters.
Uechi and Uemura (2019) showed that lowering the proportionality constant slightly from
0.528 in the 17O-excess definition does not generally affect the robustness of using 17O-excess
data for normalized relative humidity reconstruction for tropical and subtropical regions (i.e.,
between 35°S and 35°N), where precipitation δ18O values are generally >–10‰. This is because
the resulting change in 17O-excess is generally <10 ppm, i.e. close to the analytical uncertainty
of such measurements. The authors noted, however, that this is not the case if the δ18O values
are significantly lower. For example, the anti-correlation between 17O-excess and δ18O noted by
Li et al. (2015) from their measurements of meteoric waters sampled across the USA (with δ18O
values ranging from –5 to –25‰) is reduced if a slope value of 0.527 is adopted instead of 0.528.
Although initially applied to ice core data from Vostok, on the Antarctica plateau (Landais
et al. 2008), doubts have since been raised about the interpretation of 17O-excess data from
the polar regions, especially the interior of Antarctica (Shoenemann et al. 2014; Miller 2018).
As noted by Schoenemann and Steig (2016), although the use of 17O-excess measurements
has become quite common, the factors that control its spatial and temporal variability are not
fully understood. Antarctic snowfall is largely controlled by meteorological conditions over
the Southern Ocean and the penetration of marine air into the continental interior. There is
20 Miller & Pack
frequent clear‐sky precipitation and occasional, more massive falls from intrusions of maritime
air. The extreme precipitation events play a dominant role in controlling Antarctic snowfall
variability (Turner et al. 2019). Whereas the proportionality constant of 0.528 for the slope of the
ln(1 + δ17O) versus ln(1 + δ18O) relationship characterizes many triple-isotope measurements
of Antarctic snow, firn and ice cores, even for δ18O values as low as –70‰ relative to VSMOW
(Miller 2018), with correspondingly low temperatures of formation, precipitation over much
of the Antarctic plateau, including at Vostok, occurs almost daily from a clear sky (nucleation
and growth of ‘diamond dust’ ice crystals), rather than being cloud-derived. In this case, the
associated proportionality constant seems to be significantly higher than 0.528. From the Vostok
ice core measurements of Landais et al. (2008), covering the period 5–150 ka before present, an
empirical λ value of 0.5310 ± 0.0004 (95% confidence interval) can be derived (Miller 2008,
2018); this is also in accord with modern precipitation at the same locality (Miller 2018).
Measurement procedures. More recently, spectroscopic techniques that require no
chemical conversion of the water have been developed. Specifically, these are based on either
or cavity ring-down spectroscopy with laser-current-tuned cavity resonance (Steig et al. 2014),
or off-axis integrated cavity output spectroscopy (Tian et al. 2016). The attainable precision of
such instrumentation is approaching that of the Barkan and Luz (2005) fluorination method.
Another alternative, recently proposed by Affek and Barkan (2018), involves equilibrated
CO2–H2O isotope exchange followed by oxygen isotope exchange between the CO2 and O2
over hot platinum, prior to triple-isotope measurements of the O2. The authors reported that
the accuracy and precision of this procedure are similar to those of the fluorination method
described by Barkan and Luz (2005).
STANDARDIZING Δ′17O DATA FROM ROCKS AND MINERALS

Pack et al. (2016) advocated that VSMOW-SLAP scaling should be applied to oxygen
triple-isotope measurements of rocks and minerals, using the recommendation of Schoenemann
et al. (2013) as devised for standardizing Δ′17O data from waters. The principal advantage of
this approach is that measurements of waters and rocks are then reported on the same scale.
However, it does require that any laboratory undertaking high precision Δ′17O measurements
of rocks and minerals must also have the capability of making similar measurements on
waters. Few laboratories currently have such a capability. There is also the implicit assumption
that the water and silicate fluorination procedures—although different—will be characterized
by the same (low) processing errors. In only a few studies reporting oxygen triple-isotope
measurements of rocks or minerals were the data calibrated by direct measurements on
VSMOW and SLAP (Kusakabe and Matsuhisa 2008; Ahn et al. 2012; Tanaka and Nakamura
2013; Pack et al. 2016; Sharp et al. 2016; Wostbrock et al. 2020).
The Δ′17O values of silicates can currently be determined to a precision of 10 ppm or less
(standard deviation, 1σ), as reported in several recent studies (e.g., Pack et al. 2016; Sharp
et al. 2016; Kim et al. 2019; Miller et al. 2020; Wostbrock et al. 2020). This is significantly
better than for the corresponding δ17O and δ18O measurements, as a result of the high degree
of correlation between the respective errors in δ17O and δ18O. Variations of the latter shift the
associated data point along a mass-dependent fractionation line on the ln(1 + δ17O) versus
ln(1 + δ18O) plot, causing little or no change to the corresponding Δ′17O value. The accuracy of
Δ′17O data, however, is dependent on the accuracy of calibration of the δ17O and δ18O values of
the ‘working standard’ O2 gas to VSMOW, together with any corrections needed (as identified
and quantified from measurements of SLAP) to compensate for compression of the δ17O and
δ18O scales and to ensure that Δ′17O0.528 of SLAP is zero, i.e., relative to a reference line of
slope 0.528 and having no offset from VSMOW on the ln(1 + δ17O) versus ln(1 + δ18O) plot.
Even for the highest precision measurements, there is as yet no consensus on the Δ′17O
values of widely used silicate standards UWG-2 garnet, San Carlos olivine and NBS quartz
on the VSMOW-SLAP scale. This illustrates the challenging nature of such measurements.
Pack et al. (2016), in an inter-laboratory comparison and with calibration to VSMOW-SLAP,
reported a Δ′17O0.528 value of –36 ± 7 ppm (1σ) for San Carlos olivine as measured at one
institution, with –39 ± 7 ppm being recorded by the other laboratory in that study. Clearly,
these values are indistinguishable, within experimental precision. In contrast, Sharp et al.
(2016) reported that their oxygen triple-isotope measurements of San Carlos olivine gave a
Δ′17O0.528 value of –54 ± 8 ppm (1σ), calibrated to measurements of VSMOW and SLAP.
The reason for the discrepancy between that result and the –37 ± 7 ppm (weighted mean
value) reported by Pack et al. (2016) on the same calibrated scale is not clear. More recently,
Wostbrock et al. (2020) reported oxygen triple-isotope measurements of San Carlos olivine,
UWG-2 garnet and NBS 28 quartz, calibrated to the VSMOW-SLAP scale. The respective
Δ′17O0.528 values were given as –58 ± 5 ppm, –71 ± 5 ppm and –59 ± 4 ppm (1σ). The San
Carlos olivine and UWG-2 data are clearly different from (more negative) those reported by
Pack et al. (2016) and Miller et al. (2020), with the respective discrepancies being 20 ppm
and 25 ppm respectively. The Wostbrock et al. (2020) result for NBS 28 is also more negative,
by a similar amount, than as obtained at the two institutions in the study reported by Miller
et al. (2020). Possible explanations for such discrepancies are discussed by Pack (2021, this
volume) and by Wostbrock and Sharp (2021, this volume).
As an alternative to the need for every laboratory to calibrate directly to the VSMOW-
SLAP scale, Miller et al. (2020) suggested that a two-point empirical reference line defined by
measurements of silicates which differ in δ18O by a similar (or greater) amount than VSMOW
and SLAP could be used for the measurements. Two such samples were described (referred to
as the Khitostrov Rock Standard, KRS, and the Stevns Klint Flint Standard, SKFS), covering
a δ18O range of 59.1‰, based on provisional measurements of –25.20 ± 0.09‰ (1σ) for KRS
and 33.93 ± 0.24‰ for SKFS. Δ′17O data of rock and mineral samples reported relative to
this reference line are independent of whether the δ17O and δ18O data are reported relative to
VSMOW or to the ‘working standard’ O2, of any isotopic composition. This confers significant
advantages for inter-laboratory comparisons. For converting the data into the VSMOW-SLAP
reference frame, it is necessary to calibrate the position of the KRS-SKFS line accurately
line on the ln(1 + δ17O) versus ln(1 + δ18O) plot relative to VSMOW. Miller et al. (2020)
reported that such measurements conducted at two laboratories gave very good agreement.
The accuracy of those calibrations is dependent, however, on the accuracy of the calibration of
San Carlos olivine to the VSMOW-SLAP scale as reported by Pack et al. (2016).
In addition to reporting their measured δ18O and Δ′17O values for San Carlos olivine,
UWG-2 garnet and NBS 28 quartz on the VSMOW-SLAP scale, Wostbrock et al. (2020)
also proposed a higher δ18O quartz standard, 18.070 ± 0.136‰ (1σ), designated NM-Q. The
corresponding Δ′17O value, on the VSMOW-SLAP scale, was given as –81 ± 5 ppm. Although
not spanning as large a δ18O range as KRS and SKFS (Miller et al. 2020), all these materials
are useful for laboratories not able to calibrate directly to the VSMOW-SLAP scale.
SOME EXAMPLES OF THE APPLICATION OF

OXYGEN TRIPLE-ISOTOPE RATIO MEASUREMENTS
We provide here an indication of how oxygen triple-isotope ratio measurements have
contributed to a diverse range of research areas in the geosciences. The list includes only a
selection of examples and is by no means exhaustive. Several of the topics are discussed in
detail elsewhere in this volume.
22 Miller & Pack
Corrections to δ13C measurements of CO2

As noted above, the contribution of 12C17O16O isotopologues to the CO2 m/z 45 ion beam
current needs to be quantified and subtracted from the measured total value, during stable
isotope ratio measurements of CO2. Different data reduction procedures have long been
adopted for this correction, notably based on either that proposed by Santrock et al. (1985) or
by Allison et al. (1995). The latter procedure was subsequently recommended by the IAEA.
Differences between the algorithms lead to small but potentially significant discrepancies in
the magnitude of the resulting correction. Furthermore, neither allows changes to the assigned
values of λ (0.516 and 0.5, respectively) in the mass-dependent relationship between the 17O/16O
and 18O/16O ratios. Nor can non-mass-dependent distributions be accommodated. Röckmann
and Brenninkmeijer (1998) implemented a simple and approximate additional correction for
δ13C measurements of tropospheric CO, which was known to contain a non-mass-dependent
enrichment of 17O (Huff and Thiemens 1998; Röckmann et al. 1998). The authors estimated
that not taking account of the enhanced 17O abundance resulted in a systematic error of the
order of 0.08 to 0.25‰ in the corresponding δ13C values. Assonov and Brenninkmeijer (2003)
suggested that discrepancies between published values for 17RVPDB-CO2 in different correction
algorithms were a major source of the δ13C biases. The authors also suggested that the value of
the mass-dependent fractionation exponent λ should be set to 0.528, on the basis that H2O – CO2
oxygen isotope exchange probably controls the triple-isotope composition of CO2 in nature,
and natural waters fit an array characterized by a λ value of 0.528 (Meijer and Li, 1998). Brand
et al. (2010) also endorsed the adoption of 0.528 for λ and proposed a simplified 17O correction
procedure which the authors suggested was sufficiently accurate for many cases. Non-mass-
dependent oxygen triple-isotope distributions were not accommodated, however. Miller et al.
(2007) modified the Allison et al. (1995) algorithm to allow for user-defined values of λ and
Δ′17O. Kaiser (2008) gave a very detailed appraisal of 17O correction procedures and showed
that the correction could be framed in terms of relative isotope ratio differences (δ values),
although inaccuracy would still result from assigning an inappropriate value of λ.
More recently, measurements of the equilibrium θCO2–H2O value for oxygen isotope
exchange show that the value is 0.5229 ± 0.0015 (SEM multiplied by Student’s t factor for a 95%
confidence limit) at 23 °C (Hofmann et al. 2012); an identical value at 25 °C (0.5229 ± 0.0001)
was obtained by Barkan and Luz (2012). Furthermore, the same θ value applies to CO2 formed
from phosphoric acid digestion (McCrea 1950) of calcite at 25 °C (Wostbrock et al. 2020). It
seems, therefore, that 0.523 is the most appropriate λ value for use in 17O correction algorithms
for δ13C measurements of CO2 and also for carbonates equilibrated with water characterized by
δ18O ≈ Δ′17O ≈ 0‰. Freshwater carbonates or eggshell calcite would deviate from such trends,
however, therefore independent assessment of δ17O needs to be made for such materials.
With the development of gas source mass spectrometers of sufficiently high resolution,
isotope measurements on atomic ions formed from fragmenting the CO2+ molecular ion can
now be used, in principle, to determine the carbon and oxygen triple-isotopic composition of
CO2, without chemical processing or corrections for mass interferences (Adnew et al. 2019).
However, because the signal intensities are very small, long measurement times are required.
At the present state of development, 12 hours of measurement are required to determine the
Δ′17O of CO2 to a precision of the order of 37 ppm, with λRL = 0.528. Nevertheless, this
technique demonstrates a capability that offers new possibilities for the future.
Comparisons of the oxygen triple-isotope compositions of the Earth and Moon
It is widely accepted that the Moon was most likely formed from debris produced by
a Mars-sized body, named Theia, colliding with the proto-Earth. Whether or not the Moon
has an identical oxygen triple-isotope composition to that of Earth’s mantle has important
implications for refining this (and other) models of the Moon’s origin. The first high precision
oxygen triple-isotope measurements to investigate this question were by Wiechert et al. (2001),
using CO2 laser-assisted fluorination and with BrF5 as the fluorinating reagent. Previous reports,
as noted by the authors, gave a Δ17O range of 0.30‰ for lunar samples. Such a range is of
similar magnitude to the Δ17O distinction between the Earth and Mars (0.321 ± 0.013‰, 1σ),
as inferred from measurements of Shergotty–Nakhla–Chassigny (SNC) meteorites (Franchi
et al. 1999); similarly between the Earth and the asteroid Vesta (–0.26 ± 0.08‰), as inferred
from measurements of Howardite–Eucrite–Diogenite (HED) meteorites (Clayton and Mayeda
1996)6. From measurements on a lunar collection of mare basalts, KREEP basalts, highland
rocks, volcanic glasses, breccias, and a single lunar meteorite, Wiechert et al. (2001) found
that all the lunar oxygen isotope compositions plotted within ±0.016‰ (2σ) on a single mass-
dependent fractionation line on the δ17O versus δ18O plot. The array was identical—within
analytical uncertainties—to that produced from 11 terrestrial minerals with δ18O values
ranging from 0 to 12‰ and forming a δ17O versus δ18O slope of 0.5245. Thus, Δ17O was
defined as δ17O − 0.5245 δ18O for that particular study.
Spicuzza et al. (2007) similarly found that their measurements of lunar basalts and soils
(plotted in 103 ln(1 + δ17O) versus 103ln(1 + δ18O) format) showed no statistically significant
deviation from an empirically-derived terrestrial reference line (details as given above). Δ′17O
values of all the lunar samples were within uncertainty of the reference line measurements
and averaged 0.008 ± 0.022‰ (2σ). Consistent with this finding, and with that of Wiechert et
al. (2001), an investigation by Hallis et al. (2010) of the oxygen triple-isotope characteristics,
titanium content and modal mineralogy of five different types of lunar basalts revealed no
Δ′17O deviation (0.006 ± 0.021, 2σ) for the average of the lunar basalts, relative to a reference
line obtained from the set of eclogite garnets described by Rumble et al. (2007).
Herwartz et al. (2014) reported the surprising finding that their measurements of lunar
basalts gave a Δ′17O value distinctive from that of mantle xenoliths and MORB (from seven
different localities). Correcting the published data for recalibration to San Carlos olivine on
the VSMOW-SLAP scale (Pack et al. 2016), but keeping λRL = 0.5305 as assigned by the
authors for Δ′17O reporting, the Earth mantle-derived samples gave a weighted average Δ′17O
value of –49 ± 0.003 ppm (1σ) whereas –37 ± 0.008 ppm was the comparable result obtained
from replicate measurements of the (three) lunar basalts. This remarkable finding was disputed
by Young et al. (2016), who performed a similar comparison, using seven lunar samples and
one lunar meteorite, together with a suite of terrestrial igneous samples. The Δ′17O difference
between the lunar and terrestrial data sets was reported to be −1 ± 5 ppm (2 × SEM), i.e.
indistinguishable. For Δ′17O reporting, Young et al. (2016) used a reference fractionation line
with λRL = 0.528 and passing through the San Carlos olivine datum point instead of VSMOW.
A possible reason for the discrepancy between Herwartz et al. (2014) and Young et al.
(2016) about whether or not there is an oxygen isotopic distinction between lunar samples
and ‘bulk silicate Earth’, as represented by terrestrial, mafic rocks, has been revealed by more
recent studies, which involved more extensive suites of lunar samples. Greenwood et al.
(2018) included 17 lunar whole-rocks, representing all main lithological units, together with
14 lunar mineral separates. Terrestrial mafic rocks were represented by 20 basalts and a mantle
xenolith, together with previously-published data on high-3He/4He olivines, measured in the
same laboratory. From the oxygen triple-isotope data, it was seen that the set of lunar samples
as a whole was characterized by a 3 to 4 ppm, statistically resolvable, Δ′17O difference relative
to the terrestrial basalts. However, no such distinction was found between the terrestrial
olivines and lunar rocks.
6
The corresponding Δ′17O0.528 value for Mars is 0.286 ± 0.016‰ (1σ). From a recent compilation of oxygen
isotope data from early-formed differentiated meteorites (Greenwood et al. 2017), the Δ′17O0.528 value for
Vesta, as determined using results from (26) eucrite or diogenite falls only (thereby minimizing the effects of
terrestrial weathering), is –0.252 ± 0.007‰ (1σ).
24 Miller & Pack
The latest investigation of the Earth–Moon Δ′17O comparison is by Cano et al. (2020).
Their oxygen triple-isotope measurements of a range of lunar lithologies (and reported on
the VSMOW-SLAP scale) showed that the Earth and Moon have slightly different Δ′17O, if
measurements of all (23) lunar samples were averaged to determine a ‘bulk silicate Moon’
value. This procedure gave Δ′17O = –0.056 ± 0.010‰ (1σ), whereas the ‘bulk silicate Earth’
value was –0.060 ± 0.004‰ (22 samples). The magnitude of the difference is in accord with
the findings of Greenwood et al. (2018). What is particularly noteworthy, however, is that
lunar samples analyzed by Cano et al. (2020) showed nearly three times the Δ′17O variability
(0.0103‰, 1σ) when compared with Earth (0.0037‰). Furthermore, the mean Δ′17O value
for low-Ti lunar basalts showed a more positive Δ′17O value than that of the ‘bulk silicate
Earth’ samples—as did Herwartz et al. (2014)—with the magnitude of the discrepancy being
comparable to the difference reported by Herwartz et al. (2014). Cano et al. (2020) reported
that their oxygen isotope data variations for lunar samples correlated with lithology; it was
proposed that the differences were evidence for mixing between isotopically light vapor,
generated by the impact of Theia, and the outermost portion of the early lunar magma ocean.
Thus, whereas details of the ‘Giant Impact’ hypothesis of the Moon’s formation are still
being refined, high precision oxygen triple-isotope measurements during the past two decades
have contributed significantly to these developments and stimulated the emergence of new ideas.
Investigating the climate of ‘Snowball Earth’ from hydrothermal rocks
A novel application of Δ′17O systematics was described by Herwartz et al. (2015).
Noting that the oxygen isotopic compositions of hydrothermally-altered rocks originate partly
from interactions with the associated aqueous fluid, oxygen triple-isotope measurements
of hydrothermally-altered Proterozoic rocks were used to estimate the δ18O value of the
associated water. The calculations were based on the assumption that the rocks would define
a mixing trend with the meteoric waters on a Δ′17O versus δ′18O plot and that the waters
actually describe a parabolic array in δ17O versus δ18O space, thus giving resolvable mixing
trajectories. The isotopic composition of the water may then be estimated from the intersect
between the mixing trend and the meteoric waters array on this plot. For waters derived from
Paleoproterozoic (age ~2.3 – 2.4 Ga) ‘snowball Earth’ glaciers at low paleo-latitudes (<35°N),
a δ18O value of –43 ± 3‰ was derived. Precipitation characterized by such depletion of 18O
currently occurs only in central Antarctica. A similar analysis of a Neoproterozoic example
(age ~0.6 – 0.7 Ga) indicated a meltwater δ18O value of –21 ± 3‰, implying less extreme
climate conditions at similar paleo-latitude. As noted by the authors, such estimates essentially
represent ‘single snapshots’ of ancient water samples and may not be representative of peak
‘snowball Earth’ conditions. A further point is that the derived δ18O value is dependent on the
calibration of the triple-isotope composition of the ‘working standard’ O2 relative to VSMOW.
If the assigned δ17O value was 0.050‰ too low, as inferred from Pack et al. (2016), then the
calculated δ18O values of the meteoric waters will need to be adjusted accordingly. For the
Paleoproterozoic example mentioned above, the revised value is –38 ± 3‰.
The same methodology has been adopted in more recent studies. For example, Zakharov
et al. (2017, 2019) also showed that the δ18O of glacial meltwaters during two separate and
individually dated episodes of snowball Earth glaciation was approximately –40‰ relative to
VSMOW. The authors noted that the presence of Paleoproterozoic glacial diamictites deposited
at low latitudes on different continents indicated that three or four worldwide glaciations
occurred between 2.45 and 2.22 Ga before present. Subglacial hydrothermal alteration was
induced by intrusions of high-Mg and high-Fe gabbros during the early Paleoproterozoic rifting
on the Baltic Shield, which at the time was located at low latitudes. The low δ18O values of
hydrothermally altered rocks associated with these intrusions were attributed to high-temperature
isotopic exchange with glacial meltwater, indicating the presence of glacial ice globally.
Zakharov and Bindeman (2019) reported that oxygen triple-isotope analyses of

hydrothermally altered rocks (well-preserved pillow structures, hyaloclastites and komatiitic
basalts) of age 2.43–2.41 Ga from the Vetreny belt, Baltic Shield, Russia, provide a record of
high-temperature water-rock interaction induced by contemporaneous seawater. Comparison
with modern oceanic crust examples implied that the Vetreny belt examples formed in
equilibrium with seawater-derived fluids of Δ′17O value very close to zero, i.e. that early
Paleoproterozoic seawater was characterized by a Δ′17O value indistinguishable from that of
modern-day seawater. Further information on the use of oxygen triple-isotope measurements
in hydrothermal systems is given by Herwartz (2021, this volume).
Insights from Δ′17O measurements on basalts, shales and fluvial sediments
Several novel applications of high precision measurements of oxygen triple-isotope ratios
to the Earth sciences have recently been presented in the literature. Here, three examples are
described, briefly.
Quantifying the crustal component in ocean island basalts. A consequence of plate
tectonics is that subducted oceanic crust contributes to the formation of ocean island basalt
(OIB), although its actual fraction in the mantle source is not well established. Cao et al. (2019)
proposed that, on the basis of their theoretical calculations, Δ′17O values of olivine should be
unaffected by crystallization and partial melting. This feature permits the use of oxygen triple-
isotope ratio measurements of olivine to identify subducted oceanic crustal material in the
mantle source region from which OIB is believed to originate. A quantitative assessment of the
respective fractions of subducted ocean sediments and hydrothermally altered oceanic crust
in OIB mantle source was made by the authors, using Δ′17O data from the literature. It was
estimated that the fraction of subducted oceanic crust was as high as 22.3% in some examples,
although the affected region in the respective mantle plume is likely to be limited.
Constraints on the emergence of continents—evidence from the shale record.
Bindeman et al. (2018) used oxygen triple-isotope measurements of shales, from every
continent and spanning an age of 3.7 billion years, to provide constraints on the emergence of
continents over time. A stepwise total decrease of 80 ppm in the average Δ′17O value across the
Archaean–Proterozoic boundary was reported, using a reference line slope of 0.5305 to define
Δ′17O. However, because of δ18O variations in the dataset, the change in Δ′17O is reduced when
reporting relative to a reference line of slope 0.528. An increase in δ18O value of 10‰ results
in a Δ′17O change of –25 ppm when switching from a reference line slope of 0.528 to 0.5305.
Thus, changes in Δ′17O should always be discussed in the context of changes of δ18O. In this
case, however, defining Δ′17O using a reference line of slope 0.528 clearly does not obliterate
the trend observed by Bindeman et al. (2018). It was suggested that this was most probably
caused by a shift in the nature of water–rock interactions, from near-coastal in the Archaean to
predominantly continental (and diverse in δ18O) from the Proterozoic era, and accompanied by
a decrease in average surface temperatures. Furthermore, this shift may have coincided with
the onset of a modern hydrological cycle, owing to the rapid emergence of continental crust
with a diversity of hydrologic conditions (continents with near-modern average elevation and
with mountain ranges, spreading across a range of latitudes) and aerial extent, approximately
2.5 billion years ago. For further details, see Bindeman (2021, this volume).
Fluvial sediments—insights into weathering processes. Bindeman et al. (2019)
conducted an investigation of oxygen triple-isotope ratios in fluvial sediments, using samples
from 45 rivers worldwide, cumulatively representing ~25% of the area—from the tropics to the
polar regions—associated with continental transport into the oceans. Δ′17O determinations of
the clay-sized fraction indicated that the values were essentially controlled almost exclusively
by the associated meteoric waters at the respective temperature of weathering, together with
26 Miller & Pack
minor effects related to evaporation. The majority of the clays were consistently characterized
by high-δ18O signatures, regardless of the bedrock type, as weathering involves very high
water/rock ratios. The clays appeared to be in oxygen isotopic equilibrium with local meteoric
waters. Lack of significant isotopic variation between clays from different climatic regions
was attributed to the opposing effects of temperature on clay–water fractionation, on the one
hand, and the global relationship between temperature and δ18O of the local meteoric waters
(e.g., Dansgaard 1964; Rozanski et al. 1993). Bindeman et al. (2019) concluded that the Δ′17O
values of shales in the geological record provide a measure of the evolving global hydrologic
cycle during continental emergence; also of a decrease in global mean annual temperatures
or diagenetic conditions, together with decreasing ocean mass caused by rehydration of the
mantle from the subduction of hydrated, low-δ18O, high-Δ′17O slabs.
Quantifying gross photosynthetic oxygen production in the oceans
Oxygen triple-isotope measurements of dissolved O2 in marine environments have been
widely adopted during the past two decades for estimating gross photosynthetic O2 production
(i.e., the rate prior to any respiratory O2 consumption) in oceanic surface waters, on temporal
and spatial scales. Such information contributes to an understanding of both the present‐day
global carbon cycle and also to predictions of how carbon cycling might respond to future
climate forcing. The principle was originally suggested by Luz and Barkan (2000) and uses
the distinction in triple-isotope composition between dissolved oxygen of atmospheric origin
(non-mass-dependent, Δ′17O0.528 = –0.469 ± 0.007‰) and that produced during photosynthesis
(mass-dependent and essentially the same as that of the water). The Δ′17O value of dissolved O2
depends on: (1) the rate of air–water gas exchange which, in the absence of biological activity,
tends towards an equilibrium value with air; (2) the rate of in situ production of photosynthetic
O2, which tends to increase Δ′17O of dissolved O2 to a maximum value (close to zero) and
equivalent to that of the substrate water. Using measurements of surface seawater oxygen triple-
isotope ratios, together with estimates of the air–sea gas transfer of O2 via a windspeed-based
parameterization, it is possible to estimate the rate of photosynthetic O2 contribution to the
surface mixed layer budget. As respiration has no impact on the measured isotopic anomaly in
the mixed layer, the measured rate equates to the gross photosynthetic production (GPP).
Various revisions to the calculation procedure have been proposed (e.g., Kaiser 2011;
Kaiser and Abe 2012; Prokopenko et al. 2011), to improve the accuracy of the GPP estimates.
Manning et al. (2017) adapted the method to incorporate the local isotopic composition of the
water. Advances that this use of oxygen triple-isotope measurements has contributed to improved
understanding of biological carbon cycling have been reviewed by Juranek and Quay (2013).
Oxygen triple-isotope ratios in bio-apatite: a proxy for CO2 partial pressure in paleo
atmospheres
Gehler et al. (2011) described the use of oxygen triple-isotope analysis to identify
diagenetic changes of δ18O in the skeletal apatite of small mammals. The method is based
on the Δ′17O of atmospheric oxygen being partially transferred from inhaled air O2, via body
water, to the skeletal and tooth apatite of terrestrial mammals—especially small mammals
(body mass < ~1 kg). In larger mammals, low specific metabolic rates result in a lower fraction
of oxygen inhaled via breathing, relative to oxygen from other sources in the body water. The
authors reported that remnant negative Δ′17O values were detected in the apatite of rodent tooth
enamel of Eocene to Miocene age. The distinctive isotopic pattern was not present in dentine
of the same teeth, however; this was presented as evidence for diagenetic alteration.
Because the Δ′17O value of air O2 is coupled with the partial pressure of atmospheric CO2
and gross primary productivity (Luz et al. 1999; Bao et al. 2008), oxygen triple-isotope analyses
of bioapatite from fossil mammals can also be used as a proxy for paleo-CO2 partial pressure
(Pack et al. 2013). Tooth enamel was used, as it is less prone to diagenetic alteration than is
dentine or bone material. Examples from Cenozoic (age <65 Ma) fossil small mammals agree
(within uncertainty) with other proxies and geochemical modelling for pCO , if gross primary
2
production was similar to present-day values. The uncertainty intrinsic to the proxy is mainly
due to uncertainties in physiological parameters: the total water flux, the metabolic rate and
the evaporative water flux.
Gehler et al. (2016) adopted the same principle to investigate pCO during the Paleocene–
2
Eocene transition (56 Ma ago), which is known from other proxies to have been accompanied
by a rapid temperature rise, contemporaneous with a large negative carbon isotope excursion.
The latter indicates a massive release of carbon-containing gas(es) into the atmosphere,
although the carbon source and speciation were not well documented. Oxygen triple-isotope
measurements of mammalian tooth enamel indicated that the sudden rise in atmospheric
temperature during the Paleocene–Eocene transition was not accompanied by the elevated
concentrations (>~2,500 ppm) of CO2 needed to explain the temperature profile. Instead, the
low 13C/12C ratios during the Paleocene–Eocene thermal maximum were most likely caused
by a massive release of seabed methane to the atmosphere.
CONCLUSIONS AND OUTLOOK

The application of oxygen triple-isotope ratio measurements, long established for the
characterization of meteorites and other extraterrestrial samples, has also contributed valuable
new insights to a diversity of research investigations involving the lithosphere, hydrosphere,
cryosphere and atmosphere of Earth. In some cases, it is the presence and magnitude of non-
mass-dependent distributions of the isotopes that has provided unique tracer information;
in other instances it is variations in isotopic composition resulting from mass-dependent
fractionation that facilitated new inferences. In much of the rock record, oxygen triple-isotope
compositions seem to be consistent with equilibrium θ values for SiO2 –water exchange at
various temperatures. The high levels of precision currently possible on Δ′17O determinations
of rocks, minerals and waters, in conjunction with a robust and consistent reporting
framework, allow remarkably small differences to be identified. Despite these advances, the
calibration of rock and mineral data to a water-based reporting scale (VSMOW-SLAP), to a
degree of accuracy commensurate with the associated precision, continues to be challenging.
At present, additional measurements are needed on the oxygen triple-isotope compositions of
widely-used silicate standards for δ18O measurements (such as San Carlos olivine, UWG-2
garnet and NBS 28 silica), from which consensus Δ′17O values will eventually be agreed.
Furthermore, a two-point silicate calibration, with widely-separated δ values that encompass
most of the range encountered in natural samples, is highly desirable. As a first step, inter-
laboratory comparison of two potential silicate standards for this purpose have recently been
made, resulting in agreement of the Δ′17O0.528 values to within 3 ppm. Meanwhile, recent high
precision measurements of the corresponding value for ‘bulk silicate Earth’, exemplified by
San Carlos olivine as a proxy, suggest that it lies within the range –36 to –58 ppm; a recent
model prediction suggests that the value should be close to –40 ppm. A challenge for the
cosmochemistry research community is to agree on a definition of Δ′17O (or, indeed, whether
to just continue with the long-established Δ17O definition) for standardizing the reporting of
measurements on meteorites and, especially, compilations of such measurements in reference
databases. Clearly, there is no single ‘correct’ or ‘incorrect’ definition; it is for meteoriticists
to debate the respective merits of various alternative options and to reach a consensus view.
28 Miller & Pack
ACKNOWLEDGEMENTS
We appreciate the thoughtful and constructive reviews by Mark H. Thiemens and Ryoji
Tanaka. Editorial comments by Ilya Bindeman also led to improvements in the manuscript, for
which we are grateful. During the course of many years, we have benefitted from numerous
discussions, guidance and insights from friends and colleagues in the oxygen triple-isotope
research community. To all those, we offer our sincere appreciation.
REFERENCES
Adnew GA, Hofmann MEG, Paul D, Laskar A, Surma J, Albrecht N, Pack A, Schwieters J, Koren G, Peters W,
Röckmann T (2019) Determination of the triple oxygen and carbon isotopic composition of CO2 from atomic
ion fragments formed in the ion source of the 253 Ultra high-resolution isotope ratio mass spectrometer. Rapid
Commun Mass Spectrom 33:1363–1380
Affek HP, Barkan E (2018) A new method for high-precision measurements of 17O/16O ratios in H2O. Rapid Commun
Mass Spectrom 32:2096–2097
Ahn I, Lee JI, Kusakabe M, Choi B-G (2012) Oxygen isotope measurements of terrestrial silicates using a CO2-laser
BrF5 fluorination technique and the slope of terrestrial fractionation line. Geosci J 16:7–16
Allison CE, Francey RJ, Meijer HAJ (1995) Recommendations for the reporting of stable isotope measurements of
carbon and oxygen in CO2 gas. IAEA-TECDOC825:155–162
Angert A, Rachmilevitch S, Barkan E, Luz B (2003) Effects of photorespiration, the cytochrome pathway, and the
alternative pathway on the triple isotopic composition of atmospheric O2. Global Biogeochem Cycles 17:1030
Angert A, Cappa CD, DePaolo DJ (2004) Kinetic 17O effects in the hydrologic cycle: Indirect evidence and
implications. Geochim Cosmochim Acta 68:3487–3495
Assonov S, Brenninkmeijer CAM (2003) On the 17O correction for CO2 mass spectrometric isotopic analysis. Rapid
Baertschi P (1950) Isotopic composition of the oxygen in silicate rocks. Nature 166:112–113
Baertschi P, Silverman SR (1951) The determination of relative abundances of the oxygen isotopes in silicate rocks.
Geochim Cosmochim Acta 1:317–28
Baker L, Franchi IA, Maynard J, Wright IP, Pillinger CT (2002) A technique for the determination of 18O/16O and
17 16
O/ O isotopic ratios in water from small liquid and solid samples. Anal Chem 74:1665–1673
Bao H, Thiemens MH, Farquhar J, Campbell DA, Lee CC-W, Heine K, Loope DB (2000a) Anomalous 17O
compositions in massive sulphate deposits on the Earth. Nature 406:176–178
Bao H, Campbell DA, Bockheim JG, Thiemens MH (2000b) Origins of sulphate in Antarctic dry-valley soils as
deduced from anomalous 17O compositions. Nature 407:499–502
Bao H, Michalski GM, Thiemens MH (2001) Sulfate oxygen-17 anomalies in desert varnishes. Geochim Cosmochim
Acta 65:2029–2036
Bao H, Lyons J, Zhou C (2008) Triple oxygen isotope evidence for elevated CO2 levels after a Neoproterozoic
glaciation. Nature 453:504–506
Bao H, Cao X, Hayles JA (2016) Triple oxygen isotopes: Fundamental relationships and applications. Annu Rev Earth
Planet Sci 44:463–492
Barkan E, Luz B (2003) High-precision measurements of 17O/16O and 18O/16O of O2 and O2/Ar ratio in air. Rapid
Commun Mass Spectrom 17: 2809–2814
Barkan E, Luz B (2005) High precision measurements of 17O/16O and 18O/16O ratios in H2O. Rapid Commun Mass
Spectrom 19:3737–3742
Barkan E, Luz B (2007) Diffusivity fractionations of H216O/H217O and H216O/H218O in air and their implications for
isotope hydrology. Rapid Commun Mass Spectrom 21:2999–3005 (Erratum: doi:10.1002/rcm.3233)
Barkan E, Luz B (2012) High-precision measurements of 17O/16O and 18O/16O ratios in CO2. Rapid Commun Mass
Spectrom 26:2733–2738
Bender M, Sowers T, Labeyrie L (1994) The Dole effect and its variations during the last 130,000 years as measured
in the Vostok ice core. Global Biogeochem Cycles 8:363–376
Bhattacharya SK, Thiemens MH (1988) Isotopic fractionation in ozone decomposition. Geophys Res Lett 15:9–12
Bhattacharya SK, Savarino J Thiemens MH (2000) A new class of oxygen isotopic fractionation in photodissociation
of carbon dioxide: Potential implications for atmospheres of Mars and Earth. Geophys Res Lett 27:1459–1462
Bigeleisen J (1952) The effects of isotopic substitution on the rates of chemical reactions. J Phys Chem 56:823–828
Bigeleisen J, Göppert-Mayer M (1947) Calculation of equilibrium constants for isotopic exchange reactions. J Chem
Phys 15:261–267
Bigeleisen J, Wolfsberg M (1957) Theoretical and experimental aspects of isotope effects in chemical kinetics. Adv
Chem Phys 1:15–76
Bindeman IN (2021) Triple oxygen isotopes in evolving continental crust, granites, and clastic sediments. Rev
Mineral Geochem 86:241–290
Bindeman IN, Schmitt AK, Evans DAD (2010) Limits of hydrosphere-lithosphere interaction: Origin of the lowest
known δ18O silicate rock on Earth in the Paleoproterozoic Karelian rift. Geology 38:631–634
Bindeman IN, Serebryakov NS, Schmitt AK, Vazquez JA, Guan Y, Azimov PYa, Astafiev BYu, Palandri
J, Dobrzhinetskaya L (2014) Field and microanalytical isotopic investigation of ultradepleted in 18O
Paleoproterozoic “Slushball Earth” rocks from Karelia, Russia. Geosphere 10:308–339
Bindeman IN, Zakharov DO, Palandri J, Greber ND, Dauphas N, Retallack GJ, Hofmann A, Lackey JS, Bekker A
(2018) Rapid emergence of subaerial landmasses and onset of a modern hydrologic cycle 2.5 billion years ago.
Nature 557:545–548
Bindeman IN, Bayon G, Palandri J (2019) Triple oxygen isotope investigation of fine-grained sediments from major world’s
rivers: Insights into weathering processes and global fluxes into the hydrosphere. Earth Planet Sci Lett 528:115851
Blackett PMS (1925) The ejection of protons from nitrogen nuclei, photographed by the Wilson method. Proc R Soc
London A 107:349–360
Boering K, Jackson T, Hoag KJ, Cole AS, Perri MJ, Thiemens M, Atlas E (2004) Observations of the anomalous
oxygen isotopic composition of carbon dioxide in the lower stratosphere and the flux of the anomaly to the
troposphere. Geophys Res Lett 31:L03109
Brand WA, Assonov SS, Coplen TB (2010) Correction for the 17O interference in δ13C measurements when analyzing
CO2 with stable isotope mass spectrometry (IUPAC technical report). Pure Appl Chem 82:1719–1733
Brinjikji M, Lyons JR (2021) Mass-independent fractionation of oxygen isotopes in the atmosphere. Rev Mineral
Geochem 86:197–216
Cano EJ, Sharp ZD, Shearer CK (2020) Distinct oxygen isotope compositions of the Earth and Moon. Nat Geosci
13:270–274
Cao X, Liu Y (2011) Equilibrium mass-dependent fractionation relationships for triple oxygen isotopes. Geochim
Cosmochim Acta 75:7435–7445
Cao X, Bao H, Gao C, Liu Y, Huang F, Pen Y, Zhang Y (2019) Triple oxygen isotope constraints on the origin of ocean
island basalts. Acta Geochim 38:327–334
Chakraborty S, Davis RD, Ahmed M, Jackson TL, Thiemens MH (2012) Oxygen isotope fractionation in the vacuum
ultraviolet photodissociation of carbon monoxide: Wavelength, pressure, and temperature dependency. J Chem
Phys 137:024309
Chakraborty S, Yanchulova P, Thiemens MH (2013) Mass-independent oxygen isotopic partitioning during gas-phase
SiO2 formation. Science 342:463–466
Clayton RN (1955) Variations in oxygen isotope abundances in rock minerals. Ph.D Dissertation, California Institute
of Technology, Pasadena, California
Clayton RN (2002) Self-shielding in the solar nebula. Nature 415:860–861
Clayton RN, Mayeda TK (1963) The use of bromine pentafluoride in the extraction of oxygen from oxides and
silicates for isotopic analysis. Geochim Cosmochim Acta 27:43–52
Clayton RN, Mayeda TK (1983) Oxygen isotopes in eucrites, shergottites, nakhlites, and chassignites. Earth Planet
Sci Lett 62:1–6
Clayton RN, Mayeda TK (1988) Formation of ureilites by nebular processes. Geochim Cosmochim Acta 52:1313–1318
Clayton RN, Mayeda TK (1996) Oxygen isotope studies of achondrites. Geochim Cosmochim Acta 60:1999–2017
Clayton RN, Mayeda TK (2009) Kinetic isotope effects in oxygen in the laboratory dehydration of magnesian
minerals. J Phys Chem A 113:2212–2217
Clayton RN, Grossman L, Mayeda TK (1973) A component of primitive nuclear composition in carbonaceous
meteorites. Science 182:485–488
Clayton RN, Onuma N, Mayeda TK (1976) A classification of meteorites based on oxygen isotopes. Earth Planet Sci
Lett 30:10–18
Cliff SS, Thiemens MH (1997) The 18O/16O and 17O/16O ratios in atmospheric nitrous oxide: A mass-independent
anomaly. Science 278:1774–1776
Cliff SS, Brenninkmeijer CAM, Thiemens MH (1999), First measurement of the 18O/16O and 17O/16O ratios in
stratospheric nitrous oxide: A mass-independent anomaly. J Geophys Res Atmos 104:16171–16175
Coplen TB, Hopple JA, Böhlke JK, Peiser HS, Rieder SE, Krouse HR, Rosman KJR, Ding T, Vocke RD Jr, Révész
KM, Lamberty A, Taylor P, De Bièvre P (2002) Compilation of minimum and maximum isotope ratios of
selected elements in naturally occurring terrestrial materials and reagents. US Geol Survey Water-Resources
Investigations Report 01–4222
Craig H (1957) Isotopic standards for carbon and oxygen and correction factors for mass-spectrometric analysis of
carbon dioxide. Geochim Cosmochim Acta 12:133–149
Craig H (1961) Standard for reporting concentrations of deuterium and oxygen-18 in natural waters. Science
133:1833–1834
Dansgaard W (1964) Stable isotopes in precipitation. Tellus 16:436–468
Dauphas N, Schauble EA (2016) Mass fractionation laws, mass-independent effects, and isotopic anomalies. Annu
Rev Earth Planet Sci 44:709–83
Dunn PJH, Malinovsky D, Goenaga-Infante H (2020) Calibration hierarchies for light-element isotope delta reference
materials. Rapid Commun Mass Spectrom 34:e8711
30 Miller & Pack
Epstein S, Mayeda T (1953) Variation of O18 content of waters from natural sources. Geochim Cosmochim Acta 4:213–24
Epstein S, Buchsbaum R, Lowenstam H, Urey HC (1951) Carbonate-water isotopic temperature scale. Geol Soc Am
Bull 62:417–26
Farquhar J, Thiemens MH, Jackson T (1998) Atmosphere-surface interactions on Mars: Δ17O measurements of
carbonate from ALH 84001. Science 28:1580–1582
Franchi IA, Wright IP, Sexton AS, Pillinger CT (1999) The oxygen-isotopic composition of Earth and Mars. Meteorit
Fujita K, Abe O (2006) Stable isotopes in daily precipitation at Dome Fuji, East Antarctica. Geophys Res Lett 33:L18503
Gehler A, Tütken T, Pack A (2011) Triple oxygen isotope analysis of bioapatite as tracer for diagenetic alteration of
bones and teeth. Palaeogeogr Palaeoclim Palaeoecol 310:84–91
Gehler A, Gingerich PD, Pack A (2016) Temperature and atmospheric CO2 concentration estimates through the
PETM using triple oxygen isotope analysis of mammalian bioapatite. PNAS 113:7739–7744
Giauque WF, Johnston HL (1929) An isotope of oxygen, mass 17, in the Earth’s atmosphere. J Am Chem Soc
51:3528–3534
Gonfiantini R (1978) Standards for stable isotope measurements in natural compounds. Nature 271:534–536
Greenwood RC, Burbine TH, Miller MF, Franchi IA (2017) Melting and differentiation of early-formed asteroids:
The perspective from high precision oxygen isotope studies. Chemie der Erde 77:1–43
Greenwood RC, Barrat J-A, Miller MF, Anand M, Dauphas N, Franchi IA, Sillard P, Starkey NA (2018) Oxygen isotopic
evidence for accretion of Earth’s water before a high-energy Moon-forming giant impact. Sci Adv 4:eaao5928
Hallis LJ, Anand M, Greenwood RC, Miller MF, Franchi IA, Russell SS (2010) The oxygen isotope composition,
petrology and geochemistry of mare basalts: Evidence for large-scale compositional variation in the lunar
mantle. Geochim Cosmochim Acta 74:6885–6899
Harms AV, Gröning M (2017) International Atomic Energy Agency reference sheet for VSMOW2 and SLAP2
international measurement standards (revision 1, 2017-07-11). IAEA, Vienna
Heidenreich JE, Thiemens MH (1986) A non-mass-dependent oxygen isotope effect in the production of ozone from
molecular oxygen: The role of molecular symmetry in isotope chemistry. J Chem Phy 84:2129–2136
Herwartz D (2021) Triple oxygen isotopes in variations in Earth’s crust. Rev Mineral Geochem 86:291–322
Herwartz D, Pack A, Friedrichs B, Bischoff A (2014) Identification of the giant impactor Theia in lunar rocks. Science
344:1146–1150
Herwartz D, Pack A, Krylov D, Xiao Y, Muehlenbachs K, Sengupta S, Di Rocco T (2015) Revealing the climate of
snowball Earth from Δ17O systematics of hydrothermal rocks. PNAS 112:5337–5341
Hofmann MEG, Horváth B, Pack A (2012) Triple oxygen isotope equilibrium fractionation between carbon dioxide
and water. Earth Planet Sci Lett 319-320:159–164
Hofmann MEG, Horváth B, Schneider L, Peters W, Schützenmeister K, Pack A (2017) Atmospheric measurements of
Δ17O in CO2 in Göttingen, Germany reveal a seasonal cycle driven by biospheric uptake. Geochim Cosmochim
Acta 199:143–163
Holden NE, Coplen TB, Böhlke JK, Tarbox LV, Benefield J, de Laeter JR, Mahaffy PG, O’Connor G, Roth E, Tepper
DH, Walczyk T, Wieser ME, Yoneda S (2018) IUPAC periodic table of the elements and isotopes (IPTEI) for the
education community (IUPAC Technical Report). Pure Appl Chem 90:1833–2092
Horibe Y, Shigehara K, Takakuwa Y (1973) Isotope separation factor of carbon dioxide-water system and isotopic
composition of atmospheric oxygen. J Geophys Res 78:2625–2629
Huff AK, Thiemens MH (1998) 17O/16O and 18O/16O isotope measurements of atmospheric carbon monoxide and its
sources. Geophys Res Lett 25:3509–3512
Hulston JR, Thode HG (1965) Variations in S33, S34 and S36 contents of meteorites and their relation to chemical and
nuclear effects. J Geophys Res 70:3475–3484
Johnston JC, Thiemens MH (1997) The isotopic composition of tropospheric ozone in three environments. J Geophys
Res 102:25395–25404
Juranek LW, Quay PD (2013) Using triple isotopes of dissolved oxygen to evaluate global marine productivity.
Annu Rev Mar Sci 5:503–524
Kaiser J (2008) Reformulated 17O correction of mass spectrometric stable isotope measurements in carbon dioxide
and a critical appraisal of historic ‘absolute’ carbon and oxygen isotope ratios. Geochim Cosmochim Acta
72:1312–1334 (Erratum 73:4616)
Kaiser J (2011) Technical note: Consistent calculation of aquatic gross production from oxygen triple isotope
measurements. Biogeosci 8:1793–1811
Kaiser J, Abe O (2012) Reply to Nicholson’s comment on ‘Consistent calculation of aquatic gross production from
oxygen triple isotope measurements’ by Kaiser (2011). Biogeosci 9:2921–2933
Kim NK, Kusakabe M, Park C, Lee JI, Nagao K, Enokido Y, Yamashita S, Park SY (2019) An automated laser
fluorination technique for high-precision analysis of three oxygen isotopes in silicates. Rapid Commun Mass
Spectrom 33:641–649
Krankowsky D, Bartecki F, Klees GG, Mauersberger K, Schellenbach K, Stehr J (1995) Measurement of heavy
isotope enrichment in tropospheric ozone. Geophys Res Lett 22:1713–1716
Krankowsky D, Lämmerzahl P, Mauersberger K (2000) Isotopic measurements of stratospheric ozone. Geophys Res
Lett 27:2593–2595
Kroopnick P, Craig H (1972) Atmospheric oxygen: Isotopic composition and solubility fractionation. Science
175:54–55
Kusakabe M, Matsuhisa Y (2008) Oxygen three-isotope ratios of silicate reference materials determined by direct
comparison with VSMOW-oxygen. Geochem J 42:309–317
Lämmerzahl P, Röckmann T, Brenninkmeijer CAM, Krankowsky D, Mauersberger K (2002) Oxygen isotope
composition of stratospheric carbon dioxide. Geophys Res Lett 29:1582
Landais A, Barkan E, Luz B (2008) Record of δ18O and 17O-excess in ice from Vostok Antarctica during the last
150,000 years. Geophys Res Lett 35:L02709
Levin NE, Raub TD, Dauphas N, Eiler JM (2014) Triple oxygen isotope variations in sedimentary rocks. Geochim
Li S, Levin NE, Chesson LA (2015) Continental scale variation in 17O-excess of meteoric waters in the United States.
Liang M-C, Mahata S (2015) Oxygen anomaly in near surface carbon dioxide reveals deep stratospheric intrusion.
Nat Sci Rep 5:11352
Luz B, Barkan E (2000) Assessment of oceanic productivity with the triple-isotope composition of dissolved oxygen.
Science 288:2028–2031
Luz B, Barkan E (2010) Variations of 17O/16O and 18O/16O in meteoric waters. Geochim Cosmochim Acta 74:6276–6286
Luz B, Barkan E, Bender ML, Thiemens MH, Boering KA (1999) Triple-isotope composition of atmospheric oxygen
as a tracer of biosphere productivity. Nature 400:547–550
Macpherson CG, Hilton DR, Day JMD, Lowry D, Gronvold K (2005) High-3He/4He, depleted mantle and low-δ18O,
recycled oceanic lithosphere in the source of central Iceland magmatism. Earth Planet Sci Lett 233:411–427
Manning CC, Howard EM, Nicholson DP, Ji BY, Sandwith ZO, Stanley RHR (2017) Revising estimates of aquatic
gross oxygen production by the triple oxygen isotope method to incorporate the local isotopic composition of
water. Geophys Res Lett 44:10511–10519
Martin E, Bindeman IN (2009) Mass-independent isotopic signatures of volcanic sulfate from three supereruption ash
deposits in Lake Tecopa, California. Earth Planet Sci Lett 282:102–114
Martin E, Bekki S, Ninin C, Bindeman I (2014) Volcanic sulfate aerosol formation in the troposphere. J Geophys Res
Atmos 119:12660–12673
Matsuhisa Y, Goldsmith JR, Clayton RN (1978) Mechanisms of hydrothermal crystallization of quartz at 250°C and
15 kbar. Geochim Cosmochim Acta 42:173–182
Mattey D, Macpherson C (1993) High precision oxygen isotope microanalysis of ferromagnesian minerals by
laserfluorination. Chem Geol 105:305–318
Mauersberger K, Lämmerzahl P, Krankowsky D (2001) Stratospheric ozone isotope enrichments—revisited. Geophys
Res Lett 28:3155–3158
McCrea JM (1950) On the isotopic chemistry of carbonates and a paleotemperature scale. J Chem Phys 18:849–857
McKinney CR, McCrea JM, Epstein S, Allen HA, Urey HC (1950) Improvements in mass spectrometers for the
measurement of small differences in isotope abundance ratios. Rev Sci Instrum 21:724–730
Meijer HAJ, Li WJ (1998) The use of electrolysis for accurate δ17O and δ18O isotope measurements in water. Isotopes
Environ Health Stud 34:349–369 (Erratum 35:142)
Meija J, Coplen TB, Berglund M, Brand WA, De Bièvre P, Gröning M, Holden NE, Irrgeher J, Loss RD, Walczyk T, Prohaska
T (2016) Isotopic compositions of the elements 2013 (IUPAC Technical Report). Pure Appl Chem 88:293–306
Michalski G, Scott Z, Kabiling M, Thiemens MH (2003) First measurements and modeling of Δ17O in atmospheric
nitrate. Geophys Res Lett 30:1870–1873
Michalski G, Bohlke JK, Thiemens M (2004) Long term atmospheric deposition as the source of nitrate and other salts
in the Atacama desert, Chile: New evidence from mass-independent oxygen isotopic compositions. Geochim
Michalski G, Bockheim JG, Kendall C, Thiemens M (2005) Isotopic composition of Antarctic Dry Valley nitrate:
Implications for NOy sources and cycling in Antarctica. Geophys Res Lett 32:L13817
Miller MF (2002) Isotopic fractionation and the quantification of 17O anomalies in the oxygen three-isotope system:
An appraisal and geochemical significance. Geochim Cosmochim Acta 66:1881–1889
Miller MF (2008) Comment on “Record of δ18O and 17O-excess in ice from Vostok Antarctica during the last
150,000 years” by Landais A, Barkan E, Luz B (2008) [Geophys Res Lett 35:L02709]. Geophys Res Lett
35:L23708
Miller MF (2018) Precipitation regime influence on oxygen triple-isotope distributions in Antarctic precipitation and
ice cores. Earth Planet Sci Lett 481:316–327
Miller MF, Franchi IA, Sexton AS, Pillinger CT (1999) High precision δ17O isotope measurements of oxygen from
silicates and other oxides: Method and applications. Rapid Commun Mass Spectrom 13:1211–1217
Miller MF, Franchi IA, Thiemens MH, Jackson TL, Brack A, Kurat G, Pillinger CT (2002) Mass-independent
fractionation of oxygen isotopes during thermal decomposition of carbonates. PNAS 99:10988–10993
32 Miller & Pack
Miller MF, Röckmann T, Wright IP (2007) A general algorithm for the 17O abundance correction to 13C/12C
determinations from CO2 isotopologue measurements, including CO2 characterised by ‘mass-independent’
oxygen isotope distributions. Geochim Cosmochim Acta 71:3145–3161
Miller MF, Greenwood RC, Franchi IA (2015) Comment on “The triple oxygen isotope composition of the Earth
mantle and understanding Δ17O variations in terrestrial rocks and minerals” by Pack and Herwartz (2014) [Earth
Planet Sci Lett 390:138–145]. Earth Planet Sci Lett 418:181–183
Miller MF, Pack A, Bindeman IN, Greenwood RC (2020) Standardizing the reporting of Δ′17O data from high
precision oxygen triple-isotope ratio measurements of silicate rocks and minerals. Chem Geol 532:119332
Navon O, Wasserburg GJ (1985) Self-shielding in O2—a possible explanation for oxygen isotopic anomalies in
meteorites? Earth Planet Sci Lett 73:1–16
Nier A (1947) A mass spectrometer for isotope and gas analysis. Rev Sci Instrum 18:398–411
Pack A (2021) Isotopic fingerprints of atmospheric molecular O2 in rocks, minerals, and melts. Rev Mineral Geochem
86:217–240
Pack A, Toulouse C, Przybilla R (2007) Determination of oxygen triple isotope ratios of silicates without cryogenic
separation of NF3—technique with application to analyses of technical O2 gas and meteorite classification.
Rapid Commun Mass Spectrom 21:3721–3728
Pack A, Herwartz D (2014) The triple oxygen isotope composition of the Earth mantle and understanding Δ17O
variations in terrestrial rocks and minerals. Earth Planet Sci Lett 390:138–145
Pack A, Gehler A, Süssenberger A (2013) Exploring the usability of isotopically anomalous oxygen in bones and teeth
as paleo-CO2-barometer. Geochim Cosmochim Acta 102:306–317
Pack A, Tanaka R, Hering M, Sengupta S, Peters S, Nakamura E (2016) The oxygen isotope composition of San
Carlos olivine on the VSMOW2-SLAP2 scale. Rapid Commun Mass Spectrom 30:1495–1504
Pack A, Höweling A, Hezel DC, Stefanak MT, Beck A-K, Peters STM, Sengupta S, Herwartz D, Folco L (2017)
Tracing the oxygen isotope composition of the upper Earth’s atmosphere using cosmic spherules. Nat Commun
8:15702–15708
Prokopenko MG, Pauluis OM, Granger J, Yeung LY (2011) Exact evaluation of gross photosynthetic production
from the oxygen triple-isotope composition of O2: Implications for the net-to-gross primary production ratios.
Geophys Res Lett 38:L14603
Redlich O (1935) Eine allgemeine Beziehung zwischen den Schwingungsfrequenzen isotoper Molekeln. Z Phys
Chem B 28:371–382
Richet P, Bottinga Y, Javoy M (1977) A review of hydrogen, carbon, nitrogen, oxygen, sulphur, and chlorine stable
isotope fractionation among gaseous molecules. Annu Rev Earth Planet Sci 5:65–110
Robert F, Rejou-Michel A, Javoy M (1992) Oxygen isotopic homogeneity of the Earth: New evidence. Earth Planet
Sci Lett 108:1–9
Röckmann T, Brenninkmeijer CAM (1998) The error in conventionally reported 13C/12C ratios of atmospheric CO due
to the presence of mass independent oxygen isotope enrichment. Geophys Res Lett 25:3163–3166
Röckmann T, Brenninkmeijer CAM, Saueressig G, Bergamaschi P, Crowley JN, Fischer H, Crutzen PJ (1998) Mass-
independent oxygen isotope fractionation in atmospheric CO as a result of the reaction CO + OH. Science 281:544–546
Rozanski K, Araguás-Araguás L, Gonfiantini R (1993) Isotopic patterns in modern global precipitation. In:
Climate change in continental isotopic records. Swart PK, Lohmann KC, Mckenzie J, Savin S (eds) Am
Geophys Union, Geophys. Monograph Series 78:1–36
Rumble D (2018) The third isotope of the third element on the third planet. Am Mineral 103:1–10
Rumble D (III), Hoering TC (1994) Analysis of oxygen and sulfur isotope ratios in oxide and sulfide minerals by spot
heating with a carbon dioxide laser in fluorine atmosphere. Acc Chem Res 27:237–241
Rumble D (III), Farquhar J, Young ED, Christensen CP (1997) In situ oxygen isotope analysis with an excimer laser
using F2 and BrF5 reagents and O2 gas as analyte. Geochim Cosmochim Acta 61:4229–4234
Rumble D, Miller MF, Franchi IA, Greenwood RC (2007) Oxygen three-isotope fractionation lines in terrestrial
silicate minerals: An inter-laboratory comparison of hydrothermal quartz and eclogitic garnet. Geochim
Santrock J, Studley SA, Hayes JM (1985) Isotopic analyses based on the mass spectrum of carbon dioxide. Anal
Chem 85:1444–1448
Savarino J, Bekki S, Cole-Dai J, Thiemens MH (2003) Evidence from sulfate mass independent oxygen isotopic compositions
of dramatic changes in atmospheric oxidation following massive volcanic eruptions. J Geophys Res 108:4671
Schauble EA, Young ED (2021) Mass dependence of equilibrium oxygen isotope fractionation in carbonate, nitrate,
oxide, perchlorate, phosphate, silicate, and sulfate minerals. Rev Mineral Geochem 86:137–178
Schoenemann SW, Schauer AJ, Steig EJ (2013) Measurement of SLAP2 and GISP δ17O and proposed VSMOW-
SLAP normalization for δ17O and 17Oexcess. Rapid Commun Mass Spectrom 27:582–590
Schoenemann SW, Steig EJ (2016) Seasonal and spatial variations of 17Oexcess and dexcess in Antarctic precipitation:
Insights from an intermediate complexity isotope model. J Geophys Res Atmos 121:11215–11247
Schoenemann SW, Steig EJ, Ding Q, Markle BR, Schauer AJ, (2014) Triple water-isotopologue record from WAIS Divide,
Antarctica: Controls on glacial-interglacial changes in 17Oexcess of precipitation. J Geophys Res Atmos 119:8741–8763
Schueler B, Morton J, Mauersberger K (1990) Measurement of isotopic abundances in collected stratospheric ozone
samples. Geophys Res Lett 17:1295–1298
Sharp ZD (1990) A laser-based microanalytical method for the in situ determination of oxygen isotope ratios of
silicates and oxides. Geochim Cosmochim Acta 54:1353–1357
Sharp ZD, Wostbrock JAG (2021) Standardization for the triple oxygen isotope system: Waters, silicates, carbonates,
air, and sulfates. Rev Mineral Geochem 86:179–196
Sharp ZD, Gibbons JA, Maltsev O, Atudorej V, Pack A, Sengupta S, Shock EL, Knauth LP (2016) A calibration of the
triple-isotope fractionation in the SiO2–H2O system and applications to natural samples. Geochim Cosmochim
Acta 186:105–119
Spicuzza MJ, Day JMD, Taylor LA, Valley JW (2007) Oxygen isotope constraints on the origin and differentiation of
the Moon. Earth Planet Sci Lett 253:254–265
Starkey NA, Jackson RM, Greenwood RC, Parman S, Franchi IA, Jackson M, Fitton JG, Stuart FM, Kurz M, Larsen LM
(2016) Triple oxygen isotopic composition of the high-3He/4He mantle. Geochim Cosmochim Acta 176:227–238
Steig E, Gkinis V, Schauer A, Schoenemann S, Samek K, Hoffnagle J, Dennis K, Tan S (2014) Calibrated high-
precision 17O-excess measurements using cavity ring-down spectroscopy with laser-current-tuned cavity
resonance. Atmos Meas Tech 7:2421–2435
Sun T, Bao H (2011a) Non-mass-dependent 17O anomalies generated by a superimposed thermal gradient on rarefied
O2 in a closed system. Rapid Commun Mass Spectrom 25:20–24
Sun T, Bao H (2011b) Thermal-gradient-induced non-mass-dependent isotope fractionation. Rapid Commun Mass
Tanaka R, Nakamura E (2013) Determination of 17O-excess of terrestrial silicate/oxide minerals with respect to
Vienna Standard Mean Ocean Water VSMOW. Rapid Commun Mass Spectrom 27:285–297
Tanaka R, Nakamura E (2017) Silicate–SiO reaction in a protoplanetary disk recorded by oxygen isotopes in
chondrules. Nat Astron 1:0137
Thiemens MH (2013) Introduction to chemistry and applications in nature of mass independent isotope effects special
feature. PNAS 110:17631–17637
Thiemens MH, Lin M (2019) Use of isotope effects to understand the present and past of the atmosphere and climate
and track the origin of life. Angew Chem Int Ed 58:6826–6844
Thiemens MH, Lin M (2021) Discoveries of mass independent isotope effects in the solar system: Past, present and
future. Rev Mineral Geochem 86:35-95
Thiemens MH, Heidenreich JE (1983) The mass independent fractionation of oxygen—a novel isotope effect and its
cosmological implications. Science 219:1073–1075
Thiemens MH, Jackson T, Mauersberger K, Schueler B, Morton J (1991) Oxygen isotope fractionation in stratospheric
CO2. Geophys Res Lett 18:669–672
Thiemens MH, Jackson T, Zipf EC, Erdman PW, van Egmond C (1995a) Carbon dioxide and oxygen isotope
anomalies in the mesosphere and stratosphere. Science 270:969–972
Thiemens MH, Jackson T, Brenninkmeijer CAM (1995b) Observation of a mass independent oxygen isotopic
composition in terrestrial stratospheric CO2, the link to ozone chemistry, and the possible occurrence in the
Martian atmosphere. Geophys Res Lett 22:255–257
Thiemens MH, Chakraborty S, Jackson TL (2014) Decadal Δ17O record of tropospheric CO2: Verification of a
stratospheric component in the troposphere. J Geophys Res Atmos 119:6221–6229
Thirlwall MF, Gee MAM, Lowry D, Mattey DP, Murton BJ, Taylor RN (2006) Low δ18O in the Icelandic mantle
and its origins: Evidence from Reykjanes Ridge and Icelandic lavas. Geochim Cosmochim Acta 70:993–1019
Tian C, Wang L, Kaseke KF, Bird BW (2018) Stable isotope compositions (δ2H, δ18O and δ17O) of rainfall and
snowfall in the central United States. Nat Sci Rep 8:6712
Tian C, Wang L, Novick KA (2016) Water vapor δ2H, δ18O and δ17O measurements using an off-axis integrated cavity
output spectrometer—sensitivity to water vapor concentration, delta value and averaging-time. Rapid Commun
Tian C, Wang L, Tian F, Zhao F, Jiao W (2019) Spatial and temporal variations of tap water 17O-excess in China.
Turner J, Phillips T, Thamban M, Rahaman W, Marshall GJ, Wille JD, Favier V, Winton VHL, Thomas E, Wang Z,
van den Broeke M, Hosking JS, Lachlan-Cope T (2019) The dominant role of extreme precipitation events in
Antarctic snowfall variability. Geophys Res Lett 46:3502–3511
Uechi Y, Uemura R (2019) Dominant influence of the humidity in the moisture source region on the 17O-excess in
precipitation on a subtropical island. Earth Planet Sci Lett 513:20–28
Uemura R, Barkan E, Abe O, Luz B (2010) Triple isotope composition of oxygen in atmospheric water vapor.
Urey HC (1947) The thermodynamic properties of isotopic substances. J Chem Soc (London) 562–581
Urey HC, Lowenstam HA, Epstein S, McKinney CR (1951) Measurement of paleotemperatures and temperatures of
the upper cretaceous of England, Denmark, and the southeastern United States. Geol Soc Am Bull 62:399–416
34 Miller & Pack
Valley JW, Kitchen N, Kohn MJ, Niendorf CR, Spicuzza MJ (1995) UWG-2, a garnet standard for oxygen isotope
ratios: Strategies for high precision and accuracy with laser heating. Geochim Cosmochim Acta 59:5223–5231
Verkouteren RM, Klinedinst DB (2004) Value assignment and uncertainty estimation of selected light stable isotope
reference materials: RMs 8543–8545, RMs 8562–8564, and RM 8566. NIST Spec Publ 260–149, 2004 edition
Wen J, Thiemens MH (1990) An apparent new isotope effect in a molecular decomposition and implications for
nature. Chem Phys Lett 172:416–420
Wen J, Thiemens MH (1991) Experimental and theoretical study of isotope effects on ozone decomposition. J
Geophys Res 96 (D6):10911–10921
Wiechert UH, Hoefs J (1995) An excimer laser-based micro analytical preparation technique for in-situ oxygen
isotope analysis of silicate and oxide minerals. Geochim Cosmochim Acta. 59:4093–4101
Wiechert U, Halliday AN, Lee D-C, Snyder GA, Taylor LA, Rumble D (2001) Oxygen isotopes and the Moon-
forming giant impact. Science 249:345–348
Wostbrock JAG, Sharp ZD (2021) Triple oxygen isotopes in silica-water and carbonate-water systems. Rev Mineral
Geochem 86:367–400
Wostbrock JAG, Cano E, Sharp ZD (2020) An internally consistent triple oxygen isotope calibration of standards for
silicates, carbonates and air relative to VSMOW2 and SLAP2. Chem Geol 533:119432
Yeung LY, Hayles JA (2021) Climbing to the top of Mount Fuji: Uniting theory and observations of oxygen triple
isotope systematics. Rev Mineral Geochem 86:97–135
Yeung LY, Hayles JA, Hu H, Ash JL, Sun T (2018) Scale distortion from pressure baselines as a source of inaccuracy
in triple-isotope measurements. Rapid Comm Mass Spectrom 32:1811–1821
Yeung LY, Young ED, Schauble EA (2012) Measurements of 18O18O and 17O18O in the atmosphere and the role of
isotope-exchange reactions. J Geophys Res 117:D18306
Young ED, Fogel ML, Rumble D (III), Hoering TC (1998) Isotope-ratio-monitoring of O2 for microanalysis of
18 16
O/ O and 17O/16O in geological materials. Geochim Cosmochim Acta 62:3087–3094
Young ED, Galy A, Nagahara H (2002) Kinetic and equilibrium mass-dependent isotope fractionation laws in nature
and their geochemical and cosmochemical significance. Geochim Cosmochim Acta 66:1095–1104
Young ED, Yeung LY, Kohl IE (2014) On the Δ17O budget of atmospheric O2. Geochim Cosmochim Acta 135:102–125
Young ED, Kohl IE, Warren PH, Rubie DC, Jacobson SA, Morbidelli A (2016) Oxygen isotopic evidence for vigorous
mixing during the Moon-forming giant impact. Science 351:493–496
Zakharov DO, Bindeman IN (2019) Triple oxygen and hydrogen isotopic study of hydrothermally altered rocks
from the 2.43–2.41 Ga Vetreny belt, Russia: An insight into the early Paleoproterozoic seawater. Geochim
Zakharov DO, Bindeman IN, Serebryakov NS, Prave AR, Azimov PYa, Babarina II (2019) Low δ18O rocks in the
Belomorian belt, NW Russia, and Scourie dikes, NW Scotland: A record of ancient meteoric water captured by
the early Paleoproterozoic global mafic magmatism. Precambr Res 333:105431
Zakharov DO, Bindeman IN, Slabunov AI, Ovtcharova M, Coble MA, Serebryakov NS, Schaltegger U (2017) Dating
the Paleoproterozoic snowball Earth glaciations using contemporaneous subglacial hydrothermal systems.
Geology 45:667–670
Zel’dovich VaB, Maksimov IA (1976) Gas diffusion: Its dependence on nuclear spin. Soviet J Exper Theoret Phys 43:39–41
Vol. 86 pp. 35–95, 2021 2
Discoveries of Mass Independent Isotope Effects

in the Solar System: Past, Present and Future
Mark H. Thiemens
Department of Chemistry and Biochemistry
University of California San Diego
La Jolla, California 92093
USA
mthiemens@ucsd.edu
Mang Lin
State Key Laboratory of Isotope Geochemistry
Guangzhou Institute of Geochemistry,
Chinese Academy of Sciences
Guangzhou, Guangdong 510640
China
University of Chinese Academy of Sciences

Beijing 100049
China
linm@gig.ac.cn
THE BEGINNING OF ISOTOPES

Discovery and chemical physics
The history of the discovery of stable isotopes and later, their influence of chemical and
physical phenomena originates in the 19th century with discovery of radioactivity by Becquerel
in 1896 (Becquerel 1896a–g). The discovery catalyzed a range of studies in physics to develop
an understanding of the nucleus and the properties influencing its stability and instability
that give rise to various decay modes and associated energies. Rutherford and Soddy (1903)
later suggested that radioactive change from different types of decay are linked to chemical
change. Soddy later found that this is a general phenomenon and radioactive decay of different
energies and types are linked to the same element. Soddy (1913) in his paper on intra-atomic
charge pinpointed the observations as requiring the observations of the simultaneous character
of chemical change from the same position in the periodic chart with radiative emissions
required it to be of the same element (same proton number) but differing atomic weight.
This is only energetically accommodated by a change in neutrons and it was this paper that
the name “isotope” emerges. The discovery of the positive ray spectrograph, later termed mass
spectrometer by Aston (1919) was a major breakthrough in understanding not only nuclear
decay and stability but also led to the first calculations of different physical chemical properties
by isotopic substitution. Lindemann (1919) tackled the problem as to how isotopic substitution
might alter a chemical property, in this specific case the differential vapor pressures of lead of
masses 206 and 207. It was not known that vibrational properties of the isotopically substituted
species were different, and the calculation required an independent approach. At this time
quantum mechanics was developed to the extent that harmonic oscillations could be utilized
1943-2666/21/0086-0002$10.00 (online)
36 Thiemens & Lin
in energy calculations. In the Planck Einstein hypothesis atomic heat may be quantified
with a harmonic oscillation at frequency ν and with quantized absorption and emission of
energy modified to permit the oscillation absorption to be via continua. In his statistical
mechanical approach, it was determined that there may exist a difference from a second order
effect, that would be small and arising from the mass differences in vibrational energies.
This second order effect is due to a difference in “Nullpunktsenergie” or zero-point energy (ZPE).
This is one of the earliest applications of stable isotopes in physics and is the groundwork
for the later developments in isotope chemistry by Urey and colleagues. In the same year,
Lindemann and Aston (1919) demonstrated the effect of isotope substitution in a paper
“The Possibility of Chemical Separation of Isotopes” revealing that the effects are general and
encompass thermodynamics, diffusion, chemical separation from equilibrium differences, and
gravity resulting in a range of applications in physics, chemistry and biology.
The work cited above launched investigations into stable isotope effects and applications
that continue up to the writing of this Chapter. In a recent review paper by Thiemens and Lin
(2019) some of the early applications of stable isotopes are discussed. The Lindemann work
on isotope effects, particularly in the prediction of the isotopic differences in zero-point energy
was of fundamental importance in quantum chemistry at the time and is also the very basis
of many applications in isotope geochemistry. As discussed by Thiemens and Lin (2019), the
use of quantum effects to discover the oxygen isotopes in the Department of Chemistry of
the University of California Berkeley (Giauque and Johnston 1929a,b) was considered a very
early proof of the validity of the quantum theory. The atmospheric absorption bands of oxygen
were recognized as possessing a structure that was inconsistent with classical interpretations.
It was noted that the absorption bands arising from the 1.6-eV excitation levels of O2 to low
lying rotational bands. Most of these originate from the one expected for a non-polar diatomic,
two rotating nuclei system with only alternate rotational bands populated. The lines are P and
R types are doublets. However, besides the A band, there is an A′ ghost (or Planet B), the same
as A band but energetically different. This ghost arises due to the asymmetry of the rotational
bands and appearance of all lines from the 18O–16O as compared to symmetric 16O–16O.
The symmetry observation is a quantum mechanical effect and an important observation for
quantum theory and the discovery of 18O. It was quickly discovered as a result of the isotopic
asymmetry that there is an oxygen isotope of mass 17. The spectroscopy required state of the
art (in 1929) measurements and were done at the Mt. Wilson Observatory.
The line position of oxygen isotopic absorption lines at different frequencies in the
Schumann Runge bands (175 nm < λ < 205 nm) were used by Cicerone and Mccrumb (1980)
some 50 years later to calculate the transmission functions of ultraviolet (UV) light through
the earth’s atmosphere. It was shown that with increasing optical length there is “isotopic self-
shielding” in the atmosphere, the first suggestion of self-shielding. They abandoned the effect
as subsequent isotopic exchange obliterates the self-shielding effect. Of special relevance to
this Chapter is that in the first demonstration of a mass independent oxygen isotope effect by
Thiemens and Heidenreich (1983) there was a suggestion that the symmetry effect and self-
shielding in the early solar system could be an important process for production of the oxygen
isotopic anomalies, though based upon new measurements and mechanistic understanding of
the chemical physics of the quantum chemical isotope effects render shielding improbable.
Further experiments and investigation of the effect lead to greater insight into the quantum
chemistry (Heidenreich and Thiemens 1986) and deeper understanding of chemical reaction
isotopic selection mechanisms. It suggests that symmetry isotope effects may be general
effects in nature, which will be thoroughly reviewed and discussed later in this chapter.
The field of isotope chemistry and applications in nature initiates in the 1930s, in a series
of papers by Urey. Following the discovery of oxygen isotope by Giauque and colleagues at
Berkeley, Urey initiated a search for isotopes of hydrogen. A full historical account of this
Mass Independent Isotope Effects in the Solar System 37
discovery is given by Brickwedde (1982). It should be recognized that at the time of the
discovery of oxygen isotopes and Urey’s interest in hydrogen isotopes the neutron discovery
by Chadwick had not been published. The effect of isotopic species on the Balmer α-lines of
atomic hydrogen and deuterium were calculated from the Balmer series formula and searched
for by atomic discharge. By measurement of these lines as a function of temperature above and
at the liquification temperature of H2 (14 K) the distinct α-line of deuterium was recognized, and
deuterium was discovered (Urey et al. 1932a,b). The near simultaneous discovery of the isotopes
of oxygen and hydrogen catalyzed numerous studies of chemically manifested isotope effects.
A limit to the first measurements of natural samples was the ability to measure the
variation of isotope variations at high precision. Most studies involving the physical chemistry
of processes were done for processes producing large isotopic variation, such as evaporation,
condensation, diffusion or equilibria. The role of isotopes in defining biological chemical
mechanisms was recognized immediately and experiments were done using isotope spiking
techniques. In the first measurement of any isotope in meteorites, Manian et al. (1934)
reported the ratio of O16:O18 [isotope notation at that time, prior to introduction of the
δ notation by Craig (1953)] in stone meteorites. The motivation was to use oxygen isotope
ratio measurements to understand the source region of stony meteorites which at the time
was thought to be interstellar space as deduced from the high heliospheric velocity (Manian
et al. 1934). In this early work extraction of oxygen from the silicates was achieved using
carbon tetrachloride at high temperature and chemical conversion to water that was ultimately
electrolytically decomposed and the mass spectrometric measurements were made on O2. The
oxygen isotopic composition of the meteorites Mocs (Transylvanian L6), Knyinya (L5) and
Homestead (L5) were found to be the same as a terrestrial granite from Stoningtom, Maine.
Early measurements of hydrogen isotopes immediately after their discovery by Bradley and
Urey (1932) found the composition in obsidian from a Crater of Mt. Kilauea, Devonian water,
and hydrogen from helium bearing gas to be identical. The general consensus at this time was
that there is no variation in elemental isotopic variations in terrestrial or meteoric matter.
The dawn of stable isotope geochemistry
Following the paper of Manian et al. (1934) there were isotopic measurements of the
isotopes of natural materials that showed no variation. It was largely accepted that these
materials are homogenized at a level such that any isotope effect was minimal. Spectroscopic
measurements of vibrational frequency differences and calculations of equilibrium constants
revealed that ΔG for these exchange reactions produce equilibrium partition functions that
were too small to be measured.
The inability to measure isotope ratios at high precision disallowed any isotope study of
natural samples. To specifically address this, Nier (1947) developed a new mass spectrometer
that was specifically designed to measure isotope ratios in light elements at much high precision
not achievable by other spectrometric means. The new mass spectrometer differed in that it
was all metal, with better vacuum characteristics achieved. Rather than measuring ion beams
of individual masses separately to report as a ratio of current, the beams’ currents are collected
simultaneously in two faraday collectors and passed through a decadal resistance circuit to
precisely null out the currents. In addition, this allowed different gains on each collector to allow
gain in the minor isotope signal. A final aspect of the new approach was that the measurement of
an unknown sample isotope ratio was measured and switchover to a standard of know composition
was done to allow even greater precision, much like the dual beam spectrophotometer invented
by Beckman some 7 years previously. The Nier paper provided the first opportunity to measure
isotope ratios commensurate with the expected variation in nature.
In the same year as the publication of Nier (1947), Urey (1947) and Bigeleisen and
Mayer (1947) reported methodologies for a more precise calculation of equilibrium constants
for exchange reactions. As described by Urey (1947) the technique takes advantage of the
38 Thiemens & Lin
ability to use gas phase molecules where high precision spectroscopic data was available.
With the determination of the isotopically substituted species vibrational frequencies the
exchange between two species may be determined. The approach ideally requires knowing
the energy between the product dissociated state and the lowest energy of the ground
state (Nullpunktsenergie). Since the ZPE energy could not be known a methodology for
determination of it was developed.
The basis for many isotope effects relies on application of determination of the energy
partitioning of the isotopic species this ZPE approach is general in its application as will be
discussed. In the approach of Urey (1947), using the vibrational frequency, partition function
(Q) for an individual molecule may be determined. In particular, the vibrational partition
function Qvib is treated as e−u/2/(1− e−u), where u equals hc ω/kT by considering a simple
harmonic oscillator behavior. The ZPE and the energy difference between the potential energy
well and ground vibrational state, is therefore included in this consideration. For example, the
isotopic partition function ratio of water is expressed as:
H 218O Q2
 1 .0667  at 298K  (1)
H 216 O Q1
The physical interpretation of this the isotope partitioning between the separated atom and
its ground state with a 66.7 parts per thousand (or per mil, ‰) preference for the 18O/16O ratio
to be in the bound water molecule. Similarly, Q2/Q1 for CO2 is:
1
 C18O2  2 Q2 (2)
 16  1
.1172  at 298K 
 C O 2  Q1
Note the factor in the exponent has a factor of 1/2 in the exponent which accounts for
the 2 exchangeable oxygen atoms in CO2. The equilibrium constant of oxygen isotopic
exchange between CO2 and H2O then is simply the ratio of Equations (2) to (1), i.e.,
1.1172/1.0667 = 1.047. In current notation this is a δ18O for CO2 as 47‰ greater than the
water it exchanges with, not far for the present-day atmospheric CO2 isotopic composition and
reasonable given that spectroscopic measurements were significantly less precise in 1947 than
today. Since the partition function ratio is a function of temperature, the isotopic exchanges
between different molecules depend on temperature. As Craig recalled (https://www.balzan.
org/en/prizewinners/harmon-craig/rome-23-11-1998-craig), Urey suddenly realized that there
was a geological thermometer on his hands.
AN OVERVIEW OF STABLE ISOTOPE GEOCHEMISTRY

Mass dependent effects and applications
The previous section has discussed the origins of isotopes and isotope effects. Though it was
nearly a half century later, the calculation of the differential vapor pressure of isotopes (Lindemann
1919; Lindemann and Aston 1919) in 1919 and their separation factors from chemical equilibria
lead to an extensive array of applications once the isotope ratio mass spectrometer design of Nier
(1947) was available. For example, the temperature dependent isotope separation of isotopically
substituted water (oxygen or hydrogen) in evaporation and condensation provides the only
mechanism by which we can determine the global earth temperatures over many relevant time
periods such as glacial-interglacial times. Figure 1 is an example showing the ice core isotopic
composition retrieved from Dome Fuji at Antarctica and corresponding isotopically determined
temperatures over glacial–interglacial cycles in the past 0.36 million years (Uemura et al. 2012).
Figure 1. Dome Fuji ice core isotopic composition and temperatures. Data from Uemura et al. (2012). The
oxygen and hydrogen isotopes in the top figure are used to calculate the temperatures in bottom figure.
Without stable isotopes the temperature change on this time period could not be quantitatively determined.
If not for this application, the ability to understand the earth’s climate system and its
change would be severely restricted if not impossible. The development of the technique by
which the temperatures of the oceans over long time periods for oxygen isotopic measurements
of foraminiferal carbonate tests in the Urey’s laboratory at Chicago has also led to a deeper
understanding of e.g., climate change and heat transfer of the ocean’s circulation conveyor
belt. Although it is now widely acknowledged that kinetic isotopic fractionation occurs during
many biomineralization processes (the so-called “vital effects”), the oxygen isotopic variation
in foraminiferal remains a powerful tool to reconstruct the fluctuation of Earth’s climate
system on astronomical timescales as noted by Emiliani in his pioneering works (Emiliani
1955). Figure 2 is an example showing the covariation of benthic foraminifera δ18O values and
alkenone‐derived sea surface temperature in the Pliocene (Caballero-Gill et al. 2019), a period
exhibiting glacial–interglacial cycles linked to variations in Earth’s obliquity.
Figure 2. Southwest Pacific benthic foraminifera δ18O values and alkenone-derived sea surface tempera-
ture records. Data from Caballero-Gill et al. (2019).
Another simple isotope effect derives from the differential mass dependent isotope
effect on isotopes by gravity. It is of course well known that Newtonian physics dictates that
due to gravity, there is a force between the mass of two bodies. For isotopes and earth this
mass effect manifests itself at the top of the atmosphere due to the escape of atoms at the
top of the atmosphere, a process governed by kinetic velocities, the planet’s required escape
velocity, and mean free path. In equilibrium, the velocity distribution of molecules follows
40 Thiemens & Lin
the Maxwell–Boltzmann distribution and the most probable velocity v0 equals to (2kT/m)1/2,
where k is the Boltzmann constant, T is the local temperature, and m is the mass of the
escaping molecules. The escape velocity of the planet ve equals to (2GM/r)1/2, where G is the
gravitational constant, M is the mass of the planet, and r is the distance from the center of
planet. The ratio of ve2 to v02 is defined as a dimensionless parameter (λ) that is governed by
the gravitational potential energy and molecule thermal kinetic energy:
GMm (3)

r kT
The escape flux of molecules from a planet (Φ) is derived from the kinetic theory of gases:
nc  0
 1    exp    (4)
2 
where nc is the number density of escaping molecules at the exobase. The Equation (4) is
known as the Jeans escape formula, named after James Jeans, the pioneer who calculated
this process (Jeans 1904). There are other important atmospheric escape mechanisms such
as hydrodynamic escape, sputtering, charge exchange, photochemical escape and ion-drag
effects, but the Jeans escape is a fundamental one that depicts a general picture for the
evolution of planetary atmospheres as a function of atmospheric composition, planet size, and
exobase temperature. The Jeans escape formula clearly shows that lighter isotopes escape to
space faster than those with heavier isotopes, and a mass dependent isotope effect therefore
occurs during the Jeans escape. If the planetary atmosphere has been escaping over a long
time (billions of years), light isotopes will be preferentially removed, and the atmosphere
will become enriched in heavy isotopes. The relatively large ratio of 15N/14N in the Martian
atmosphere was firstly detected by the Viking Lander, providing evidence for long-term
erosion of the Martian atmosphere (Mcelroy et al. 1976). The isotopic composition of volatiles
is a crucial tracer for tracking the evolution of Martian atmosphere, climate, and habitability.
Such works are still under way by Martian meteorite isotopic measurements and in situ
observations such as the Curiosity and MAVEN missions.
In geochemistry it is by now well established that due to the development of the calculation
of the equilibrium constant in an exchange reaction (Bigeleisen and Mayer 1947; Urey 1947),
one may very precisely determine the temperature of formation of a rock containing two
different minerals that have equilibrated with each other. The temperature dependence of the
equilibrium constant is determined by laboratory measurements allowing the measurement
of the oxygen isotopic composition of the two minerals to determine the temperature
(e.g., review by Thiemens 2006).
In the identification of mass independent isotope effects, the basis for mass dependent
effects should be resolved. Classically, the energy is determined by summing over all available
energy states Qi. These typically are translational, rotational, vibrational, and electronic.
Figure 3 shows the different energy partitions and their ultimate reliance upon nuclidic mass
differences for a given element’s isotopes.
The energy partitioning by translation is that of motion, velocity and energy collisions
resulting in a Boltzmann distribution of velocities or energies. The rotational energies are
basically expressed in terms of the angular moment of inertia I. These two energy forms are
conventionally described in classical forms as they are very accurate above room temperature.
The vibration of a molecule depends upon the spring constant, which reflects the nature of
the chemical bond and is usually described in quantum mechanical equations. The nuclear
term is a small energy term that arises only in the case where the nucleus is odd Z and
possesses a nuclear spin and hyperfine magnetic moment. The overlap integral of the hyperfine
moment with the magnetic moment of the electrons, providing a hyperfine coupling (nuc).
Figure 3. The partitioning of energy between different contributions: translation, rotation, vibration, nu-
clear hyperfine and electronic.
The isotopic contributions of this effect are reviewed by Buchachenko (2001, 2013, 2018) and
Thiemens (2006). The electronic component is generally not considered as a consequence of
the Born–Oppenheimer approximation. This derives from the relative sizes of the electrons
and nucleons and the fact that changes in chemistry from electronic changes occur on a time
scale that nuclear positional change is too small to result in an isotope effect.
An important observation is that when you look at the ratio of two isotopes, of masses
m and m*, given that they both are embedded in the same equation, there is a cancellation of
all terms except mass (or reduced mass μ). Thus in the translational ratio, the magnitude of
the fractionation is governed by a power of 3/2 dependency, rotational by 1, and vibrational,
1/2. The same mass relation would hold for gravity. It is noted that symmetry numbers σ are
considered in rotational partition function calculations, but this term is cancelled out in many
cases and in their magnitude in energy are much less than vibrational energy. The role of
symmetry in isotope effects will be discussed later, especially in its role of reaction selectivity.
The result is that all of the variations are mass dependent. Geochemical reactions are governed
in large part by exchange, and thus a vibrational energy effect. The commonly used expression
of this from a geochemical perspective is a three-isotope plot to express a mass fractionation
line. In Figure 4 the mass fractionation line is shown for terrestrial and lunar samples and the
slope ca. 1/2 reflects the mass effect, with mass difference of 1 for δ17O and 2 for δ18O.
4.5
Terrestrial Fractionation Line
4.0
δ'17O (‰)
3.5
Figure 4. Triple oxygen isotope compositions of
3.0 basalt and glass from the Earth and Moon. Data
Lunar from Cano et al. (2020).
2.5
Terrestrial
2.0
4.0 5.0 6.0 7.0 8.0 9.0
δ'18O (‰)
42 Thiemens & Lin
As will be discussed in many Chapters of this volume, at the precision of present days’
measurements, the actual slope of a mass dependent fractionation line varies due to the small
variations in fractionation factors of many kinetic processes and the different masses of the
oxygen bearing species, ranging from the bare nuclides for 16O, 17O, 18O to large organic
species that may range into hundreds of amu (atomic mass unit). The advance in analytical
technique allows a better understanding of our Earth using mass dependent isotope effects
with 17O, which was not normally done some 20 years ago, but a first example of the use of
high precision triple oxygen isotopic analysis was measurements of atmospheric molecular
oxygen, which possesses a miniscule mass independent isotope component deriving from
the O3–O2–CO2 coupling with photosynthesis and primary productivity (Luz et al. 1999).
Model calculations further quantify that ~1/3 of the 17O deficit in O2 relative to the reference
water fractionation line are associated with mass independent reactions (Young et al. 2014).
The high precision definition of the slope in a multi-isotope system has also found one of its
most significant applications in sulfur that has been used to track the origin and evolution of
life. The ability to precisely recognize a UV driven mass independent process in the early earth
from the background biogeochemical signal was critical (Farquhar et al. 2000). Much of the
theoretical and experimental basis for the assignment of the source of anomalies derived from
work done for oxygen. Studies from both systems are now commonly used between them
to amplify understanding of the fundamental contributory processes. We will focus on mass
independent oxygen isotope effects in the rest of this chapter.
Mass independent isotope effects and applications
Traditionally only one isotope ratio had been used in isotope ratio measurements,
e.g., δ18O for natural samples, with δ17O not considered. One of the first recognitions that
mass independence could be used to identify other processes, specifically nuclear is from
sulfur isotope studies. Hulston and Thode (1965b) pointed out that physical and chemical
processes are all mass dependent as shown in Figure 3. Deviation from that relationship then
must result from a nuclear process and might be observed in meteoritic material. Sulfur has
four stable isotopes (32S, 33S, 34S, 36S) and with a mass range of 4 amu it has the potential to
be a uniquely precise and sensitive nuclear and chemical diagnostic tool. In the case of sulfur,
its chemical change will reflect the mass dependent processes relations. Such phenomena,
the relation δ33S/δ34S = 0.5, was shown for the first time by Hulston and Thode (1965b). The
relation δ36S/δ34S = 1.9 also defines mass fractionation lines. It was shown in the same year
that measurements of iron meteorites of long exposure age (> billion years) have excess
33
S and 36S produced by spallation reactions of high energy galactic cosmic rays with iron
(Hulston and Thode 1965a). The deviation from the mass dependent fractionation line then
is a direct measure of cosmic ray exposure, which was subsequently used in other iron (Gao
and Thiemens 1989), stony (Gao and Thiemens 1993b) and carbonaceous chondritic (Gao
and Thiemens 1993a) meteorites to better define the spallation yields of the nuclides, and the
grouping of the individual meteoritic groups.
The first observation of a mass independent isotope effect of oxygen was in the first
condensates in the early solar system, the calcium aluminum rich inclusions (CAI) in the
Allende meteorite (Clayton et al. 1973). In Figure 5 it may be seen that rather than the
mass dependent fractionation line of δ17O = 0.52δ18O (Fig. 4) expected from the application
of the different physical and chemical process of Figure 3 applied to a common terrestrial
environment, a relation of δ17O = δ18O is observed. As discussed by Clayton et al. (1973),
this unexpected deviation could only be explained in two ways: either as an equal simultaneous
alteration of 17O and 18O or the addition of pure 16O. Following (Hulston and Thode 1965a),
it was assumed that since chemistry is ruled out this must be nuclear, and consistent with
nucleosynthetic processes, particular supernovae injecting pure 16O.
Figure 5. Triple oxygen isotope compositions of (a) CAIs (normalized to a carbonaceous chondrite refer-
ence standard; Clayton et al. 1973), (b) O3 and residual O2 in O3 experiments (normalized with respect to
starting oxygen; Thiemens and Heidenreich 1983), (c) SiO2 in SiO + OH experiments [with respect to mass
dependent SiO2 (starting SiO and calculated O bulk)] (Chakraborty et al. 2013).
A decade later Thiemens and Heidenreich (1983) demonstrated that the simple process
of ozone formation from O2 dissociation produces the same slope observed by Clayton et al.
(1973) and was the first demonstration of a chemically produced mass independent isotope
effect in any element. In Figure 5b, it is observed that in the reaction, isotopic mass balance
is maintained and thus a seemingly simple chemistry is involved. Given that this was the first
mass independent fractionation, there was no theoretical basis for the effect. After careful
consideration of the experiments, supernova injection was eliminated. The conclusion of
Thiemens and Heidenreich (1983) was that there are two possibilities. First, it may be a
consequence of isotopic symmetry with equal 16O17O, 16O18O reaction rates versus 16O16O.
This recognizes that as discussed in the section on the discovery of oxygen isotopes (Giauque
and Johnston 1929a,b) there is a doubling of the asymmetric species in the rotational states
44 Thiemens & Lin
compared to symmetric and consequently that may be a relevant factor. Secondly, there could
be an effect of self-shielding in O2, as had been suggested for CO in interstellar molecular
cloud. As discussed, Cicerone and Mccrumb (1980) had already suggested that self-shielding
may occur in the earth’s atmosphere from self-shielding of O2 leading to ozone formation.
Stratospheric ozone isotopic measurements with ~400‰ 18O enrichments by Mauersberger
(1981) seemed consistent with this premise. Later reassessment of the data however led
to retraction of this stratospheric data (Mauersberger et al. 2001). The atmospheric ozone
hypothesis from self-shielding was later abandoned as it was found that the fate of the
anomalous oxygen atom is to undergo isotopic exchange between the product oxygen atom
and molecular oxygen removing any shielding anomaly.
For the nebular, self-shielding in CO to occur the δ17O = δ18O effect arises as a consequence
of the natural abundance of oxygen isotopes. Optical absorption occurs with a radial dependency
as a factor of e−σcl where σ is the absorption cross section, c the molecular number density
and l the path length. The product cl is the column density and is typically the unit used in
atmospheric and astrophysical sciences. In the most simplified and ideal case cross sections for
all isotopologues are the same but well separated at a given wavelength region and state (with
a constant incident radiation intensity for all species), photolysis rates of isotopologues only
depend on abundances of isotopologues. Since the abundance of 16O is orders of magnitude
higher than 17O and 18O, a δ17O = δ18O effect is obtained. In a more realistic case, the optical
depth becomes important (Lyons et al. 2014, 2020). It becomes opaque to light transmission in
a region of the column where 17O and 18O are not (Fig. 6). The results are that there is a limited
optical region where the δ17O = δ18O occurs because photons that can be absorbed by 16O are all
shielded and there is therefore no interaction between 16O and photons. The relation terminates
when 18O becomes opaque and there will be a region where pure 17O photolyzes followed by
no absorption at all. It should be noted that there is only a restricted optical region through
the column of gas where the slope 1 effect occurs. The radial dependency in self-shielding
produces a region where the δ17O = δ18O occurs, but it is an interior zone in a photochemical
column. If the CAI data (Fig. 5a) was explained by CO self-shielding, it is only this layer
where the required isotopic composition exists and must somehow be purely extracted from
the surrounding layers. This restriction is particularly important in evaluating the plausibility
of some self-shielding scenarios (e.g., molecular cloud chemistry and photochemistry). In the
isotopic cross section are not equal to each other the slope 1 will not occur anywhere.
In a later paper, Navon and Wasserburg (1985) modeled shielding of O2 in a nebular
environment and concluded that isotopic exchange may effectively remove the shielding
effect. They also discussed the potential role of CO in trapping the isotopic anomalies
and suggested experimental and theoretical studies of both carbon and oxygen isotopes in
CO self-shielding. The theory of isotopic self-shielding in CO was exhumed by Clayton
(2002) though with no new measurements or models. The processes associated with
understanding the mechanisms for the ultimate disproof/proof of this theory rest upon basic
chemical and quantum chemical physics. In the original theory it was assumed that there
was no chemical process that could produce a mass independent isotopic composition, and
ultimately experiment and theory provided new insight into mass independent isotope effects
via symmetry dependent reactions. In the case of self-shielding there is a basic assumption
that is no isotope effect in the dissociation process itself. Experiments have shown that the
assumption is incorrect and simultaneously shown that there is much to be learned about the
actual process of not only CO, but any gas phase molecule. We will come back to the detailed
experimental results and underlying chemical physics in the next section.
Figure 6. Schematic illustration showing the CO isotopic self-shielding process and variation of triple
oxygen isotope composition over path length. In the CO self-shielding model, it is assumed that absorption
lines of C16O, C17O and C18O are well separated and therefore the energies of photons these isotopologues
absorbed differ. The reaction rates of C16O, C17O and C18O are controlled by the availability of UV photons
absorbed by these isotopologues, which ultimately depends on the abundance of these isotopologues in the
column due to shielding. It is assumed that there is no isotope effect in the photodissociation process. UV
photons absorbed by C16O can only pass through a short distance due to the high number density of C16O
in the column. After that, only C17O and C18O can access required UV photons for photodissociation and
therefore the δ17O/δ18O = 1 slope occurs. At some distance further, C18O dissociation is restricted by C18O
shielding, leading to a region where pure C17O dissociation occurs. At the end there is an opaque region
of no dissociation at all. The distance shown in the schematic illustration is not to scale. The δ17O/δ18O = 1
slope zone is larger but still restricted to a specific optical region.
As the self-shielding theory cannot explain the O3 experiment results (Thiemens and
Heidenreich 1983), the other proposed mechanism in the original paper, symmetry-driven
isotopic fractionation becomes central. The reaction mechanism for the sequestering
of the isotope effect is of immediate importance for very many isotope geochemistry and
cosmochemistry applications, especially for synthesis of the meteoritic oxygen isotope
anomalies. For the source of the δ17O = δ18O isotopic anomaly in ozone formation and reactions
in general, a mechanism based upon chemical reaction dynamic theory was proposed by
Heidenreich and Thiemens (1986). There are a number of competing theoretical approaches,
but all ultimately are reliant upon the role of isotopic symmetry.
In a gas phase chemical reaction, there are three separate steps. First, reaction of an atom
with a molecule to ABC*:
A + BC ↔ ABC*→ ABC (R1)
The species ABC* is a short live species (10−13 sec) that must remove its excess energy or
re-dissociate back into A + BC or AB + C. This is accomplished by collision within a restricted
number of vibrations and most ABC* species re-dissociate. The result is that given the short
lifetime and number of collisions, within a short period of time there is a steady state is
established and the lifetime is driven by the number of pathways to stabilization. As discussed by
Heidenreich and Thiemens (1986), the number of states are the vibrational and rotational levels.
Taking oxygen isotopes as an example, the species could be 16OX16O, 16OX17O and 16OX18O,
where X denotes that this may be any element. At the isotope level, the numbers of vibrational
46 Thiemens & Lin
states are the same, but wavelength shifted. Meanwhile, there are doubling of the rotational
states for 16OX17O and 16OX18O. Heidenreich and Thiemens (1986) discusses the case of ozone.
The doubling of states and its origin has been detailed in a previous section on the discovery of
oxygen isotopes. The increased state enhances lifetime and in turn stabilization probability. The
result is an enhanced probability for isotopically asymmetric ozone to stabilize and the product
formation rate is based upon symmetry rather than mass, which produces a δ17O = δ18O effect.
There are three points of significance. First this is a general effect as oxygen predominantly
coordinates species. Second, the effect is based upon symmetry and lifetime of the excited state
and occurs for all gas phase reactions and not a specially ozone reaction feature. Finally, a most
important feature is that in a gas phase reaction such as (R1), during stabilization, even though
the oxygen atom may have a unique isotopic composition such as from shielding, the selectivity
and rapidity of the exchange reaction will reset the original composition. This chemical feature
will be developed and discussed further in ensuing sections.
In atmospheric chemistry it is well known that ozone is a major driving species controlling
the oxidation state of the atmosphere. Ozone chemical and photochemical reactions in the
stratosphere are drivers of a preponderance of oxidation reactions and its dissociation in the
stratosphere not only provides a UV screen for the troposphere the heat generation creates
the stability of the stratosphere and hence, a role in climate. Tropospheric ozone via its
dissociation and subsequent reaction of the product atomic oxygen with water provides the
hydroxyl radical that though low in its concentration (a million molecules per cm3) its high
reactivity mechanistically renders it a dominant oxidant species of reduced species. Ozone,
with its large mass isotopic anomaly is consequently identifiable in all atmospheric oxygen
bearing molecules and captures reaction source and process information that would not be
otherwise identifiable. Figure 7 is an updated triple oxygen isotope plot showing atmospheric
species acquiring signatures of mass independent fractionation effects. It is noted that there is
a miniscule mass independent isotope component in O2 as well as tropospheric CO2 resulted
Figure 7. Oxygen isotopic compositions of atmospheric MIF (Mass Independent Fractionation) molecular
species that have been measured to date. Note that ozone in the atmosphere has δ18O values that are twice
that of CAI and a D17O of approximately 48‰, also double CAI.
from stratospheric O3–O2–CO2 cycling and its magnitudes is mainly controlled by the fluxes of
stratosphere-to-troposphere transport, photosynthesis and primary productivity. The stability of
most of the ultimate oxidation products especially sulfate is generally high and consequently
of utility in probing many of the Earth’s major cycles (C, N, O, S), both present and past as
well as Mars and meteorites. The major discoveries in a wide range of geoscience research
fields using mass independent isotope effects and their applications have been reviewed earlier
(Thiemens 1999, 2006, 2012; Bao 2015; Crockford et al. 2019; Thiemens and Lin 2019).
Here we focus on existing advances and unsolved problems in fundamental chemical physics
of mass independent isotope effects and triple oxygen isotope systematics of the solar system.
In the next section we discuss the physical chemistry of mass independent oxygen reactions
in two broad categories and where they occur in the nature. This includes chemical reactions that
produce new products and photodissociation that break bonds. Both processes are prevalent in
the Earth’s and Martian atmospheres as well as the solar nebula. Each chapter of this book draws
upon these basic phenomena and, our ability to understand these effects at the basic level will
elevate our ability to interpret isotopic measurements. This is particularly so for photodissociation
where wavelength dependencies are known but cannot be modeled at an appropriate level.
FUNDAMENTAL CHEMICAL PHYSICS OF

MASS INDEPENDENT ISOTOPE EFFECTS:
WHAT IS KNOWN AND WHAT NEEDS TO BE KNOWN
Bond formation processes
The process of gas phase bond formation is one of extraordinary complexity. For
a simple reaction such as (R1), there is an array of parameters that must be included in
predicting the products at the isotopic level. These include the energy partitioning between
translation, vibration, rotation and electronic. Furthermore, angular distribution of collisions
as a function of their energy and angular approach cross section are important. The three-
dimensional geometry of the reaction and products hyper spherical potential energy surfaces
must be known and the dynamics are important. Many of the important processes occur on
femtosecond (10−15 seconds) time scales. The limit in computation is largely in the many body
problem, beyond H2, the number of interactions becomes computationally formidable, and
approximations are typically required. When the system of interest is at the isotopic level, the
difficulties become greater due to the requirement of higher resolution of all relevant processes.
For mass independent processes in chemical reaction theory, a chemical minireview
of mass independent chemistry by Thiemens and Lin (2019) has detailed the details of the
development of the chemical physics of mass independent chemistry. For the geochemical
application of mass independent oxygen chemistry to natural systems there are key points that
are significant, which will be thoroughly discussed in the ensuing sections.
The role of symmetry
The role of symmetry as compared to mass as a reaction control has led to a field of new
applications. As noted, the role of oxygen is important as it typically coordinates other species,
both in the gas phase and solid. Its role in the early solar system during condensation has been
predicted based upon this feature and Chakraborty et al. (2013) experimentally demonstrated
the reactions:
O + SiO → SiO2 (R2)
OH + SiO → SiO2 + H (R3)

48 Thiemens & Lin
Reactions (R2) and (R3) are isotopically symmetry dependent, an effect that appears
in the product OSiO and subject to the same selection process with δ17O/δ18O = 1 as in
ozone formation (Figs. 5c and 8). The role of symmetry also extends beyond oxygen and
includes sulfur (Fig. 8). Recognizing that another test of symmetry may be available in sulfur,
Bainssahota and Thiemens (1989) performed experiments to test this. It was shown for the
symmetry dependent reaction:
SF5 + SF5 → S2F10 (R4)
a purely symmetric molecule is created and a mass independent isotopic anomaly was observed.
The effect of energy and mechanism of creation of the reactive S2F10 was studied and it was
shown that the S2F10* state was responsible for the anomaly. Consequently, the symmetry effect
is not restricted to oxygen. Babikov (2017) has provided mechanistic details for atmospheric
generation of sulfur anomalies for a host of symmetry dependent reactions. These reactions
were not seriously considered in most of previous Archean sulfur studies, which were mainly
focused on photodissociation (Thiemens and Lin 2019). Recent models (Babikov et al. 2017;
Harman et al. 2018; Liu et al. 2019) have demonstrated that these symmetry dependent reactions
may be important in the Archean atmosphere. Most recently, Lin et al. (2018) employed a
5 sulfur isotope approach (4 stable sulfur and radioactive 35S isotopes) in sulfate aerosols and
revealed that the structure of the mass independent sulfur isotopes may require two distinct mass
independent processes, apart from mass dependent processes, to define both the Archean and
present day sulfur record. This includes both photochemical and potentially symmetry driven
sulfur reactions, echoing with theoretical predictions by Babikov (2017). A pilot experimental test
of symmetry-dependent isotope effects in elemental sulfur recombination reactions was carried
out by Lin and Thiemens (2020). A first hint that alternative reactions for mass independent
sulfur are needed was observed and discussed by Shaheen et al. (2014) who observed that the
largest mass independent sulfur anomaly in aerosol sulfate was in 1998–99 during a year of no
volcanoes and new mechanisms of production and sulfur sources are required. Additionally, it is
observed that there are very few years devoid of anomalous sulfur in the Earth’s atmosphere and
more understanding is needed to interpret the modern and Archean records.
Figure 8. Symmetry molecules that acquire MIF signatures during their formation and are experimentally
observed. Titanium is shown in Figure 10.
As discussed, the role of symmetry as a significant factor was included in the original
work (Thiemens and Heidenreich 1983) and specifically associated with the transition state
a few years later (Heidenreich and Thiemens 1986). To experimentally resolve the role of
symmetry, Yang and Epstein (1987a,b) performed an extensive set of experiments using
isotopically enriched (17O, 18O) of varying enrichment factors and reaction extent to define
the role of symmetry when the terminal enrichment is higher and also for enrichments
with double enrichments of normally minor species of molecular oxygen (17O18O, 18O18O).
The experiments were consistent with the conclusions of Heidenreich and Thiemens (1986)
that defined the transition state O3* as the source of the mass independent selection process.
In a series of follow up work of Yang and Epstein’s experiments, again using isotopic spikes
of the nuclidic species of 16O,17O, and 18O, nearly all of the isotopic variants were investigated
(Morton et al. 1989; Mauersberger et al. 1993, 1999; Wolf et al. 2000; Janssen et al. 2001).
In these works, the role of symmetry was underscored (Fig. 9). The fully symmetric
16 16 16
O O O (666) and 18O18O18O (888) possess similar rate constants as compared to the
asymmetric variants. Extremely low reaction rates (10−2 of 16O16O16O) were found in
symmetric 16O18O16O (686) and 18O16O18O (868). In general, the asymmetric isotope species
are all enriched compared to symmetric as reviewed by Thiemens and Lin (2019). The effect is
clearly seen when the rate constants for the individual isotopic species are measured.
Figure 9. The variation of ozone formation rates with respect to isotopic enrichment. 666 represents 16O16O16O.
Data from Janssen et al. (2001). Note the preference for asymmetric (A; red) vs symmetric (S; blue) species.
As will be discussed the data displayed above has been most important in developing a basic
physical chemical model for the symmetry dependent ozone effect. The experimental effect of
pressure (Morton et al. 1990; Thiemens and Jackson 1990) and temperature (Morton et al. 1990)
on the isotopic enrichment are key parameters in not only developing a fully quantum mechanical
understanding of the ozone symmetry effect, but most relevant for this paper is their application
in nature. The inverse isotope effect observed by Thiemens and Jackson (1990) is particularly
important. For example, in the case of the SiO + OH work an asymmetric product is observed
with equal 17,18O enrichment (Fig. 5c) (Chakraborty et al. 2013). The mass independent deviation
from the mass dependent fractionation line (conventionally expressed as Δ17O = δ17O − 0.52 δ18O,
with Δ17O = 0‰ for mass dependence) (Fig. 7) is measured to be 10‰ at near room temperature.
An inverse isotope effect produces a larger isotope effect at higher temperature rather than
e.g., in isotope exchange reactions are smaller. With increasing temperature, they converge on an
equilibrium exchange factor of 1. For the SiO reaction, under nebular condensation temperatures
(1700 K or higher) the Δ17O would be significantly higher and extend across the meteorites range.
The temperature effects on symmetry dependent isotope effects may also play an important role
in sulfur recombination reactions, which are strictly thermochemical gaseous reactions relevant
to many high temperature planetary processes such as impacts and volcano activities as discussed
by Lin and colleagues (Lin et al. 2018; Lin and Thiemens 2020).
50 Thiemens & Lin
Since the first discovery of the mass independent effect in ozone formation (Heidenreich
and Thiemens 1983), though deceivingly simple in appearance, the data shown in Figure 5b
has escaped a fully accountable theory. The paper of Heidenreich and Thiemens (1986) first
introduced the source of the effect as symmetry based, the short-lived transition state, state
density and number of states. Measurements of the rates of isotopically substituted species and
their pressure and temperature dependencies provide a quantitative basis for modeling efforts
along with the isotopic partitioning observed in Figure 5b. Work by Hathorn and Marcus (1999,
2000) provided the first models for these parameters that used the transition state stabilization
process and the number of states that exit, or couple to the stable ozone product. The limit, which
still exists, is that though the parameters above impressively account for most the isotopomeric
species, the model must include an empirical fit (an η value) to the magnitude of the isotope
effect. The theory is consistent with RRKM (Rice–Ramsperger–Kassel–Marcus) theory that
quantifies energy sharing in the transition state between rotational/vibrational states of the
isotopic species with a preference included due to state density (Gao and Marcus 2001).
The extraordinary difficulty of providing a full quantum mechanical treatment of the ozone
isotope effect is another fundamental problem to be solved. Babikov and his colleagues have
attempted at a near fully quantum mechanical treatment of the effect rather than an RRKM
statistical mechanical approach. In the first papers in this series, the difference in approach is
that Babikov et al. (2003a,b) employed the full three-dimensional surface treatment in the model
along with the difference in zero-point energy and also later work by Gao and Marcus (2001).
The results of the calculation show that there is resonance in the asymmetric metastable states
that are sparse in the symmetric. The use of the potential energy surfaces allows for decay of the
densely packed metastable states through three exit channels for O2 + O. With the advancement in
computational power, better three-dimensional potential energy surfaces may be determined, and
this was incorporated into a more sophisticated quantum mechanical approach to isotopic ozone
formation. In this treatment, the new parameters to be included are ZPE, scattering resonances,
and tunneling (Teplukhin and Babikov 2018a,b; Teplukhin et al. 2018). These most recent new
works take advantage of the new computational ability to calculate three-dimension potential
energy surfaces at much higher resolution and to define the resonances of the metastable
state. Shaped resonances below the centrifugal barrier and populated by tunneling have been
introduced as a new consideration in the generation of the isotope effect and mechanisms to
evaluate these properties described. What is clear is that the isotope community has facilitated in
the development of quantum mechanical processes that could not otherwise be detected, much
as in the case of the discovery of hydrogen and oxygen isotope. From a geochemical application
perspective, the new theories underscore that the isotope effect in ozone formation is general and
applies to other isotopic systems, including non-ozone species.
Besides oxygen and sulfur, a recent work by Robert et al. (2020) has extended the observed
species capable of a symmetry driven isotope effect to include titanium. The experiments were done
in a plasma (2450 MHz) of TiCl4/C5H12 in a flow system. The solid products of organic titanium rich
grains were collected from the quartz walls and measure with a NanoSIMS (Nanoscale Secondary
Ion Mass Spectrometer). The remarkable experiments show that there is a massive isotope effect
in 50Ti, with a mass independent component observed in the grains that extends from −200 to
1200‰ (Fig.10). It is concluded that the results are interpretable on a symmetry basis, either
via the symmetry based reaction recombination reaction, or, as a consequence of the scattering
reactions with a higher degree of scattering for the encounter between two indistinguishable
isotopes (i = j) and distinguishable (i ≠ j) where i and j are a given nuclide (Reinhardt and Robert
2018). This is a significant advancement not only in theory but also in deeper understanding of the
formation processes of the solar system and planets.
Figure 10. Mass independent fractionation of titanium isotopes in TiCl4/C5H12 plasma experiments. Data
from Robert et al. (2020). Note the slope one line.
Whether the oxygen isotopic composition of the Earth and Moon is the same is one of key
questions to be answered for constraining the impact energy of the giant impact model. In a recent
work, a mass independent character in lunar samples has been suggested by Cano et al. (2020).
By investigating the slight difference in Δ17O values among various mineral phases, they were
able to show that an anomalously depleted Δ17O reservoir may exist and contrary to commonly
held theory the oxygen isotopic composition of the moon and earth may not be the same (Fig. 11).
Specifically, secondary effects derived from the condensation of silicate cloud generated by the
impact of Theia produce the anomalous reservoir at the surface of lunar magma ocean. The mass
independent effects in SiO + OH reactions (Fig. 5c) (Chakraborty et al. 2013) are involved. In the
future, a more extensive set of lunar samples and enhanced statistics, plus further pushing the
precision of Δ17O measurements are vital in the pursuit of deeper details of lunar origin.
Figure 11. Δ17O values of different lunar rocks and the bulk silicate Earth. The predicted Δ17O values of Theia
and isotopically anomalously light silicate vapor is schematically shown. Modified from Cano et al. (2020).
For isotope effects associated with bond formation, as a consequence of symmetry, mass
independent oxygen isotope effects are observed in all atmospheric species and most of them are
derived ultimately from ozone. In the solar nebula, the condensation of the first solids are likely
recorders of the first process of nucleation and gas to particle formation as a consequence of the
coordination of oxygen. In this process due to the transition state fragility and selection rules,
the process is the record of the earliest events leading to the solids of the solar system and in the
moon forming event (Cano et al. 2020). The role of symmetry is a general phenomenon in gas
phase reactions and may be observed in other elements, including Titanium (Robert et al. 2020)
52 Thiemens & Lin
and sulfur effects (Bainssahota and Thiemens 1989) produced in the Archean (Babikov 2017;
Harman et al. 2018; Lin et al. 2018; Liu et al. 2019; Lin and Thiemens 2020). Though state-
of-the-art understanding of the chemical mechanism of mass independent isotope effects has
advanced considerably, the complexity of the calculations at state-of-the-art computation has
limits. For the future, models that can predict a priori what the effects would be would be a
major advancement and a plus up for guiding experiments that are difficult, especially for gas to
solid reactions of relevance in the solar nebula (Chakraborty et al. 2013) and applications in the
Archean and present day atmosphere (Lin et al. 2018; Lin and Thiemens 2020).
Bond breaking isotope effects
The Chapman cycle is a series of reactions accounting for the O3 production and
destruction in the stratosphere (R5–R8):
O2 + hn → 2O (R5)
O + O2 + M → O3 + M (R6)
O + O3 → 2O2 (R7)
O3 + hn → O2 + O(1D) (R8)
where hv represents a single photon of light and O(1D) represents an electronically excited state
energetically above the ground state atomic oxygen O(3P). It was not until the paper of Cicerone
and Mccrumb (1980) that the issue of stable isotope effects that occur during photodissociation
was quantitatively considered. The model was for self-shielding in the earth’s atmosphere by
O2. A recalculation of the 18O16O shielding effect was done line by line through the relevant
absorption region of the O2 absorption spectrum and shown that this level of detail is needed to
determine the precise photochemical dissociation coefficient Ji for the 32O2 and 34O2 molecules
(Blake et al. 1984; Omidvar and Frederick 1987). The coefficients vary by approximately two
orders of magnitude through the mesosphere and most of the stratosphere (Fig. 12). The results
between the two line by line calculations differ considerably and exemplify the difficult in the
modeling. It was shown however that in spite of the massive isotope effect, the rapidity of the
isotope exchange process removes the anomaly (Kaye and Strobel 1983).
Figure 12. Calculated profiles of odd-oxygen production rates from 16O16O (dashed line) and 18O16O
(solid lines; note that two different cross sections were employed) in the first consideration of isotopo-
logues photodissociation in the terrestrial atmosphere. Modified from Cicerone and Mccrumb (1980).
Though isotopic self-shielding of O2 does not allow the sequestration of an isotopic

anomaly in a stable species, the potential for a mass independent tracer in the stratosphere
from the photolysis of ozone does afford a possibility since it is known that the formation of
ozone through the oxygen recombination Reaction (R6) produces a large mass independent
isotopic composition. Though the early balloon-borne mass spectrometer measurement of a

massive 18O enrichment was an artifact, subsequent return sample measurements have shown that
stratospheric ozone is mass independent and consistent with experiments (e.g., Mauersberger
et al 2001). Photodissociation of ozone enriched in the heavy isotopes (17O16O16O, 18O16O16O
vs. 16O16O16O) from Reaction (R8) has the effect of producing a reservoir of isotopically
enriched, highly reactive 17O(1D) and 18O(1D) atoms. It was recognized by Yung et al. (1991)
that as a consequence of the high energy of the O(1D) atoms, the barrier to exchange with CO2
via the following reaction is lowered:
17,18
O(1D) + C16O2 ↔ 16O(1D) + C17,18O2 (R9)
The isotopic exchange reaction (R9) is rapid and the ozone isotopic anomaly is passed on to
CO2, which was observed by Thiemens et al. (1991) that stratospheric balloon collected CO2 is
mass independent in composition with a deviation from mass dependent components by greater
than 10‰ above 30 km. This was the first observation of a mass independent oxygen isotopic
composition in any non-meteorite sample. Most importantly, it was shown in later stratospheric
and mesospheric sampling using a cryogenically whole air sampler (CWAS) that the isotopic
composition of CO2 varies across the entire upper atmosphere (Thiemens et al. 1995). The work
also measured the concentration of N2O and CH4 in the same samples and an inverse relation is
observed with the δ18O of the CO2 (Thiemens et al. 1995). This relationship is a direct proof of
the involvement of the atomic oxygen atom as the isotope effect in (R6) results from the reaction
of the anomalous ozone derived atom and a branch of reactions between O(1D) and N2O and CH4:
O(1D) + CH4 ↔ OH + CH3 (R10)
O(1D) + N2O ↔ O2 + N2 (R11)
Reactions (R10) and (R11) are both sink reactions for removal of these significant
greenhouse gases.
A significant point for not only atmospheric chemistry is that one creates a range of
mass independent reservoirs in the same place and time that vary in the Δ17O. The anomaly is
transferred between different molecules in the stratosphere (Lyons 2001). A consequence is
that in e.g., the solar nebula, the isotopic composition between the various meteorite classes
in Δ17O in one model or another is thought to be a spatial difference which has always placed
major constraints on evolutionary models. In the case of shielding for example, the requirement
is that in the self-shielding depth character, there is only a restricted layer that produces the
required δ17O = δ18O effect (Fig. 6) and, stored immediately as a stable ice product to avoid
the back isotopic exchange that would remove the isotopic anomaly by the process modeled
by Navon and Wasserburg (1985). Somehow, in spite of its large UV absorption coefficient
for photo destruction, it must be transported across solar nebular without loss of its signature.
If, however there is a formation process that creates a steady state reservoir, mixing is not
necessary. Figure 7 shows that in a given cubic centimeter of air there are at least three oxygen
mass independent isotopic reservoirs simultaneously present (CO2, O2, and O3). It should be
noted that due to the scale, a counter negative Δ17O reservoir in O2 is not apparent as others.
The Δ17O value is smaller compared to others but in a reservoir that constitutes 20.95% of the
entire atmosphere. The nebular processes will be discussed later.
When the mass independent effect was first observed by Thiemens and Heidenreich
there was no laboratory-based experiments that quantified isotope effects at high precision.
Bhattacharya and Thiemens (1988) provided the first experimental observations of the
ozone photodissociation process, which was a small mass independent isotope effect with a
slight deviation from mass dependence. Chakraborty and Bhattacharya (2003) were able to
54 Thiemens & Lin
demonstrate that the effect varies with wavelength and is mass dependent in the visible and mass
independent in the UV actinic regions. In the experiments, there are also competing processes
that are simultaneous processes that need to be separated. First there is the primary interest of
ozone photolysis reaction (R8), but it is confounded by the effects of the reaction with atomic
oxygen (R7). To calculate the isotope effects from modeling of spectral data is highly complex
and a semi analytical approach was developed by Liang et al. (2004) for simple molecules and
was later applied specifically to ozone photolysis and stratospheric chemistry (Miller et al. 2005;
Liang et al. 2006). The difficulty of both the experiments and theories was discussed by Cole
and Boering (2006). The next level down in isotopic resolution was to model the dissociation
pathways during ozone photolysis of the isotopomers and their electronic branching for
asymmetric vs symmetry effects (Ndengue et al. 2012, 2014). For applications in the Antarctic
and for resolution of the physical chemical processes, the ozone isotope effects for ozone
trapped in ice were measured (Bahou et al. 1997). The work was of significant for applications
in nature as it resolved the isotopic perturbations of absorption properties in ice (such as polar
work) and the rates at which a species trapped in ice can photodissociate. The advantage of
solid matrix photochemical experiments such as this described by Bahou et al. (1997) is that
following the photolysis the products are trapped consequently avoiding the problem discussed
above of complications arising from secondary reactions. McCabe et al. (2005) reported isotopic
measurements of nitrate in ice photodecomposition to correct values reported in South Pole
nitrate (McCabe et al. 2007). Besides interpretation of the decades long nitrate oxygen isotopes
South Pole snow pit samples, the natural and laboratory ice photodissociation experiments are
relevant today in the troposphere and polar regions as well as ice as a reservoir in the early
solar system. In the self-shielding models, it is assumed that there is no photolysis of ice or
interference from other molecules trapped in the ice or coated on surfaces.
The photodissociation process of ozone is clearly complex and an up to date analysis of
the isotopic fractionation processes is provided by Huang et al. (2019). The complexity of
detailing a fully comprehensive detail of the bond breakage remains a forefront of physical
chemistry problem that increases in resolution with time. In the case of ozone, Wen and
Thiemens (1990, 1991) adopted an approach that was thermal as compared to photochemical
to compare the results of the differential mechanism of an O–O2 bond breakage. In a thermal
dissociation process the mechanism of energy acquisition leading to the bond scission differ.
In the thermal case there is an energy rise form the thermal Boltzmann driven molecular
ensemble which raises the energy levels up to dissociation. The energy jumps are quantized
as they are rotational states and obey quantum mechanical properties for their population and
transfer. The mechanism of energy transfers and notably the transition from quantized energy
states to dissociate involves selection rules. In these works, the simple thermal dissociation
of the ozone molecule was studied by thermally dissociation at 90 and 110 °C following the
reaction yield and isotope results as a function of time (Fig. 13)
The results are not that expected for a thermal dissociation process. The decomposition
process produces a slope one compared to 1/2 expected for a thermal bond breakage, and the
product O2 is enriched in 17O, 18O, the reverse of the value expected from conventional kinetic
isotope effects. In analysis of the data, unimolecular compositional theory of Troe (1977) was
applied and incorporated isotopic kinetic modeling developed by Kaye (1986). The unimolecular
theory was that for low pressure steady state energy transfer. The vibrational energy transfer
is included in molecules of variable densities of state. Such considerations also include a
variable height of the centrifugal barrier in determination of the transition probabilities. Also
considered are the low-pressure limit of energy transfer and the non-equilibrium population
of excited states and weak energy transfers. In the unimolecular decomposition theory, one
predicts a mass dependent process with an expected 1/2 slope. Given that thermal energy
transfer processes are common in natural systems and bond breakage, for example mineral
formation and equilibration, one does not see this reversal in isotope effect. What is needed is
Figure 13. Triple oxygen isotope compositions of O2 and residual O3 in the thermal dissociation experi-
ments (Wen and Thiemens 1991). The δ17O = δ18O effect is observed but completely reversed from that of
ozone formation as shown in Figure 5b.
a model such as that of Babikov and colleagues designed to study the isotopic partitioning at
the quantum level and the concomitant fractionation processes and determine the microscopic
reversibility. This is another example where isotopic measurements exceed what is quantum
mechanically computationally possible, but potentially tractable with suitable approximations.
Research into the subject of isotope effects during a gas phase thermal decomposition are of
potential application for atmospheres, e.g., the atmosphere of Venus at a surface temperature
of 467 °C with the presence of gas phase species such as sulfur oxide and likely oxygen radical
intermediates is of relevance. Further investigations into the underlying chemical physics
would add insight into the isotopic selectivity, energy dependence, and quantum dynamics
of reactions that occur into and out of the hyper spherical surfaces of ozone and by extension
other molecules. These are facets that are not adequately known and needed at present.
Physical chemical details of photodissociation general process
The intricacies of photodissociation are complex and there exist many mechanisms by
which it occurs. There are many varieties of photochemistry and photoionization and not all
need be discussed here. A review of a subset was presented by Thiemens et al. (2012). Many
of the photodissociation effects applied in planetary and astrophysical environments are of a
specific type of pre-dissociation. For example, relevant molecules for multi-isotope systems
where mass independent isotope effects may occur in pre-dissociation include O2, O3, CO2,
CO, SO2, and OCS. For these reactions, the process in general is outlined below.
In Figure 14, a general schematic of photodissociation is shown. The y-axis is the
potential energy and x-axis is the internuclear distance between atoms. The ground state of
the molecule (the lowest state in vibration and rotation) is shown and its energy is labeled as
ZPE. The vibrational levels differ for the different energies of the molecules for the isotopic
species as discussed in detail. The higher mass is a lower frequency and stronger energy and
thus for a given vibrational line on a given state, the heavy isotope will be lower in potential
energy. At the top of the ground state potential energy curve, the asymptote of the curve is the
dissociation energy (De) of the molecule of interest. It is important to note that the distance
from the ZPE to De is the energy of bond breakage, but it is not photochemically achieved
by climbing states to De. In Figure 14a, the process in a simple system occurs by absorption
at that energy to an excited state molecule, which reflects a less stable, different electronic
configuration of the ground state molecule. They have spectroscopic notation that define
this by the electronic configuration and e.g., if they are single, doublet of triplet. The ground
states upon absorption at a given wavelength transits to the excited state, which also possesses
vibrational and rotational levels with isotopic lines. The energy associated with this is generally
driven by the transition dipole moment and it occurs at a given inter atomic separation. The
excitation, as the vertical upward arrow in the figure must cross the potential energy surface
56 Thiemens & Lin
Figure 14. Schematic potential energy curves of ground, excited, and repulsive states and pathways of
predissociation. (a) Simple predissociation with only one excited state and repulsive curve; (b) Indirect
predissociation with ground state exciting to an excited state, crossing to a second state then repulsive
state. Modified from Heays et al. (2017).
of the excited state. The probability of intersection is determining by Frank Condon factors. If
the ground and excite states lie one above another exactly at the R distance of the well of the
curves, this is the greatest Frank Condon Factor region. Figure 14b shows that potential energy
surfaces of different excited electronic states may cross each other. There is a high probability
that a non-radiative transition between these states occurs if their vibrational levels overlap.
This is a non-adiabatic process. An isotopically substituted molecule possesses slightly different
vibrational energy levels compared to the original molecule because of the change of mass.
The shift of vibrational levels caused by isotopes may significantly change the probability of
resonance between two different electronic states as the intersystem crossing is highly sensitive
to positions of vibrational energy levels. If the energy state transits to a repulsive potential
energy surface (as shown by curved arrows in Fig. 14), the molecule normally dissociates and
an isotope effect can be observed in photodissociation products because of different transition
probabilities, i.e., photodissociation rates. A small change of mass in isotopically substituted
molecules may lead to a large isotope effect, which is difficult to precisely predict due to
uncertainties in calculating the vibrational energy levels at isotopic levels.
Specific examples of isotope effects in dissociation: carbon dioxide
Carbon dioxide is the most abundant molecule in the Martian atmosphere and given
the thin optical opacity, most of the atmosphere is exposed to dissociative UV energies.
The dissociation of N2 into atomic nitrogen in part facilitates its escape from the gravitational
energy well and fractionates the residual nitrogen in the atmosphere, resulting in a high
15
N/14N ratio. In the Earth’s atmosphere, N2 photolysis does not occur until an altitude of
approximately 120 km and peaks at 175 km, providing a source of atomic nitrogen to the
thermosphere (Thiemens et al. 2012). In the Martian atmosphere, CO2 photolysis is significant
and consequently of interest in interpreting the Martian atmospheric record, present and past.
Reconstructing atmospheric CO2 concentration in the early Mars and the escape rate of carbon
and especially the UV-driven photochemical product O atom to space (Tian et al. 2009) via
isotopic measurements in the Martian atmosphere and regolith is a necessary component
in understanding the climate, habitability, and potential existence of life in Noachian Mars.
Carbonates and sulfates from Martian meteorites are known to possess mass independent
oxygen isotopic compositions that reflect the ozone, water CO2 cycles and reflect the evolution
of those interactions over time (Farquhar et al. 1998; Farquhar and Thiemens 2000; Shaheen
et al. 2015). Terrestrial atmospheric carbonates are also known to possess a mass independent
isotopic composition that includes the heterogeneous chemistry of ozone, CO2 and water in the
10 nm surface film of the aerosols (Shaheen et al. 2010).
In the Martian atmosphere, CO2 is exposed to UV light and subject to photodissociation.

The necessary O3–O2 photochemical isotopic parameters for modeling the Martian
atmosphere are reasonably well known as discussed in this paper and other reviews (Thiemens
2006; Thiemens and Lin 2019). The fractionation factor associated with CO2 at a relevant
UV wavelength was reported by Bhattacharya et al. (2000). In this work, line dissociations at
185 and 254 nm and continuum at 120–160 nm were measured for the isotopic fractionation
factor. The different wavelengths allow for exploration of different electronic states to be
identified. Ideally one would measure nm by nm across the absorption spectrum, but laboratory
experiments are restricted to a small group of light sources, none of which are tunable across
wavelength. Synchrotron radiation is the only tunable UV source which requires that the entire
separation and collection apparatus must be constructed at a beam line and disassembled
following allotted time sources. At shorter wavelengths, such as for CO dissociation, there
are no windows that are transparent and windowless experiments must be done involving
differential pumping while allowing sample collection. Consequently, most isotope chemistry
research is done at select lines from laboratory emission or continuum sources. Figure 15a
shows the relevant electronic states that are accessed in the experiments (Bhattacharya et al.
2000). The ground state of the molecule is linear and a singlet (1Σg + ) with three vibrational
modes. A given wavelength utilized in the experiments is shown by the upward arrow in the
figure. This shows what energy of the excited state molecules is accessed and most probable.
For example, at 184.9 nm, the Franck Condon region of excitation intersects favorable with the
1
B2 state but minimally with the 3B2 state. The CO2 molecule has different vibrational modes
and Figure 15a represents a cut through the three-dimensional surface.
The experimental results of Bhattacharya et al. (2000) (Fig. 15b) are striking for several
reasons. First, there is a clear wavelength dependency. For the Kr-continuum source lamp
(120–160 nm) the effect is mass dependent with a depletion in δ17O and δ18O with respect to
the makeup CO2 gas by 50‰ on average for the product O2 and CO. Secondly, at 184.9 nm,
Figure 15. (A) Schematic potential energy surfaces of ground and selected excited states of CO2 that are
relevant to CO2 photolysis experiments by Bhattacharya et al. (2000). Modified from Mahata and Bhat-
tacharya (2009a). (B) Triple oxygen isotope compositions of photolysis products in the experiment. Modi-
fied from Bhattacharya et al. (2000). In Panel B, the remarkable observation is the change from the open
balloon Hg lamp to solid mercury lamp in oxygen isotopes by only variation of the 13C abundance.
58 Thiemens & Lin
there is essentially a pure 17O effect, with a δ17O enrichment of between 90 and 140‰ and none
in δ18O. At 184.9 nm, as shown in Figure 15a, the excitation is to a lower vibrational energy
level of the 1B2 state. This energy level crosses to the 3B2 state (see the shaded circle in Fig. 15a)
and is above the dissociation energy level of the triplet state, consequently photodissociation
occurs, though it is a forbidden crossing. In such crossing the probability of transition may be
calculated, though it is not achievable at the isotopic level for a tri-atomic molecule due to the
requirement of knowing the three-dimensional potential energy surfaces at high resolution. In
general, and as shown by the Babikov models for ozone formation, there are resonance between
excited states, and they derive by the overlap in energy. In each state (1B2, 3B2) these states are
primarily the vibrational-rotational levels. If the overlap they cross and in this case dissociate,
the wavelength difference must be very small (less than 1 cm−1). The first calculation for this
effect was done for N2 by Muskatel et al. (2011) and this sensitivity was shown in detail for the
state crossings. The larger dependence of wavelength is shown in Figure 15b as the 123.6 nm
region experiments show mass dependence instead of the 17O enrichment at 184.9 nm.
If the crossing is the source of the isotopic effect, there should be a highly resolved
wavelength dependence. Rather than altering the light source in which the absorption features
are different, the same 184.9 nm experiments were duplicated with the only difference being
that 13C-enriched CO2 (99%) was used. The only difference in the experiments was the shift in
vibrational frequency from predominantly 12CO2 to 13CO2. In Figure 15b it is observed that the
isotope effect is enormous; from a pure 17O effect of 100‰ (Δ17O) to a mass dependent change
of ca. 150‰ (δ18O) at Δ17O = 0. This clearly shows that the crossing is the source of the effect
and a very sensitive effect of vibrational states. It may not be argued that there is a secondary
effect of ozone formation as this large difference was achieved only by the change in carbon
isotopic abundance. Such isotopically sensitive inter-system crossing revealed by experiments
using isotopically enriched molecules is also observed in quadruple sulfur isotope systematics
such as CS2 polymerization reactions (Zmolek et al. 1999) and SO2 photolysis reactions (Franz
et al. 2013) and recently discussed by Thiemens and Lin (2019).
As shown in Figure 3, nuclear spin energies are small and conventionally negligible
in contributing to chemical energy. For isotopes possessing a nuclear spin, magnetic, or
hyperfine, coupling between unpaired electrons and magnetic nuclei in paramagnetic species
such as radicals may spin-flip the radicals and allow the reaction between spin-allowed and
spin-forbidden channels and produce a magnetically modulated isotope effect (Buchachenko
2001, 2013, 2018). Bhattacharya et al. (2000) also suggested that given that the enrichment is
in 17O which possess a nuclear spin the effect may be a nuclear spin, or hyperfine enrichment
process. Mahata and Bhattacharya (2009a) developed a set of experiments over a wider range
of experimental parameters and suggest that the effect is likely hyperfine induced. In the
experiments the carbon and oxygen isotopic compositions of the products were measured
and the enrichment ratio of 17O/13C is 2.2 ± 0.2, which is attributed to the ratio of the nuclear
spins of oxygen/carbon and their g-factors (dimensionless magnetic moments). In a follow
up experiment (Mahata and Bhattacharya 2009b) the temperature relation for the process
was studied and a non-mass dependent isotopic distribution with increase in temperature is
observed and attributed to a change associated with the transition dipole moment and change
in Boltzmann distribution, with an underlying hyperfine effect.
In a full first principles approach by Schmidt et al. (2013) the isotopic structure and
dissociative processes in the 150–210 nm were modeled. The success of the model is observed
in the close agreement with independent measurements of the cross sections. The structure
of the molecule is noted from the involvement in the geometry change of the bending motion
of the molecule and linkage from spin-orbit coupling to the deep wells at bent geometries in
the 21A′′and 11A′′ potential energy surfaces connections to the dissociative state. The model
compares to the isotopic measurements and potentially a detailed model to explain the data.
As discussed by Bhattacharya et al. (2000) the CO2 photodissociation experiments and

massive isotope effect observed at 184.9 nm may be operative in the Martian atmosphere. Later
measurements of carbonates and sulfates from Martian SNC meteorites demonstrated that the
interactions of the atmosphere and surface of Mars sequester a record of these interactions. After
Thiemens and Heidenreich (1983)’s suggestion that the ozone δ17O = δ18O effect is responsible for
the CAI isotopic compositions this is the first suggestions of mechanistically how an atmospheric
(or nebular) chemically produced mass independent signal may be transferred to solids. As in
the case of the discovery of isotopes and isotope effects in nature, the ultimate success of these
applications resides in the development of deeper understanding of the basic physical chemistry.
PHOTODISSOCIATION IN THE EARLY SOLAR SYSTEM AND

SELF-SHIELDING MODELS
Triple oxygen isotope composition in CAI is a window for understanding the formation
and evolution of the early solar system and planets. Thiemens and Heidenreich (1983)
discussed the mechanisms by which a chemically produced mass independent isotope effect
may arise. One was based upon symmetry influences, which has ultimately proven to be a
commonly accepted basis. Secondly, they suggested that it could also arise from self-shielding
in the solar nebula as an oxygen isotope abundance produces a δ17O = δ18O composition due
to the large difference in abundance of 16O and 17,18O (Fig. 6). A basis was suggested that
self-shielding had reputably been observed in CO in interstellar molecular clouds and that
during a T-Tauri phase of the sun, UV radiation may be enhanced by a factor of 104. It was
also suggested that self-shielding in N2 and H2 could produce the high 15N, 2H enrichments
observed in chondritic meteorites. Models for the early solar system by Navon and Wasserburg
(1985) modeled the fate of the oxygen atom post self-shielding. As a rapidity of the radical
exchange, the isotope effect produced in the oxygen product was removed, which is consistent
in the case of ozone and the effect arising from the gas phase symmetry driven chemical
reaction. This is an important fatal point for self-shielding, irrespective of self-shielding is that
the fate of the product, be it isotope exchange or the subsequent chemical reaction forming a
solid, or destruction of that product must be fully considered at an isotopic level. As discussed
in the previous section CO2 is an example of how the dissociation process itself gives rise to a
major isotope effect. Each of these limits is discussed in the ensuing discussion.
Self-shielding models
Nearly 20 years after the suggestion of self-shielding in the nebula, Clayton (2002)
resurrected the self-shielding model and in a commentary suggested self-shielding as
the source of the CAI oxygen isotopic anomalies. The basis for CO self-shielding was the
observation of self-shielding in interstellar molecular clouds and an active T-Tauri phase sun
as discussed in Thiemens and Heidenreich (1983). The paper however did not address the
issue of how the signal was not erased by isotopic exchange as the fate of the product oxygen
was not considered. It was suggested that as known in the astrophysics community, for self-
shielding to occur one needs to occur, the 12CO and 13CO abortion lines are separated by
45 cm−1 allowing differential absorption due to the large difference in the isotopic abundances
of approximately 100:1. For CO this is known to occur at 105 nm, but not at other wavelengths
as they either do not have differential absorption or the relevant UV region is opaque due to
H2 absorption. This also assumes that all absorbed light leads to dissociation and there is no
leakage back to the ground state (or a quantum yield of 1).
To circumvent the issue of isotope exchange, shielding was suggested as taking place on
the edges of the solar nebular and past 10 AU where temperatures are sufficiently low that the
product 18O,17O atoms are frozen as water ice and the fate of isotopic exchange presumably
avoided (Yurimoto and Kuramoto 2004; Lyons and Young 2005; Sakamoto et al. 2007;
60 Thiemens & Lin
Young 2007a,b). In these models, CO self-shielding could have occurred at the inner annulus or
surface of the solar protoplanetary disk, or the molecular cloud prior to collapsing to form the
solar protoplanetary disk. The UV source may also have originated from outside the nebula and
clustered massive star formation in nearby regions or the sun itself. The ice created in the outer
edges of the solar system is then transported intact where it is transported intact to the inner solar
system where the water is chemically converted to meteorites of varying isotopic composition,
from 16O-delepted new-PCP (poorly characterized phase) in Acfer-094 to 16O-enriched CAI
and Acfer-214 a006 inclusion (Fig. 16). As discussed by Lyons et al. (2009) the mechanism
by which the anomaly is safely transferred and preserved in a silicate is not known or if it is
possible. The occurrence of aqueous alteration in carbonaceous chondrites is well known (e.g.,
Clayton and Mayeda 1984; Young 2001; Benedix et al. 2003; Airieau et al. 2005; Tyra et al.
2007) but there remain uncertainties in the detailed mechanisms and concomitant isotope effects.
The unaddressed problems with self-shielding are partitioned into three areas, all based on the
basic physical chemistry of the isotope effects. These include: (1) the CO photodissociation
process itself; (2) unfractionated storage in water ice and its isotopic stability; (3) conversion of
water ice after transport through a turbulent nebula to mineral phases.
Figure 16. Oxygen isotopic compositions of MIF species in the solar system that have been measured to date.
Mass independent isotope effects in CO photodissociation

The isotope effects in CO photodissociation when the models were proposed were
unknown. There is a basic assumption in self-shielding models that there is no additional
isotope effect in photodissociation and the isotopic fractionation in self-shielding is only linked
to the abundance of isotopes. To specifically examine this issue, Chakraborty and colleagues
(Chakraborty et al. 2008, 2012, 2018) experimentally tested the model directly. These are
the only experiments that directly test self-shielding. In self-shielding models (Clayton 2002;
Yurimoto and Kuramoto 2004; Lyons and Young 2005; Sakamoto et al. 2007; Young 2007a,b)
and radio astronomical measurements of CO in interstellar molecular clouds (Vandishoeck
and Black 1988), it is known that for self-shielding to occur, it is restricted to the absorption
region around 105 nm. At this wavelength there is the 12,13C16,17,18O2 line spacing available
for differential isotopic absorption to occur (Fig. 6) and a hydrogen shielding window.
In addition, there is a window in H2 absorption available for photolysis. Therefore, at 105 nm
and that region, self-shielding is potentially viable. At 107 nm however, the isotopic lines are
only separated by 1 cm−1 and 13C16O, 12C17O, 12C18O and overlap (Vandishoeck and Black
1988) and may not self-shield and consequently must be distinct in their isotopic fractionation
products compared to 105 nm. Figure 17 is the CO potential energy surface diagram showing
various excited states and relevant wavelengths. The CO photodissociation experiments
were carried out using synchrotron radiation at the Advanced Light Source at the Lawrence
Berkeley National Laboratory. At these wavelengths light is opaque to transmission through
widows and a windowless photocell is required. The differentially pumped system designed
for this is described in detail by Chakraborty et al. (2008). A flow system is employed to
maintain an infinite reservoir of CO and there is also a cryogenic system to trap the products
for transport the products is needed. The results across six different wavelengths and three
different temperatures in a series of CO photodissociation experiments are summarized in
Figure 18 for the product atomic oxygen.
The CO photodissociation experiment results are clear. The 105.17 and 107.61 nm results
lie along the same slopes (Figs. 18a,b). At 107.61 nm self-shielding is not possible as the
isotope lines 12C16O, 12C17O, 12C18O overlap. The beam profile determination also allows
resolution of the 107.61 vs 105.17 nm dependence. The identical isotopic composition of the
products for the two wavelengths rules out self-shielding. A second important aspect is that the
slope of the products (Fig. 18) is not the required δ17O/δ18O = 1 slope for either the meteorites
(Figs. 5a and 16) or self-shielding models (Fig. 6) and is very reproducibly so.
The temperature dependence was further probed with a new photolysis system that allows
access to lower temperatures (80K) (Chakraborty et al. 2016; Chakraborty et al. 2018). In the
experiments, careful measurement of the beam profile was done so that the isotope effect and
the states accessed are assigned at high specificity. Figure 18 also shows that there is a massive
variation with temperature and varies with electronic state by more than 1000‰ for Δ17O and
as high as nearly 4000‰ for δ18O. The very strong temperature suggests that the involvement
of rotational states involved as both the line width and rotational state population are strong
functions of temperature.
Figure 17. Schematic potential energy surfaces of CO excited states relevant to CO photolysis experiments.
Modified from Chakraborty et al. (2008, 2012).
62 Thiemens & Lin
Figure 18. A summary of triple oxygen composition of product atomic O in CO photodissociation experiments.
Wavelength and temperature dependencies are shown. Data obtained from Chakraborty et al. (2008, 2012,
2018). Based on the self-shielding theory and isotopologue cross-sections, a shielding effect should be proven in
105.17 nm (panel A) but not in the 107.61 experiments (panel B). It is however observed that 107.61 nm experi-
ments exhibit the same behavior as in 105.17 nm experiments, inconsistent with an origin in shielding.
The experimental data have further shown the effect of wavelength on δ17O/δ18O and Δ17O
(Fig. 18). The wavelength-dependent variation of the photolytic isotopic products is apparent
and not explainable by self-shielding. In fact, the data is consistent with the isotope effect in CO
dissociation being driven by the process of dissociation post absorption and not light filtration.
The comparison of the 105.17 and 107.71 nm experiments outlined above is consistent with this.
As will be discussed, the use of cross sections only does not capture the isotope effect as it does
not reveal the factors that influence the processes and are dominated by the overall isotopically
selective bond breakage features captured by the photolytic bond breakage.
Another evidence that self-shielding does not dominate the mass independent isotope
effects in CO photodissociation experiments comes from the pressure dependency. The process
of self-shielding is driven by the column density of the gas molecules and for astrophysical
applications of self-shielding it is the column density that is most significant (Vandishoeck and
Black 1988). The column density (cl) in an absorption process is the local number of molecules
per cm3 (molecule number density, c) times the light optical path length (l; unit: cm). The light
absorbance is exponentially dependent of this feature and is expressed as:
I
 e  cl (5)
I0
where I0 is the initial photon flux of the light source, I is the photon flux at a given distance
l, and σ is the cross section (cm2). For the different isotopically substituted molecules the I/I0
vary differently as they are a function of the isotopic abundances of 16O, 17O and 18O and are
incorporated in the c term above and shown in Figure 6. The variation in the isotope with
pressure then should capture this effect of shield as a linear change in pressure is expressed
in the exponent for the isotopic species. There are disagreeing comments on the first
CO photodissociation experiment (Chakraborty et al. 2008), pointing out that the deviation of
δ17O/δ18O from unity might be an optical depth effect (Federman and Young 2009; Lyons et
al. 2009b; Yin et al. 2009) and were answered in Chakraborty et al. (2009). In the follow-up
CO experiments, the pressure was varied over two orders of magnitude and as a function of
temperature and wavelength and the oxygen and carbon isotopes were measured (Chakraborty
et al. 2012, 2018). The experimental pressures are done for those that allow the windowless
experiments that are differentially pumped. If self-shielding could have dominated the isotopic
fractionation in the experiments, an increasing δ17O/δ18O slope along with pressure (i.e., column
density) was expected. It is however observed that over a large pressure range the δ17O/δ18O
values are consistent or varying in a way different the prediction of the self-shielding model
(Fig. 19). In addition, they are not a value of 1 required for meteorites. There is no pressure
dependency observed over nearly 2 orders of magnitude pressure range as expected from
shielding and the exponential dependency upon pressure given in Equation (5) and illustrated
in Figure 6. Again, the shielding (105.17 nm) and non-shielding (107.61 nm) wavelength gave
the same slope which rules out shielding as it does not occur. The results cannot be explained
as an overlap of states within beam as the beam profile is measured, reported and accounted
for. Following the Clayton (2001) paper re- suggesting self-shielding of CO, these are the only
experiments that directly measure the isotope effect in CO dissociation.
A final problem for self-shielding is also experimentally shown by Chakraborty et
al. (2018). For the first time the 13C of the carbon monoxide product of the photolysis was
measured. As discussed and shown in model calculations of self-shielding (Chakraborty et
al. 2018), experimental results for the experiments are not that expected for self-shielding.
The calculations were done at wavelengths where the isotopic cross sections are known.
The problem of the overlap of discrete CO absorption bands within the beam (full width at half
maximum: ca. 2 nm) raised by previous comments (Chakraborty et al. 2009; Federman and
Young 2009; Lyons et al. 2009b; Yin et al. 2009) are carefully considered in the experiment
and calculation. For carbon, the difference is as much as 3000‰ in δ13C! For oxygen,
Figure 19. Variation of δ17O/δ18O in different pressures (column densities) in CO photodissociation experi-
ments. Data obtained from Chakraborty et al. (2012, 2018). The unshielded wavelength (107.61 nm; Panel A)
and shielded (105.17nm; Panel B) produced similar slopes over a factor of e−5 difference in column density.
64 Thiemens & Lin
the difference in δ18O is more than 1000‰. Their variation patterns (especially for δ18O) along
with wavelength also differ. For the self-shielding models for the meteorites, a 10‰ error in
isotopic variation will not account for the data and consequently the calculations themselves
have an inherent error bar that has never been adequately discussed. It arises in large part from
the assumptions that the cross section and light filtration accounts for all of the isotope variation,
which is not substantiated. The second major problem which gives rise to more than a 1000‰
variation error is that the cross-section measurements at the isotope level vary the modeling
results by this much simply because of the precision and accuracy of the ability to measure them.
For the self-shielding of carbon monoxide all parameters associated with photodissociation
have been measured. The pressure, wavelength, temperature effects do not agree with self-
shielding models, the only parameters that are variable. In the next section, further experimental
and modeling results of relevance will be given that also show that the process cannot be
applied to the solar nebula.
High resolution view of isotopic photodissociation
The foregoing section details the process of photodissociation of from experimental
CO, O3, CO2 results and models for dissociation. The process of dissociation is complex as
schematically illustrated in Figure 14. The process of the actual breaking of the bond following
the intimal absorption of light involves a series of isotope selective process. The consequence
if there is a filtration effect of the light source leading to absorption of light, this leads in
most cases an excitation to another electronic state. The selection of the dissociation isotopic
products occurs after this process. A consequence is that for models that utilize cross sections
to calculate the dissociation do not incorporate these highly selective processes that are the
largest measured isotope effects known. To quantum mechanically incorporate all of the
processes shown in Figure 14, each step in the nuclear, electronic and time effects must be
included and incorporated. The difficulty at the isotope level is that all of the spectroscopic
factors and the potential energy surfaces must be known at high precision. For the molecules
of interest and that have been measured, these have not been measured and consequently
cannot be measured. For CO for example, the cross sections for several isotopic species have
not been measured and must be calculated for some. The real problem however is that the
cross sections do not incorporate the isotope effect as will be shown. It is possible that a unity
δ17O/δ18O slope may be obtained in a complex system where all processes are mixed together
(in solar nebular or laboratory chamber photochemical studies), either by self-shielding or by a
combination of various isotope effects. However, the CO photodissociation experiments imply
that this is likely not be the case. We must understand the chemistry at an isotopic level in each
step otherwise all modeling efforts embed large uncertainties.
To further our ability to interpret isotope results for photodissociation, the problem of
the requirement of a suitable quantum mechanical model must be overcome. In modeling
many kinetic atmospheric reactions, the rate constants are needed for more than 400 reactions.
It is often the case that these are not all known and one then adopts a rate constant based upon
another reaction that is demonstrably sufficiently similar that its rate constant may be used.
For the consideration of CO dissociation, N2 is isoeletronically the same as CO (Fig. 20) and
all of its spectroscopic features and isotope lines have been measured. It is an ideal substitute
for CO. Figure 20 shows the potential energy surfaces of these two molecules.
The first full treatment of dissociation process at the isotope level was done by Muskatel
et al. (2011). The dissociation does not occur from the ground state. The dissociative surface
in each case is shown in Figure 20 by the golden line. When an electronic crossing to this state
occurs, it dissociates as it is an unbound (repulsive) as compared to all of the other states which
are bound (Rydberg and valence). Valence and Rydberg states for a given molecule refer to
different bonding characters of the excited state. A Rydberg state, in general, refers to the energy
of an excited state for a given atom separation that moves a single electron away from a core
Figure 20. Schematic potential energy surfaces of (A) N2 and (B) CO. The two molecules are isoelectronic
with one another, as shown by the schematic electronic structures of two molecules embedded in the figure.
Modified from Thiemens et al. (2012).
electronic structure. A historically interesting review of Rydberg states is given by Mulliken

(1964). The valence state is an excited energy state that may be of the same energy as a Rydberg
state, but its chemical bonding character is different that the Rydberg state (Lefebvre-Brion
and Field 2004). The bonding is generally weak and the shape of their potential energy curve
is shallower and the distance scale of the chemical species is further apart. Given that these
electronic states are of the same energy, but different bonding character, the states may cross
between each other. The valence state, e.g., in nitrogen is the state that leads to the crossing to
the repulsive surface (dissociation) and as Muskatel et al. (2011) discuss the valence state is the
gateway state to dissociation. However, it is not connected to the ground state by absorption
of energy and it is only by coupling to the excited Rydberg state that initiates the dissociation
process. In the first step of the Rydberg states have a force constant and atom-atom separation
is highly effective, which is known as a Franck Condon region. Following access to this state,
it may cross to the valence state, a highly selective process, and then dissociates. In the first step
of the process in both cases it is the same. For nitrogen for example the ground state absorbs a
photon and goes to the excited states G3Πu and c1Πu. There are multiple embedded states that
mix. It is in this step that crossing to the gateway valence state and exit to the dissociation state
there is a very high selectivity occurring. The probability of crossing from one energy surface
to another (diabatic crossing) is a very specific process as the energy (vibration and rotational)
levels between the two states must match. As discussed, these also have isotopic lines associated
with them. The consequence is that the energy levels are very sensitive to the energy separations
and must match and this then selects which isotope lines are most closely aligned.
Muskatel et al. (2011) calculated the selection processes between states by solution of the
Schrodinger equation for both the electronic and nuclear motions of the Rydberg and valence
states as well as the time dependence and observed strong and selective isotope effects in the N2
predissociation step. Figure 21 is schematic diagram showing the alignments between states of
differing symmetry and isotopic composition. In the Σ symmetry (Fig. 21a), the red (b′) and black
(e′) curves overlap at the left side of the surfaces. In the close up, the 14N14N, 14N15N, 15N15N
lines are shown and for the mass range of 28–30 amu. The same is shown in the Π symmetry
(Fig. 21b) for different parts of the energy surfaces. In the Σ symmetry, the overlap is for e′(0)
and b′(18) levels, and in the Π symmetry, the overlap is for b(5) and c(o) levels. For 14N15N, there
is an essentially exact resonance between b(5) and c(o) levels in the Π symmetry, as highlighted
by a yellow shaded area in Figure 21b. These crossings facilitated by the shift of vibrational
states resulted from isotope substitution are the ones that access the gateway valence state for
66 Thiemens & Lin
Figure 21. Schematic potential energy surfaces of N2 excited states of (A) Σ and (B) Π symmetry. The reso-
nance of 14N15N at c(o) with b(5) states is highlighted by the yellow shaded area. Modified from Muskatel
et al. (2011). The overlap between the electronic states and their isotopic lines is required to cross between
surfaces. Greater overlap leads to enhanced crossings between states.
an isotopologue to dissociate. It is this feature that gives rise to the isotope effect seen for CO in
Figure 18. At 105 and 107 nm the CO excitation states (Fig. 17) are similar to the ones for N2
shown by Muskatel et al. (2011) (Fig. 21) and follow a similar path to dissociation. At different
wavelengths different electronic surfaces are accessed and the fractionation effects differ.
The model is applicable to the CO2 dissociation process shown in Figure 15. The significant
changes in oxygen isotope effects for CO2 molecules substitution of 13C for 12C derives from
the shift of vibrational rotational lines as shown in Figure 21. This fits the concept of isotopic
selectivity in dissociation, but not self-shielding. We do not attempt to completely preclude
the self-shielding isotope effect in an optically deep photochemical column, but we argue that
isotopic selectivity during predissociation plays a pivotal role in explaining many features shown
in photodissociation experiments and must be included which it presently is not. Figure 19
showed that for over a 2 order of magnitude range in CO pressure the δ17O/δ18O remained
constant where one expects a change in that ratio for shielding. Existing experimental results
unambiguously suggest that the isotope effect is sensitive to wavelength due to perturbation in
predissociation rather than pressure due to self-shielding. A larger variation of pressure (column
density) may be carried out in future experiments to further verify this interpretation.
The final test of the isotopically selective dissociation effect was N2 photodissociation
experiment carried out by Chakraborty and colleagues (Chakraborty et al. 2014, 2016). Using a
modified version of the synchrotron experiment for CO, N2 was photodissociated. Wavelength
and temperature dependency were done and repeated to test models. The results (Fig. 22) show
that there is a massive isotope effect at the wavelength expected from the gateway isotope
selective process model by Muskatel et al. (2011). The room temperature curve (red) appears to
be small, but in fact varies between more than 100‰ to above 2000‰. A very strong temperature
effect is observed and at the otherwise same conditions. Lowering the temperature causes the
effect to increase from 2000‰ to 12000‰. The only effect for this change is at the rotational
level, and is a decrease in the line width of the rotational lines (making them sharper) and a
lowering in the population of higher rotational levels. This makes the crossing even more sharp
and selective but would not alter self-shielding. The experiments have been duplicated and beam
profiles known as are the relevant non-adiabatic processes.
There is another important facet of Figure 22 that is significant and was a specific test of
the curve crossing model. At 90 nm (111,111 cm−1) there is an extraordinary enrichment in 15N
observed. The choice for this wavelength was specifically focused on a wavelength that would
underline the crossing effect on the isotope process. The calculated and experimental comparison
of the N2 dipole moments across wavelengths were determined by Spelsberg and Meyer (2001)
and shown in Figure 23. Comparison of the experimental and observed dipole moments in
Figure 23 shows that at 111,111 cm−1 (90 nm) exhibits disagreement between experimental and
calculated dipole moments (shaded area). The dashed lines are cuts through the potential energy
surfaces and show that the most diabatic crossings occur here and thus be a perturbation not
captured by models must be the source of the disagreement. It also would be the region where
the most diabatic crossings would simultaneously occur and differ from other wavelengths.
The selective crossings and multiple pathways should produce greater access to exit channels.
The largest isotope effect would occur at this specific wavelength, as observed (Fig. 22).
The isotope effect observed at 111,111 cm−1 by the enormous enrichment and may not be
attributable to experimental errors as the value was highly reproduced by replicated experiments.
Figure 22. A result summary of N2 photodissociation experiments. Data from Chakraborty et al. (2014,
2016). Panel (A) shows the enormous change in isotope ratio with only a change in temperature. All other
parameters are exactly the same. Panel (B) shows calculated self-shielding effects expected for the experi-
ments. No assumptions are made. It may be seen that shielding does not account for the data. It also shows
that if everything is held constant and the different published cross sections are used there is a variation of
nearly 10,000‰. This is in fact the error in the shielding calculation.
68 Thiemens & Lin
Self-shielding final test and errors

There are two final tests of the self-shielding model. First is shown in Figure 22b. A self-
shielding model was used with three different sets of N2 isotopic cross sections (Spelsberg
and Meyer 2001; Liang et al. 2007; Muskatel et al. 2011; Li et al. 2013; Heays et al. 2014) to
compare both with one another and determine how the best cross sections available differ in
self-shielding results and how they compare to one another (Chakraborty et al. 2014, 2016).
This is a measure of error of the models. Nothing was changed except the cross sections. It is
observed that in this internal test of shielding with everything constant except application of
isotopic cross sections from three groups that the difference may be as large as 20,000‰! The
best comparison shows that the disagree by 1000‰. This observation is a clear demonstration
of the best that a shielding model can do given the error associated with measuring cross
sections. To apply a self-shielding model over a range photolysis experiments and ascribe to
or not from self-shielding one would need an error of 10‰ or less to test the slope variation.
In the case of CO, the cross sections, especially for C17O are not all experimentally determined
as for N2 and the error may exceed 1000‰, even for a simple model. In the CO self-shielding
models none of the published plots show error bars.
Finally Figure 22 is a recognition that the large isotope effects are post photon absorbance.
The plot shows experimentally determined isotope effects in the dissociation curve crossings
and the massive peak at 90 nm (Fig. 22) is where the effect of curve crossings is a maximum
(Figs. 21 and 23) that is not captured in any of the self-shielding models. This is further
confirmation that the isotope effect of dissociation is post absorption. Any model reliant
upon only cross section may not capture the actual dissociation process and isotope effects
(Fig. 22b). Theoretical and experimental determination of a highly accurate and precise cross
sections at the isotopic level is vital.
Figure 23. N2 dipole strength distribution (solid lines). Dotted lines are slices through electronic surfaces of ex-
cited electronic states. The disagreement between experimental and calculated results at 111,111 cm−1 (90 nm) is
highlighted by a shaded area. Modified from Spelsberg and Meyer (2001). This area is where the three different
states come closest and perturbation of one state upon the other should be the greatest. The prediction was that this
should show the largest isotope effect. This is clearly the case as shown in Figure 22a.
Nebular fate of water and water ice from photodissociation

Most astronomical photochemistry occurs on the edges of proto nebulae or large molecular
clouds. In all self-shielding models for oxygen isotopes in meteorites, the product oxygen atom
is trapped as water ice and is the molecular species that converts by an undefined chemical
mechanism into silicates following the intact transport across the nebulae. Ice occurs in this
environment predominantly on grain surfaces. Not considered in the models is the issue of water
ice stability during its nebular traverse or immediately following its trapping on a grain surface.
As a first consideration is survival of the oxygen isotopes of ice formed in the presence of the
stellar UV field that photolyzes the parental CO at 105 nm as H2O ice also photo decomposes.
In a review paper of photodissociation and photoionization of astronomically important
molecules, Heays et al. (2017) have reviewed the fate of water in the background UV field
(Fig. 24a). The water absorption spectrum is shown, and it is observed that the cross section
for water has a rich and strong absorption spectrum in the region of 105 nm (Fig. 24b).
The UV absorption of water is competitive with that of CO (Fig. 24c), and consequently
the stability of water ice in a catalytic chemical environment must be considered. Besides
being dissociative at the same region where CO occurs, the water spectra and dissociative
process has an onset at 190 nm and consequently is dissociative over a larger wavelength
range. From the spectra shown in Figure 24a, one expects with increasing UV output towards
Figure 24. (a) Background UV fields relevant to the solar system. (b) Cross-section of H2O vapor. The shift of
H2O ice cross-section is schematically shown in the ca. 140-180 nm region as a blue dashed curve. (c) Cross-
section of CO. The wavelength for CO isotopic self-shielding is at 105 nm and compared to water that is assumed
to be the stable host of the oxygen anomalies that will produce CAI. Modified from Heays et al. (2017) and van
Dishoeck et al. (2013).
70 Thiemens & Lin
Figure 25. Schematic diagram defining the reaction net-

work leading to the formation of water ice which is in near-
ly all cases on a silicate surface. The reactions that have
been demonstrated to acquire mass independent isotope
effects are highlighted by shaded boxes. Modified from van
Dishoeck et al. (2013).
longer wavelength increasing the probability of water dissociation at 190 nm. Due to the
photodissociation of water, the repository molecule from self-shielding is clearly not stable
in regions with strong UV radiation. The storage of the anomalous oxygen isotopes from
CO shielding was hypothesized to circumvent the exchange issue raised by Navon and
Wasserburg (1985) and Kaye and Strobel (1983). The ice storage is nearly exclusively of
grain surfaces initially. If the formation of water ice was in the outer edges of the solar
system where CO photodissociation occurred, water ice then has to be considered stable
and the oxygen isotopic composition maintained until it transfers from the outer edges of
the solar system to the inner solar system where it becomes incorporated into a silicate and
unfractionated by the chemical transformation process. The model across this many AU
requires zero secondary alteration of the water ice. Figure 24b shows the cross section of
H2O ice compared to H2O gas. The difference is wavelength shifted only ~20 nm which
places it in the region of UV field for CO and H2O photodissociation and consequently water
ice is not stable. Amorphous or crystalline water ice undergoes undefined photodissociation.
To solve these issues, Young (2007) considered the transport and formation rates of water
ice in a box model and showed that the conversion of CO to water ice was on a time scale of
105 years, and the transport from the disk surface to midplane between 105–106 years. It was
suggested that the buildup of water ice could have been in the midplane where UV radiation
was not as strong as the disk surface, and the isotopic composition of water ice in the midplane
at ca. 20K could be well preserved. A spatially sophisticated disk model simulating water vapor
distribution in protoplanetary disks shows rich water directly beneath the photon dominated region
of the disk (Du and Bergin 2014). Recent observations by ALMA (Atacama Large Millimeter/
submillimeter Array) on the protoplanetary disk of the young star V883 Ori also reveal that water
ice accompanied with complex organic molecules can even exist in the photodissociation region
of the protoplanetary disk (Lee et al. 2019). These however are not isotopic measurements and do
not include the effects of radial distance reprocessing that occurs in transit the dark cold region.
Although there are observational and computational evidences that water ice is not
completely unstable in the protoplanetary disk, water ice dissociation processes in the
radiation field that CO dissociates, which occurs on surfaces of many bodies in the outer solar
system (Johnson and Quickenden 1997; Watanabe et al. 2000; Yabushita et al. 2008), remain
a critical factor to consider, especially at an isotopic level. On icy satellites for example, water
ice photolyzes and at temperatures above 20K the recombination to water is slow and the
H + H→H2 reaction on a surface dominates. The hydrogen is lost from the system by diffusion
and Jeans escapes. The sticking coefficient is low and amplifies H2 loss and oxygen remains as
either an oxidant of reduced species, or formation of O2 and OH. The OH is spectroscopically

observed and it too has a large photo destruction cross section at 105 nm. It has also been
observed that the D/H of the residual water is enriched and the oxygen isotope ratios should
also be affected. The photolytic of production of H2 also is consistent with the observation of
vibrationally excited H2 (ν = 5) by sensitive multibeam microwave detection systems (Green
et al. 2009). The chemical dynamics of the H and OH on surficial ice has been modeled for
amorphous and crystalline ice (Andersson et al. 2006; Al-Halabi and Van Dishoeck 2007).
It is also well known that water ice on grains as a consequent of the UV and IR radiative
fields desorbs thermally, adding to its instability (Fraser et al. 2001). Another factor not fully
considered in models at the triple oxygen isotopic level is the actual process by which the water
is actually formed during deposition on the grain surface as ice that may be photodissociated.
It is the pseudo steady state of water formation and dissociation processes that leads to the
“stability” of water ice in the protoplanetary disk. The process is highly complex and is not a
single step, rather many other processes are involved in the steady state radiative environment.
Figure 25 illustrates the complexity of the process. It should be noted that the transfer of
the atomic oxygen from CO photodissociation to the ice there are many intermediates that
includes reactive free radicals (OH, O, HO2) which may react either with surrounding reduced
species in the gas phase or with the surface itself. Note also that there is a buildup on the
surface of ozone which results in the mass independent isotope effect and, loses the original
isotopic composition as discussed. O2 + H and OH + OH reactions shown in the Figure 25 also
produce mass independent isotope effects in H2O2 (Savarino and Thiemens 1999; Velivetskaya
et al. 2016, 2018), which is likely driven by symmetry similar to the O + O2 reaction. If there
is hydrogen leakage from the system, then the water ice will derive its mass independence
from the water by way of ozone and not self-shielding. The mass dependent and independent
isotope effects in these processes are not included or not fully considered in many models.
Dominguez (2010) modelled some of these reactions and predicted Δ17O of molecular cloud
H2O to be ca. 20‰ without considering any CO self-shielding effect.
At steady state, besides the formation of ice and the photo destruction of the water vapor
and ice, there is also the loss of ice by sublimation (or thermal desorption) that can occur
between 10 and 30K when involved in ice complexes with CO ice (van Dishoeck et al. 2013).
Temperatures below those considered a safe haven for ice in nebular self-shielding models.
Adding to the complexity of the water system is the adsorption of CO onto the ice surfaces
to form formaldehyde (H2CO) which is stable and also desorbs from the system (Noble et
al. 2012), likely fractionating the residual water ice from which it was synthesized. Carbon
monoxide adsorption on the surface of water ice grains is in steady state equilibrium between
adsorption and removal by sublimation and photochemical desorption. The process has been
studied in great detail and e.g., using isotopically (12C,13C) labeled CO layer have detailed the
process and shown it occurs via an indirect sub-surface process (Fig. 26), which is heavily
suppressed in the CO–H2O ice (Bertin et al. 2012). The water and CO ices are also the progenitor
of interstellar methanol (CH3OH), which may be linked to grain formation and subsequent
chemical processing is significant (Fuchs et al. 2009) and will be discussed in the next paragraph.
The chemical and physical altering of the water ice and formation of new molecules is known
to be an agent of catalytic chemical production in the interstellar environment. Young (2007)
carried out a relatively large astrochemical reaction network (>500 species and >7600 reactions)
to consider these reactions and showed that these side-reactions would not influence the
isotopic composition of the large water reservoir. A limitation is that all reactions are presumed
to be mass-dependent by estimating reaction rates of isotopologues using reduced masses
for collisions between reactants. As shown in Figure 25, there are several mass independent
reactions even in a relatively small reaction network. It is clear that the role of these well-known
and other unidentified reactions that may lead to significant mass independent isotope effects in
a much more complex reaction network should be carefully evaluated.
72 Thiemens & Lin
Figure 26. (a) Schematic diagram of the CO ice photodesorption experiment (Bertin et al. 2012): By label-
ing the top layer with 13CO (with varying thickness), the authors found that the desorbed gaseous CO is
always from the surface layer of CO ice (< 2 monolayer equivalents for CO, MLeq). A synchrotron mono-
chromatic beam was used in the experiment. (b) Proposed scheme of the photodesorption mechanism:
CO molecules in the subsurface of CO ice were excited to its A1Π state by UV photons, and the energy is
subsequently transfer to the neighboring CO molecules. The CO molecule in CO ice surface is ejected after
receiving enough energy. Modified from Bertin et al. (2012).
The fate of water ice in different astronomical environments is highly complex and there
is a rich literature on observations in molecular clouds, proto stellar environments, cold and
warm. There are relevant laboratory experiments that help define the role of all contributing
processes. This includes the chemistry rates for the reactions shown in Figure 25 as well as
photo and thermal desorption, ice properties of crystalline and amorphous ice, photo destruction
and chemical reactions within the ice (van Dishoeck et al. 2013), especially the interaction
between H2O and CO to form CH3OH as outlined earlier. Many of these processes have been
defined by laboratory measurements and models of satellite interferometric observations of
D2O/HDO in ice in protosolar cores (Furuya et al. 2016). The measurements show that the
D2O/HDO ratios are not that expected by simple ice formation and require secondary chemical
evolution to account for the observations. The processing is schematically shown in Figure 27.
The models and application of the isotopic fractionation process show the formation of a water
ice layer from stage 1 and the layer of pure water ice with a lower D2O ratio than HDO, to the
second stage with a chemical evolution and associated fractionation associated with all of the
processes mentioned above and re-partitioning of the deuterium (Furuya et al. 2016). This is
further evidence that storage in ice is not stable and the processes of formation/sublimation
and chemistry within the ice is substantial and interactive with the surrounding gas during
transport. The hydrogen isotope effect also provides observational constraints for testing the
validity of current models that have already considered isotope effects. Most of the individual
contributing factors have been described in the literature (Johnson and Quickenden 1997;
Watanabe et al. 2000; Fraser et al. 2001; Andersson et al. 2006; Al-Halabi and Van Dishoeck
2007; Yabushita et al. 2008; Bertin et al. 2012; Noble et al. 2012; van Dishoeck et al. 2013).
Figure 27. Schematic diagram showing

the formation of H2O-dominated and
CO and CH3OH-rich layers on grains
that significantly influence the partition-
ing of deuterium. Modified from Furuya
et al. (2016).
With the ability to use satellite data, our understanding of these processes has advanced
greatly. Using the far infrared spectrometer on the Herschel Space Observatory it was possible
to use the emission lines from the cold water vapor around the young star TW Hydrae to map out
the distribution of ice (Hogerheijde et al. 2011). The authors being mapped the water/ice lines
and the observations amplify understanding of the processes associated with ice formation.
The emission lines are measured and derived from the cold water vapor associated with the ice
coated dust grains at the edge of the disk. In Figure 28a the number density of H2 is shown in a
cut through the protoplanetary disk, and Figure 28b is the associated thermal structure. Note that
the blue contours in Figure 28b is the layer of maximum water vapor concentration. Figure 28c
is most important as it shows logarithmic variation of the water molecules/water ice. The
equilibrium between the photodissociation of water to the photo desorption of water providing
an equilibrium water column (Woitke et al. 2009; Hogerheijde et al. 2011) even though it is
well below the freezing point of water. It is not until deeper into the disk that the temperature is
lower and the amount of photons become too low for photo desorption to occur. Figure 29 shows
the effect of photons on water/ice number density distribution in a cut of a solar nebula and a
component of the equilibrium (Bethell and Bergin 2009). The self-shielding in this case is not
for CO or isotopes but rather the effect of water opacity on its own gas/solid phase processes.
The left panel is the complete consideration which includes its self-shielding of photons.
Figure 28. Distribution of (a) H2, (b) dust temperature, and (c) water vapor and ice in the TW Hydrae pro-
toplanetary disk. Modified from Hogerheijde et al. (2011). It is to be noted that there is a significant amount
of water vapor into the nebula before it is effectively made into ice.
74 Thiemens & Lin
Figure 29. Schematic diagram showing the effect of H2O self-shielding on distribution of H2O in solar
nebular. Modified from Bethell and Bergin (2009).
In the upper layer there are sufficient photons to photodissociate water and deeper in there
remains water but less dissociation due to the decrease of photons. These observations
somehow support the theoretical prediction (e.g., Young 2007; Du and Bergin 2014) that water
may be “stable” in some region of the protoplanetary disk. However, it does show although
the isotopic composition of the ice reflects the CO dissociation process in any way and that
requires most importantly, experimental data of ice photochemical processes for all oxygen
isotopes and, across all relevant wavelengths.
There are aspects of the grand titration associated with the transport of the δ17O = δ18O
ice and reacted with nebular CO to form secondary products to consider. This includes, but
not limited to (1) the solar nebula is turbulent; and (2) the ice preserved must be formed and
transported intact. Ciesla (2014) has provided an analysis of the fate during formation of ice
and its cycling in the environment and subsequent transport. The work shows the difficulty
from effects of nebular diffusivity, different temperature regimes, gas densities, chemical
compositions and photons. Ciesla (2010) discuss the turbulence of the nebula as a significant
driving force in condensation. The requirement is that the ice particles before doing the titration
to form the many meteorite classes with the preserved is that ice oxygen isotopic composition
must survive intact for 105–106 years and transported through the different regimes totally intact.
Indeed, Young (2007) considered both chemistry and vertical/radial mixing and showed that
the transport of the δ17O = δ18O water ice produced by CO photodissociation and self-shielding
from the disk surface to midplane is on a time scale of 105 years. Furuya et al. (2016) however
showed that the chemical evolution of water on grains that may lead to additional isotope effects
could be less than 105 years (Fig. 27). These reactions have been considered by Young (2007)
as mass dependent reactions. Nevertheless, potential mass independent reactions shown in
Figure 25 remains not well considered. Whether such reactions significantly alter water triple
oxygen isotopic composition is an open question and needy of relevant experimental work to
progress. Dominguez (2010) has suggested that δ17O = δ18O water could have been produced
without any CO self-shielding effect if some of these mass independent reactions are considered.
The sum of the foregoing discussion is that the fate of water formed from self-shielding might
not be an isotopically stable reservoir due mostly to ice photolysis and subsequent complicated
chemical reactions accompanied with isotopic partitioning. A much more comprehensive
reaction network that consider all possible mass independent isotope effects is needed. As will
be discussed, the oxygen isotopic composition of the meteoritic oxygen isotopes places stringent
limits on the source of oxygen and how much it may vary from secondary alteration.
Summary
In the foregoing sections the development of the theory and understanding of the triple
oxygen isotope chemistry in chemical recombination reactions and photo dissociation have
been discussed. Much of the discussion has centered on the oxygen isotopic record stored
in the meteorite record and consequences for understanding the earliest formation processes
in the solar system. The basic physical chemical principles apply to all processes on earth as
well and molecules that have isotopic compositions thought to derive from photochemical
shielding. The experimental and theoretical chemical, photo and quantum mechanical basis
since the discovery of the mass independent isotope effect (Thiemens and Heidenreich 1983)
have advanced considerably and have aided in providing a mechanistically better understanding
of these processes in nature. Though not relevant to this special issue, the understanding of
the Archean processes using sulfur isotopes have been included in this developmental process.
For meteoritic oxygen isotopes the discussion has led to basically two fundamental physical
chemical processes: dissociation and bond formation.
The intrinsic effects associated with CO self-shielding and the total steps render it
unlikely. First, the isotope effects associated with the bond breakage process do not coincide
with observed experiments. Secondly, following photolysis, a dissociation effect is immaterial
as the reaction that leads to the formation of the solids containing oxygen will eliminate the
isotopic record of the shielding in the short-lived transition state of the combination reaction.
The consequence is that the most important step in the formation of the solar system, the gas
to particle conversion process creates a new isotopic signature that is based upon this step
exclusively. The fractionation produces a δ17O = δ18O composition in the condensates (Fig. 5).
This is consistent with both contemporary experimental and physical chemical theory.
Another uncertainty arises from the inability to store the isotopic anomaly in a species
that is stable such as water ice. The assumption is that this ice species will retain the signal
of self-shielding, at thousands of ‰ enrichment with near perfect equal 17O,18O enrichment
requires transport from the outer edge of the nebulae to the region where it will melt, evaporate
and transfer to the local reservoir where it first produces a CAI condensate. Relevant chemical
mechanisms are proposed and incorporated in the models for this process, but are significantly
incomplete, especially for a thorough consideration of the associated isotope effects across
wavelengths. Experiments are required to support the envisioned scenarios. Large oxygen
isotopic anomalies of Fe3O4 in new-PCP from one meteorite may originate from other sources
and may not be a sufficient evidence to reconstruct the isotopic composition of water in the
protoplanetary disk. This issue will be further discussed in the next section.
Most importantly, non-CAI components comprise >99 % of solar system materials: the
bulk meteorite compositions (Fig. 30). The basic premise is that there are varying mixes between
a presumably single valued ice and an inner nebular component. The isotopic difference of two
reservoirs are hundreds of per mil. This model for example requires that one mix a component
of heavy isotope water ice that survives the transport and mixes it with a titration precision
so tight that one can differentiate between eurcrites and diogenites at −0.24‰ and Angrites at
−0.04‰! The same is true for all of the features shown in Figure 30. The titration must capture
the difference of a 1 to 2‰ between difference between the ordinary chondrites (H, L, LL) and
also the original δ17O = δ18O CAI line, all in a turbulent environment and on a short time scale.
Alternatively, if the small difference among these materials is interpreted as a “noise” during
mixing, the overall small variability of oxygen isotopic composition in all planetary materials
that formed over a large distance in the disk implies an extremely thorough mixing on a large
spatial scale. Large temporal and spatial variabilities of water isotopic compositions (hundreds
of per mil) were however shown in models that carefully considered vertical and radial mixing.
76 Thiemens & Lin
Figure 30. Triple oxygen isotopic composition of the solar system. Left panel: The close up of ma-
jor meteorites in the solar system. Adapted and modified from Northern Arizona Meteorite Laboratory.
Right Panel: A scaled-down version of Figure 16. Note that with Bulk earth as a starting composition, the
formation from the SiO + OH reaction includes nearly all meteorite types and requires no mixing except for
the back mixing of CAI which are <1% of the mass and of no influence to the bulk of the solar system (see
discussion in Section “A New Model for Triple Oxygen Isotopes, Meteorites, and the Origin of the Solar
System”). In a self-shielding and mixing model, all of the individual classes require very precise mixing
on an ice far outside the figure at very high positive enrichments and of contestant value with a reservoir
that is far outside the figure in the third quadrant near δ17O = δ18O = −50‰ or even farther. The mixing
and conversion of the ice to solid must be sufficiently precise to account, e.g., the differences between the
H, L, LL, and R chondrites. That process also assumes no isotope effect in the gas to particle conversion.
A requirement in the transport phenomena and concomitant photochemical, nebular

dynamical processes is for the δ17O = δ18O source to be derived from self-shielding.
This fractionation process only occurs in a restricted optical region of the actual shielding
zone at a given time, and at greater optical depth 18O becomes opaque and 17O is selectively
dissociated (Fig. 6). At low opacity, there is a range in δ17O/δ18O between 0.5 and 1.0, with
no effect at the shortest depth. In additional, the product must be transported intact to the zone
of condensation as discussed earlier. These processes have been modeled and suggested to
be plausible, though actual isotope effects in the reaction network are not comprehensively
characterized nor include relevant fractionation processes.
Even all problems outlined above can be overcome, there are final photochemical
requirements. There are built in assumptions in the basic photochemical spectral requirements
not discussed but required is in the CO absorption process. The slope one is only be attained if
the cross section σ and quantum yields for 16O, 17O and 18O are exactly identical and their
absorption lines are well separated. As shown in Figure 22 a difference in experimental
measurements of σ magnifies to more than a factor of 1000‰ in the product of photolysis due
to the exponential e−σcl absorption relation and, the large differences in the isotopic abundances
in 16O and 17,18O. No error bars including this effect for CO have been reported. The issue
is amplified if an isotopic line must be calculated, this folds in another error which must be
included. Finally, the likelihood that all cross sections are the same is limited by perturbations
of the electronic states. The CO states at the 105 nm region (Fig. 17) are highly congested
between excited states and perturbation is most probable at these wavelengths producing
anomalous cross sections. Box 1 is a summary of the outstanding issues concerned with
self-shielding in the nebula and other environments. Some issues in Box 1 related to solid
formation will be further discussed in the ensuing session.
BOX 1
SELF-SHIELDING PROBLEMS
1. There remain no experiments that provided clear unambiguous support for models of
self-shielding, only the contrary.
2. The self-shielding models relies on unavailable high precision cross sections for the
important minor species and must be calculated.
3. The error in self-shielding models is driven by the accuracy of the cross sections. For mol-
ecules of well-known measured lines, the error has been shown to be hundreds if not thou-
sands of ‰. This is due to the difficulty of obtaining the cross sections for the minor isotopes.
4. The photochemical isotope effects in post absorption that lead to dissociation have been
experimentally shown at self-shielding wavelengths to not produce the required isotope
fractionation to explain any meteorite observations.
5. It has been experimentally and theoretically shown by different groups that the actual
dissociation process is highly mass independent and does not retain the self-shielding
record required for explaining the δ17O/δ18O = 1 line.
6. The storage of the oxygen anomaly irrespective of its isotopic composition in water ice
is not retained. The cross section for water and ice water photolysis at CO shielding
wavelengths is large and ice dissociates and loses its isotopic signature.
7. The ice must be transported over astronomical unit length scales through the protosolar
nebula and be 100% stable.
8. Once the ice has been delivered to the site of creating a stable solid non ice product,
there remains no step wise chemical mechanistic models to establish how this record is
converted from ice to silicate or a perovskite species without any fractionation.
9. During the overall process, a highly precise titration must be done to add exactly the
right proportions of an ice from beyond 20 A.U. to an inner solar system region for solid
formation. This must be done in a way that creates the meteoritic class identities, that are
statistically isotopically separable at a sub ‰ level. These reservoirs may be differing
more than 1000, and the ice value must be assumed to be constant.
10. The ice that is created for the δ17O = δ18O slope via shielding only arises from a slice through
the optical region between no effect and where 18O becomes opaque. There is no mecha-
nism for how this 3-D slice is extracted from the turbulent nebula and transported intact.
11. The observation of mineral phases that are thought to represent reaction of ice of heavy
isotope water composition with troilite to create Fe3O4 require the ice to be delivered with
complete conservation of its original signature. The anomaly may simply be created in situ
by the oxidation of troilite with OH from water that is the same composition of the bulk
meteorite. It is the process itself that creates the anomaly and requires no transport or pres-
ervation. Relevant reactions are observed in atmospheric gases and solids as well as water.
12. The assumption of the starting point at as a solar value has not been validated by models.
Recent models do not match Genesis isotopic or abundance measurements and conclude
that the link between protosolar and photosphere has not been established yet. The as-
sumption of the CAI point as being solar remains inferred.
78 Thiemens & Lin
SOLID FORMATION IN THE EARLY SOLAR SYSTEM

Simultaneous formation of the first solids and their isotopic anomalies
The final aspect is perhaps ultimately the most significant. Even if an isotopic anomaly does
get captured in ice and heroically survives to arrive at the very highly restricted space and time
required, it is a moot point. It is suggested that exchange of oxygen atoms in minerals between water
at high temperature in the inner disk accounts for the anomalous oxygen isotopic compositions
in planetary materials. If oxygen atoms in original minerals are normal (δ17O/δ18O = 0.5),
the final isotopic composition will superimpose a slope ½ species and a slope 1 may not be
obtained, depending on the sizes of two reservoirs which restricts the process. Alternatively,
the gas-to-particle formation step is the very first chemical reaction that occurred in the solid
formation and should be included as it is one of the most significant steps in solid formation.
As discussed in the section on ozone formation mechanism the well-known features of a gas phase
reaction of any variety is that to proceed to a stable product there is a transient metastable state.
During the combined process for collision of reactant molecules or atom–molecule, a transition
state is formed that either stabilizes, or, re dissociates. In the reactions to be discussed, the ratio
of stabilization to falling apart is approximately a million to one. For the mass independent ozone
reactions discussed this time scale to be around 10−11 seconds. It is this process where the high
selectivity that gives rise to the mass independent isotope effect. Indeed, the actual time scale to
populate the states may be even shorter, which is on the order of 10 femtoseconds for N2 (Ajay
et al. 2018). The consequence of this is that irrespective of the anomaly being present in the ice,
there must ultimately be a chemical reaction between that (or any) oxygen species that leads
to a solid. That process proceeds through a transition state and eliminates the original signal
on a 10−11 s time scale per reacting molecule. The consequence is that the critical factor for the
meteoritic oxygen isotopes, the gas to solid formation process becomes of particularly important
and controls the ultimate product. As shown in Figures 5a and 5b, the slope 1 line of CAI is
identically produced in the O + O2 reaction. In the ensuing years since that discovery as discussed
in this paper and reviews (Thiemens 2006, 2019; Thiemens and Lin 2019), the mass independent
isotope effect is now sufficiently documented to detail the process in the early solar system.
The generation of the slope 1 is in the transition state and the isotopic symmetry/asymmetry control
is the source of mass independent isotope effect. It is terminal position of 16O16O16 vs 16O16O17,18O
that ultimately produces the slope 1 effect. It is this factor plus the position of oxygen on the
periodic chart that leads to oxygen typically coordinating other elements, e.g., SiO4, SiO2, Al2O3,
Fe2O3, Fe3O4, SO4, CO3, TiO3, TiO4, NO3, ClO4, and PO4 which encompasses a predominance of
the geochemical minerals. During condensation and nebular conditions, the conditions are ideal
for creation of the slope one effect in the earliest minerals as well as the bulk of the minerals.
The symmetry effect is general and was noted in sulfur (Bainssahota and Thiemens 1989; Lin and
Thiemens 2020) and has a qualitative quantum chemical explanation (Babikov 2017; Babikov
et al. 2017). A massive effect has now been extended to the heavier mass range (titanium) and
a massive effect has been experimentally proven by Robert et al. (2020) as shown in Figure 10.
The mass independent effect of titanium isotopes was interpreted using a theoretical framework
developed by Reinhardt and Robert (2018). According to Reinhardt and Robert (2018), the mass
independent effect is not caused by isotope exchange reactions or differences in the interaction
potentials. Using trajectory calculations and classical mechanics, these authors show that the
complex lifetimes are different if the complexes are formed by reactions involving identical (e.g.,
16 16
O O) or non-identical (e.g., as 17O16O) isotopes.
An important step is connecting the production of the oxygen anomalies to the gas-to-
particle formation process. Though the role of symmetry for importance in the early solar
system has been recognized since the first publication on the mass independent oxygen
isotopes via chemical processes, a vital key to the next step is experimental observation in
a cosmochemically relevant process. The difficulty is that such experiments must be done
in a system that insures gas phase reactions in a controlled manner. A first experimental
demonstration of the production of a mass independent isotopic composition arising from such
a chemical reaction was shown by Kimura et al. (2007). Using a high temperature (>1000 oC)
flow system with precursor iron (from Fe(CO)5) and Silane (SiH4) in the presence of a third
body such as He and with different oxidants, including O2, H2O and N2O in the presence
of an electrical discharge to produce reactive oxygen species, a mass independent isotopic
composition was observed in the solids produced, with Δ17O extending to greater than 4‰.
This is consistent with prediction of a symmetry-based production mechanism for both Fe
and Si. The products collected in this system have unreacted Fe and Si present which makes
it impossible to state the magnitude of the effect and theory. Even for the well-studied ozone
case state of the art quantum mechanically based models do not allow for determination of a
single stage fractionation factor. The Δ17O value of 4‰ is a lower limit.
In a more recent work, Chakraborty et al. (2013) used a laser-based ablation system to
create controlled amounts of gas phase SiO and determine its isotopic fractionation to SiO2 via
O (O2) and OH. SiO is the dominant nebular gas leading to the formation of the first condensates
and subsequent ones leading to silicates. It is a key step in the very first formational processes
in the solar system. In the experiments, pure SiO was laser ablated at 248 nm providing a
reproducible number density of gas phase SiO molecules. Oxidants were provided as OH
as this is a likely oxidant is the solar nebula. The results are shown in Figure 5c. In these
experiments a mass independent fractionation is observed. As in previous experiments there
are contributions from more than one reaction, however in this case all of the rate constants
for relevant reactions are known and the isotopic contribution for a given reaction may be
determined. Consequently, the mass independent component in the reaction network can be
unambiguously extracted. The Figure 5c shows the contribution for the SiO + OH reaction,
arguably one of the most likely early reactions in the solar system. It is seen in the figure that
the slope one required for the meteorites is observed. All variables are known and clearly
defined and thus the assumption for this as a mechanism for production of the meteoritic
anomalies is consistent with the most recent physical chemical models for production of the
anomaly and experiments. In the experiment, the H2/O2 ratio is significant to acquiring a unity
slope as that ratio determines the production of OH radical in the laser system. In the solar
nebular, if the SiO + OH reaction dominates the production of SiO2 solid, a pure unity slope
can be directly observed. A detailed chemical model quantitatively accounting for various
SiO2 production pathways is needed in the future. The take-away point is that the assumption
that a reaction of mass dependent nebular species may produce a δ17O = δ18O composition
in a solid and is likely a meteoritic mineral precursor has been experimentally observed.
It has been known since the classic paper of Grossman (1972) that a first condensate
should be corundum at around 1758 degrees, followed by perovskite (CaTiO3) and melilite.
In each case the oxygen is in a position where a symmetry factor is relevant. The experiments
of Chakraborty et al. (2013) show that the slope 1 effect is a feature associated with gas to
particle conversion. It also shows that the effect arises and the original oxygen species in ice
covered grains of shielding is lost. In the solar nebula this would suggest that the origin is near
the bulk earth and Mars. In Figure 16, triple oxygen isotope composition of the putative self-
shielding precursor water from self-shielding (Fe3O4 in new-PCP) is shown as well as different
terrestrial species. In the model, water is brought to the inner solar system and reacts with
troilite either in the pre-condensation of on a planetary surface. The probability of the water
possessing the anomaly has been discussed in the previous section. It is stated that these are
the heaviest isotopic enrichments in the solar system though stratospheric ozone extends to the
same enrichment factors is not discussed. Sakamoto et al. (2007) suggested that the anomalous
feature arises from the net stoichiometric reactions with either troilite or native nickel:
80 Thiemens & Lin
3FeS + 4H2O ↔ Fe3O4 + 3H2S + H2 (R12)
3Fe + 4H2O ↔ Fe3O4 + 4H2 (R13)
In this reaction the water is presumed to lie along an extension to the end member
somewhere in the outer solar nebula (as high as 1000‰ in δ18O) (Lee et al. 2008; Lyons et al.
2009) and reacts passing along the anomaly to the Fe3O4, which under the ambient conditions is
the thermodynamically stable phase. The water δ18O is presumably between 150‰ and 200‰.
In the model of Lyons et al. (2009), the δ18O value of water produced by self-shielding that
is responsible for creation of the Fe3O4 is at approximately 200‰ (Fig. 31a). A difficulty is
that the required isotopic composition (δ17O = δ18O) in water is not widely observed and the
two lines (1: CO photodissociation line segments in which δ17O/δ18O slope may not be one
based on experimental results; 2: the δ17O = δ18O line of O3 formation that likely involved in
the processes) converge for only a limited time. Even in a self-shielding model, for much of the
time the δ17O/δ18O ratio in product water is not exactly 1 (Fig. 31a). Furthermore, the ca. 200‰
for δ18O of Fe3O4 in new-PCP (Acfer 094) is required to produce for a minute period of time in
this model. Figure 31b further shows that the variation of ice along a hypothesized slope 1 line
varies considerably. The δ17O and δ18O values depending on UV light intensity, radial distance
and time varies from 0‰ to ca. 1000‰. To achieve the proper value for new-PCP of Acfer
094 at 200‰ it has been suggested there may be mixing of ice at different values must occur.
At present measurements are too scarce to evaluate the variability induced by the mixing
effect. Photodesorption is included as a factor but it is not stated how. The isotope effect and its
wavelength dependence has not been measured for oxygen. Water photolysis is included but only
as a rate of oxygen atom formation not isotopes. For the model to explain the Acfer 094 new-PCP
measurements there are restrictions in mixing, radial distance, light intensity, time, isotope effects
in water dissociation and mixing of ice proportions that may differ by hundreds to thousands of
per mil to acquire the right isotopic composition of ice. Also not included directly in the models
are the chemical reactions that actually produce the anomalous ice along with their isotopic
fractionations or as discussed side reactions with e.g., CO. A significant point is that the chemical
production of water is thought to occur on grain surfaces and proceeds by many reactions:
Figure 31. (a) Model results of time-dependent triple oxygen isotopic compositions of H2O at 30 AU from
the protosun at the disk midplane (Lyons et al. 2009). (b) Three isotope plot of H2O ice at 125 AU from the
protosun at the disk midplane simulated by Lee et al. (2008) with different G0 value, a parameter scaling
the strength of the local FUV radiation field relative to the standard interstellar radiation field. Modified
from Lyons et al. (2009) and Lee et al. (2008).
These reactions have been suggested by Tielens and Hagen (1982), Hasegawa et al. (1992),
Cuppen and Herbst (2007), Nunn (2015). This reaction network to actually form ice will
not necessarily store the shielding record in ice. Young (2007) did consider mass dependent
fractionation processes in a relatively large reaction network in the self-shielding model, but the
isotope effect was not treated in an isotopically relevant selective chemical reaction mechanism.
For the new-PCP results of Sakamoto et al. (2007) shown in Figure 16, the matrix of the
material is shown near the origin in the triple oxygen isotope plot at the terrestrial fractionation
line of the host meteorite Acfer 094, an ungrouped primitive carbonaceous chondrite
(Fig. 30). To account for the data measured in the Fe3O4, the ice composition has numerous
restrictions. Apart from the ice compositions, the aqueous alteration reaction is not well known.
An analogy is serpentinization in which water reacts with iron-bearing rocks to form oxide,
a mass dependent process. In this case, the oxygen isotopic composition of iron oxides
is determined by oxygen in the large water reservoir. However, it is noted that the new-PCP
are tiny grains. Elemental mapping suggests that most new-PCP grains are less than 7 μm2,
with the largest 160 μm2 (Sakamoto et al. 2007). This is very similar to the observation of
iron sulfate on terrestrial aerosols measured by NanoSIMS (Li et al. 2017), a reaction product
of SO2 and iron oxides formed in the aqueous aerosol surface of some 10s of nanometers.
Therefore, water that directly reacted with carbonaceous chondrite to form new-PCP may not be
a large reservoir, but potentially a thin surface layer. Therefore, an alternate chemically plausible
to interpret the anomalous oxygen in the new PCP comes from heterogeneous reactions on liquid
layers of terrestrial aerosols, as thoroughly discussed by Thiemens (2018). It is known that in
such reactions, the isotopic anomaly from gaseous ozone is passed along from the 10 nm liquid
water layer to the mineral surface. Most carbonate minerals on the Earth are isotopically normal
due to their interaction with the large isotopically normal water reservoir. Shaheen et al. (2010)
have observed that atmospheric crustal-derived CaCO3 possess a significant mass independent
oxygen isotopic composition ranging from 0 to more than 3‰ in Δ17O (Fig. 7). The aerosols
were collected as a function of particle size and observed that the sub-micron particles have the
smallest value and the micron sized particle the largest. Laboratory experiments defined that the
odd oxygen with isotopic anomaly from ozone or OH is transferred to the surface carbonate in
the water-solid thin layer. The same scenario would occur on the troilite surface in the presence of
the oxidizing water of isotopic composition of new-PCP in Acfer 094. It is well known that water
on a surface such as FeS or FeS2 undergoes dissociation via electron transfer reaction producing
OH. This is the actual radical that oxidizes the reduced species and not direct reaction with water.
A recent paper has performed a density functional theory/plane wave calculation for pyrite.
The initial step of the surface, which will be the case for troilite is that the two Fe(II) sites on the
surface react with water and produce Fe(III)–OH− (Dos Santos et al. 2016). This species then
in reducing environments will successively produce Fe3O4, and these reactions are all subject
to symmetry isotopic fractionation effects. In this case, the anomalous oxygen did not originate
from water per se but reactive oxidizing species with mass independent isotopic signatures.
Indeed, the reaction sequence referenced above by Sakamoto et al. (2007) is not a mechanistic
reaction but rather the net stoichiometric reaction of several steps. For the two Reactions (R12)
and (R13), it is important to note that water does not react with FeS directly and the initiating step
is from an OH reaction, followed by a series of oxidation steps that ultimately produce Fe3O4
(Dos Santos et al. 2016). If the water in the troilite had a composition similar to the host meteorite
the anomaly in Fe3O4 would be produced from the reverse of that from self-shielding produced
exotic water. The simplest explanation is that the troilite begins with oxidation from OH derived
from the host meteorite near the terrestrial fractionation line. OH is produced on surfaces by
water dissociation (Dzade et al. 2016). For the beginning of the oxidation processes leading
to the thermodynamically stable Fe3O4, the reaction is very much analogous to the SiO + OH
reaction and proceeding by successive oxidations such as the intermediate:
82 Thiemens & Lin
Fe + OH ↔ FeOH ↔ FeO + 1/2 H2 (R14)
FeO + OH ↔ OFeO + H (R15)
The Reaction (R14) would produce a mass independent fractionation as observed in the
experiment of Chakraborty et al. (2013). As seen in Figure 16, the magnitude of the Fe3O4
fractionation is close to that observed in the stratosphere and troposphere in ozone, which derives
from O + O2 with an extent as high as 150‰ for δ18O and >26‰ for Δ17O. It should be noted that
for the mass independent fractionation process, there is an inverse temperature effect where a
higher temperature enhance the isotope effect (Morton et al. 1990). In that case, for the oxidation
of troilite oxidation, if the process were at e.g., 200–400 degrees the magnitude of the effect in
oxygen would be more than sufficient to produce a fractionation from above mentioned reactions
and seen in Figures 5 and 16. The high temperature effect is also observed in the most recent titanium
experiment that was carried out at ~1000K and the mass independent isotopic effect mimics
the observation from meteorites (Fig. 10) (Robert et al. 2020). Atmospheric ozone (produced
from symmetry-dependent reactions) and nitrate (inheriting the isotopic signature from ozone)
are also nearly overlapping with the meteoritic Fe3O4 as seen in the figure. Consequently, this
process does not require a photochemical process, preservation of an isotopically labeled ice,
and transport trans nebula and quantitative reaction without fractionation. The process is a
single step starting with water at the mass fractionation line where the bulk meteorite lies.
It had been suggested that the ice to solid occurs at high temperature and no isotope effect
occurs. The symmetry reactions have a negative temperature effect, and the isotope effect in
forming the solids at high temperatures produce a larger effect rather than no effect. In this
case it is simply the oxidation process, as in experiments and the atmosphere and no reservoir
mixing is required. The reactions are known and the basic physical chemical symmetry relations
known at a level to model, and requires no major assumptions.
Therefore the bottom line for the presence of heavy water is that it is not required and
the indirect evidence in magnetite is not an anomaly from water. Based upon experiments,
atmospheric species isotopic composition and chemical reaction theory is most plausible that
this is locally produced and hence not subject to secondary processes. The reaction leading to
the stable products would be the relevant host phase.
Oxygen isotopic composition of the Solar System
Ultimately the main objective is that a model for well-known oxygen isotopic composition
of meteorites shown in Figure 16. A prediction of mixing models, inculding self sheilding is
that a reservoir of oxygen exisits at a δ17O = δ18O value of approximately −40‰ or lighter
(−50‰ is more likely) (Krot et al. 2019). This CAI reservoir was predicted to be the sun.
The Genesis spacecraft mission collected solar wind for two years and returned samples for a
high precion measurement of the elemental and isotopic composition of the solar wind for a
large portion of solar wind elements. In the return samples, the first after Apollo, the oxygen
isotopes of the solar wind were measured by MegaSIMS, an instrument that combines the
advantages of secondary ion microscopy with accelerator mass spectrometry (McKeegan et
al. 2011). At the low concentrations of the solar wind an electrostatic concentrator was used to
concentrate the solar wind for a greater quantity of solar wind material. A series of correction
factors for instrumental and Genesis concentrator electrostratic fractionations is applied and
the data obtained for the measurements after correction is shown in Figure 16. It may be
seen that the actual solar wind does not lie at the point predicted by self shielding models at
δ17O = δ18O = −50‰ and is at δ18O = −102‰ and δ17O = −80‰. In McKeegan et al. (2011),
it was speculated that there is a fractionation in the formation of the solar wind from the solar
photosphere and the sun lies along a slope one line, with the 40‰ fractionation factor difference
in δ18O attributed to solar wind formation. Heber et al. (2012) modeled and deterimened the
solar wind isotopic fractionation factor for helium (63‰/amu), neon (4.2‰/amu) and argon
(2.6‰/amu). For a charge state of + 6 for oxygen (which may also possesses a component
of + 7), a fractionation between sun and solar wind must be known and compared to what is
expected to test if the inferred value for the sun is correct (Fig. 16).
In a more recent study (Laming et al. 2017), oxygen isotopes were modeled using a
electromagneic model and applying a pondermotive inetractive force in the solar chromosphere,
with maintainence of the first adiabtic invariant in the lower corona. Laming et al. (2017)
developed a solar wind formation mechanism specifically to understand oxygen concentrations
and isotopes. In their calculations they have determined the isotopic fractionation factor
between the bulk solar wind and the photosphere for low mass dependent fractionation and
high mass depeendent fractionation prcesses that typify solar wind energy regimes. It is found
for the low mass regime the 16O/18O value ranges between 0.8–0.9 %/amu and for high MDF,
1.57–1.62 %/amu compared to the measured Genesis value at 2.2 %/amu. In their models,
the calculated range of 25Mg/26Mg and 15N/14N values agree with the Genesis measurements.
The oxygen isotopes do not however, thus there is some component missing in the model
or different values for input parameters are needed. In Laming et al. (2017), the elemental
abundances have also been calculated by the ponderomotive model and compared to Genesis
measurements (Fig. 32). It is observed that the model results for the element concentrations
with respect to solar are close and do the best at low values of the First Ionization Potential,
or FIP. A most important point is that the modeled elemental abundance of oxygen does not
agree with the model by a significant factor, as highlighted in Figure 32. It is suggested that
the photospheric values assumed are too small, and the assumed values do not agree with those
of von Steiger and Zurbuchen (2016). Part of the issue is concerned with fractionation factors
in slow and fast wind regimes and their relation to the first ionization potential. The polar
coronal holes are of particular importance. The bulk oxygen isotopic composition of the sun
and solar system remains unknown. Laming et al. (2017) conclude that more data and analysis
is needed to solve the existing inability to model both the oxygen isotopic and elemental
compositions of Genesis. “Once this has been achieved then a full assessment of whether the
solar photospheric values may be used as a proxy for the pre-solar nebula.”
Figure 32. Left panel: The close up of triple oxygen isotopic composition of selected CAIs. The fraction-
ation processes of CAI after formation are schematically shown by green dotted line. FUN CAIs distribute
in the region with green dotted lines. CCAM and PCM stands for “carbonaceous chondrite anhydrous
mineral line” and “primitive chondrule mineral line”, respectively. Modified from Krot et al. (2019). HAL
data from Lee et al. (1980). Right Panel: A scaled-down version of Figure 16.
84 Thiemens & Lin
A NEW MODEL FOR TRIPLE OXYGEN ISOTOPES, METEORITES,

AND THE ORIGIN OF THE SOLAR SYSTEM
As reviewed, the history of models on the origin of the meteoritic isotopic anomalies
has several categories, ranging from the original nucleosynthetic, to self-shielding, to the
origin arising during the bond formation that occurs during the actual formation of the first
solids in the proto solar environment due to experimentally and theoretically determined
symmetry dependent reactions. A nucleosynthetic case for the anomalies has been ruled out.
Self-shielding, while still utilized has been discussed in great detail in the forgoing section
is unlikely. This then leaves a mass dependent isotopic fractionation process associated with
formation of solids as a source of the meteoritic oxygen isotopic class differences. Here we
analyze existing measurements and propose a new Chemical Mechanism Model for production
of meteorite oxygen isotopic anomalies and the origin of the solar system.
Revisiting triple oxygen isotopes of CAI
Many of the models for solar system formation revolve around CAI and their formation.
These objects as well as chondrules are among the most primitive and oldest objects in the solar
system and the interest is warranted. This leads to the outstanding issues with the source of
meteoritic oxygen isotopic compositions arising from self-shielding/ice transport and addition
to an inferred solar system oxygen at δ17O = δ18O = −60‰. The short comings suggest that an
alternative route to synthesis of the observations shown in Figure 16 is warranted. It is suggested
that the early nebula was at the point in three isotope space resembling the Calcium Aluminum
Inclusions (CAI) that is subsequently titrated with ice created in the nebular fringes. As discussed
by Laming et al. (2017), the inability to model the oxygen elemental and isotopic data of Genesis
solar wind sufficiently well to conclude with certainty that the solar wind is fractionated and to
an exact extent that it intersects with the CAI line. The consequence is that the link between solar
nebular processes and the sun is not solidly established. Though CAI studies have been a Rosetta
stone for development of nebular formation, chemistry, and alteration understanding, they are
15% of a rare meteorite class (CV) that constitute about 0.84% of meteorite falls (Sears 1998)
and in the interest of material balance for nebular oxygen, the other classes of meteorites need to
be included and how many reservoirs and processes are needed. If one assumes the 18O/17O ratio
of the solar system was on the CAI line, it is estimated to be 5.2 ± 0.2, which is quite different
from the galactic value of 4.1 (Young et al. 2011). A better evaluation of the triple oxygen isotope
composition of our solar system for a deepening understanding of their origins is needed.
In the case of oxygen, it is reasonably well established that based on the original paper
by Lee et al. (1980) there is a select group of FUN (Fractionated Unknown Nuclear) CAI that
possesses isotopic anomalies in oxygen that differ from other CAI and requires a very different
interpretation. The range of FUN triple oxygen isotope composition is schematically shown
in Figures 16 and 33. Lee et al. (1980) interpret the data as starting with the original CAI at
δ17O = δ18O = −40‰ and resembling most CAI. The CAI then undergo an extensive heating that
results in evaporative material loss and a 25‰ amu−1 mass-dependent fractionation. A variable
amount of this fractionation moves the individual core along a mass fractionation line, with
Allende inclusion HAL undergoing a greater fractionation (red stars in Fig. 33). Subsequent to
that heating event the CAI core undergoes an exchange with a gas reservoir that converge upon a
reservoir that also intersects with the normal CAI line at a point that is very near δ17O = δ18O = 0‰
suggesting that this is a major nebular gas phase reservoir. This finding is later supported by
further triple oxygen isotope and detailed mineralogy and petrology studies on FUN CAI (Krot
et al. 2010, 2014). These processes are schematically shown in Figure 33.
Krot et al. (2019) in a recent review have studied other inclusions (“normal CAI”) and
carefully documented the complex mineralogy to define the different processes associated with
its formation some 3–5 × 106 years before the complete melting of some CAI and meteorite
formation. Grossite (CaAl4O7) bearing CAI with the canonical initial 26Al/27Al ratio (5 × 10−5)
is a record of the antiquity of these CAI. The Δ17O composition of these CAI is −24 ± 2‰
(Krot et al. 2019). In the ensuing 3–5 million years the CAI will exchange and the most labile
oxygen species will exchange the most. Δ17O values of many CAI with Wark–Lovering rims vary
between −40 and −5‰ (Krot et al. 2019). As exchange proceeds and planetary body become
molten and have incomplete aqueous exchange, their Δ17O values range between 0 and −2‰ as
observed in CO and CV meteorites (Krot et al. 2019). The time scale is a few million years post
CAI formation. These values (Fig. 33) cover the full range of CAI isotopic composition shown
in Figure 16. Figure 33 clearly illustrates the end member, the minerals of exchange, and the
final product of exchange being the aqueous exchange near δ17O = δ18O = 0‰. The data are also
clear that there is a reservoir at δ17O = δ18O = ca. −50‰ and it is well produced but only resides
in a very minor phase with respect to the volume of the other meteorites.
Figure 33. Left panel: The close up of triple oxygen isotopic composition of selected CAIs. The fraction-
ation processes of CAI after formation are schematically shown by green dotted line. FUN CAIs distribute
in the region with green dotted lines. CCAM and PCM stands for “carbonaceous chondrite anhydrous
mineral line” and “primitive chondrule mineral line”, respectively. Modified from Krot et al. (2019). HAL
data from Lee et al. (1980). Right Panel: A scaled-down version of Figure 16.
Material balance of bulk meteorites oxygen isotopes

It has been suggested in the self-shielding model that the ice produced from CO
photolysis may vary depending on time, UV field and vary from 1000‰ downward (Lee et
al. 2008) (Fig. 31). The many problems associated with this have been discussed. The heavy
isotope effect observed by Sakamoto et al. (2007) does not require a heavy ice production and
may be a consequence of the oxidation process itself with the starting species at Δ17O = 0‰.
As discussed, it is well known that the Δ17O in atmospheric nitrate and ozone is near that
of the “heavy water” point (new-PCP) and would overlap at slightly higher temperatures.
Ozima et al. (2007) noticed that CAI (and some other meteorites with large non-zero Δ17O
values) is a minor component and the mean oxygen isotope composition of the solar system
should be characterized by Δ17O = 0‰. As shown in Figure 30, the δ18O composition of
86 Thiemens & Lin
99% of meteorites lie near the point δ18O = 5‰ and δ17O = 2.5‰. This is not considering the CAI.
The entire range of bulk meteorites is only ca. 4‰ near terrestrial fractionation line (Δ17O = 0‰).
If one uses NASAs meteorite collection (>20,000 Falls) as a very rough guide as to the amount
of material, approximately 92% are H, L, LL and carbonaceous 4%. Most carbonaceous
chondrites (expect for CI) possess negative Δ17O values and the lowest was found in CV
(ca. −6‰). Besides carbonaceous chondrites the other meteorites: ureilites, acapulcoites,
lodranites, angrites, aubrites, pallasites, howardites, eucrites and diogenites all have slightly
negative Δ17O values and altogether are only 2–3% of falls. Aubrites and eucrites have terrestrial
values (Δ17O = 0‰). In addition, Comet 81/P Wild 2 bulk composition (McKeegan et al. 2006)
is located near δ18O = 5‰ and δ17O = 2.5‰ and only a small fractionation (polymineralic
refractory grain) in the Stardust mission returned sample possess isotopic signature similar
to CAI (δ17O = δ18O = −40‰) (Fig. 16). δ18O values of 67P/Churyumov-Gerasimenko dust
and CO2 measured by the Rosetta Mission are also near δ18O = 5‰ (3 ± 50‰ and 10 ± 15‰,
respectively) (Hassig et al. 2017; Paquette et al. 2018), although there are large uncertainties
and δ17O values are unable to measure. We acknowledge that an isotopic balance cannot be
quantitatively done since 1) these numbers reflect the survivability of falls and the most fragile
ones are least abundant 2) this is a number of falls but one cannot say that this is equivalent to a
mass balance 3) in a true mass balance Earth, Mars, Venus (and Vesta) should be included with
earth at Δ17O = 0‰, Mars Δ17O = 0.3‰, Vesta Δ17O = −0.2‰ and Venus unknown. The current
safest statement with respect to the bulk meteorite, comet and planet compositions is that there
are both positive and negative values and the observed range is modest.
The basic considerations for a chemical local production of the meteorite oxygen isotopes
we note the following:
1. The presence of what has been observed is considered evidence of a heavy water
may be produced locally and isotopically similar to components in the terrestrial
atmosphere where the reactions are all known and isotopically characterized. The
observed meteoritic composition may be explained by chemical reactions of isotopic
compositions of water that is the same as the bulk meteorite, which is near terrestrial.
2. Based upon CAI and anomalous FUN (e.g., HAL) CAI oxygen isotopic measurements
and the back exchange captured in minerals of CAI converge on the major solar
system reservoir at a value near Δ17O = 0‰, this point is the most likely isotopic
value to consider as a starting value to explain bulk isotopic values. Given that
bulk Earth, moon, Mars and HED meteorites, comets, CAI and “heavy water”
lines lie near 5‰, this is reasonable value to assume as a nebular reservoir. Given
the data spread in several of the datum, this might vary by 1–2‰.
3. The solar component at δ17O = δ18O = −50‰ has yet to have a model that is capable
of quantitatively justifying the large correction applied to Genesis measurement and
Genesis inferred. Present models may not account for either elemental or isotopic
abundances (Laming et al. 2017). The link between solar nebular processes including
CAI may not be directly coupled.
If the bulk isotopic values observed in meteorites can be explained by a chemical model,
the requirement for multiple isotopic reservoirs that mix vanishes. It is well known from several
independent observations that in the symmetry driven reactions, it has been known since the
very first experiments that in the reaction both positive and negative reservoirs are created with
complete mass balance observed. The positive ozone reservoir is exactly matched by the negative
O2 reservoir (Fig. 5). There is no mixing required as it is simply one chemical reaction with
extensive theoretical and experimental measurements by numerous laboratories. Indeed, the
mass independent effect can be positive or negative according to the type of chemical reaction
sampling the intermedia complex (Robert et al. 2020). This has been observed for titanium but
not for oxygen yet. In a boarder view, a complete mass balance is not a necessary requirement.
The positive and negative Δ17O reservoir observed in the solar system could have been due to
different chemical reactions, which are required future identification though.
If the solar nebula is defined as at a value of Δ17O = 0‰ and approximately somewhere with
δ O between 3‰ and 6‰ and we consider a symmetry dependent reaction that produces positive
18
and negative reservoirs, the possible consequences may be see in Figures 5 and 30. The results show
that if we use the measured experiments of Chakraborty et al. (2013) for the reaction SiO + OH,
one of the first reactions in gas to particle conversion process, for an equal partitioning between a
nebular value centered around δ18O = 5 and δ17O = 2.5‰ (presumed starting nebula), all meteorite
classes, Earth, Mars, Vesta, comets, CAI positive terminus, “water ice”, chondrules fall within
this range. The consequence is that there is a requirement of only one reservoir. The entire solid
bodies of the inner solar system are created by one process. Transport, storage and mixing are not
required. The chemical reaction is required as SiO does not react with CO or H2O but rather OH is
considered the dominant reaction. The fractionation factor has been experimentally measured and
the results are interpretable in terms of known theory and a wide range of experimental tests by
multiple groups of the symmetry dependent reaction. The use of equal partitioning may be varied
and still incorporate all members. In the schematic arrow shown in Figure 30, we have employed
a 10‰ fractionation factor, a lower limit. The value may be higher.
Another important point to be made is the Earth’s atmosphere. As shown in the Figure 7
there is a δ17O = δ18O fractionation factor observed in the Earth’s atmosphere as those in the
solar system (Fig. 16). It should be noted that atmospheric ozone isotope effect is double
the magnitude of CAI (Fig. 16). It has more variation from the pure effect because of its
exchange reactions in oxidations and secondary photolysis in the stratosphere. Its origin
is molecular oxygen, not bulk earth. With the exception of CAI, most meteorites are not
plotted in Figure 16 but in an enlarged version in Figure 30 as their range is too small to
be seen on Figure 16. An important point to be made with respect to Figures 16 and 30 is
the mass balance. In Figure 16, there is a >100‰ enrichment factor in ozone (up to 150‰).
As shown in Figure 7 this very large positive effect has a slightly negative counterpart in
a large reservoir (O2). The >100‰ enrichment effect arises in the stratosphere in ozone
formation, which has a concentration of on average about 3 ppm. Tropospheric ozone has
the same large enrichment, and a concentration on average of 40 ppb. The troposphere is
ca. 80% and stratosphere ca. 10% of the Earth’s atmosphere by mass. The balance is between
the ozone and in part, molecular oxygen. As discussed by Miller and Pack (2021, this
volume), there is a mass independent Δ17O of −0.467‰ in O2 that is the counter balance.
It is not exact because the ozone enrichment is passed to CO2 in the stratosphere as need first
in return balloon and rocket samples by the Thiemens group (Thiemens et al. 1991, 1995)
and is maintained in the troposphere (Thiemens et al. 2014) (Fig. 7). This has been modeled
by Hoag et al. (2005) and is used to determine global primary productivity. The CO2 loses its
signal in steady state by exchange with water in the stomata of green plants as first detected
by Luz et al (1999). In a recent model, Young et al (2014) quantified the mass independent
component of the negative Δ17O value to be ca. 1/3 while the remaining signature is a result
of mass-dependent fractionation due to respiration. Mass-dependent fractionation processes
can lead to small Δ17O values but cannot account for all non-zero Δ17O components. Overall,
the Earth’s atmosphere is in a steady state with simultaneous positive and negative Δ17O
reservoirs. This is a natural proof that in a single interactive environment positive and negative
Δ17O reservoirs may be produced and does not require mixing of preserved reservoirs from
different places and production times. In Box 2, we summary several critical points of our
Chemical Mechanism Model for production of meteorite oxygen isotopic anomalies.
88 Thiemens & Lin
BOX 2
SUMMARY OF CHEMICAL MECHANISM MODEL FOR
PRODUCTION OF METEORITE OXYGEN ISOTOPIC ANOMALIES
1. There is only one reservoir required and no mixing. The basis for the production is the
slope one effect discovered for ozone and experimentally observed in SiO + OH reac-
tion, a likely first gas-to-solid formation reaction in the nebula. The process requires
no photons. The reaction is the step that determines isotopic composition of the solids
independent of the source of the oxygen, consistent with gas phase isotopic models
and experiments
2. The fractionation factor associated with the fractionation factor assumes the nebula has
isotopic composition on the earths bulk solid fractionation line at about δ18O = 5‰ and
δ17O = 2.5‰. The value is at the Nexus of bulk earth, enstatite chondrites, aubrites, CAI
and FUN CAI back reaction intersections
3. If the nebula starts a value at δ18O = 5‰ and δ17O = 2.5‰ and assumes the fractionation
factor measured for the SiO + OH of 10‰ (ΔΔ17O, θ = ln α17/ln α18 = 1) from that point,
one produces two reservoirs with the values dependent upon reservoir sizes. If both
are e.g., equal in equal amounts all reservoirs are overlapped: Earth, Mars, Vesta, all
meteorite classes, comets, CAI, chondrules. Other proportions will do the same.
4. At higher temperature, the isotope effect increases as it is known to be an inverse iso-
tope effect. The effect is still δ18O = δ17O. The high temperature condition is relevant to
the environment where the first solids were formed in the solar system.
5. The earth’s atmosphere shows the same simultaneous partitioning into positive and nega-
tive Δ17O reservoirs without mixing and δ18O extends past 100‰; twice that of CAI
6. The nebular reservoir is at a point in part set by the oxygen isotopes in individual min-
erals in CAI and FUN inclusions. The starting composition of the CAI has no influence
on the bulk nebula and the isotope composition. At high temperatures however, the first
condensates to form CAI lie along the slope one line and might be a minor but inter-
esting process that chemically produces CAI as proposed by Nobel Laureate Marcus
(2004). In this model CAI are produced from the nebula as proposed and not by the
sun’s inferred composition.
7. The solar wind measured value at δ18O = −102.3‰ and δ17O = −80.5‰ or the inferred
value at δ18O = δ17O = −60‰ does not play a role. As concluded by Laming et al. (2017)
the link between the solar wind and solar value and the protosolar nebula has not been
proven. The present model depends on the value of the nebula at the time of back reac-
tion of CAI at δ18O = 5‰, δ17O = 2.5‰.
Overall, the present model has avoided the need to create reservoirs and transport ice on
grains and mix at ultra-high δ18O precision by undefined processes. The major components of
the model have been experimentally measured and are consistent with the underlying quantum
chemical models by different groups. The measurements and models of the present and past
atmosphere on Earth and Mars is a testing ground for how reactions in nature occur, transfer
chemically and are preserved. The mass independent isotopic anomaly of water in sulfate on
Mars has been preserved for billions of years’ time scale for example (Thiemens 2006).
For the future, experiments that lead to a deeper understanding of the gas-to-particle
conversion process would be of high value. Isotopic measurements of the oxidation process
for different elements and oxidants will be of importance, especially for the magnitude of
effects as observed in present-day’s terrestrial atmosphere. Theory cannot predict the models
due to lack of precise details of the relevant potential energy surfaces. This will lead to
better understanding of the first condensates and CAI. The catalytic oxidation on surfaces is
fundamental to grain and larger and is not understood at the quantum level. There is a need
for studies of the temperature effect of the mass independent fractionation processes at high
temperature, especially near condensation temperatures. It is expected to be larger and the
reverse of classical mass dependent isotope effects. The CAI do not play a role in influencing
the isotopic composition of the meteorites, but they play a highly valuable role in defining the
nebular composition and the timing of the major events. Future studies that examine the role
of chemical processes in this process might be illuminative.
The new model simplifies the larger details in the meteoritic oxygen isotopic formational
process. The case for photochemical shielding has been discussed and the details of the
obstacles for the models listed, but not all included. The present model based on experiments,
theory and atmospheric and meteoritic observations reduces many long-standing barriers and
at the same time presents a range of new experiments that would deepen understanding and
are needed in the future.
ACKNOWLEDGEMENTS
We thank François Robert and Edward Young for their critical reviews and Andreas Pack
and Ilya Bindeman for their editorial handling and suggestions. All greatly improved this paper.
M.H.T. is supported by the Chancellors Associates fund. M.L. is supported by Key Research
Program of Frontier Sciences from Chinese Academy of Sciences (ZDBS-LY-DQC035).
REFERENCES
Airieau SA, Farquhar J, Thiemens MH, Leshin LA, Bao H, Young ED (2005) Planetesimal sulfate and aqueous
alteration in CM and CI carbonaceous chondrite. Geochim Cosmochim Acta 69:4167–4172
Ajay JS, Komarova KG, Remacle F, Levine RD (2018) Time-dependent view of an isotope effect in electron-nuclear
nonequilibrium dynamics with applications to N2. PNAS 115:5890–5895
Al-Halabi A, Van Dishoeck EF (2007) Hydrogen adsorption and diffusion on amorphous solid water ice. Mon Not R
Astron Soc 382:1648–1656
Andersson S, Al-Halabi A, Kroes GJ, van Dishoeck EF (2006) Molecular-dynamics study of photodissociation of
water in crystalline and amorphous ices. J Chem Phys 124:064715
Aston FW (1919) A positive ray spectrograph. Philos Mag 38:707–714
Babikov D (2017) Recombination reactions as a possible mechanism of mass-independent fractionation of sulfur
isotopes in the Archean atmosphere of Earth. PNAS 114:3062–3067
Babikov D, Semenov A, Teplukhin A (2017) One possible source of mass-independent fractionation of sulfur isotopes
in the Archean atmosphere of Earth. Geochim Cosmochim Acta 204:388–406
Babikov D, Kendrick BK, Walker RB, Schinke R, Pack RT (2003a) Quantum origin of an anomalous isotope effect
in ozone formation. Chem Phys Lett 372:686–691
Babikov D, Kendrick BK, Walker RB, Pack RT, Fleurat-Lesard P, Schinke R (2003b) Formation of ozone: Metastable
states and anomalous isotope effect. J Chem Phys 119:2577–2589
Bahou M, SchriverMazzuoli L, CamyPeyret C, Schriver A (1997) Photolysis of ozone at 693 nm in solid oxygen.
Isotopic effects in ozone reformation. Chem Phys Lett 273:31–36
Bainssahota SK, Thiemens MH (1989) A mass-independent sulfur isotope effect in the nonthermal formation of
S2F10. J Chem Phys 90:6099–6109
Bao H (2015) Sulfate: A time capsule for Earth’s O2, O3, and H2O. Chem Geol 395:108–118
Becquerel H (1896a) Sur les radiations émises par phosphorescence. CR Hebd Seances Acad Sci 122:420–421
Becquerel H (1896b) Sur les radiations invisibles émises par les corps phosphorescents. CR Hebd Seances Acad Sci 122:501
Becquerel H (1896c) Sur quelques propriétés nouvelles des radiations invisibles émises par divers corps
phosphorescents. CR Hebd Seances Acad Sci 122:559
Becquerel H (1896d) Sur les radiations invisibles émises par les sels d’uranium. CR Hebd Seances Acad Sci 122:689–694
Becquerel H (1896e) Sur les propriétés différentes des radiations invisibles émises par les sels d’uranium, et du
rayonnement de la paroi anticathodique d’un tube de Crookes. CR Hebd Seances Acad Sci 122:762
Becquerel H (1896f) Émission de radiations nouvelles par l’uranium métallique. CR Hebd Seances Acad Sci 122:1086
90 Thiemens & Lin
Becquerel H (1896g) Sur diverses propriétés des rayons uraniques. CR Hebd Seances Acad Sci 123:1086–1088
Benedix GK, Leshin LA, Farquhar J, Jackson T, Thiemens MH (2003) Carbonates in CM2 Chondrites: Constraints on
alteration conditions from oxygen isotopic compositions and petrographic observations. Geochim Cosmochim
Acta 67:1577–1588
Bertin M, Fayolle EC, Romanzin C, Oberg KI, Michaut X, Moudens A, Philippe L, Jeseck P, Linnartz H, Fillion JH
(2012) UV photodesorption of interstellar CO ice analogues: from subsurface excitation to surface desorption.
Phys Chem Chem Phys 14:9929–9935
Bethell T, Bergin E (2009) Formation and survival of water vapor in the terrestrial planet-forming region. Science
326:1675–1677
Bhattacharya SK, Thiemens MH (1988) Isotopic fractionation in ozone decomposition. Geophys Res Lett 15:9–12
Bhattacharya SK, Savarino J, Thiemens MH (2000) A new class of oxygen isotopic fractionation in photodissociation
of carbon dioxide: Potential implications for atmospheres of Mars and Earth. Geophys Res Lett 27:1459–1462
Bigeleisen J, Mayer MG (1947) Calculation of equilibrium constants for isotopic exchange reactions. J Chem Phys 15:261–267
Blake AJ, Gibson ST, Mccoy DG (1984) Photodissociation of 16O18O in the atmosphere. J Geophys Res-Atmos
89:7277–7284
Bradley CA, Urey HC (1932) The relative abundance of hydrogen isotopes in natural hydrogen. Phys Rev 40:0889–0890
Brickwedde FG (1982) Urey, Harold and the discovery of deuterium. Phys Today 35:34–39
Buchachenko (2001) Magnetic isotope effect: Nuclear spin control of chemical reactions. J Phys Chem A 105:9995–10011
Buchachenko (2013) Mass independent isotope effect. J Phys Chem B 117:2231–2238
Buchachenko (2018) Magnetic isotopes as a means to elucidate Earth and environmental chemistry. Russ
Chem Rev 87:727–740
Caballero-Gill RP, Herbert TD, Dowsett HJ (2019) 100-kyr paced climate change in the Pliocene warm period,
Southwest Pacific. Paleoceanogr Paleocl 34:524–545
Cano EJ, Sharp ZD, Shearer CK (2020) Distinct oxygen isotope compositions of the Earth and Moon. Nat Geosci 13:270–274
Chakraborty S, Bhattacharya SK (2003) Oxygen isotopic fractionation during UV and visible light photodissociation
of ozone. J Chem Phys 118:2164–2172
Chakraborty S, Ahmed M, Jackson TL, Thiemens MH (2008) Experimental test of self-shielding in vacuum ultraviolet
photodissociation of CO. Science 321:1328–1331
Chakraborty S, Ahmed M, Jackson TL, Thiemens MH (2009) Response to Comments on “Experimental test of self-
shielding in vacuum ultraviolet photodissociation of CO”. Science 324:1516
Chakraborty S, Davis RD, Ahmed M, Jackson TL, Thiemens MH (2012) Oxygen isotope fractionation in the vacuum
ultraviolet photodissociation of carbon monoxide: Wavelength, pressure, and temperature dependency. J Chem
Phys 137:024309
Chakraborty S, Yanchulova P, Thiemens MH (2013) Mass-independent oxygen isotopic partitioning during gas-phase
SiO2 formation. Science 342:463–466
Chakraborty S, Muskatel BH, Jackson TL, Ahmed M, Levine RD, Thiemens MH (2014) Massive isotopic effect in
vacuum UV photodissociation of N2 and implications for meteorite data. PNAS 111:14704–14709
Chakraborty S, Jackson TL, Rude B, Ahmed M, Thiemens MH (2016) Nitrogen isotopic fractionations in the low
temperature (80 K) vacuum ultraviolet photodissociation of N2. J Chem Phys 145:114302
Chakraborty S, Rude B, Ahmed M, Thiemens MH (2018) Carbon and oxygen isotopic fractionation in the products
of low-temperature VUV photodissociation of carbon monoxide. Chem Phys 514:78–86
Cicerone RJ, Mccrumb JL (1980) Photo-dissociation of isotopically heavy O2 as a source of atmosphere. Geophys
Res Lett 7:251–254
Ciesla FJ (2010) Residence times of particles in diffusive protoplanetary disk environments. I. Vertical Motions.
Astrophys J 723:514–529
Ciesla FJ (2014) The phases of water ice in the solar nebula. Astrophys J Lett 784:L1
Clayton RN (2002) Solar System—Self-shielding in the solar nebula. Nature 415:860–861
Clayton RN, Mayeda TK (1984) The oxygen isotope record in Murchison and other carbonaceous chondrites.
Earth Planet Sci Lett 67:151–161
Clayton RN, Grossman L, Mayeda TK (1973) Component of primitive nuclear composition in carbonaceous
Cole AS, Boering KA (2006) Mass-dependent and non-mass-dependent isotope effects in ozone photolysis: Resolving
theory and experiments. J Chem Phys 125:184301
Craig H (1953) The geochemistry of the stable carbon isotopes. Geochim Cosmochim Acta 3:53–92
Crockford PW, Kunzmann M, Bekker A, Hayles J, Bao H, Halverson GP, Peng Y, Bui TH, Cox GM, Gibson TM, Wörndle
S (2019) Claypool continued: Extending the isotopic record of sedimentary sulfate. Chem Geol 513:200–225
Cuppen HM, Herbst E (2007) Simulation of the formation and morphology of ice mantles on interstellar grains.
Astrophys J 668:294–309
Du F, Bergin EA (2014) Water vapor distribution in protoplanetary disks. Astrophys J 792:2
Dos Santos EC, Silva JCD, Duarte HA (2016) Pyrite oxidation mechanism by oxygen in aqueous medium. J Phys
Chem C 120:2760–2768
Dominguez G (2010) A heterogeneous chemical origin for the 16O-enriched and 16O-depleted reservoirs of the early
solar system. Astrophys J Lett 713:L59–L63
Dzade NY, Roldan A, de Leeuw NH (2016) DFT-D2 simulations of water adsorption and dissociation on the low-
index surfaces of mackinawite (FeS). J Chem Phys 144:174704
Emiliani C (1955) Pleistocene temperatures. J Geol 63:538–578
Farquhar J, Thiemens MH (2000) Oxygen cycle of the Martian atmosphere-regolith system: d17O of secondary phases
in Nakhla and Lafayette. J Geophys Res-Planet 105:11991–11997
Farquhar J, Thiemens MH, Jackson T (1998) Atmosphere-surface interactions on Mars: d17O measurements of
Farquhar J, Bao HM, Thiemens M (2000) Atmospheric influence of Earth’s earliest sulfur cycle. Science 289:756–758
Federman SR, Young ED (2009) Comments on “Experimental test of self-shielding in vacuum ultraviolet
photodissociation of CO”. Science 324:1516
Franz HB, Danielache SO, Farquhar J, Wing BA (2013) Mass-independent fractionation of sulfur isotopes during
broadband SO2 photolysis: Comparison between 16O- and 18O-rich SO2. Chem Geol 362:56–65
Fraser HJ, Collings MP, McCoustra MRS, Williams DA (2001) Thermal desorption of water ice in the interstellar
medium. Mon Not R Astron Soc 327:1165–1172
Fuchs GW, Cuppen HM, Ioppolo S, Romanzin C, Bisschop SE, Andersson S, van Dishoeck EF, Linnartz H (2009)
Hydrogenation reactions in interstellar CO ice analogues A combined experimental/theoretical approach. Astron
Astrophys 505:629–639
Furuya K, van Dishoeck EF, Aikawa Y (2016) Reconstructing the history of water ice formation from HDO/H2O and
D2O/HDO ratios in protostellar cores. Astron Astrophys 586:A127
Gao YQ, Marcus RA (2001) Strange and unconventional isotope effects in ozone formation. Science 293:259–263
Gao X, Thiemens MH (1989) Multi-isotopic sulfur isotope ratios (d33S, d34S, d36S) in meteorites. Meteoritics 24:269–269
Gao X, Thiemens MH (1993a) Isotopic composition and concentration of sulfur in carbonaceous chondrites. Geochim
Gao X, Thiemens MH (1993b) Variations of the isotopic composition of sulfur in enstatite and ordinary chondrites.
Giauque WF, Johnston HL (1929a) An isotope of oxygen, mass 18. Nature 123:318–318
Giauque WF, Johnston HL (1929b) An isotope of oxygen of mass 17 in the earth’s atmosphere. Nature 123:831–831
Green JA, Caswell JL, Fuller GA, Avison A, Breen SL, Brooks K, Burton MG, Chrysostomou A, Cox J, Diamond PJ,
Ellingsen SP (2009) The 6-GHz multibeam maser survey-I. Techniques. Mon Not R Astron Soc 392:783–794
Grossman L (1972) Condensation in primitive solar nebula. Geochim Cosmochim Acta 36:597–619
Harman CE, Pavlov AA, Babikov D, Kasting JF (2018) Chain formation as a mechanism for mass-independent
fractionation of sulfur isotopes in the Archean atmosphere. Earth Planet Sci Lett 496:238–247
Hasegawa TI, Herbst E, Leung CM (1992) Models of gas–grain chemistry in dense interstellar clouds with complex
organic-molecules. Astrophys J Suppl S 82:167–195
Hässig M, Altwegg K, Balsiger H, Berthelier JJ, Bieler A, Calmonte U, Dhooghe F, Fiethe B, Fuselier SA, Gasc S,
Gombosi TI (2017) Isotopic composition of CO2 in the coma of 67P/Churyumov–Gerasimenko measured with
ROSINA/DFMS. Astron Astrophys 605:A50
Hathorn BC, Marcus RA (1999) An intramolecular theory of the mass-independent isotope effect for ozone. I. J Chem
Phys 111:4087–4100
Hathorn BC, Marcus RA (2000) An intramolecular theory of the mass-independent isotope effect for ozone. II.
Numerical implementation at low pressures using a loose transition state. J Chem Phys 113:9497–9509
Heays AN, Ajello JM, Aguilar A, Lewis BR, Gibson ST (2014) The high-resolution extreme-ultraviolet spectrum of
N2 by electron impact. Astrophys J Suppl S 211:28
Heays AN, Bosman AD, van Dishoeck EF (2017) Photodissociation and photoionisation of atoms and molecules of
astrophysical interest. Astron Astrophys 602:A105
Heber VS, Baur H, Bochsler P, McKeegan KD, Neugebauer M, Reisenfeld DB, Wieler R, Wiens RC (2012) Isotopic
mass fractionation of solar wind: evidence from fast and slow solar wind collected by the Genesis Mission.
Astrophys J 759:121
Heidenreich JE, Thiemens MH (1983) A non-mass-dependent isotope effect in the production of ozone from
molecular-oxygen. J Chem Phys 78:892–895
molecular-oxygen - the role of molecular symmetry in isotope chemistry. J Chem Phys 84:2129–2136
Hoag KJ, Still CJ, Fung IY, Boering KA (2005) Triple oxygen isotope composition of tropospheric carbon dioxide as
a tracer of terrestrial gross carbon fluxes. Geophys Res Lett 32:L02802
Hogerheijde MR, Bergin EA, Brinch C, Cleeves LI, Fogel JK, Blake GA, Dominik C, Lis DC, Melnick G, Neufeld D,
Panić O(2011) Detection of the water reservoir in a forming planetary system. Science 334:338–340
Huang CH, Bhattacharya SK, Hsieh ZM, Chen YJ, Yih TS, Liang MC (2019) Isotopic fractionation in photolysis of
ozone in the Hartley and Chappuis Bands. Earth Space Sci 6:752–773
Hulston JR, Thode HG (1965a) Cosmic-ray-produced 36S and 33S in metallic phase of iron meteorites. J Geophys Res
70:4435–4442
92 Thiemens & Lin
Hulston JR, Thode HG (1965b) Variations in 33S, 34S and 36S contents of meteorites and their relation to chemical and
nuclear effects. J Geophys Res 70:3475–3484
Janssen C, Guenther J, Mauersberger K, Krankowsky D (2001) Kinetic origin of the ozone isotope effect: a critical
analysis of enrichments and rate coefficients. Phys Chem Chem Phys 3:4718–4721
Jeans J (1904) The Dynamical Theory of Gases. Cambridge University Press, Cambridge
Johnson RE, Quickenden TI (1997) Photolysis and radiolysis of water ice on outer solar system bodies. J Geophys
Res-Planet 102:10985–10996
Kaye JA (1986) Theoretical-analysis of isotope effects on ozone formation in oxygen photochemistry. J Geophys
Res-Atmos 91:7865–7874
Kaye JA, Strobel DF (1983) Enhancement of heavy ozone in the Earth’s atmosphere. J Geophys Res-Oceans 88:8447–8452
Kimura Y, Nuth JA, Chakpaborty S, Thiemens MH (2007) Non-mass-dependent oxygen isotopic fractionation in
smokes produced in an electrical discharge. Meteorit Planet Sci 42:1429–1439
Krot AN, Nagashima K, Ciesla FJ, Meyer BS, Hutcheon ID, Davis AM, Huss GR, Scott ERD (2010) Oxygen isotopic
composition of the sun and mean oxygen isotopic composition of the protosolar silicate dust: Evidence from
refractory inclusions. Astrophys J 713:1159–1166
Krot AN, Nagashima K, Wasserburg GJ, Huss GR, Papanastassiou D, Davis AM, Hutcheon ID, Bizzarro M (2014)
Calcium-aluminum-rich inclusions with fractionation and unknown nuclear effects (FUN CAIs): I. Mineralogy,
petrology, and oxygen isotopic compositions. Geochim Cosmochim Acta 145:206–247
Krot AN, Nagashima K, Simon SB, Ma C, Connolly HC, Huss GR, Davis AM, Bizzarro M (2019) Mineralogy,
petrography, and oxygen and aluminum–magnesium isotope systematics of grossite-bearing refractory
inclusions. Geochemistry-Germany 79:125529
Laming JM, Heber VS, Burnett DS, Guan Y, Hervig R, Huss GR, Jurewicz AJ, Koeman-Shields EC, McKeegan KD,
Nittler LR, Reisenfeld DB (2017) Determining the elemental and isotopic composition of the pre-solar nebula
from Genesis Data analysis: The case of oxygen. Astrophys J Lett 851:L12
Lee T, Mayeda TK, Clayton RN (1980) Oxygen isotopic anomalies in Allende inclusion Hal. Geophys Res Lett 7:493–496
Lee JE, Bergin EA, Lyons JR (2008) Oxygen isotope anomalies of the Sun and the original environment of the solar
system. Meteorit Planet Sci 43:1351–1362
Lee JE, Lee S, Baek G, Aikawa Y, Cieza L, Yoon SY, Herczeg G, Johnstone D, Casassus S (2019) The ice
composition in the disk around V883 Ori revealed by its stellar outbust. Nat Astron, 3:314–319
Lefebvre-Brion H, Field R (2004) The spectra and dynamics of diatomic molecules. Academic Press. Paperback
ISBN 9780124414563
Li X, Heays AN, Visser R, Ubachs W, Lewis BR, Gibson ST, van Dishoeck EF (2013) Photodissociation of interstellar
N2. Astron Astrophys 555:A14
Liang MC, Blake GA, Yung YL (2004) A semianalytic model for photo-induced isotopic fractionation in simple
molecules. J Geophys Res-Atmos 109:D10308
Liang MC, Irion FW, Weibel JD, Miller CE, Blake GA, Yung YL (2006) Isotopic composition of stratospheric ozone.
J Geophys Res-Atmos 111:D02302
Liang MC, Heays AN, Lewis BR, Gibson ST, Yung YL (2007) Source of nitrogen isotope anomaly in HCN in the
atmosphere of Titan. Astrophys J 664:L115–L118
Lin M, Zhang XL, Li MH, Xu YL, Zhang ZS, Tao J, Su BB, Liu LZ, Shen YA, Thiemens MH (2018) Five-S-isotope
evidence of two distinct mass-independent sulfur isotope effects and implications for the modern and Archean
atmospheres. PNAS 115:8541–8546
Lin M, Thiemens MH (2020) A simple elemental sulfur reduction method for isotopic analysis and pilot experimental
test of symmetry-dependent sulfur isotope effects in planetary processes. Geochem Geophys Geosyst 21:
e2020GC009051
Lindemann FA (1919) Note on the vapour pressure and affinity of isotopes. Philos Mag 38:173–181
Lindemann FA, Aston FW (1919) The possibility of separating isotopes. Philos Mag 37:523–534
Li WJ, Xu L, Liu X, Zhang J, Lin YT, Yao X, Gao HW, Zhang DZ, Chen JM, Wang WX, Harrison RM,
Zhang XY, Shao LY, Fu PQ, Nenes A, Shi ZB (2017) Air pollution–aerosol interactions produce more
bioavailable iron for ocean ecosystems. Sci Adv 3, e1601749
Liu P, Harman CE, Kasting JF, Hu YY, Wang JX (2019) Can organic haze and O2 plumes explain patterns of
sulfur mass-independent fractionation during the Archean? Earth Planet Sci Lett 526:115767
Lyons JR (2014) Photodissociation of CO isotopologues: Models of laboratory experiments and implications
for the solar nebula. Meteorit Planet Sci 49:373–393
Lyons JR (2020) An analytical formulation of isotope fractionation due to self-shielding. Geochim Cosmochim
Acta 282:177–200
Lyons JR (2001) Transfer of mass-independent fractionation in ozone to other oxygen-containing radicals in the
atmosphere. Geophys Res Lett 28:3231–3234
Lyons JR, Young ED (2005) CO self-shielding as the origin of oxygen isotope anomalies in the early solar nebula.
Nature 435:317–320
Lyons JR, Bergin EA, Ciesla FJ, Davis AM, Desch SJ, Hashizume K, Lee JE (2009a) Timescales for the evolution of
oxygen isotope compositions in the solar nebula. Geochim Cosmochim Acta 73:4998–5017
Lyons JR, Lewis RS, Clayton RN (2009b). Comments on “Experimental test of self-shielding in vacuum ultraviolet
Mahata S, Bhattacharya SK (2009a) Anomalous enrichment of 17O and 13C in photodissociation products of CO2:
Possible role of nuclear spin. J Chem Phys 130:234312
Mahata S, Bhattacharya SK (2009b) Temperature dependence of isotopic fractionation in CO2 photolysis. Chem Phys
Lett 477:52–56
Manian SH, Urey HC, Bleakney W (1934) An investigation of the relative abundance of the oxygen isotopes 16O:18O
in stone meteorites. J Am Chem Soc 56:2601–2609
Marcus RA (2004) Mass-independent isotope effect in the earliest processed solids in the solar system: A possible
chemical mechanism. J Chem Phys 121:8201–8211
Mauersberger K (1981) Measurement of heavy ozone in the stratosphere. Geophys Res Lett 8:935–937
Mauersberger K, Morton J, Schueler B, Stehr J, Anderson SM (1993) Multi-isotope study of ozone - implications for
the heavy ozone anomaly. Geophys Res Lett 20:1031–1034
Mauersberger K, Erbacher B, Krankowsky D, Gunther J, Nickel R (1999) Ozone isotope enrichment: Isotopomer-
specific rate coefficients. Science 283:370–372
Mauersberger K, Lammerzahl P, Krankowsky D (2001) Stratospheric ozone isotope enrichments-revisited. Geophys
Res Lett 28:3155–3158
McCabe JR, Boxe CS, Colussi AJ, Hoffmann MR, Thiemens MH (2005) Oxygen isotopic fractionation in the
photochemistry of nitrate in water and ice. J Geophys Res-Atmos 110: D15310
McCabe JR, Thiemens MH, Savarino J (2007) A record of ozone variability in South Pole Antarctic snow: Role of
nitrate oxygen isotopes. J Geophys Res-Atmos 112:D12303
Mcelroy MB, Yung YL, Nier AO (1976) Isotopic composition of nitrogen—implications for past history of Mars
atmosphere. Science 194:70–72
McKeegan KD, Aléon J, Bradley J, Brownlee D, Busemann H, Butterworth A, Chaussidon M, Fallon S, Floss C, Gilmour
J, Gounelle M (2006) Isotopic compositions of cometary matter returned by Stardust. Science 314:1724–1728
McKeegan KD, Kallio AP, Heber VS, Jarzebinski G, Mao PH, Coath CD, Kunihiro T, Wiens RC, Nordholt JE, Moses
RW, Reisenfeld DB (2011) The oxygen isotopic composition of the Sun inferred from captured solar wind.
Science 332:1528–1532
Miller MF, Pack A (2021) Why measure 17O? Historical perspective, triple-isotope systematics and selected
applications. Rev Mineral Geochem 86:1–34
Miller CE, Onorato RM, Liang MC, Yung YL (2005) Extraordinary isotopic fractionation in ozone photolysis.
Morton J, Schueler B, Mauersberger K (1989) Oxygen fractionation of ozone isotopes 48O3 through 54O3. Chem Phys
Lett 154:143–145
Morton J, Barnes J, Schueler B, Mauersberger K (1990) Laboratory studies of heavy ozone. J Geophys Res-Atmos
95:901–907
Mulliken RS (1964) The Rydberg states of molecules. Parts I–V. J Am Chem Soc 86:3183–3197
Muskatel BH, Remacle F, Thiemens MH, Levine RD (2011) On the strong and selective isotope effect in the UV
excitation of N2 with implications toward the nebula and Martian atmosphere. PNAS 108:6020–6025
Navon O, Wasserburg GJ (1985) Self-shielding in O2—a possible explanation for oxygen isotopic anomalies
in meteorites. Earth Planet Sci Lett 73:1–16
Ndengue SA, Schinke R, Gatti F, Meyer HD, Jost R (2012) Ozone photodissociation: isotopic and electronic branching
ratios for symmetric and asymmetric isotopologues. J Phys Chem A 116:12271–12279
Ndengue S, Madronich S, Gatti F, Meyer HD, Motapon O, Jost R (2014) Ozone photolysis: Strong isotopologue/
isotopomer selectivity in the stratosphere. J Geophys Res-Atmos 119:4286–4302
Nier AO (1947) A mass spectrometer for isotope and gas analysis. Rev Sci Instrum 18:398–411
Noble JA, Theule P, Mispelaer F, Duvernay F, Danger G, Congiu E, Dulieu F, Chiavassa T (2012) The desorption of
H2CO from interstellar grains analogues. Astron Astrophys 543:A5
Nunn M (2015) The Oxygen Isotopic Composition of Water in the Inner Solar System. PhD University of California,
San Diego
Omidvar K, Frederick JE (1987) Atmospheric odd oxygen production due to the photodissociation of ordinary and
isotopic molecular-oxygen. Planet Space Sci 35:769–784
Ozima M, Podosek FA, Higuchi T, Yin QZ, Yamada A (2007) On the mean oxygen isotope composition of the Solar
System. Icarus 186:562–570
Paquette JA, Engrand C, Hilchenbach M, Fray N, Stenzel OJ, Silen J, Ryno J, Kissel J, Team C (2018) The oxygen
isotopic composition (18O/16O) in the dust of comet 67P/Churyumov-Gerasimenko measured by COSIMA on-
board Rosetta. Mon Not R Astron Soc 477:3836–3844
Reinhardt P, Robert F (2018) On the mass independent isotopic fractionation in ozone. Chem Phys 513:287–294
Robert F, Tartese R, Lombardi G, Reinhardt P, Roskosz M, Doisneau B, Deng ZB, Chaussidon M (2020) Mass-
independent fractionation of titanium isotopes and its cosmochemical implications. Nat Astron 4:762–768
94 Thiemens & Lin
Rutherford E, Soddy F (1903) Radioactive change. Philos Mag 5:576–591

Sakamoto N, Seto Y, Itoh S, Kuramoto K, Fujino K, Nagashima K, Krot AN, Yurimoto H (2007) Remnants of the
early solar system water enriched in heavy oxygen isotopes. Science 317:231–233
Savarino J, Thiemens MH (1999) Mass-independent oxygen isotope (16O, 17O, 18O) fractionation found in Hx, Ox
reactions. J Phys Chem A 103:9221–9229
Schmidt JA, Johnson MS, Schinke R (2013) Carbon dioxide photolysis from 150 to 210 nm: Singlet and triplet
channel dynamics, UV-spectrum, and isotope effects. PNAS 110:17691–17696
Sears DWG (1998) The case for rarity of chondrules and calcium-aluminum-rich inclusions in the early solar system
and some implications for astrophysical models. Astrophys J 498:773–778
Shaheen R, Abramian A, Horn J, Dominguez G, Sullivan R, Thiemens MH (2010) Detection of oxygen isotopic
anomaly in terrestrial atmospheric carbonates and its implications to Mars. PNAS 107:20213–20218
Shaheen R, Abaunza MM, Jackson TL, McCabe J, Savarino J, Thiemens MH (2014) Large sulfur-isotope anomaly in
nonvolcanic sulfate aerosol and its implications for the Archean atmosphere. PNAS 111:11979–11983
Shaheen R, Niles PB, Chong K, Corrigan CM, Thiemens MH (2015) Carbonate formation events in ALH 84001 trace
the evolution of the Martian atmosphere. PNAS 112:336–341
Soddy F (1913) Intra-atomic charge. Nature 92:399–400
Spelsberg D, Meyer W (2001) Dipole-allowed excited states of N2: Potential energy curves, vibrational analysis, and
absorption intensities. J Chem Phys 115:6438–6449
Teplukhin A, Babikov D (2018a) Several levels of theory for description of isotope effects in ozone: Symmetry effect
and mass effect. J Phys Chem A 122:9177–9190
Teplukhin A, Babikov D (2018b) Properties of Feshbach and “shape”-resonances in ozone and their role in
recombination reactions and anomalous isotope effects. Faraday Discuss 212:259–280
Teplukhin A, Gayday I, Babikov D (2018) Several levels of theory for description of isotope effects in ozone: Effect
of resonance lifetimes and channel couplings. J Chem Phys 149:164302
Thiemens MH (1999) Atmosphere science—Mass-independent isotope effects in planetary atmospheres and the early
solar system. Science 283:341–345
Thiemens MH (2006) History and applications of mass-independent isotope effects. Annu Rev Earth Planet Sci 34:217–262
Thiemens MH (2018) The discovery of chemically produced mass independent isotope effects: The physical chemistry
basis and applications to the early solar system, planetary atmospheres, and the origin of life. Acceptance Speech
of the Leonard Medal July 2017. Meteorit Planet Sci 54:231–248
Thiemens MH, Heidenreich JE (1983) The mass-independent fractionation of oxygen—a novel isotope effect and its
possible cosmochemical implications. Science 219:1073–1075
Thiemens MH, Jackson T (1990) Pressure dependency for heavy isotope enhancement in ozone formation. Geophys
Res Lett 17:717–719
Thiemens MH, Lin M (2019) Use of isotope effects to understand the present and past of the atmosphere and climate
and track the origin of life. Angew Chem Int Edit 58:6826–6844
Thiemens MH, Jackson T, Mauersberger K, Schueler B, Morton J (1991) Oxygen isotope fractionation in stratospheric
CO2. Geophys Res Lett 18:669–672
Thiemens MH, Jackson T, Zipf EC, Erdman PW, Vanegmond C (1995) Carbon-dioxide and oxygen-isotope anomalies
in the mesosphere and stratosphere. Science 270:969–972
Thiemens MH, Chakraborty S, Dominguez G (2012) The physical chemistry of mass-independent isotope effects and
their observation in nature. Annu Rev Phys Chem 63:155–177
Thiemens MH, Chakraborty S, Jackson TL (2014) Decadal D17O record of tropospheric CO2: Verification of a
stratospheric component in the troposphere. J Geophys Res-Atmos 119:6221–6229
Tian F, Kasting JF, Solomon SC (2009) Thermal escape of carbon from the early Martian atmosphere. Geophys Res
Lett 36:L02205
Tielens AGGM, Hagen W (1982) Model-calculations of the molecular composition of inter-stellar grain mantles.
Astron Astrophys 114:245–260
Troe J (1977) Theory of thermal unimolecular reactions at low-pressures 2. Strong collision rate constants—
Applications. J Chem Phys 66:4758–4775
Tyra MA, Farquhar J, Wing BA, Benedix GK, Jull AJT, Jackson T, Thiemens, MH (2007). Terrestrial alteration of
Antarctic CM chondrite meteorites in a suite of Antarctic CM chondrites: evidence from oxygen and carbon
isotopes. Geochim Cosmochim Acta 71:782–795
Uemura R, Masson-Delmotte V, Jouzel J, Landais A, Motoyama H, Stenni B (2012) Ranges of moisture-source temperature
estimated from Antarctic ice cores stable isotope records over glacial-interglacial cycles. Clim Past 8:1109–1125
Urey HC (1947) The thermodynamic properties of isotopic substances. J Chem Soc:562–581
Urey HC, Brickwedde FG, Murphy GM (1932a) A hydrogen isotope of mass 2 and its concentration. Phys Rev 40:1–15
Urey HC, Brickwedde FG, Murphy GM (1932b) A hydrogen isotope of mass 2. Phys Rev 39:164–165
van Dishoeck EF, Herbst E, Neufeld DA (2013) Interstellar water chemistry: from laboratory to observations. Chem
Rev 113:9043–9085
Vandishoeck EF, Black JH (1988) The photodissociation and chemistry of interstellar CO. Astrophys J 334:771–802
Velivetskaya TA, Ignatiev AV, Yakovenko VV, Vysotskiy SV (2018) Experimental studies of the oxygen isotope
anomalies (D17O) of H2O2 and their relation to radical recombination reactions. Chem Phys Lett 693:107–113
Velivetskaya TA, Ignatiev AV, Budnitskiy SY,Yakovenko VV, Vysotskiy SV (2016) Mass-independent fractionation of oxygen
isotopes during H2O2 formation by gas-phase discharge from water vapour. Geochim Cosmochim Acta 193:54–65
von Steiger R, Zurbuchen TH (2016) Solar metallicity derived from in situ solar wind composition. Astrophys J 816:13
Watanabe N, Horii T, Kouchi A (2000) Measurements of D2 yields from amorphous D2O ice by ultraviolet irradiation
at 12 K. Astrophys J 541:772–778
Wen J, Thiemens MH (1990) An apparent new isotope effect in a molecular decomposition and implications for
nature. Chem Phys Lett 172:416–420
Wen J, Thiemens MH (1991) Experimental and theoretical-study of isotope effects on ozone decomposition. J
Geophys Res-Atmos 96:10911–10921
Woitke P, Thi WF, Kamp I, Hogerheijde MR (2009) Hot and cool water in Herbig Ae protoplanetary disks A challenge
for Herschel. Astron Astrophys 501:L5–L8
Wolf S, Bitter M, Krankowsky D, Mauersberger K (2000) Multi-isotope study of fractionation effects in the ozone
formation process. J Chem Phys 113:2684–2686
Yabushita A, Hama T, Iida D, Kawanaka N, Kawasaki M, Watanabe N, Ashfold MNR, Loock HP (2008) Release
of hydrogen molecules from the photodissociation of amorphous solid water and polycrystalline ice at 157 and
193 nm. J Chem Phys 129:044501
Yang JM, Epstein S (1987a) The effect of the isotopic composition of oxygen on the non-mass-dependent isotopic
fractionation in the formation of ozone by discharge of O2. Geochim Cosmochim Acta 51:2011–2017
Yang JM, Epstein S (1987b) The effect of pressure and excitation-energy on the isotopic fractionation in the formation
of ozone by discharge of O2. Geochim Cosmochim Acta 51:2019–2024
Yin QZ, Shi X, Chang C, Ng CY (2009) Comments on “Experimental test of self-shielding in vacuum ultraviolet
Young ED (2001) The hydrology of carbonaceous chondrite parent bodies and the evolution of planet progenitors.
Philos Trans R Soc Lond A 359:2095–2110
Young ED (2007a) Time-dependent oxygen isotopic effects of CO self shielding across the solar protoplanetary disk.
Young ED (2007b) Geochemistry - Strange water in the solar system. Science 317:211–212
Young ED, Gounelle M, Smith RL, Morris MR, Pontoppidan (2011) Astronomical oxygen isotopic evidence for
supernova enrichment of the solar system birth environment by propagating star formation. Astrophys J 729:43–55
Yung YL, Demore WB, Pinto JP (1991) Isotopic exchange between carbon-dioxide and ozone via O(1d) in the
stratosphere. Geophys Res Lett 18:13–16
Yurimoto H, Kuramoto K (2004) Molecular cloud origin for the oxygen isotope heterogeneity in the solar system.
Science 305:1763–1766
Zmolek P, Xu XP, Jackson T, Thiemens MH, Trogler WC (1999) Large mass independent sulfur isotope fractionations
during the photopolymerization of 12CS2 and 13CS2 . J Phys Chem A 103:2477–2480
Vol. 86 pp. 97–137, 2021 3
Climbing to the Top of Mount Fuji: Uniting Theory and

Observations of Oxygen Triple Isotope Systematics
Laurence Y. Yeung
Department of Earth, Environmental and Planetary Sciences
Rice University
Houston, TX 77005, U.S.A.
lyeung@rice.edu
Justin A. Hayles
Jacobs, Astromaterials Research and Exploration Science, Johnson Space Center
National Aeronautics and Space Administration
Houston, TX 77058, U.S.A.
justin.a.hayles@nasa.gov
INTRODUCTION
The near-simultaneous discovery of both minor isotopes of oxygen in 1929 was a watershed
moment for modern science, to say nothing of its impacts on isotope geochemistry. At the time,
oxygen was the international standard for atomic weight, as it had been for over twenty-five
years. However, chemists and physicists had grown fond of different definitions: physicists used
the weight of the 16O atom, while chemists used half the weight of atmospheric oxygen (O2) to
define the precise weight of 16 atomic mass units. While they usually avoided direct conflicts,
these contrasting definitions found an unexpected impasse in the near-infrared absorption
spectrum of atmospheric oxygen: it contained a series of faint lines that could not be explained
by any known atmospheric constituents (Dieke and Babcock 1927; Mulliken 1928).
The key breakthrough came when Herrick Johnston and William Giauque, who were
separately studying the thermodynamics of oxygen, realized that small amounts of heavier
oxygen isotopes in O2, 17O and 18O, could generate those weak absorptions. With no prior
precedent for these isotopes aside from Ernest Rutherford’s α-particle-driven transmutation of
14
N into 17O (Rutherford 1919), Johnston and Giauque relied on the theoretical mass dependence
of vibrational frequencies and spectroscopic selection rules from quantum mechanics—then
still a young theory—to predict where in the infrared 16O18O and 16O17O would absorb.
Their predictions were quantitative and definitive (Giauque and Johnston 1929a,b). Together
with subsequent observations by Malcolm Dole, George Lane and others (Dole 1935; Dole and
Jenks 1944; Dole et al. 1947; Epstein and Mayeda 1953; Lane and Dole 1956)—the discovery
of 17O and 18O led the International Union of Pure and Applied Chemistry to redefine atomic
weight relative to carbon-12 in 1962, which is still valid today.
Oxygen triple-isotope geochemistry owes its existence to this discovery, which emerged from
a clever application of theory to a puzzling observation. Isotope geochemists today still rely on
theory to fill this same role: to predict and explain observations of nature, using that nexus to make
fundamental advances of broader scientific relevance. However, the targets have evolved; our ability
to measure variations in oxygen-isotope abundances in natural materials has been transformed by
high-precision isotope ratio mass spectrometry and innovative analytical techniques (Nier 1947;
McCrea 1950; Sharp 1990; Baker et al. 2002). These methods have improved in the past few
decades to yield parts-per-million (ppm) levels of precision in oxygen-isotope ratios for a wide
1943-2666/21/0086-0003$05.00 (online)
98 Yeung & Hayles
variety of natural materials (Luz et al. 1999; Luz and Barkan 2000; Barkan and Luz 2005; Pack
et al. 2013; Pack and Herwartz 2014). Concurrently, the explosive growth of computing power,
particularly the recent advances in parallel computing, have made first-principles electronic
structure calculations more accessible than ever to the geochemist: large chemical systems once
considered intractable are being studied routinely in silico. How well can theory and observations
work together toward new insights in these much more exacting times?
Oxygen triple-isotope geochemistry focuses in two general areas, both of which have
been informed and advanced by theory. The mass-independent isotopic fractionation of
oxygen preserved in early solar-system condensates (Clayton et al. 1973) and atmospheric
ozone (Mauersberger 1981; Thiemens and Heidenreich 1983) stimulated a new wave of
theoretical work in the 1990s and 2000s to understand the physical-chemical origins of
those anomalies, leading to the discovery of other mass-independent isotope fractionation
mechanisms for oxygen and other elements (Röckmann et al. 1998; Hathorn and Marcus
1999; 2000; Farquhar et al. 2001; Gao and Marcus 2001; Clayton 2002; Babikov et al. 2003;
Marcus 2004; Schinke et al. 2006; Bergquist and Blum 2007; Sun and Bao 2011). These
signals have led to fundamental discoveries about the formation and evolution of the solar
system, Earth history, and biogeochemical cycling (Thiemens 2006). Other chapters in this
volume will cover those developments, so we will focus here on the second major area of
oxygen triple-isotope geochemistry, that of natural variations in mass-dependent fractionation.
The role of theory in this area is clear: it provides independent, physics-based reference
points for oxygen triple-isotope partitioning. Such reference points are invaluable because of
the myriad ways in which oxygen isotopes can be fractionated, transported, and mixed to yield
the parts-per-million (ppm) level variability that is routinely observed and interpreted (Juranek
and Quay 2013; Herwartz et al. 2014; Luz et al. 2014; Young et al. 2016; Bindeman et al. 2018;
Sharp et al. 2018); theory can help distinguish the signal from the noise. Theoretical reference
points can also lead to analytical innovations and the characterization of unusual processes
(Richet et al. 1977; Thiemens 2006; Eiler 2007; Bao et al. 2015; Hayles et al. 2017).Even
when imperfect, a theoretical prediction constitutes a vital third approach to high-precision
isotope geochemistry that has a different set of biases from experiments and observations.
Are current theoretical methods sufficiently accurate to benchmark oxygen triple-isotope
geochemistry? Analogous investigations in the clumped-isotope community (Wang et al. 2004;
Schauble et al. 2006; Guo et al. 2009; Piasecki et al. 2016, 2018) suggest that they are not
far off: robust features of isotope partitioning have been observed across theory, experiment,
and observations to within several tens of ppm. Some disagreements in isotopic fractionation
factors have persisted, however (e.g., for acid digestion fractionation: Guo et al. 2009; Murray
et al. 2016; Müller et al. 2017; Petersen et al. 2019; Swart et al. 2019; Zhang et al. 2020),
resulting in barriers to reproducibility and (in our view) limits the certainty with which
isotopic records can be interpreted. Breaking through these barriers to find agreement at the
single-digit ppm level would be both a triumph for theoretical approaches and transformative
for oxygen triple-isotope geochemistry. This article aims to evaluate how close the state of the
art is to this target, and hopefully to guide the way forward.
In this article, we will first cover basic concepts and notation relevant to oxygen triple-
isotope geochemistry. Second, we will examine what theory predicts for oxygen triple-isotope
variability in chemical processes. Third, we will examine the systematic biases that may be
present in theoretical approaches, with special attention paid to first-principles electronic
structure calculations. Fourth, we will consider the current limits of analytical accuracy and
the complications introduced by physical effects in real systems. Finally, we will revisit the
triple-isotope mass dependence of carbonate acid digestion as a case study of how theory and
experiment can work together to improve both each other and ultimately also our understanding
of a process that is vital for the emerging area of carbonate-based paleohydrology.
Uniting Theory and Observations of Oxygen Triple Isotope Systematics 99
BASIC CONCEPTS
The equations
Molecules as physical entities can be viewed as collections of balls and springs, with the
balls being nuclei of atoms and the springs being the bonds that tie them together. The strength
of chemical bonds is tied primarily to electrons, which are each about 1/30,000th the mass of
an oxygen atom. Consequently, changing the mass of an oxygen atom has little effect on basic
molecular structure or chemistry in most cases. This simplification is known as the Born-
Oppenheimer approximation (Born and Oppenheimer 1927). Exchanging an oxygen atom for
a heavier isotope, however, will change the frequencies at which a molecule naturally vibrates:
heavier masses connected to the same springs will yield lower frequencies of vibration. It is these
subtle variations in vibrational frequencies in isotopically substituted molecules, which have
subtle effects on molecular enthalpy and entropy, that drive a majority of chemical isotope effects.
Bigeleisen, Goeppert-Mayer, and Urey are credited with deriving expressions describing
the thermodynamics of isotopically substituted gas-phase molecules (Bigeleisen and Goeppert-
Mayer 1947; Urey 1947). They used the reduced partition function ratio (RPFR) between two
isotopic variants of molecules as a building block to describe isotope-exchange reactions.
For the exchange of a single isotope,
 *i   e Ui / 2   1  e Ui 

*
3 N 6
s
RPFR     Ui / 2   Ui* 
 *K (1)
i 1  i  MMI  e  ZPE  1  e  EXC s
Here, νi is the ith harmonic vibrational frequency of the molecule (of 3N – 6 total for nonlinear
molecules, 3N − 5 for linear molecules), Ui = hνi / kBT (or hcνi / kBT if νi is expressed in
wavenumbers, where c is the speed of light), N is the number of atoms in the molecule,
h is Planck’s constant, kB is Boltzmann’s constant, and T is the temperature in Kelvin.
This equation represents the thermodynamic preference to form an isotopically substituted
molecule (denoted above by an asterisk) compared to the isotopically “normal” molecule in an
idealized equilibrium between the species and free atoms (e.g., 16O16O + 18O ⇌ 16O18O + 16O).
The RPFR can thus be thought of as the equilibrium constant K for the reaction scaled by the
ratio of isotopologue symmetry numbers s/s*, e.g., s = 1 for 16O16O and s* = 2 for 16O18O.
The equilibrium 18O/16O isotopic fractionation between species A and species B in the
reaction A + B* ⇌ A* + B, in which a single isotope is exchanged, is written as

 18
18

O / 16 O
A
 


 
O / 16 O   18 RPFR
atomic  (2)
18
 AB   A


18

O / 16 O
B
  18 RPFR B



18

O / 16 O  
atomic 
For exchange of n isotopes of the same atom in a molecule, the reduced partition function ratio
in Equation (1) is raised to the 1/n power, i.e., (RPFR)1/n.
Many excellent discussions of Equation (1) exist (Schauble 2004; Liu et al. 2010), so we will
highlight only a few elements most germane to this discussion. First, the expression has inherent
assumptions: it is derived assuming that bonds are perfect springs holding atoms together (the
so-called harmonic oscillator approximation) and that bond lengths are constant (the so-called
rigid-rotor approximation). Clearly, these two approximations cannot simultaneously be true,
but they simplify the math considerably, and are good approximations of many real systems at
Earth-surface temperatures. Moreover, the approximations allow one to separate the energies
100 Yeung & Hayles
associated with molecular vibrations (ZPE and EXC subscripts) from those involving translations
and rotations (MMI subscript). The ZPE (zero-point energy) term corresponds to the ratio of
“occupancies” of the lowest-energy vibrational states ν0 and ν0*, i.e., how often an isotopically
substituted molecule would be in that state compared to its unsubstituted counterpart at a given
temperature. The EXC term corresponds to the occupancies of the higher vibrational states.
The MMI term is deceptively simple because it relates the ratio of masses and moments of inertia
of the molecule (from classical mechanics) to a ratio of vibrational frequencies (from quantum
mechanics) through the Teller–Redlich isotopic product rule (Redlich 1935). Indeed, without
the harmonic-oscillator and rigid-rotor approximations, Equation (1) would be much more
complicated. There are many notable instances in which these approximations are insufficient
for describing isotope fractionation; we will return to these later.
In liquids, solutions, and solids—which are in many cases the phases most relevant to
oxygen triple isotopes—the potential range of atom–atom interactions broadens considerably
compared to an isolated gas-phase molecule. Oxygen atoms in aqueous oxides (e.g., HCO3,
SO42−, and H4SiO4) interact meaningfully with the solvent, so one expects aqueous isotopic
fractionation of oxygen to depend on the strength and number of these interactions; in addition
to influences on vibrations of the solute, there are simply more relevant vibrational modes i to
consider in Equation (1), and the “sphere of influence” of an isotopic substitution is larger.
These interactions cause narrow, gas-phase vibrational absorption lines to shift and
broaden (Fig. 1). Weak long-range interactions, including those arising from long-range order
in solids, can comprise an important component of the RPFRs in condensed-phase systems.
Schauble (2004) introduced the following equation, based on the work of Kieffer (1982), to
represent the reduced partition function ratio in terms of a continuum of vibrations in solids:
 max  e U * / 2  
3/ 2
  0 ln  U * 
 
 g  d 
* *
 m 1
RPFR   *  exp  1 e  
(3)
 U / 2
 
m  n  max e

 ln    g    d 
 0
 1 e 
U

 
Here, m and m* are the masses of the normal and rare (commonly heavy) isotope being
exchanged, respectively, n is the number of atoms exchanged in a unit cell, and g is the
vibrational density of states, which is proportional to the number of vibrational modes in the
frequency window between ν and ν + dν. The upper frequency limit of the integrals is the
highest vibrational frequency νmax in the crystal. Note that crystals have exceedingly large
molecular weights, rendering insignificant the effect of occasional isotopic substitutions on
translations and rotations. The related terms in the Teller–Redlich isotopic product rule thus
cancel, leaving only the leading (m/m*)3/2 term.
Because many of the terms in Equation (3) are either uncertain or poorly constrained,
one often must approximate it. Elcombe and Pryor (1970) showed that the continuous crystal
spectrum of the ionic solid CaF2 can be well-approximated by using a set of discrete frequencies
as constraints on lattice vibrations simulated to satisfy known physical properties (e.g., the
dielectric and elastic constants). This approach has not yet been utilized for predictions of triple
oxygen-isotope fractionation, although it has been used to examine bulk 18O/16O fractionation
factors in minerals (Bottinga 1968; Kieffer 1982; Chacko et al. 1991; Schauble et al. 2006).
Kinetic isotope effects are typically approached through the lens of adiabatic transition-
state theory (Evans and Polanyi 1935; Eyring 1935) using the schematic depicted in Figure 2.
Transition-state theory presumes that reactions proceed first through a pre-equilibrium between
the reactants and an activated complex (i.e., a transition state, denoted by a “‡” superscript)
followed by a unidirectional transformation of that complex into the products. The
transformation of reactant A to product B through a transition state TS would thus be written:
Figure 1. Broadening of vibrational transitions in condensed phases. Data from NIST Mass Spectrometry
Data Center (2020).

Figure 2. Schematic of reaction pathway for a generic O-atom exchange reaction.
A ⇌ [TS]‡ → B
The transition state is a saddle point on the potential energy surface, meaning that all nuclear
vibrations are bound except that which defines the bond being made or broken. The axis that
describes the path of that bond is known as the reaction coordinate (see Fig. 2). The rate of a
reaction can be written in terms of the partition functions Q of the reactants (i.e., the product
of the partition functions for each reactant) and the transition state:
k BT Q ‡
‡
E
k  tun e k BT
(4)
h Qreactants
Here, ηtun is a correction for nuclear tunneling (a uniquely quantum-mechanical effect discussed
below) and Q refers to the nuclear partition functions only. The exponential term represents
the effects of the electronic activation barrier, ΔE‡. Assuming a full separation of nuclear and
electronic motions (i.e., the Born–Oppenheimer approximation; see above), ΔE‡ is the same
for all isotopologues because they all lie on the same electronic potential energy surface. The
kinetic 18O/16O fractionation factor is thus (Bigeleisen 1949; Bigeleisen and Wolfsberg 1958):
102 Yeung & Hayles
18
k 18
tun RPFR ‡
18
(5)
18

kin 
16
k 16
tun 18
RPFR reactants
where k is the reaction-rate coefficient and
 e U i / 2   1  e U i 
‡* ‡
 ‡*j 3 N 7   ‡*
i 
RPFR ‡*      
 ‡j i  j   ‡i  MMI  e Ui / 2 
‡ 
 1  e Ui‡*


(6)
ZPE   EXC
The reduced partition function ratio for the transition state, RPFR‡, is nearly identical to
that in Equation (1) except that the vibrational mode j that corresponds to the transforming
bond is removed from the ZPE and EXC terms. That mode is unbound because it defines the
reaction coordinate (Fig. 2); thus it is not free to vibrate in the transition state; consequently, it
has an imaginary frequency, νj‡. The MMI term, however, expressed using the Teller–Redlich
product rule, includes this imaginary frequency, so the ratio νj‡*/νj‡ remains outside the product.
In addition, the ηtun terms in Equation (5) also depend on νj‡; they are approximately equal to
1 + Uj‡2/24 (Wigner 1932; Bigeleisen and Wolfsberg 1958). Note that these expressions also
employ the harmonic-oscillator and rigid-rotor approximations.
Quantifying mass-dependent fractionation
Triple-isotope variations can be quantified in many ways, but we will restrict the present
discussion to several recurring terms used in the mass-dependent fractionation literature: κ, θ,
and λ (Bao et al. 2016; Dauphas and Schauble 2016). The three terms cover the spectrum of
theoretical to empirical quantities, with κ being theoretically straightforward but difficult to
verify by experiment, and λ being observationally straightforward but the least directly related
to theory. Each has its utility for different objectives in oxygen triple-isotope geochemistry.
The triple-isotope κ value is characteristic of isotope-exchange equilibrium between
an isotopically substituted species and a free atom. It is defined as the ratio of the natural
logarithms of the RPFRs (Cao and Liu 2011):
ln 17 RPFR A
 (7)
ln 18 RPFR A
This term, while abstract, has the benefit of putting all oxygen triple-isotope variations
against a common reference state (atomic oxygen, which has no chemical isotopic
preferences). Under the approximations used in Equation (1), it is bounded in the triple-
oxygen system, in the high-temperature limit, to a maximum value of 0.5305 because
κ(T→∞) = (1/m16–1/m17)/(1/m16–1/m18), in which the subscripts denote the isotope masses.
Most oxygen-containing species thus far studied have temperature-dependent κ values lying
between 0.527 and 0.530 (Young et al. 2002; Cao and Liu 2011; Hayles et al. 2017, 2018).
Individual contributions to κ from can be separated by the relation (Hayles et al. 2017):
complete 
  ln 
i i
18
i
(8)
 ln  i
18
i
in which Πβi = RPFR, and βi refers to the contributions from individual vibrational modes or
multiplicative corrections to the RPFR (e.g. anharmonic corrections). Each pair of 17βi and 18βi
values thus yields a κi value for that mode. The contribution from the imaginary frequency of
a transition state in Equation (6) in this framework is defined as βIF‡ = νj‡*/νj‡. Corrections for
deviations from the harmonic-oscillator and rigid-rotor assumptions can also be incorporated
this way, allowing one to understand their influence on κcomplete (see below).
A more empirically tractable quantity is the triple-isotope exponent θ, which describes

the triple-isotope fractionation between two species or phases, rather than a single species or
phase relative to an atom. The quantity θ is derived from the relation 17αA–B = (18αA–B)θ for
fractionation factors and is conventionally defined in logarithmic form, i.e.,
ln 17  A  B (9)
A  B 
ln 18  A  B
Because uncertainties in 17αA–B and 18αA–B values covary in laboratory measurements, θA–B values
for many processes can be derived from laboratory measurements to high precision (Angert et
al. 2003, 2004; Helman et al. 2005; Luz and Barkan 2005; Barkan and Luz 2007; Pack and
Herwartz 2014; Sharp et al. 2016; Sengupta and Pack 2018; Stolper et al. 2018; Wostbrock et al.
2018, 2020; Ash et al. 2020); thus, θA–B values have been indispensable as a mediator between
experiment and theory. To first order, θA–B values are insensitive of temperature (Young et al.
2002), but experimental and theoretical work has shown that a subtle temperature dependence
can be observed in many cases (Cao and Liu 2011; Pack and Herwartz 2014; Casado et al. 2016;
Sharp et al. 2016; Hayles et al. 2018; Stolper et al. 2018; Wostbrock et al. 2018).
The triple-isotope coefficient λ has multiple meanings in the literature. We will simplify
them here as λ and λRL. The former is descriptive and used to characterize experimentally
observed relationships between isotope ratios (e.g., the meteoric water line has λ = 0.528).
In the simplest case, λ describes the differences in δ′ values between two species, but more
often it describes a best-fit slope for set of measured data, i.e.,
 17 O (10)

 18 O
The latter quantity λRL, however, is the reference slope against which Δ′17O values are reported:
 17 O  17 O   RL   18 O (11)
Both of these uses serve empirical ends, and do not necessarily represent discrete
processes. However, λRL is sometimes tied to λ, θ, or κ values for convenience. One prominent
example is the high-temperature limit of κ (i.e., 0.5305; Pack et al. 2013; Pack and Herwartz
2014). We will use λRL = 0.528 to report data.
The relationship between observed λ values and the more fundamental θ values relies on an
assumed physical model of the system, and thus can vary. For example, for a well-mixed system
described by Rayleigh fractionation, one can define λRayleigh (γ in Angert et al. 2003) as
1  18   (12)
 Rayleigh 
1  18 
whereas for a diffusion-limited Rayleigh system (He and Bao 2019; Li et al. 2019)
 
/2
1 18
 diff  rxn  (13)

1  
1/ 2
18
Using triple-isotope observations to constrain chemical or thermodynamic quantities (e.g.,

formation temperature) thus requires that one evaluate and understand the physical constraints
on a given problem, often through its extended geochemical context.
104 Yeung & Hayles
Note that theory meets the “real world” in these three quantities. The experimentalist will
almost always measure α and λ values, which, through a suitable model of the system under
study, may allow them to infer θA–B values. The theorist will always calculate a set of RPFRs first
before obtaining α, κ, and θA–B values for fundamental processes. Interpreting subtle variations
in triple-oxygen compositions in natural systems thus requires an understanding of how accurate
each of these approaches is, or at least how well they are able to reproduce each other.
Natural variability in the triple-oxygen exponents θ
In pursuit of an intuition about triple-oxygen isotope systematics, one may wish to
interrogate the bounds of θ values allowed by theory: what are typical ranges for equilibrium
and kinetic processes, and when can θ values deviate from these ranges? For equilibrium
A  B can be expressed in terms of κ,
processes, eq α, and 18RPFR values (Cao and Liu 2011):
18
ln 18 RPFR B
A  B  A    A   B 
eq (14)
ln 18  A  B
This expression highlights the larger physically allowable range in eq A  B values compared to κ
values, which have a maximum value of 0.5305 at high temperatures (see above). When κA ≠ κB,
eq
A  B values diverge for small O/ O fractionations, i.e., when ln(18αA–B) = ln 18RPFRA–ln
18 16
18
RPFRB is near zero. This divergence allows eq A  B to take any value in an isotope-exchange
equilibrium in principle; fractionation “crossovers,” where ln(αA–B) changes sign as a function
of temperature (e.g., hematite–water equilibrium near 60 °C), have been highlighted before
as regions where eq A  B values can deviate strongly from the limits of κ and yield seemingly
anomalous isotope effects (Skaron and Wolfsberg 1980; Kotaka et al. 1992; Hayles et al. 2017).
Nevertheless, the typically expected range in eq A  B values is between 0.525 and 0.530 at Earth-
surface temperatures and above, slightly larger than the range for κ values (Bao et al. 2016).
The kinetic triple-isotope exponent Akin B has a typical range that is even larger. Young et al.
(2002) first showed, using both transition-state theory and Rice–Ramsperger–Kassel–Marcus
theory for unimolecular decomposition (Marcus 1952), that kinetic triple-oxygen isotope
effects depend on whether the isotopic substitution is part of the bond being altered. If oxygen-
atom motions are only relevant along the reaction coordinate (cf. Fig. 2), then the Akin B value
approaches the mass dependence of classical particles in motion, i.e., the relative velocities
of 16O, 17O, and 18O, approaching Akin B = 0.516 as the masses of the participant fragments
increases or the temperature increases. Within the framework of κ values, Equation (6) implies
that κcomplete approaches κIF (the imaginary-frequency contribution to κcomplete) when an isotopic
substitution only affects motion along the reaction coordinate (Hayles et al. 2017):
 17  ‡   17 ‡ 
ln  16 ‡j  ln  16 ‡ 
j 
 IF   18 ‡    18 ‡  (15)
 j    
ln  16 ‡  ln  16 ‡ 
 j    
Here, the reduced mass μ, i.e., derived from 1/μ = Σ (1/m), corresponds to that of the two
fragments participating in the vibration that becomes the reaction coordinate. In contrast, if
the oxygen-atom vibrations are all orthogonal to the reaction coordinate, then Akin B approaches
eq
A  B. One might then surmise that if oxygen atoms participate in some vibrational modes of
both the reactants and transition state, then Akin B lies somewhere in between the classical limit
and eq A  B (Dauphas and Schauble 2016). We note that Bao et al. (2015) showed that because
eq
AB is not bounded—e.g., it adopts unusual values near isotopic crossovers—Akin B can also
adopt any value. Nevertheless, this simplified view obscures an important and non-intuitive
feature of kinetic fractionation for oxygen triple isotopes: Akin B is not bounded, even for well-
behaved reactions far from where the kinetic isotope fractionation changes sign.
The loss of one vibrational degree of freedom in Equation (6) relative to Equation (1) is the
ultimate origin for the larger potential variability of Akin B relative to its equilibrium counterpart.
Importantly, the ratio of imaginary frequencies, βIF‡, is always less than one (Bigeleisen and
Wolfsberg 1958). If the substituted oxygen atom participates in some vibrational modes of both
the reactants and products, then the remaining, real portion of RPFR‡ in Equation (6) is greater
than one (i.e., βreal‡ > 1). This disparity in βi values on either side of unity creates an “internal
crossover” for κ‡: the denominator of Equation (8) for the transition state, ln18βIF‡ + ln18βreal‡,
is predisposed to be close to zero. κ‡ itself is not bounded, and instead is predisposed to appear
anomalous. Higher values of 18βreal‡, arising from lower reaction temperatures and/or substitutions
that shift high-frequency vibrations, tend to drive κ‡ toward a canonically more “normal” range.
To illustrate this concept, we plot predicted values of κ‡ for a variety of model transition states
as a function of 18RPFR‡ = 18βIF‡ × 18βreal‡ in Figure 3. A κ‡ crossover can be seen at 18RPFR‡ = 1
(i.e., ln18βIF‡ + ln18βreal‡ ~ 0), with anomalous nearby κ‡ values. Given this predisposition for high
variability in κ‡, then, the values of Akin B typically measured and calculated, between 0.516–0.53
(Bao et al. 2016), are surprising. The absence of anomalous Akin B values in the literature imply
that bounds on its value are imposed by chemical and not mathematical limits. For example,
the lower zero-point energy of isotopically substituted reactants may mitigate the expression of
anomalous 18RPFR‡ values as anomalous kinetic isotope effects.
Figure 3. Monte-Carlo sampling of 18RPFR‡ = 18βIF‡ × 18βreal‡ and κ‡ values for a model transition state.
In this model, the harmonic vibration of a diatomic molecule M1‑O is used to obtain 18βreal‡, which
characterizes the real vibrations inEquation (6). The force constant for that bond is varied from that of
CO to 1/8th that of CO. The imaginary frequency contribution to Equation (6),18βIF‡, is here equal to
(μM1 – O + M2/μM1 – O* + M2)1/2, i.e., a function of the reduced mass along the decomposition mode only. Both M1
and M2 are varied from 1–300 amu for Monte Carlo sampling. For comparison, the harmonic vibration of
O2 (ν = 1580 cm−1) 18βreal‡ = 1.17 at 25 °C.
Nuclear tunneling is thought to be negligible for oxygen isotopes, leading to the

generalizations above (cf. Eqns. 4, 5). However, electron–nuclear tunneling may be relevant in
electron-transfer (ET) reactions, which are ubiquitous in biology (Bertini et al. 1994). In this
mechanism, electron tunneling through the reaction barrier results in nuclear reorganization,
especially at low temperatures (De Vault and Chance 1966; Hopfield 1974); the nuclear
motions in this case are a response to the reaction, not the cause of it (sensu stricto). Its mass
dependence is not obvious at first glance.
Two mechanisms of ET reaction are relevant: inner-sphere and outer-sphere. Inner-
sphere electron transfer involves bond-making and/or breaking, whereas outer-sphere electron
transfer does not. The former mechanism is characteristic of enzyme catalysis and has oxygen-
isotope effects arising from both transition-state properties and tunneling (Roth et al. 2004;
Mukherjee et al. 2008). The latter type describes all long-range electron transfers in biological
systems (Bertini et al. 1994) and has isotope effects arising primarily from electron–nuclear
tunneling. We explore the mass dependence of the latter mechanism below.
106 Yeung & Hayles
Buhks, Jortner, and coworkers (Buhks et al. 1981a,b) described outer-sphere electron
transfer in terms of an electronic coupling term (i.e., the probability of an ET event) and
a Franck–Condon overlap factor (i.e., for a given geometry in a given system). The latter
term encompasses the Franck–Condon principle, which states that an electronic transition is
most likely to occur between states that resemble each other on the potential energy surface.
For example, the half-reaction O2 + e− → O2− is most likely to occur when the O–O bond
length in O2 (1.21Å on average) is equal to the equilibrium bond length of O2− (1.28Å).
Note that changes in the O–O bond length occur naturally as the O2 molecule vibrates,
modulating this overlap (Fig. 4, left). The equations describing the tunneling effects on the
reaction O2* + O2− → O2−* + O2 are:
4   d   r  p
2
Y
 
(16)
 Up  U
 r coth     p coth r 4
 4 
2  r 
2
(17)
 d 
2

h
ln  Y  Y * (18)
Here, (Δr)2 is the squared change in equilibrium bond length between reactant (subscript r)
and product (subscript p). For harmonic oscillators, the (Δr)2 term and the constants cancel
when θ values are computed, rendering the difference between νr and νp (i.e., the change
in bond strength upon reaction) the most important factor that determines θET for the ET
component of these reactions.
We compared the ET and equilibrium θ values for the reaction O2* + O2− → O2−* + O2
using known harmonic vibrational frequencies of 1580 cm−1 and 1090 cm−1 for the neutral
and anionic 16O16O species, respectively (Celotta et al. 1972; Huber and Herzberg 1979).
The results between 0 °C and 400 °C are shown in Figure 4. Interestingly, the outer-sphere
electron transfer mechanism is not mass-anomalous, as one might have guessed for an electron
tunneling-driven reaction, or from the dependence of α on (Δr)2. Instead, θET is well behaved
and less than θeq over the entire temperature range. Therefore, a kinetically limiting outer-
sphere ET pathway has a mass dependence in the range of more typical kinetic isotope effects:
the range in θET values is still between 0.52 and 0.53 in this temperature range.
Figure 4. Effects of outer-sphere electron transfer in the reaction O2 + O2 →

* −
O2−*
+ O2. (Left) Schematic
depicting outer-sphere ET through electron–nuclear tunneling. (Right) Comparison of θ values between
0 °C and 400 °C predicted for outer-sphere ET and isotopic equilibrium.
The preceding discussion emphasizes the relatively small range of triple-isotope

fractionation exponents observed, and the broader range allowed by theory in geochemical
systems. These ranges set a target on how accurate theoretical predictions and measurements
need to be in order to interpret isotopic trends unequivocally. Systems showing higher variability
in θ values such as those influenced by isotopic crossovers, photochemistry, surfaces, and/or
low temperatures (Skaron and Wolfsberg 1980; Horita and Wesolowski 1994; Abe 2008; Balan
et al. 2009; Sun and Bao 2011; Eiler et al. 2013; Hayles et al. 2017) may have slightly lower
accuracy requirements, but their influence in nature is no less important to understand.
A note about anharmonicity
The assumptions present in Equations (1, 3, 6, 15) are nontrivial, so an exploration of their
effects on calculated triple-isotope fractionation is warranted. The harmonic-oscillator and rigid-
rotor approximations, in particular, represent idealized behaviors of atoms and bonds, and are
generally justified if molecules primarily occupy the lowest vibrational state (e.g., under mild
conditions). Relaxing the harmonic-oscillator and rigid-rotor assumptions allows vibrational
and rotational modes to influence each other and centrifugal forces to affect bond lengths.
More subtle molecular motions, such as hindered internal rotations, may also be relevant
(Ellingson et al. 2006). Collectively, these are known as anharmonic effects. While accounting
for these effects adds considerable complexity to calculations, it also yields a more accurate
description of the physical system and more accurate isotopic fractionations in principle.
Richet et al. (1977) and Liu et al. (2010) found that the most significant anharmonic
effect for isotopic fractionation is that associated with the lowest vibrational state.
While this anharmonic correction can be calculated from spectroscopic constants for some
simple molecules, a more typical treatment is to replace the ZPE term in in Equation (1)
with an anharmonic zero-point energy obtained directly from electronic structure programs.
The resulting expression for RPFR, incorporating this first-order correction, is thus:
 e  ZPEanh / kT   *   1  e Ui 
*
RPFR anharm    ZPEanh / kT    i   (19)

e  i   i   1  e Ui* 
  MMI EXC
where ZPEanh is the total anharmonic vibrational zero-point energy of the molecule. These
effects are most pronounced for molecules bearing light elements such as hydrogen.
While the accuracy benefits of anharmonic corrections are apparent for individual RPFRs
and α values, there is a certain degree of error cancellation for mass-dependent processes.
The accuracy improvements for the quantities κ and θ in the oxygen triple-isotope system
therefore are not as obvious. Calculating anharmonic effects is expensive, and may be of limited
use if the end result is nearly unchanged from the harmonic result. Cao and Liu (2011) found
that The anharmonic correction typically accounts for single-digit percentages of ln(18RPFR)
for hydrogen-bearing molecules and even less for non-hydrogen bearing species. For small
molecules (i.e., CO2 and N2O), changes to κ values arising from anharmonic effects were
therefore found to be < 0.0001 (see Eqn. 8). The negligible importance of anharmonic effects for
κ values has not yet been evaluated for oxygen-isotope exchange reactions in condensed phases,
so it may be premature to rule out completely. We will proceed acknowledging this uncertainty.
ELECTRONIC STRUCTURE CALCULATIONS

The central consideration for applying theoretical chemistry to triple-isotope fractionation is
the electronic structure calculation. Accurate vibrational frequencies rely on an accurate calculation
of the potential energy experienced by the nuclei—i.e., the potential energy surface, so more
accurate electronic structure predictions should lead to more accurate triple-isotope fractionation
predictions. However, myriad factors influence the accuracy of the predictions: First, what is the
108 Yeung & Hayles
molecular model being investigated, and how well does it represent the real system of interest?
Second, how are the chemical (i.e., electronic) interactions being treated by the chosen electronic
structure method? Third, how are electrons being represented in the calculation, i.e., what set of
basis functions is being used to describe their distribution? A discussion of these factors, and their
potential importance for first-principles isotopic fractionation calculations, follows below.
Molecular models
In systems containing only a few atoms, such as gas-phase atmospheric reactions, the
molecular model often closely approximates the real system; thus the limitations on accuracy
in the gas phase generally revolve around the electronic structure method and basis functions
used. Key details about the electronic configuration, e.g., electron spins in open-shell
molecules, become the most important elements in the model to consider. In contrast, long-
range interactions in condensed phases that must be tackled using finite computing resources
forces the investigator to truncate solid and liquid systems in a meaningful way. The optimal
molecular model in that case is an approximation of the geochemical system that recovers
sufficient accuracy to be useful (Schauble 2004). Two basic approaches to approximation have
emerged: cluster models and lattice-dynamics models.
Cluster models simulate only a small portion of a crystal or liquid. They can be accurate if
isotopic substitutions are rare and only affect vibrational frequencies local to the substitution.
Long-range interactions (e.g., acoustic lattice vibrations) are omitted in order to simulate
the vibrational modes near the isotopic substitution more accurately. Rustad and Liu, among
others, have utilized these models extensively for isotopic equilibrium calculations in solvated
and crystalline species (Rustad et al. 2008, 2010; Li and Liu 2015; Hayles et al. 2018).
Each cluster contains the local crystal (or liquid) structure around a subunit that contains
several “shells” of atoms around the target chemical species (e.g., bonds around a central
SiO4 tetrahedron in quartz). For mineral models, the edge of the cluster is truncated in some
way, either by replacing the broken bonds with hydrogen to generate a neutral cluster (where
appropriate), or by embedding an ionic cluster within an appropriate array of point charges.
The minimum number of shells/bonds around the isotopic substitution is generally taken to
be three—one fixed outer shell and at least two inside that are free to vibrate (Rustad et al.
2008, 2010)—with larger clusters and more vibrating shells yielding higher accuracy with
diminishing returns in exchange for more computational time (i.e., scaling roughly with the
number of electrons raised to the 3rd or 4th power). Energies and vibrational frequencies are
calculated using electronic structure methods (e.g., commercial software packages Gaussian,
Q-Chem, and Turbomole, and open-source or freeware software GAMESS, NWChem, and
Quantum Espresso: Schmidt et al. 1993; Valiev et al. 2010; Furche et al. 2014; Shao et al.
2015; Frisch et al. 2016; Giannozzi et al. 2017). The discrete results are inputted into the
Bigeleisen–Geoppert–Mayer–Urey formula (Eqn. 1) to obtain isotopic fractionation factors.
An alternative approach to cluster models is to simulate a more extended crystal structure,
and thus the long-range interactions, using a lattice-dynamics approach (Elcombe and Hulston
1975; Dove et al. 1992; Schauble et al. 2006; Yeung et al. 2012; Blanchard et al. 2017). A larger
number of atoms is simulated, often with periodic boundary conditions, to recover the broad
spectrum of lattice vibrations relevant to the phase. To mitigate the added computational burden
of simulating more atoms, the electronic structure calculation is simplified further by assuming
that only valence electrons of each atom interact with each other. The electrons in core atomic
orbitals are replaced with a pseudopotential, a stand-in for the electronic potential energy of
the atomic core; many open-source databases are available online. Potential shortcomings of
this method include its semi-empirical nature. To date, this approach has been used by several
groups for triple-oxygen isotope predictions (Cao et al. 2019; Schauble and Young 2021, this
volume). Lattice dynamics-based isotopic fractionation predictions for the carbonate clumped-
isotope system are smaller than those computed using the cluster method, indicating that the
two methods may yield meaningfully different results (Schauble et al. 2006; Zhang et al. 2020).
Ultimately, a theoretical calculation can only be as accurate as its molecular model.

Ways of verifying the accuracy of a model include comparing its harmonic vibrational frequencies
against experimental harmonic frequencies; however, only observed spectroscopic frequencies
(which include anharmonicity) are typically available for condensed phases. The electronic
structure calculation method often introduces additional bias, so uncertainties remain in many
cases (see below). Hayles et al. (2018) used two clusters of different sizes to test the convergence
of their results, an approach that we recommend as a practical test of systematic errors. However,
the omission of long-range lattice modes in all cluster models could still result in errors.
Electronic structure methods: different ways of treating chemical interactions
Electronic structure calculations generally fall under two approaches. The first is
wavefunction theory, which aims to describe a molecule by its electronic wavefunction;
that wavefunction is used in the Schrödinger wave equation to obtain molecular properties.
The most elementary computational approach to obtaining molecular wavefunctions is the
Hartree–Fock (HF) method, which assumes that electrons only interact with positively-
charged nuclei and a mean field of negative charge arising from other electrons (Jensen 2017).
Pairwise and many-body electron–electron interactions (electron correlation) are omitted in
that energy calculation. Moreover, relativistic effects relevant to transition metals are ignored.
Clearly, these assumptions comprise a severe approximation of the real behavior of electrons,
resulting in low accuracy in all but the simplest, closed-shell systems. Treatment of electron
correlation is possible, often by either perturbing the mean-field approximation using a
correlation potential (e.g., second-order Møller–Plesset perturbation; MP2) or by allowing
electronic configurations other than the ground state to influence the ground-state energy (e.g.,
configuration interaction/CI or coupled cluster/CC approaches). Increasing the number of
configuration interactions gives a predictable march toward increased accuracy—an important
consideration if molecular structures cannot be linked to observables (e.g., transition states).
However, these corrections are computationally expensive, especially in larger systems,
because each electron and configuration must be individually considered and computed;
continued advances in implementation are reducing the cost of these calculations (Peng et
al. 2016; Vogiatzis et al. 2017). At present they are not widely used in isotopic calculations,
although they may be affordable for small gas-phase systems.
The second and most-often used electronic structure method in geochemistry is density
functional theory (DFT). It conceptualizes the electronic configuration as the electron density
in three dimensions in space (i.e., number of electrons per unit volume) rather than a collection
of constituent electronic wavefunctions (Becke 2014; Peverati and Truhlar 2014). Therefore,
instead of solving the many-body interaction problem directly as HF approaches do, modern
DFT methods—Kohn–Sham DFT being the most common—break up the electronic energy into
separable components comprised of (1) coulombic electron–nuclei interactions, (2) electron
kinetic energy in the absence of electron–electron interactions, (3) coulombic electron–
electron repulsion, and (4) energy associated with electron exchange and correlation1. Each is
a function of the total electron density, and can be computed separately, increasing efficiency.
The electron density is varied until energy is minimized. In principle, this approach can recover
the molecular wavefunction exactly, provided the function describing the electron density is
known. In practice, however, one can only approximate the true electron density, resulting in an
approximate description of the electronic energy (which is a functional of the electron density).
For example, one component of this approximation arises from the three-dimensional
electron density being represented on a computer by an array of values calculated at different
points in space (i.e., the integration grid). The fineness of this grid is user-specified in electronic
1
Electron exchange energy is the tendency for two electrons of the same spin to avoid each other. The effective repulsion
arises from the Pauli exclusion principle of quantum mechanics. Electron correlation energy is associated non-coulombic
electron–electron interactions arising from the wave nature of electrons.
110 Yeung & Hayles
structure programs and can have significant effects on the accuracy of the final calculation.
Recent work has shown that many default integration grids (75,302 points or fewer) can be
too coarse, and that a ultra-fine grids (99,590 points or more) are needed to obtain numerically
robust results from DFT calculations (Bootsma and Wheeler 2019).
Other limitations of DFT are more pernicious and lack such straightforward resolutions.
In particular, the last calculation in the Kohn–Sham DFT method, the electron-exchange and
correlation energy, is wrought with particular difficulty. One can choose functional forms and
the associated parameters for exchange and correlation that are based in physical constraints,
but some functionals often require empirical tuning to maximize accuracy through the use
of a benchmark molecular training set that has limited relevance to geochemistry. Other
functionals construct a hybrid between HF and DFT methods (with relevant contributions
of each optimized similarly) to exploit the advantages of each approach. Developments in
DFT focus on improving various representations of exchange and correlation, which have
yielded good accuracy in the structure and thermodynamics of an increasingly broad range of
materials; however, not all approaches are appropriate for all systems, and unlike wavefunction
methods, the path to increased accuracy is not always clear (Mardirossian and Head-Gordon
2017). The “alphabet soup” of DFT methods unfortunately yields little insight. Below we will
explore what these developments portend for the isotope geochemist, as well as investigate
their consequences for oxygen triple-isotope fractionation predictions specifically. The reader
is directed to Becke (2014) and Peverati and Truhlar (2014) for more detailed reviews on DFT.
DFT methods generally differ the most by how they treat the exchange-correlation term,
and an appropriate starting point for its discussion is the local spin density approximation
(LSDA). In the LSDA, the electron exchange and correlation energy at a given point in
space is assumed to be a function of the local electron density only, with the energy equal
to that of a “gas” of uniformly distributed electrons at that density. This first approximation
yields good molecular geometries, but poor bond- and reaction-barrier energies (Peverati
and Truhlar 2014). One can make a higher-order correction to the LSDA (akin to a Taylor
expansion about the LSDA minimum) that considers also the local change in electron density.
This family of approaches is thus known as the generalized gradient approximation (GGA)
family of methods. Further improvements can also be made, such as a consideration of not
only the first derivative (gradient) of the electron density, but also its second derivative (i.e.,
its curvature). This approach is known as a meta-GGA approach. Finally, HF and DFT can
be hybridized owing to the equivalency of their descriptions of electronic structure. Hybrid
methods incorporate the exchange energy calculated from the HF method to improve the
DFT exchange-correlation energy. The exact amount of HF exchange energy used in a hybrid
method is most often optimized against chemical properties in a benchmark molecular
database. Additional parameters used to improve the performance of long-range interactions
have also been employed, and include the development of range-separated and double-hybrid
functionals, which have shown great promise (Grimme 2006; Su et al. 2018). The most popular
DFT method in the chemical sciences is B3LYP, a semiempirical hybrid GGA functional
utilizing three fit parameters that result in 20% HF exchange (Lee et al. 1988; Becke 1993).
In contrast, the PBE and TPSS (and later RevTPSS) functionals are pure GGA and meta-GGA
functionals with no empirically fit parameters; their forms arise from more detailed treatments
of the physical behavior of electrons. Broadly, these incremental improvements to DFT have
been likened to rungs on Jacob’s ladder (Perdew and Schmidt 2001), with thermodynamic
accuracy improving as LSDA → GGA → meta-GGA. Hybrid approaches perform exceedingly
well for their training sets, and even beyond them as well; they are often placed above meta-
GGAs in the ladder, although they may sometimes perform poorly for cryptic reasons (Paier
et al. 2007). In general, the performance of DFT methods for a variety of problems, in isotope
geochemistry and beyond, is quite remarkable, despite some consternation in the theoretical
chemistry community over the use of semiempiricism (Medvedev et al. 2017).
It is worth noting that treating electron–electron repulsion in terms of electron density has
a fundamental problem: an electron cannot repel itself. Yet, the density functional describing
electron–electron repulsion would predict otherwise. In principle, proper calculation of the
electron correlation energy will cancel this interaction, but the exchange-correlation functional
is only approximate in practice. This problem is thus only partially accounted for in the
approximate treatment of electron exchange and correlation, and despite schemes to correct
for it (Perdew and Zunger 1981), the “electron self-interaction problem” in DFT remains a
sticking point for many theorists.
For the isotope geochemist, the choice of electronic structure method will affect the
accuracy of the vibrational frequencies obtained, and thus also the accuracy of isotopic
fractionation calculations. Indeed, agreement between theory and experiments for molecular
geometries and energies tends to improve as one moves up Jacob’s ladder, but the results for
vibrational frequencies are not as clear for geologic materials. Comparison of DFT-derived
harmonic frequencies with those obtained through high-level wavefunction theory for a suite
of molecules reveals biases that vary in both magnitude and sign, with some dependence on
the orbital basis set used (Kesharwani et al. 2015; Kashinski et al. 2017). This non-systematic
variability may be an artifact of how density functionals are obtained and optimized, and results
in a wide range of empirical scaling factors. In practice, direct comparisons of DFT and high-
level wavefunction calculations for portions of the system being studied, e.g., Guo et al. (2009)
may lend additional confidence in the veracity of the results; however, these calculations may not
always be available or affordable. Comparing structures and vibrational frequencies calculated
at the same theory and basis set is another approach that could mitigate the poorly characterized
systematic errors in DFT methods, but its effectiveness has not been thoroughly evaluated.
Representing electrons using basis sets
The third element to an electronic structure calculation is the set of building blocks used
to construct the electron density distribution of a molecule. These building blocks are known
as basis functions for atomic orbitals, prescribed collections of which are known as basis
sets. These electron distributions around atoms are combined in linear combinations to form
molecular orbitals. The primitive functions most often utilize Gaussian functions to describe the
radial electron density distribution (a reasonable approximation that also facilitates efficiency
in computation) plus a separable spherical-harmonic component to describe the shape of the
atomic orbital (e.g., s, p, d, and f atomic orbitals). For efficiency, core orbitals are often described
with fewer terms or contracted. Additional Gaussian functions are used to increase flexibility
in the radial distribution because electron densities can decrease with nuclear distance in a
variety of ways. Using two functions to describe this long-distance falloff yields a “double-
zeta” basis function, while using three functions yields a “triple-zeta” basis function, and so
forth. In addition, mixed atomic orbital character can be introduced (e.g., s-orbitals can take
on a little p- character when chemical bonds are made), resulting in polarization functions
that further improve optimized electronic structures. Finally, one can add diffuse functions that
focus on capturing the electron density at very large distances from the nucleus. Those are most
useful for anionic and other large molecular systems with partial charges. The reader is directed
to the review by Jensen (2013) for further details about how they are developed.
Basis sets are named descriptively, although each family tends to have its own scheme.
For example, the 6-31G* basis set [equivalent to 6-31G(d)] represents each core orbital shell
using a linear combination of six gaussian functions, while each valence orbital is represented
with three and one gaussian functions to describe the inner and outer regions of the valence
shell, respectively (Hehre et al. 1972; Hariharan and Pople 1973). Splitting orbitals into inner
and outer regions makes it a double-zeta basis set, while treating core and valence electrons
differently makes it a split-valence basis set. The addition of the asterisk or “(d)” means that
p-orbitals can polarize to incorporate d-orbital qualities.
112 Yeung & Hayles
Using a high-zeta basis set that includes at least some of the basis function extensions
listed above should lead to convergence to the “true” electron density and accurate chemical
and isotopic predictions. However, even a moderately-sized chemical system can quickly
render a large basis set impractical to use, especially when one might also be able to pair a
higher level of electronic structure theory with the smaller basis set. The rate of convergence to
the full-basis-set limit becomes key, and depends somewhat on how each basis set is derived.
These factors largely govern the choice of basis set in an electronic-structure calculation.
For example, the widely used “Pople” basis set family (e.g., 6-31G) was initially
constructed in the 1970s and 1980s from atomic orbitals obtained through HF calculations
(Ditchfield et al. 1971), with the addition of MP2 calculations for the triple-zeta basis set
6-311G. While Pople basis sets show good performance for a variety of molecules, the
omission of electron–electron correlation in HF calculations causes errors that can propagate to
molecular geometries, particularly in delocalized systems (Moran et al. 2006). The “Dunning”
or correlation-consistent basis sets (e.g., cc-pVDZ, where “pV” indicates polarized valence
and “DZ” indicates double-zeta) first introduced in the 1980s and 1990s were a significant
step up: using polarized valence orbitals derived from CI-based atomic calculations, they tend
to yield more accurate results than Pople basis sets (Dunning 1989). They are so accurate that
decades later, high-zeta Dunning basis sets are still considered a benchmark of accuracy in
wavefunction theory calculations. However, they are costly, converge slowly, and were not
optimized for DFT calculations. Subsequent improvements have yielded many other basis
set families, notably among which are the “Karlsruhe” basis sets (e.g., def2-SVP, a double-
zeta basis set where “SVP” stands for split-valence polarized; Weigend and Ahlrichs 2005).
The Karlsruhe basis sets are similar to the Dunning basis sets, but are much more efficient
and suitable for a larger portion of the periodic table. A recent DFT basis-set intercomparison
showed that, at least for weakly bound complexes, the Karlsruhe basis sets converged
toward the complete-basis-set limit faster than Dunning basis sets, with similar performance
sometimes achieved with roughly half the number of basis functions (Witte et al. 2016).
We note that basis sets are not restricted to Gaussian-type orbitals. “Slater-type” orbitals,
which are more strongly peaked near the nucleus and slower to decay at a distance, were
historically the first ones introduced to describe atomic orbitals. However, they were not as
computationally efficient or flexible as the Gaussian-based description. Also, lattice-dynamics
models use plane-wave basis functions in conjunction with core pseudopotentials (Kresse and
Furthmüller 1996; Schwerdtfeger 2011).
Finally, we will mention the basis set superposition error (BSSE), an artifact of incomplete
basis sets that may be relevant for the calculation of transition states and aqueous-phase clusters.
The error arises from the tendency for each component of a weakly bound complex to utilize
the basis functions of the other component(s), erroneously increasing the size of the basis set.
The BSSE manifests as an artifactual stabilization of bonds in weakly bound complexes and
thus bond lengths that are too short, as well as vibrational frequency shifts that are too large
upon binding (Hermida-Ramón et al. 2003; Hermida-Ramón and Graña 2007). Moreover,
while BSSE errors are well known for minimum-energy structures, they are also important for
transition states (Kobko and Dannenberg 2001). The counterpoise correction for BSSE, available
in many electronic-structure programs, often yields a sufficient first-order correction for the
error (Boys and Bernardi 1970). It involves calculating the energies of both the complex and
the isolated monomers of the complex in their complexed geometry using the complete basis set
of the complex (i.e., each monomer with a “ghost” partner). The accuracy of this correction is
also basis-set dependent, however, with larger basis sets still yielding higher accuracy (Paizs and
Suhai 1998). The BSSE is a potential hazard to the isotope geochemist using affordable (e.g.,
double-zeta) basis sets to simulate complex molecular clusters and transition states, although to
our knowledge the effect of BSSE on calculated isotopic fractionation has not been investigated.
A preliminary investigation of basis set effects on isotopic fractionation on bioinorganic

oxyheme complexes suggests that discrepancies within a family (e.g., Pople, Dunning, or
Karlsruhe) are small, but not negligible (i.e., ~1‰ in 18O/16O fractionation and ~0.001 in
θeq value; Ash et al. 2020) despite significant differences in calculated harmonic vibrational
frequencies. However, we are not aware of any systematic investigations of basis-set accuracy
for high-precision isotope measurements. Large systems, in particular, may not be approachable
with expensive triple-zeta basis sets or diffuse functions; if smaller basis sets can suffice by
capturing the vibrational frequencies well enough, then many more computationally tractable
systems may exist than commonly thought.
Key takeaways
Each of the three components mentioned above—the molecular model, electronic structure
method, and basis set—could have varied consequences for high-precision geochemical
calculations. While isotope geochemists are typically not as concerned with bulk thermodynamic
quantities (e.g., atomization or ionization energies), the practitioner should be familiar with the
potential strengths and weaknesses of the available tools: A calculation can only be as good as
the molecular model of the real system; newer or more sophisticated DFT methods may not
necessarily be more accurate because of potential over-fitting to unrelated test molecules; basis-set
size, form, and convergence rate all matter for high-precision accuracy. We believe that the elusive
intersection of theory, experiment, and reality at the top of Mount Fuji will appear not by chance,
but through the careful selection of methods and evaluation of potential biases in all approaches.
A COMPARISON OF THEORETICAL METHODS

How do the preceding methods affect isotopic fractionation calculations, and oxygen triple-
isotope calculations in particular? One might imagine that nearly all of the aforementioned
complications will affect the usefulness of calculated triple-oxygen fractionation factors if
the target accuracy is at the ppm level in Δ′17O values. However, do error cancellations in
the calculation of triple-isotope trends render these methodological imperfections negligible?
It is clear from the literature that fortuitous error cancellation occurs for θ values in certain
systems (e.g., O2-heme binding Ash et al. 2020. The harmonic-oscillator approximation used
to estimate vibrational frequencies upon isotopic substitutions renders subtle (< 10%) absolute
frequency errors less important for θ values if the relative errors are shared for both 17O and
18
O substitutions. Yet, the same cannot be said for bulk 18O/16O fractionation factors, which are
often used as an independent constraint on geochemical problems when paired with θ values
(Pack and Herwartz 2014; Young et al. 2014; Sharp et al. 2016; Hayles et al. 2019).
Approach
We investigated the methodological dependence for the triple-isotope fractionation
between crystalline quartz and SiO2 gas (which is chemically similar to quartz, stable, and
easily modeled). The molecular model for crystalline quartz is a cluster model similar to
that of Rustad et al. (2008) and Hayles et al. (2018) in which all atoms within five bonds
of the central Si, as well as some bridging Si atoms, are included to make a cluster with a
stoichiometry Si23O5212−. Starting bond lengths and geometries are taken from the American
Mineralogist Crystal Structure Database (Levien et al. 1980; Downs and Hall-Wallace 2003).
Dangling bonds at the edges, which in this case all have a Pauling bond strength of 1, are
replaced with hydrogens, making a final, neutral cluster with the stoichiometry H36Si23O52
(Fig. 5). Calculations proceed as follows: First, the hydrogen bond lengths are optimized at
the specified level of theory and basis set with all other atoms fixed in their experimentally
determined positions. Next, the internal atoms within four bonds from the center are geometry
optimized with the remaining atoms fixed in place. Finally, harmonic vibrational frequencies
114 Yeung & Hayles
Figure 5. Embedded quartz cluster used in the electronic-structure method intercomparison. “Frozen”
atoms are shown as lines, with the colors distinguishing the different atoms (gray: Si, red: O, white: H).
are obtained from the optimized, partially fixed structure. Optimized structures and harmonic
vibrational frequencies for SiO2 gas were calculated using the same theoretical method.
Quartz–SiO2(g) isotope fractionations were calculated using Equations (1, 7, 14). Adherence
to the Teller–Redlich product rule for the quartz models, i.e. MMI term of (m/m*)3/2 due to
cancellation of rotation and translation terms, was within 3 ppm in all cases.
A total of eight DFT levels of theory were studied, spanning four rungs of “Jacob’s
ladder”: pure GGA (BLYP, PBE), meta-GGA (M11-L, RevTPSS), hybrid GGA (B3LYP,
PBE0), and hybrid meta-GGA (M11, TPSSh) (Perdew et al. 1996, 2009; Adamo and Barone
1999; Staroverov et al. 2003; Tao et al. 2003; Peverati and Truhlar 2011, 2012). We note that
the M11 and M11-L functionals are optimized to reproduce observable physical quantities for
a large suite of molecules, whereas the others are optimized to reproduce the physical behavior
of electrons. Triple-zeta basis sets were not affordable for this intercomparison, so the double-
zeta orbital basis sets 6-31G(d), cc-pVDZ, and def2-SVP were used to investigate differences
between basis set types. In total, 24 unique electronic structure calculations were performed,
all with an ultra-fine integration grid (99,590 points). Then, we compared the results to a
published benchmark B3LYP calculation for a larger quartz cluster model calculated using the
triple-zeta 6-311G(d) basis set. That quartz cluster was part of a set of calculations that yielded
good agreement with experimental triple oxygen-isotope fractionations between quartz and
liquid water (Sharp et al. 2016; Hayles et al. 2018; Wostbrock et al. 2018). We performed new
calculations on SiO2 at the same level of theory and basis set for this comparison. Figure 6
shows the isotopic trends derived from these benchmark calculations compared with available
observations, which are generally within 3‰ and 0.001 in 18O/16O fractionation and θeq value,
respectively, at 25 °C and below for quartz in equilibrium with liquid water.
Results
Table 1 compares the average Si–O bond lengths obtained for the central silicate
tetrahedron in the optimized embedded quartz clusters to that measured by high-resolution
X-ray diffraction (Levien et al. 1980). Nearly all methods yield bonds that are too long by
1–3% (0.02–0.05 Å). The cc-pVDZ bond lengths are consistently 0.02 Å longer than those
for the other basis sets. While Si–O bond lengths tend to approach the experimental value
with each higher rung of Jacob’s ladder, the evolution is not monotonic. Of note are the
M11-L/6-31G(d) and M11-L/def2-SVP methods, which appear to outperform all the others to
yield average Si–O bond lengths within 0.002 Å of experiment.

Figure 6. Theoretical quartz–SiO2(g) and quartz–H2O(l) fractionation trends, derived from B3LYP/6-
311G(d) calculations on a larger embedded quartz cluster used as a benchmark for the methodological in-
tercomparison (Hayles et al. 2018). Blue points are a compilation of empirically determined quartz–H2O(l)
fractionation factors (Sharp et al. 2016; Wostbrock et al. 2018, and references therein).
Table 2 compares the highest-frequency infrared-active mode to the highest-frequency

transverse optical phonon mode (TO8) inferred from second-harmonic phonon spectroscopy
for α-quartz (Winta et al. 2018). All the calculated harmonic frequencies are too low by
as much as 9% (~100 cm−1), with higher-level calculations generally performing better.
The TPSSh functional is the only functional that does not follow this trend, although the total
spread within a basis set is ~3% (30–40 cm−1). Unlike for bond lengths, M11-L does not
outperform the other methods on this arguably more important metric.
Table 3 compares the DFT-calculated harmonic asymmetric stretch frequency for SiO2
gas against that calculated by the wavefunction method CCSD(T)/aug-cc-pVTZ. A high-level
wavefunction calculation was chosen for comparison because of a lack of experimental data on
gaseous SiO2 and high accuracy expected for this method (Kesharwani et al. 2015). DFT methods
both over- and under-estimate the asymmetric stretch frequency, with the higher-level hybrid and
hybrid meta-GGA methods performing best. Moreover, the spread in vibrational frequencies,
both between basis sets (50–60 cm−1) and across DFT methods within a basis set (~150 cm−1),
is larger than for the Si–O symmetric stretch in quartz. Other vibrational modes show similar
results. The M11-L functional shows the largest basis-set dependence, with one of the most
accurate predictions (cc-pVDZ) but also the two least accurate [6-31G(d) and def2-SVP].
Figure 7 compares the calculated quartz–SiO2 fractionations from 0 °C to 400 °C. The
spread in 18O/16O fractionation factors decreases as temperature increases, from 6–9‰ at 0 °C to
~2‰ at 400 °C. In contrast, the spread in eq quartz SiO2 values is ~0.0005 over the entire temperature
range. The cc-pVDZ basis set consistently predicts smaller 18O/16O fractionations (by ~2‰ at
0 °C, on average) and higher eq quartz SiO2values. Disparities relative to the benchmark quartz–SiO2
calculation range from 0.1 to 5‰ in 18O/16O fractionation at 0 °C and are less than 0.0003 in
eq
quartz SiO2 across the temperature range. Theory-experiment (for quartz) and DFT-wavefunction
(for SiO2) agreement in vibrational frequencies, taken in isolation, were not good predictors
of the method-benchmark agreement in 18O/16O fractionation factor. However, the accuracy of
calculated quartz–SiO2 isotopic fractionations depends on the both the direction and magnitude
of errors in vibrational frequencies. Interestingly, methods that underestimate the vibrational
frequencies in both species can show fortuitous error cancellation (e.g., PBE), but not always
(e.g., BLYP). Similarly, methods that show frequency errors differing in sign between species
can show exacerbated errors in isotopic fractionation [e.g., M11-L/6-31G(d)], but not always
116 Yeung & Hayles
Table 1. Average calculated Si–O bond lengths for the central SiO2 tetrahedron in quartz.
Experimental bond length is 1.608 Å (Levien et al. 1980).
<rSi–O> (Å)
Method Type 6-31G(d) cc-pVDZ Def2-SVP
BLYP GGA 1.641 1.664 1.646
PBE GGA 1.640 1.665 1.646
RevTPSS Meta-GGA 1.638 1.663 1.643
M11-L Meta-GGA 1.606 1.629 1.610
B3LYP Hybrid GGA 1.628 1.651 1.634
PBE0 Hybrid GGA 1.625 1.649 1.631
M11 Hybrid meta-GGA 1.632 1.651 1.635
TPSSh Hybrid meta-GGA 1.630 1.654 1.636
Table 2. Comparison of the strongest infrared-active harmonic vibrational frequencies for quartz
calculated by DFT. Experimentally determined harmonic frequency is 1171 cm−1.
νSi–O (cm−1)
BLYP GGA 1109 1072 1110
PBE GGA 1109 1070 1111
RevTPSS Meta-GGA 1118 1080 1121
M11-L Meta-GGA 1120 1097 1137
B3LYP Hybrid GGA 1144 1109 1144
PBE0 Hybrid GGA 1144 1114 1155
M11 Hybrid meta-GGA 1148 1121 1155
TPSSh Hybrid meta-GGA 1134 1100 1139
Table 3. Comparison of the harmonic asymmetric stretch frequencies (ν3) for gas-phase SiO2
calculated by DFT. The frequency calculated by CCSD(T)/aug-cc-pVTZ is 1432 cm−1.
ν3 (cm−1)
BLYP GGA 1379 1326 1376
PBE GGA 1390 1340 1390
RevTPSS Meta-GGA 1406 1353 1404
M11-L Meta-GGA 1526 1434 1522
B3LYP Hybrid GGA 1448 1389 1444
PBE0 Hybrid GGA 1472 1410 1467
M11 Hybrid meta-GGA 1490 1434 1492
TPSSh Hybrid meta-GGA 1435 1380 1433

Figure 7. Comparison of 18O/16O fractionation factors (Δln18α = ln18αcalculated–ln18αbenchmark) and θeq (Δθeq
= θeqcalculated–θeqbenchmark) for quartz–SiO2(g) equilibrium from 24 different electronic-structure method/basis
set pairs.
(e.g., M11 for 18α). This variability highlights one of the challenges in benchmarking theoretical
calculations of isotopic fractionation: not all vibrational modes were compared in Tables 2 and 3,
and indeed, one can rarely assess the accuracy of all calculated harmonic frequencies in a system.
We also evaluated how well these calculations objectively predict quartz–H2O(l)
fractionations based on the semi-empirical RPFR and κ-value fits for liquid water used in
Hayles et al. (2018). The results show broadly similar trends as for quartz–SiO2(g) (Fig. 8).
18
O/16O fractionations obtained from the cc-pVDZ basis set are consistently smaller than those
from the 6-31G(d) and def2-SVP basis sets, with little difference in performance for eq quartz SiO2
values. Perhaps remarkably, our B3LYP/6-31G(d) calculation predicts a 18O/16O fractionation
and eqquartz SiO2 value within 1.5‰ and 1 ppm, respectively, of the B3LYP/6-311G(d) benchmark
across the temperature range; the basis set effect on θ appears to be negligible within the
Pople basis set family for this quartz cluster model. Both 18O/16O fractionations and eq quartz SiO2
values systematically increase up the rungs on Jacob’s ladder, amounting to a 10‰ range
(~2‰ when omitting M11-L) at 0 °C in 18O/16O fractionation and a 0.0007 range in eq quartz SiO2.
The trends largely mirror the increased accuracy in vibrational frequencies with methodological
sophistication highlighted in Tables 2 and 3, but the fractionation factors do not converge at
high levels toward the benchmark calculation.
Finally, we evaluated the effects of harmonic vibrational scaling using the scaling factors
of Kesharwani et al. (2015). Frequencies for all isotopic species obtained from the calculations
were uniformly scaled. Only a few combinations of theory and basis could be scaled, but
general trends are still apparent; the results are shown in Figure 9. In this case, frequency
scaling appears to mitigate much of the spread in 18O/16O fractionation associated with both
the level of theory and the basis set size, although the final range of results is still ~3‰ in
18
O/16O fractionation at 0 °C. The method-to-method disparities in θ values is also improved
118 Yeung & Hayles
Figure 8. Same as Figure 7, but for quartz–H2O(l) fractionation derived from the H2O(l) RPFR of Hayles et al. (2018).
Note the change in scale to encompass larger deviations from the benchmark calculation than in Figure 7.
PBE B3LYP PBE0
4 6-31G(d) cc-pVDZ def2-SVP
2
∆ln 18α (‰)
-2
-4
0 100 200 300 0 100 200 300 0 100 200 300

8
2
∆θeq x 104
-2
-4
-6
-8
0 100 200 300 0 100 200 300 0 100 200 300
Temperature (°C)
Figure 9. Same as Figure 8, but after scaling quartz vibrational frequencies according to empirical scaling factors. Note
the change in scale reflecting improved agreement with the benchmark calculation relative to the unscaled calculations
in Figure 8, and that only a subset of the models used in this study were scaled owing to the availability of consistent
scaling factors.
by about a factor of two. However, the B3LYP results deviate further from the benchmark
upon scaling. We note that these empirically derived scaling factors depend strongly on both
the molecular test set and benchmarking method, so we consider this reduction in scatter
between methods to be provisional.
These results resemble those from another methodological intercomparison, made on
oxyheme complexes relevant for triple-isotope effects in cellular respiration (Ash et al. 2020).
In that study, three functionals (GGA, hybrid GGA, and hybrid meta-GGA types) were used
to investigate the triple-oxygen and clumped-isotope equilibrium isotope effects upon the
binding of O2 to three different heme groups. The calculated θheme–O2 values showed agreement
within < 0.001 at 25 °C, but the calculated 18O/16O fractionations differed by as much as 8‰
at 25 °C. While the varied change in bond strength upon binding drove the variable 18O/16O
fractionations, the mechanism of O–O bond weakening appeared to control the θ value.
Taken together, the most notable outcome of these comparisons is the insensitivity of
θ to DFT method and basis set. The calculated 18O/16O fractionations are sensitive to the
method used, and the systematic errors are insidious: merely changing the basis set can
yield per-mil-level changes in predicted 18O/16O fractionations for a given molecular model.
Systematically fortuitous error cancellation appears to be mostly limited to θ values, rendering
them more robust across methods and basis sets. This relative invariance implies that θ values
reflect something that is conserved between the approaches such as the nature and number of
vibrational modes involving oxygen—which shift together upon an isotopic substitution—
and/or their relative bond strengths. While the spread in raw calculated 18O/16O fractionations
continues to be problematic, the use of literature harmonic vibrational frequency scaling factors
may reduce disagreements among the theoretical methods, even when the geometries are
slightly inaccurate. Using high-quality scaling factors, the accuracy of calculated equilibrium
isotopic fractionation obtained through inexpensive methods may be able to approach that of
higher-level methods (in theory or basis set). If qualitative aspects of molecular vibrations
and the relative harmonic frequencies remain accurate, the use of low-level DFT methods
and small basis sets could provide useful insights into large molecular clusters (keeping in
mind potential systematic errors, e.g., BSSE), although higher-level calculations may still be
more accurate. Broadly, we expect that the accuracy of first-principles approaches does not yet
exceed 2–3‰ in 18O/16O equilibrium fractionations and ~0.001 in θ values, but in many cases
such accuracy is sufficient to advance the science and push the field into new areas.
ANALYTICAL CONSIDERATIONS
Boaz Luz and Eugeni Barkan led the group at Hebrew University that first demonstrated the
possibility of measuring oxygen triple-isotope compositions of natural materials to ultra-high
precision (Luz et al. 1999; Luz and Barkan 2000, 2005; Barkan and Luz 2003). Their efforts first
focused on atmospheric oxygen, but subsequent work demonstrated that ultra-high precision
could be achieved for water as well (Barkan and Luz 2005, 2007; Luz and Barkan 2010; Affek
and Barkan 2018). Improvements in measurements of CO2 and mineral phases followed, owing
to numerous analytical innovations (Hofmann et al. 2012b; Mahata et al. 2013; Pack et al. 2013;
Levin et al. 2014; Pack and Herwartz 2014; Passey et al. 2014; Barkan et al. 2015; Sharp et
al. 2016; Laskar et al. 2019; Wostbrock et al. 2020) that include a new generation of optical
isotope instruments (Berman et al. 2013; Steig et al. 2014; Prokhorov et al. 2019). State-of-the-
art analytical precision is currently between 3 and 5 parts per million in Δ′17O values.
Despite this high precision—which appears to be broadly achievable—significant inter-
laboratory triple-isotope discrepancies have persisted. The most prominent of these may be the
triple-isotope contrast between atmospheric O2 and ocean water (Luz and Barkan 2011; Tanaka
120 Yeung & Hayles
and Nakamura 2012; Young et al. 2014; Pack et al. 2017; Wostbrock et al. 2020). Labs measuring
argon-free atmospheric O2 consistently report smaller Δ′17O differences, by up to 0.07‰. Similar
disagreements exist in measurements of oxygen triple isotopes in carbonates (Passey et al. 2014;
Barkan et al. 2015; Bergel et al. 2020; Voarintsoa et al. 2020; Wostbrock et al. 2020).
While the origins of these disagreements remain enigmatic, the discrepancies may be
partially attributable to mass-spectrometric artifacts. One such artifact has been dubbed a
“pressure baseline” (He et al. 2012; Yeung et al. 2018), any unwanted signal registering in
the rare-isotope collection cups of a mass spectrometer that appear only when the analyte is
present. They arise from positively or negatively charged isobars, and are especially pernicious
in the reference gas of a dual-inlet system: if the “true” reference ion current is different than
what the instrument is registering, then the measured isotopic ratios will be inaccurate. The
resulting scale distortion is usually non-mass-dependent because it is influenced by instrument
geometry, ion-source conditions, and ion optics, and it disproportionately affects the weaker
beams (i.e., 16O17O+ relative to 16O18O+).
Scale distortion and calibration
Yeung et al. (2018) derived a formula for a correction factor ϕ, which relates the (non-
logarithmic) δ value to the relative size of the unwanted “pressure baseline” ion current (iBG)
and the intrinsic ion current due to the analyte (iion):
true
measured   REF  true
 iBG  (20)
 1  
 iion  REF
The ϕ factors for O and O-substituted analytes are typically neither equal nor mass-
17 18
dependent, so they are related to the measured Δ′17O value by the expression
   
17O
measured  17 O true  ln  17,REF    RL  ln  18,REF  (21)
 17,SAM   18,SAM 
It is important to notice that while ϕREF values are constant for a given reference gas and
mass-spectrometric setting, the ϕSAM factor changes with isotopic composition, even if the
pressure baseline is constant. The precise value of ϕSAM also depends on the abundance of the
minor ion via iion. Therefore, because ϕREF ≠ ϕSAM, this composition-dependent effect causes
inaccuracies in measured Δ′17O values. As the isotopic contrast between sample and reference
gases becomes larger, so too does the difference between ϕSAM and ϕREF, and thus the Δ′17O
error. Measured λ values would thus be affected because the composition-dependent Δ′17O
error acts to “rotate” triple-isotope slopes (Fig. 10).
How important is this effect? It can be significant. Consider a 1-V minor-isotope signal
with a 1-mV pressure baseline being reported by a 1011-Ω amplifier, which is in the range
of what has been observed before (Yeung et al. 2012, 2018; Young et al. 2014). In that case,
iBG = 10 fA with iion = 10 pA, resulting in ϕREF = 0.999. A sample with the same iBG but
δtrue = +20‰ would thus yield δmeasured = 19.98‰, a deviation of −20 ppm from the true value.
If δtrue = +40‰, then δmeasured = 39.96‰ and the deviation is −40 ppm. Such an error would
typically be regarded as insignificant for δ18O values, but it is significant for Δ′17O values.
The error is comparable to the difference between bulk silicate earth and seawater in Δ′17O values
(Pack et al. 2016) and would result in a measured λ value that is too shallow by ~0.001. In a
field where scientific conclusions are routinely based on ppm-level composition differences and
slope differences of 0.001–0.005, this otherwise subtle effect becomes quite a nuisance if gone
unnoticed. Furthermore, the problem can become more significant and more difficult to correct
when sample and reference gases have different amounts of a major isobar (e.g., Ar in O2).
Figure 10. Effect of mass-spectrometric pressure baselines on triple-isotope compositions. (Left) Sche-
matic of positive and negative pressure baselines iBG. (Right) Rotation of triple-isotope slopes, which alters
observed λ values. Reproduced with permission of John Wiley and Sons from Yeung et al. (2018) Scale
distortion from pressure baselines as a source of inaccuracy in triple-isotope measurements Rapid Com-
mun. Mass Spectrom., Vol. 32, Figs. 1 and 3, p. 1813–1814.
Some triple-isotope analysts have independently solved the problem of inter-laboratory

agreement by tying isotopic measurements to a two-point (e.g., VSLAP2-VSMOW2), rather
than single-point calibration scale (Berman et al. 2013; Schoenemann et al. 2013). The two-
point scale implicitly allows the analyst to determine ϕREF values, which are equivalent to oft-
used empirical “stretching” factors. The greatest accuracy is achieved when the two calibration
tie-points show a large difference in isotopic composition (e.g., 55.5‰ for VSLAP2-
VSMOW2); measurements of water triple isotopes across three labs showed agreement
within 8 ppm using this method, in spite of the different techniques utilized (Berman et al.
2013). Note, however, that the two materials used for triple-oxygen calibration need to have a
large and known (or at least agreed-upon) isotopic contrast, preferably >20‰ in δ18O value.
Two calibrated laboratory working gases differing in isotopic composition could serve the
same purpose with a significantly reduced analytical burden.
We will note that the VSMOW-VSLAP scale for triple-oxygen measurements is currently
assigned rather than objectively known. The λ = 0.528 value relating the two standards is
merely an accepted value; a difference between the accepted value of λ and the objectively
true value would result in a scale distortion in the VSMOW-VSLAP scale, and disagreements
between theoretical and experimental values of triple-isotope fractionation.
In practice, these considerations likely limit the accuracy of reported triple-isotope
differences to ±10 ppm, and observed λ values to ±0.001, even when labs are well calibrated
and despite higher intra-laboratory precision. While the magnitude of these potential errors
is worth estimating independently in every laboratory, we caution against over-interpreting
signals within these uncertainty ranges, especially when the interpretations rely on assumed
inter-laboratory agreement. Reporting isotopic compositions using traceable, multi-point
standardization procedures will also aid in improving data quality in future compilations.
Physical and reservoir effects in the real world
Isotopic signals of interest, of course, are seldom as simple as to be limited by theoretical
or analytical uncertainties in nature. Atoms in molecules often arrive at mineral or complex
through a convoluted physical path, fractionated from their source reservoir, and perhaps
mixed with atoms from other sources. Examples relevant to the triple-oxygen system include
the evaporation, condensation, diffusion, and mixing of water vapor or the interphase
transfer of CO2 and O2 (Schoenemann and Steig 2016; Guo and Zhou 2019; Li et al. 2019).
122 Yeung & Hayles
The theory of mass-dependent fractionation includes descriptions of these predominantly

physical effects (here loosely defined as those that do not involve bond breaking or structural
distortion). These effects cause subtle, but meaningful isotopic effects that overprint the
intrinsic chemical effects calculated by first-principles methods.
It is well known that two-endmember reservoir mixing yields curved trajectories in a
cross-plot of Δ′17O vs. δ′18O because Δ′17O values are defined on a logarithmic scale (Miller
2002). Consider the contamination of an oxygen-isotope reservoir by a secondary source
that differs by 30‰ in δ′18O, but has the same Δ′17O value: just a 10% contamination in this
case is needed to cause a ~10 ppm Δ′17O depletion relative to the pure reservoir. Smaller
δ′18O differences between reservoirs yield smaller Δ′17O deviations. Thus the presence of
secondary phases needs to be evaluated explicitly when interpreting measurements at or near
the limits of analytical precision. For example, quartz derived from silica diagenesis may
lead to heterogeneity at the sum-millimeter scale, which would lower Δ′17O values relative to
diagenetic quartz with the same δ′18O value crystallized at a single, well-defined temperature
(Marin-Carbonne et al. 2012; Stefurak et al. 2015; Sengupta and Pack 2018; Hayles et al.
2019; Liljestrand et al. 2020). In marine dissolved-oxygen studies, the presence of unusually
low Δ′17O values at low oxygen concentrations has also been attributed to mixing (Hendricks
et al. 2005; Nicholson et al. 2014), although they may also be due to biological or analytical
artifacts (Stolper et al. 2018; Yeung et al. 2018; Ash et al. 2020).
Other physical effects such as molecular diffusion can be important as well. They overprint
intrinsic chemical isotope effects in both geochemical and biogeochemical systems. In the gas
phase, the triple- and multi-isotopologue effects of diffusion in a variety of molecules can be
calculated with high accuracy using the kinetic theory of gases (Eiler and Schauble 2004; Barkan
and Luz 2007; Yeung et al. 2012, 2017; Magyar et al. 2016; Li et al. 2019): the fractionation
factor in a non-interacting gas mixture is determined by the reduced mass of the diffuser and
the medium by the relation α = (μ/μ*)1/2, where the asterisk indicates that the isotopically
substituted diffuser is used in the reduced mass calculation. For example, a calculation for
O2 diffusing through N2 yields 18α = 0.986 and θ = 0.513. The one published experiment on
this process showed 18α = 0.9849 ± 0.0001 and θ = 0.5228 ± 0.0002 (1σ) instead (Angert
et al. 2003), indicating some experimental or analytical bias. Experiments on the diffusive
fractionation of water vapor through air yielded better agreement, with 18α = 0.9725 ± 0.0003
and θ = 0.5185 ± 0.0002 (1s), while theory predicts 18α = 0.969 and θ = 0.519 (Barkan and
Luz 2007). The above approach may not be valid for a diffuser and medium that interact
meaningfully with each other (e.g., SO2 in water vapor), but there have been few investigations
related to these systems; an analogue to such systems might be gas–solid interaction, which are
known to cause unusual isotope effects at low temperatures (Eiler et al. 2013).
Aqueous-phase diffusive isotope effects are therefore less straightforward to calculate;
solute-solvent interactions are generally significant enough to merit a revision to the simple
ideal-gas reduced-mass power law. For example, the kinetic mass dependence of noble gases
diffusing through water deviates strongly from (μ/μ*)1/2 for atoms larger than He; however,
the limited data on multi-isotope trends is inconclusive as to whether θ values are anomalous
(Bourg and Sposito 2008; Tyroller et al. 2014, 2018; Seltzer et al. 2019). While experimentally
determined Δ′17O and clumped-isotope fractionations for molecular oxygen diffusion are
consistent with the reduced-mass power law above, the 18O/16O fractionation indicates that
exponent is either smaller (0.2) or the characteristic diffusion partner has a larger mass (e.g.,
corresponding to several water molecules) (Knox et al. 1992; Li et al. 2019). Diffusion of
charged species is strongly affected by the energetic cost of ion desolvation, lowering their
effective exponent as well (Richter et al. 2006; Bourg et al. 2010; Hofmann et al. 2012a).
Consequently, diffusive isotope effects manifest in many crystal growth environments (Watkins
et al. 2017), resulting in diffusion-influenced 18α and θ values in real systems.
To illustrate the importance of diffusive fractionation on more “typical” geochemical

processes, consider a two-step reaction that has a reversible first step and an irreversible
second step:
A⇌B→C
This scheme has been used to describe CO2 fixation by leaves (O’Leary 1981), respiratory
oxygen consumption (Stolper et al. 2018; Ash et al. 2020), and mineral precipitation (Watkins
et al. 2014, 2017), among many others. In all of these cases, molecular diffusion is the first
reversible step, although the physical processes are slightly different (e.g., gas-phase, in and
out of leaves through stomata vs. aqueous phase, through the crystallizing boundary layer).
Assuming that the reaction proceeds keeping B in steady state, one can show that the effective
fractionation factor αeff and exponent θeff for A is (Bao et al. 2015):
1
 eff 
  (22)
r  1  1 r
 2 
  17   
ln 17 1  ln r  17 1   1  r 
eff    2   (23)
  18 1  
ln 1  ln r  18   1  r 
18
  2  
where r is the reversibility of A ⇌ B, defined as the flux B → A divided by the flux A → B,
with r = 1 for equal fluxes. Diffusion limitation of reactions can therefore drive 18αeff and
θeff values toward those characteristic of diffusion, which are typically smaller than those
for equilibrium crystallization. Moreover, one may be tempted to infer that θeff always lies
in between that for each step, θ1 and θ2, function of r. However, the equation predicts non-
monotonic variations when θ1 is similar to θ2 and when the 18α1 and 18α2 are quite different
(Fig. 11).
0.528
0.526
18
α1 = 0.990, θ1 = 0.525
18
α2 = 0.980, θ2 = 0.527
0.524
θeff
0.522
18
α1 = 0.990, θ1 = 0.525
18
α2 = 0.900, θ2 = 0.527
0.520
0.0 0.2 0.4 0.6 0.8 1.0

Reversibility
Figure 11. Monatonic and non-monatonic trends in θeff for a two-step reaction according to Equation (23).
The upper curve corresponds to processes with similar 18O/16O fractionation factors, while the lower curve
corresponds to processes with different 18O/16O fractionation factors, all else unchanged.
124 Yeung & Hayles
CASE STUDY: KINETIC ISOTOPE FRACTIONATION

DURING CARBONATE ACID DIGESTION
The transformation of carbonate minerals into CO2 analytes is most commonly achieved
via phosphoric acid digestion (McCrea 1950). During this process, only two of the three
C-bound oxygen atoms are preserved in CO2, resulting in a kinetic isotope fractionation.
This reaction has been studied extensively, both experimentally and theoretically, for 18O/16O
as well as triple-oxygen and clumped isotope fractionation to aid in the reconstruction of the
original carbonate isotopic composition (Kim and O’Neil 1997; Kim et al. 2007; Guo et al.
2009; Passey et al. 2014; Müller et al. 2017; Petersen et al. 2019; Swart et al. 2019; Bergel et
al. 2020; Wostbrock et al. 2020; Zhang et al. 2020). Yet, despite the considerable efforts of the
community, the isotopic fractionation factors remain uncertain. We will use this as a case study
of theory versus the “real world,” focusing on triple oxygen isotopes.
The problems
Two transition-state theory investigations of triple-isotope trends in the reaction have been
published, both using the B3LYP level of theory. Guo et al. (2009) used the double-zeta 6-31G(d)
basis set to investigate H2CO3 hydrolysis and dehydroxylation, whereas Zhang et al. (2020) used
the triple-zeta basis set with diffuse functions 6-311G+(d,p) to investigate H2CO3-catalyzed,
H3O+-catalyzed, and H3PO4-catalyzed mechanisms for H2CO3 decomposition (Fig. 12).
Both investigations scaled the B3LYP vibrational frequencies to those obtained from higher-level
wavefunction methods (MP2 and coupled-cluster, respectively). These calculations yielded acid kin
values between 0.525 and 0.529 at 25 °C, depending on the mechanism, with an effective mean
acid
kin
of 0.528 and 0.527, respectively. In contrast, the calculated 18O/16O fractionation factors
are different by 0.5‰ at room temperature, with ln 18αacid = 10.7‰ and 10.2‰, respectively.
Experimental studies have yielded 18O/16O fractionation factors within this range (Kim and
O’Neil 1997; Kim et al. 2007). Guo et al. (2009) noted some variability arising from the carbonate
partner cation and crystal structure. Zhang et al. (2020) attribute the spread in experimental
18 16
O/ O fractionation factors to partial hydration of nominally anhydrous ortho-phosphoric acid,
which increases the proportion of carbonate reacting through the H3O+-catalyzed pathway.

Figure 12. Three relevant reactants (RE) and transition states (TS) for carbonic acid decomposition during
acid digestion of carbonates. Shown are structures for the (A) H2CO3-catalyzed, (B) H3O+-catalyzed, and
(C) H3PO4-catalyzed pathways. Reprinted with permission of ACS Publications, from Zhang et al. (2020),
Molecular-level mechanism of phosphoric acid digestion of carbonates and recalibration of the 13C–18O
clumped isotope thermometer. ACS Earth & Space Chemistry, Vol. 4, Fig. 1, p. 422.
The two published experimental measurements of acid kin

disagree with both the theoretical
calculations and with each other. The most direct determination of acid kin
was by Wostbrock and
Sharp (2019), who bomb-fluorinated three calcite standards (NBS-18, NBS-19/19a, and IAEA-
603) and the CO2 liberated through acid digestion of each standard. Across all three standards, they
reported consistent acid
kin
values of 0.523 using this method after standardizing against VSMOW2
and VSLAP2 using direct BrF5 fluorination of those waters. Bergel et al. (2020) determined
acid
kin
indirectly using their determinations of (1) the Δ′17O values in CO2 liberated from biogenic
molluskan aragonite that had presumably grown at equilibrium at constant temperatures and (2) the
Δ′17O values of the water in which the mollusks grew. These two quantities were used to obtain an
effective θ value between carbonate-derived CO2 and water. Using this quantity and independently
determined θeq values for carbonate–water equilibrium from theory ( eq carbonate  H 2 O
= 0.526) (Guo
et al. 2009; Hayles et al. 2018) or experiment (0.525; Gibbons et al. 2017), the value of acid kin
was
inferred to be 0.516 or 0.517. Voarintsoa et al. (2020) also found similar results for inorganically
precipitated laboratory calcite. Like the measurements of Wostbrock and Sharp (2019), these
measurements, too, were standardized to VSMOW and VSLAP through two laboratory working
standards (Barkan and Luz 2005; Affek and Barkan 2018).
Evaluation of theory and measurements
First, we will evaluate the theoretical prediction of acid
kin
. One could question the accuracy
of B3LYP calculations for predicting accurate transition-state structures (Xu et al. 2011), which
would affect the predicted kinetic isotope effects. However, both calculations yielded transition-
state structures that compared favorably with high-level wavefunction-theory based structures
and energies, which are reasonably accurate; moreover, the vibrational frequencies of the
transition states were scaled to harmonic frequencies obtained from the wavefunction methods.
Given these accuracy checks, and the relative insensitivity of calculated θ values to method
and basis set, we do not believe that electronic structure methodology is a significant source
of systematic error for θ values. However, both studies derive isotopic fractionations for acid
digestion using the following simplifying assumptions: (i) the reaction occurs in the gas phase
and (ii) the first step of the process—carbonate ion protonation—exhibits no isotopic effects.
The bias introduced by the first assumption is difficult to assess, but surface and/or solvent
stabilization almost certainly plays a role in the reaction pathway. The transition state, and
therefore the kinetic isotope effect, will be affected. Regarding the latter assumption, the three
oxygen positions are not chemically identical; only two of the three oxygens are protonated in a
reacting H2CO3 molecule. Thus a site-specific protonation isotope effect could be relevant, one
that biases the initial isotopic composition of the H-bound oxygens in H2CO3.
Both theory and experiment suggest that surface hydrolysis of calcite by water exposes
Ca(OH)(HCO3) species, which react with acid to form the H2CO3 that decomposes to CO2 and H2O
(Stipp et al. 1994; Kuriyavar et al. 2000; Al-Hosney and Grassian 2004). As a two-step reaction
with a reversible protonation by the acid HX, the process could be represented schematically by:
Ca(OH)(HCO3) + HX ⇌ Ca(OH)X + H2CO3 → CO2 + H2O
with the H2CO3 decomposition proceeding through any of the pathways previously investigated.
The protonation step may consist of two elementary steps, i.e., CO32− + H+ and HCO3− + H+,
but we will assume that these steps yield a net isotopic fractionation αH. Using Equation (23),
the effective θkin value is therefore:
  17   
ln 17  H  ln r  17 H   1  r 
   decomp  
acid,eff
kin

  18
  
ln 18  H  ln r  18 H   1  r 
   decomp  
126 Yeung & Hayles
A θ value for the first step less than 0.527 could lower acid,eff
kin
significantly from the
acid-digestion-only value if protonation is not completely reversible (i.e., r < 1). However,
a acid,eff
kin
value of 0.516–0.517 is difficult to explain. One way to display such low acid,eff
kin
values is for the associated pathway to involve essentially no oxygen-isotope vibrations

beyond the reaction coordinate. However, the presence of three potential oxygen-isotope
substitutions on H2CO3 implies that other pathways contribute to the mean acid kin
value for
H2CO3 decomposition. Another possibility would be for H2CO3 decomposition to be near an
isotope-effect crossover at room temperature. However, the measured 18O/16O fractionation for
acid digestion is relatively large (10‰), suggesting that the reaction does not have a crossover
in this temperature range. A third possibility is that the mechanisms have the right combination
of vibrational modes in the transition states to express these kinetic isotope effects, but that is
disfavored by the DFT calculations. Consequently, while acid,eff
kin
< 0.527 is plausible, a value
as low as 0.516 appears improbable. Nevertheless, the molecular model of the reaction used
in theoretical investigations deserves additional scrutiny; the simplifying assumptions may not
be sufficiently accurate to describe the acid digestion process.
We now turn our attention to the disagreement between laboratory determinations of acid kin
.
Clearly, the results (acid = 0.523 and 0.516–0.517) cannot both be correct; the origin of the
kin
discrepancy likely lies in the differences in methodology. The bomb-fluorination method of

Wostbrock and Sharp (2019) yielded poorer reproducibility in δ18O values (e.g., exceeding
±1‰ for NBS-19/19a) but good reproducibility in Δ′17O values, resulting in an uncertainty in
the carbonate-CO2 Δ′17O difference of about 16–24 ppm (i.e., the half-width of 95% confidence
interval is 8–12 ppm). This conservative estimate of the random errors yields an uncertainty
of ±0.002 in that determination of acid kin
. The indirect determination of acid
kin
by Bergel et al.
(2020) has less well constrained uncertainties given that it utilizes a mix of experimental and
theoretical results. Assuming an overall uncertainty of about ±0.001 in each θ value used
to derive acid
kin
indirectly also yields a likely uncertainty of ±0.002 in that determination.
The results are still significantly different; what other methodological differences are relevant?
The measurements are both standardized using VSMOW and VSLAP, but each lab does
so slightly differently. First, the method for analyzing the Δ′17O value of water is different
between labs: the University of New Mexico (UNM) group uses direct BrF5 fluorination of
water in a nickel micro-bomb to make an O2 analyte, while the Hebrew University group uses
CO2–H2O exchange followed by CO2–O2 exchange on a platinum sponge at 750 °C to make their
O2 analyte (Barkan et al. 2015; Affek and Barkan 2018). The latter double-exchange method
allowed both the carbonate and its growth water to be measured as O2, limiting potential errors
related to standardizing different analytes, but potentially incurring errors from two isotope-
exchange steps. Affek and Barkan (2018) showed that the method could reproduce data obtained
from CoF3 fluorination of water within analytical precision once normalized on the VSMOW-
VSLAP scale. Second, Wostbrock and Sharp (2019) report isotopic compositions relative to
VSMOW, eschewing the use of normalizing factors (Schoenemann et al. 2013), and obtain
Δ′17O = –15 ± 7 ppm (95% CI) for VSLAP. The absence of normalizing factors was justified
because their reported δ18O value for VSLAP is nearly identical to the IAEA recommended
value (–55.5‰), although we note that the absence of scale distortion for δ18O values does not
imply an absence for δ17O values. The Hebrew University group report isotopic compositions
on the VSMOW-VSLAP scale, but it is not clear whether normalizing factors were used; they
report Δ′17O = 7 ± 5 ppm (95% CI) for SLAP, but their reported δ18O value (–55.1‰) is slightly
higher than the IAEA recommended value (Barkan and Luz 2005; Luz and Barkan 2010). This
difference of 22 ± 9 ppm (95% CI) resembles the disparity between labs in the measurement of
NBS-19/19a and NBS-18: the UNM group report a Δ′17O difference of 54 ± 17 ppm (95% CI),
whereas the Hebrew University group most recently reported a Δ′17O difference of 16 ± 6 ppm
(95% CI) (Barkan et al. 2019); the offset between these measurements is 38 ± 18 ppm (95% CI).
Noting that the difference in δ18O value between NBS-19/19a and NBS-18 (21.45‰) is
twice that for acid digestion (10.2‰), the results are consistent with a scaling inconsistency
between the two labs, with UNM-measured θ values being ~0.002 higher than those derived
from the Hebrew University group. Therefore, once normalized on the VSMOW-VSLAP
scale and using conservative estimates for random errors, the acid
kin
discrepancy becomes much
smaller (0.523 ± 0.002 vs. 0.518–0.519 ± 0.002). A downward revision of the theoretically
predicted acid,eff
kin
value after inclusion of the carbonate protonation step may bring theory and
experiment into closer agreement, although the experimental acid kin
value continues to be lower.
We therefore hypothesize that the theory-experiment disagreement in acid kin
arises partially
from an oversimplification of the molecular mechanism of acid digestion and partially from
inconsistent laboratory standardization. Transition-state theory argues against acid
kin
= 0.516, but
multi-step acid digestion chemistry suggests that the effective value may be less than 0.527,
and the effects of surface and/or solvent stabilization are unknown. Additional theoretical
calculations are therefore needed to evaluate the veracity of the theoretical model. Even more
exacting experiments, wherein the 95% confidence limits of acid kin
can be reduced at least by a
factor of two, will be needed to obtain a higher-precision comparison with theory. Emerging
laser-based techniques will offer a useful test of analytical biases of these measurements
(Berman et al. 2013; Steig et al. 2014; Prokhorov et al. 2019)
CONCLUDING REMARKS
A central challenge in theoretical isotope geochemistry has been to calculate isotopic
fractionation with sufficient accuracy to predict, validate, and/or extend the useful range of
experimental studies. In oxygen triple isotopes, the challenge is more acutely focused because
the isotopic variations of interest are often near limits of analytical and theoretical accuracy.
At present, it appears that theory and observations are well matched, with comparable
uncertainties that allow one to push the other forward. However, we argue that these uncertainties
are generally underestimated—by twofold or more—owing to cryptic systematic biases in both
theory and measurements that have only recently come to light. Geochemical triple-isotope
mass-balance models seldom propagate these more realistic uncertainties forward.
Improving triple-oxygen analysis beyond its present level of accuracy will require more
than just a sustained effort to understand and correct for these remaining sources of systematic
errors. With higher precision and accuracy comes higher detail and potentially a new set of
problems, including more subtle expressions of chemical, physical, and reservoir effects in
natural systems. This dilemma is not unlike the task of measuring the length of a coastline (a
problem made famous by Benoit Mandelbrot).
However, we contend that DFT-based isotopic calculations are still underutilized in the
isotope geochemist’s toolkit. They are now affordable for a wide range of geochemically
interesting systems, including large and solvated molecular clusters, for triple-isotope
analysis and beyond. The fortuitous cancellation of errors for multi-isotope fractionation
trends, combined with the potential effectiveness of empirical harmonic frequency scaling
factors allows DFT to bring first-order insights to oxygen triple-isotope fractionation even
on a desktop computer using open-source software. For example, the pathways for isotopic
fractionation associated with incorporating small atmospheric Δ′17O signatures into minerals
through weathering and biological cycling have not been rigorously investigated, despite their
importance for triple-oxygen-based CO2 paleobarometry (Kohn 1996; Bao et al. 2008; Kohl
and Bao 2011; Pack et al. 2013; Passey et al. 2014; Bao 2015; Crockford et al. 2018; Waldeck
et al. 2019; Sutherland et al. 2020).
128 Yeung & Hayles
In addition, few triple-isotope fractionation studies have been conducted for open-shell
systems, which not only include free-radical reactions relevant to the atmospheric chemistry,
but also many interesting reactions in biochemistry. The sensitivity to electronic structure
calculation methods highlighted above is heightened in those systems, as improper accounting
of electron spin, stabilizing long-range interactions, and basis-set biases can lead to qualitative
errors in chemical structures and (presumably also) derived partition functions. Nevertheless,
a judicious choice of methods and some chemical intuition may bring new insight to oxygen
cycling and new predictions for geochemical tracers that can be tested by measurements.
ACKNOWLEDGEMENTS
This work was supported by the David & Lucile Packard Foundation Science and
Engineering Fellowship to L. Y. Y. and National Science Foundation Postdoctoral Fellowship
(grant EAR-1806124) to J. A. H.
REFERENCES
Abe O (2008) Isotope fractionation of molecular oxygen during adsorption/desorption by molecular sieve zeolite.
Adamo C, Barone V (1999) Toward reliable density functional methods without adjustable parameters: The PBE0
model. J Chem Phys 110:6158–6170
Affek HP, Barkan E (2018) A new method for high-precision measurements of 17O/16O ratios in H2O. Rapid Commun
Al-Hosney HA, Grassian VH (2004) Carbonic acid: an important intermediate in the surface chemistry of calcium
carbonate. J Am Chem Soc 126:8068–8069
alternative pathway on the triple isotopic composition on atmospheric O2. Global Biogeochem Cycles 17:1030
Ash JL, Yeung LY, Hu H (2020) What fractionates oxygen isotopes during respiration? Insights from multiple
isotopologue measurements and theory. ACS Earth and Space Chemistry 4:50–66
Babikov D, Kendrick BK, Walker RB, Pack RT (2003) Formation of ozone: Metastable states and the anomalous
isotope effect. J Chem Phys 119:2577–2589
17 16
Balan E, Cartigny P, Blanchard M, Cabaret D, Lazzeri M, Mauri F (2009) Theoretical investigation of the anomalous
equilibrium fractionation of multiple sulfur isotopes during adsorption. Earth Planet Sci Lett 284:88–93
Bao H, Lyons JR, Zhou C (2008) Triple oxygen isotope evidence for elevated CO2 levels after a Neoproterozoic
Bao H, Cao X, Hayles JA (2015) The confines of triple oxygen isotope exponents in elemental and complex mass-
dependent processes. Geochim Cosmochim Acta 170:39–50
Bao H, Cao X, Hayles JA (2016) Triple oxygen isotopes: fundamental relationships and applications. Annu Rev Earth
Barkan E, Luz B (2003) High-precision measurements of 17O/16O and 18O/16O of O2 and O2/Ar ratio in air. Rapid
Spectrom 19:3737–3742
isotope hydrology. Rapid Commun Mass Spectrom 21:2999–3005
Barkan E, Musan I, Luz B (2015) High-precision measurements of δ17O and 17Oexcess of NBS19 and NBS18. Rapid
Barkan E, Affek HP, Luz B, Bergel SJ, Voarintsoa NRG, Musan I (2019) Calibration of δ17O and 17Oexcess values of
three international standards: IAEA-603, NBS19 and NBS18. Rapid Commun Mass Spectrom 33:737–740
Becke AD (1993) Density-functional thermochemistry. III. The role of exact exchange. J Chem Phys 98:5648–5652
Becke AD (2014) Perspective: Fifty years of density-functional theory in chemical physics. J Chem Phys 140:18A301
Bergel SJ, Barkan E, Stein M, Affek HP (2020) Carbonate 17Oexcess as a paleo-hydrology proxy: Triple oxygen isotope
fractionation between H2O and biogenic aragonite, derived from freshwater mollusks. Geochim Cosmochim
Acta 275:36–47
Bergquist BA, Blum JD (2007) Mass-dependent and -independent fractionation of Hg isotopes by photoreduction in
aquatic systems. Science 318:417–420
Berman ESF, Levin NE, Landais A, Li S, Owano T (2013) Measurement of δ18O, δ17O, and 17O-excess in water by
off-axis integrated cavity output spectroscopy and isotope ratio mass spectrometry. Anal Chem 85:10392–10398
Bertini I, Gray HB, Lippard SJ, Valentine JS (1994) Bioinorganic Chemistry. University Science Books, Mill Valley, CA
Bigeleisen J (1949) The relative reaction velocities of isotopic molecules. J Chem Phys 17:675–678
Bigeleisen J, Goeppert-Mayer M (1947) Calculation of equilibrium constants for isotopic exchange reactions. J Chem
Phys 15:261–267
Bigeleisen J, Wolfsberg M (1958) Theoretical and experimental aspects of isotope effects in chemical kinetics. Adv
Chem Phys 1:15–76
Nature 557:545–548
Blanchard M, Balan E, Schauble Edwin A (2017) Equilibrium fractionation of non-traditional isotopes: a molecular
modeling perspective. Rev Mineral Geochem 81:27–63
Bootsma AN, Wheeler S (2019) Popular integration grids can result in large errors in DFT-computed free energies.
ChemRxiv. 8864204:v5.
Born M, Oppenheimer R (1927) Zur Quantentheorie der Molekeln. Annalen der Physik 389:457–484
Bottinga Y (1968) Calculation of fractionation factors for carbon and oxygen isotopic exchange in the system calcite-
carbon dioxide-water. J Phys Chem 72:800–808
Bourg IC, Sposito G (2008) Isotopic fractionation of noble gases by diffusion in liquid water: Molecular dynamics
simulations and hydrologic applications. Geochim Cosmochim Acta 72:2237–2247
Bourg IC, Richter FM, Christensen JN, Sposito G (2010) Isotopic mass dependence of metal cation diffusion
coefficients in liquid water. Geochim Cosmochim Acta 74:2249–2256
Boys SF, Bernardi F (1970) The calculation of small molecular interactions by the differences of separate total
energies. Some procedures with reduced errors. Mol Phys 19:553–566
Buhks E, Bixon M, Jortner J (1981a) Deuterium isotope effects on outer-sphere electron-transfer reactions. J Phys
Chem 85:3763–3766
Buhks E, Bixon M, Jortner J, Navon G (1981b) Quantum effects on the rates of electron-transfer reactions. J Phys
Chem 85:3759–3762
Cao X, Bao H, Gao C, Liu Y, Huang F, Peng Y, Zhang Y (2019) Triple oxygen isotope constraints on the origin of
ocean island basalts. Acta Geochimica 38:327–334
Casado M, Cauquoin A, Landais A, Israel D, Orsi A, Pangui E, Landsberg J, Kerstel E, Prie F, Doussin J-F (2016)
Experimental determination and theoretical framework of kinetic fractionation at the water vapour–ice interface
at low temperature. Geochim Cosmochim Acta 174:54–69
Celotta RJ, Bennett RA, Hall JL, Siegel MW, Levine J (1972) Molecular photodetachment spectrometry. II. The
electron affinity of O2 and the structure of O2−. Phys Rev A 6:631–642
Chacko T, Mayeda TK, Clayton RN, Goldsmith JR (1991) Oxygen and carbon isotope fractionations between CO2
and calcite. Geochim Cosmochim Acta 55:2867–2882
Clayton RN (2002) Self-shielding in the solar nebula. Nature 415:860–861
Crockford PW, Hayles JA, Bao H, Planavsky NJ, Bekker A, Fralick PW, Halverson GP, Bui TH, Peng Y, Wing BA
(2018) Triple oxygen isotope evidence for limited mid-Proterozoic primary productivity. Nature 559:613–616
De Vault D, Chance B (1966) Studies of photosynthesis using a pulsed laser: I. Temperature dependence of cytochrome
oxidation rate in chromatium. Evidence for tunneling. Biophys J 6:825–847
Dieke GH, Babcock HD (1927) The structure of the atmospheric absorption bands of oxygen. PNAS 13:670–678
Ditchfield R, Hehre WJ, Pople JA (1971) Self-consistent molecular-orbital methods. IX. An extended gaussian-type
basis for molecular-orbital studies of organic molecules. J Chem Phys 54:724–728
Dole M (1935) The relative atomic weight of oxygen in water and air. J Am Chem Soc 57:273
Dole M, Jenks G (1944) Isotopic composition of photosynthetic oxygen. Science 100:409
Dole M, Hawkings RC, Barker HA (1947) Bacterial fractionation of oxygen isotopes. J Am Chem Soc 69:226–228
Dove MT, Winkler B, Leslie M, Harris MJ, Salje EKH (1992) A new interatomic potential model for calcite:
Applications to lattice dynamics studies, phase transition, and isotope fractionation. Am Mineral 77:244–250
130 Yeung & Hayles
Downs RT, Hall-Wallace M (2003) The American Mineralogist crystal structure database. Am Mineral 88:247–250
Dunning TJ (1989) Gaussian basis sets for use in correlated molecular calculations. I. The atoms boron through neon
and hydrogen. J Chem Phys 90:1007–1023
Eiler JM (2007) “Clumped-isotope” geochemistr—The study of naturally-occurring, multiply-substituted
isotopologues. Earth Planet Sci Lett 262:309–327
Eiler JM, Schauble E (2004) 18O13C16O in the Earth’s atmosphere. Geochim Cosmochim Acta 68:4767–4777
Eiler J, Cartigny P, Hofmann AE, Piasecki A (2013) Non-canonical mass laws in equilibrium isotopic fractionations:
Evidence from the vapor pressure isotope effect of SF6. Geochim Cosmochim Acta 107:205–219
Elcombe MM, Pryor AW (1970) The lattice dynamics of calcium fluoride. J Phys C: Solid State Phys 3:492–499
Elcombe MM, Hulston JR (1975) Calculation on sulphur isotope fractionation between sphalerite and galena using
lattice dynamics. Earth Planet Sci Lett 28:172–180
Ellingson BA, Lynch VA, Mielke SL, Truhlar DG (2006) Statistical thermodynamics of bond torsional modes: Tests
of separable, almost-separable, and improved Pitzer–Gwinn approximations. J Chem Phys 125:084305
Epstein S, Mayeda T (1953) Variation of O18 content of waters from natural sources. Geochim Cosmochim Acta 4:213–224
Evans MG, Polanyi M (1935) Some applications of the transition state method to the calculation of reaction velocities,
especially in solution. Trans Faraday Soc 31:875–894
Eyring H (1935) The activated complex in chemical reactions. J Chem Phys 3:107–115
Farquhar J, Savarino J, Airieau S, Thiemens MH (2001) Observation of wavelength-sensitive mass-independent
sulfur isotope effects during SO2 photolysis: Implications for the early atmosphere. J Geophys Res: Planets
106:32829–32839
Frisch MJ, Trucks GW, Schlegel HB, Scuseria GE, Robb MA, Cheeseman JR, Scalmani G, Barone V, Petersson GA,
Nakatsuji H, Li X, Caricato M, Marenich AV, Bloino J, Janesko BG, Gomperts R, Mennucci B, Hratchian HP,
Ortiz JV, Izmaylov AF, Sonnenberg JL, Williams-Young D, Ding F, Lipparini F, Egidi F, Goings J, Peng B,
Petrone A, Henderson T, Ranasinghe D, Zakrzewski VG, Gao J, Rega N, Zheng G, Liang W, Hada M, Ehara
M, Toyota K, Fukuda R, Hasegawa J, Ishida M, Nakajima T, Honda Y, Kitao O, Nakai H, Vreven T, Throssell
K, Montgomery JA, Jr., Peralta JE, Ogliaro F, Bearpark MJ, Heyd JJ, Brothers EN, Kudin KN, Staroverov VN,
Keith TA, Kobayashi R, Normand J, Raghavachari K, Rendell AP, Burant JC, Iyengar SS, Tomasi J, Cossi M,
Millam JM, Klene M, Adamo C, Cammi R, Ochterski JW, Martin RL, Morokuma K, Farkas O, Foresman JB,
Fox DJ, (2016) Gaussian 16, Revision B.01, Gaussian, Inc., Wallingford CT.
Furche F, Ahlrichs R, Hättig C, Klopper W, Sierka M, Weigend F (2014) Turbomole. WIREs Comput Mol Sci 4:91–100
Gao YQ, Marcus RA (2001) Strange and unconventional isotope effects in ozone formation. Science 293:259–263
Giannozzi P, Andreussi O, Brumme T, Bunau O, Nardelli MB, Calandra M, Car R, Cavazzoni C, Ceresoli D,
Cococcioni M, Colonna N (2017) Advanced capabilities for materials modelling with Quantum ESPRESSO. J
Phys: Condens Matter 29:465901
Giauque WF, Johnston HL (1929a) An isotope of oxygen of mass 17 in the Earth’s atmosphere. Nature 123:831–831
Giauque WF, Johnston HL (1929b) An isotope of oxygen, Mass 18. Nature 123:318–318
Gibbons JA, Sharp ZD, Atudorei V (2017) Oxygen isotope analyses of biogenic carbonate to reconstruct early Triassic
ocean oxygen isotopic values and temperatures. American Geophysical Union Fall Meeting Abstracts:PP14A-04
Grimme S (2006) Semiempirical hybrid density functional with perturbative second-order correlation. J Chem Phys
124:034108
Guo W, Zhou C (2019) Triple oxygen isotope fractionation in the DIC–H2O–CO2 system: A numerical framework
and its implications. Geochim Cosmochim Acta 246:541–564
Guo W, Mosenfelder JL, Goddard III WA, Eiler JM (2009) Isotopic fractionations associated with phosphoric
acid digestion of carbonate minerals: Insights from first-principles theoretical modeling and clumped isotope
measurements Geochim Cosmochim Acta 73:7203–7225
Hariharan PC, Pople JA (1973) The influence of polarization functions on molecular orbital hydrogenation energies.
Theor Chim Acta 28:213–222
Hathorn BC, Marcus RA (1999) An intramolecular theory of the mass-independent isotope effect for ozone. I. J Chem
Phys 111:4087–4100
Hathorn BC, Marcus RA (2000) An intramolecular theory of the mass-independent isotope effect for ozone. II.
Numerical implementation at low pressures using a loose transition state. J Chem Phys 113:9497–9509
Hayles JA, Cao X, Bao H (2017) The statistical mechanical basis of the triple isotope fractionation relationship.
Geochem Persp Let 3:1–11
Hayles J, Gao C, Cao X, Liu Y, Bao H (2018) Theoretical calibration of the triple oxygen isotope thermometer.
Hayles JA, Yeung LY, Homann MAB, Jiang H, Shen B, Lee C-TA (2019) Three billion year secular evolution of the
triple oxygen isotope composition of marine chert.Submitted
He B, Olack G, Colman A (2012) Pressure baseline correction and high precision CO2 clumped-isotope (Δ47)
measurements in bellows and micro-volume modes. Rapid Commun Mass Spectrom 26:2837–2853
He Y, Bao H (2019) Predicting high-dimensional isotope relationships from diagnostic fractionation factors in systems
with diffusional mass transfer. ACS Earth Space Chem 3:120–128
Hehre WJ, Ditchfield R, Pople JA (1972) Self-consistent molecular orbital methods. XII. Further extensions of Gaussian—
type basis sets for use in molecular orbital studies of organic molecules. J Chem Phys 56:2257–2261
Helman Y, Barkan E, Eisenstadt D, Luz B, Kaplan A (2005) Fractionation of the three stable oxygen isotopes by
oxygen-producing and oxygen-consuming reactions in photosynthetic organisms. Plant Physiol 138:2292–2298
Hendricks MB, Bender ML, Barnett BA, Strutton P, Chavez FP (2005) Triple oxygen isotope composition of dissolved
O2 in the equatorial Pacific: A tracer of mixing, production, and respiration. J Geophys Res 110:C12021
Hermida-Ramón JM, Graña AM (2007) Blue-shifting hydrogen bond in the benzene–benzene and benzene–
naphthalene complexes. J Comput Chem 28:540–546
Hermida-Ramón JM, Karlström G, Nelander B (2003) On the influence of the basis set superposition error on
calculated vibrational frequencies. Theo Chem Acc 110:190–195
344:1146–1150
Hofmann AE, Bourg IC, DePaolo DJ (2012a) Ion desolvation as a mechanism for kinetic isotope fractionation in
aqueous systems. PNAS 109:18689–18694
Hofmann MEG, Horváth B, Pack A (2012b) Triple oxygen isotope equilibrium fractionation between carbon dioxide
and water. Earth Planet Sci Lett 319–320:159–164
Hopfield JJ (1974) Electron transfer between biological molecules by thermally activated tunneling. PNAS 71:3640–3644
Horita J, Wesolowski DJ (1994) Liquid-vapor fractionation of oxygen and hydrogen isotopes of water from the
freezing to the critical temperature. Geochim Cosmochim Acta 58:3425–3437
Huber KP, Herzberg G (1979) Constants of diatomic molecules. Van Nostrand Reinhold Company, New York, NY
Jensen F (2013) Atomic orbital basis sets. WIREs Comput Mol Sci 3:273–295
Jensen F (2017) Introduction to Computational Chemistry. John Wiley & Sons, Ltd., Hoboken, NJ
Juranek LW, Quay PD (2013) Using triple isotopes of dissolved oxygen to evaluate global marine productivity. Annu
Rev Mar Sci 5:503–524
Kashinski DO, Chase GM, Nelson RG, Di Nallo OE, Scales AN, VanderLey DL, Byrd EFC (2017) Harmonic
vibrational frequencies: approximate global scaling factors for TPSS, M06, and M11 Functional Families Using
Several Common Basis Sets. J Phys Chem A 121:2265–2273
Kesharwani MK, Brauer B, Martin JML (2015) Frequency and zero-point vibrational energy scale factors for double-
hybrid density functionals (and other selected methods): Can anharmonic force fields be avoided? J Phys Chem
A 119:1701–1714
Kieffer SW (1982) Thermodynamics and lattice vibrations of minerals: 5. Applications to phase equilibria, isotopic
fractionation, and high-pressure thermodynamic properties. Rev Geophys 20:827–849
Kim S-T, O’Neil JR (1997) Equilibrium and nonequilibrium oxygen isotope effects in synthetic carbonates. Geochim
Kim S-T, Mucci A, Taylor BE (2007) Phosphoric acid fractionation factors for calcite and aragonite between 25 and
75 °C: Revisited. Chem Geol 246:135–146
Knox M, Quay PD, Wilbur D (1992) Kinetic isotope fractionation during air–water gas transfer of O2, N2, CH4, and
H2. J Geophys Res 97:20,335–320,343
Kobko N, Dannenberg JJ (2001) Effect of basis set superposition error (BSSE) upon ab initio calculations of organic
transition states. J Phys Chem A 105:1944–1950
Kohl I, Bao H (2011) Triple-oxygen-isotope determination of molecular oxygen incorporation in sulfate produced
during abiotic pyrite oxidation (pH=2–11). Geochim Cosmochim Acta 75:1785–1798
Kohn MJ (1996) Predicting animal δ18O: Accounting for diet and physiological adaptation. Geochim Cosmochim
Acta 60:4811–4829
Kotaka M, Okamoto M, Bigeleisen J (1992) Anomalous mass effects in isotopic exchange equilibria. J Am Chem
Soc 114:6436–6445
Kresse G, Furthmüller J (1996) Efficient iterative schemes for ab initio total-energy calculations using a plane-wave
basis set. Phys Rev B 54:11169–11186
Kuriyavar SI, Vetrivel R, Hegde SG, Ramaswamy AV, Chakrabarty D, Mahapatra S (2000) Insights into the formation
of hydroxyl ions in calcium carbonate: temperature dependent FTIR and molecular modelling studies. J Mater
Chem 10:1835–1840
Lane GA, Dole M (1956) Fractionation of oxygen isotopes during respiration. Science 123:574–576
Laskar AH, Mahata S, Bhattacharya SK, Liang M-C (2019) Triple oxygen and clumped isotope compositions of CO2
in the middle troposphere. Earth Space Sci 6:1205–1219
Lee C, Yang W, Parr RG (1988) Development of the Colle-Salvetti correlation-energy formula into a functional of the
electron density. Phys Rev B 37:785–789
Levien L, Prewitt CT, Weidner DJ (1980) Structure and elastic properties of quartz at pressure. Am Mineral 65:920–930
Li X, Liu Y (2015) A theoretical model of isotopic fractionation by thermal diffusion and its implementation on
silicate melts. Geochim Cosmochim Acta 154:18–27
Li B, Yeung LY, Hu H, Ash JL (2019) Kinetic and equilibrium fractionation of O2 isotopologues during air-water gas
transfer and implications for tracing biological oxygen cycling in the ocean. Mar Chem 210:61–71
132 Yeung & Hayles
Liljestrand FL, Knoll AH, Tosca NJ, Cohen PA, Macdonald FA, Peng Y, Johnston DT (2020) The triple oxygen
isotope composition of Precambrian chert. Earth Planet Sci Lett 537:116167
Liu Q, Tossell JA, Liu Y (2010) On the proper use of the Bigeleisen–Mayer equation and corrections to it in the
calculation of isotopic fractionation equilibrium constants. Geochim Cosmochim Acta 74:6965–6983
Luz B, Barkan E (2000) Assessment of oceanic productivity with the triple-isotope composition of dissolved oxygen.
Science 288:2028–2031
Luz B, Barkan E (2005) The isotopic ratios 17O/16O and 18O/16O in molecular oxygen and their significance in
biogeochemistry. Geochim Cosmochim Acta 69:1099–1110
Luz B, Barkan E (2011) The isotopic composition of atmospheric oxygen. Global Biogeochem Cycles 25:GB3001
Luz B, Barkan E, Severinghaus JP (2014) 5.14 – The stable isotopic composition of atmospheric O2. In: Treatise on
Geochemistry (Second Edition). Holland HD, Turekian KK, (eds). Elsevier, Oxford, p 363–383
Magyar PM, Orphan VJ, Eiler JM (2016) Measurement of rare isotopologues of nitrous oxide by high-resolution
multi-collector mass spectrometry. Rapid Commun Mass Spectrom 30:1923–1940
Mahata S, Bhattacharya SK, Wang C-H, Liang M-C (2013) Oxygen isotope exchange between O2 and CO2 over hot
platinum: an innovative technique for measuring Δ17O in CO2. Anal Chem 85:6894–6901
Marcus RA (1952) Unimolecular dissociations and free radical recombination reactions. J Chem Phys 20:359–364
Marcus RA (2004) Mass-independent isotope effect in the earliest processed solids in the solar system: A possible
chemical mechanism. J Chem Phys 121:8201–8211
Mardirossian N, Head-Gordon M (2017) Thirty years of density functional theory in computational chemistry: an
overview and extensive assessment of 200 density functionals. Mol Phys 115:2315–2372
Marin-Carbonne J, Chaussidon M, Robert F (2012) Micrometer-scale chemical and isotopic criteria (O and Si) on
the origin and history of Precambrian cherts: Implications for paleo-temperature reconstructions. Geochim
McCrea JM (1950) On the Isotopic Chemistry of Carbonates and a Paleotemperature Scale. J Chem Phys 18:849–857
Medvedev MG, Bushmarinov IS, Sun J, Perdew JP, Lyssenko KA (2017) Density functional theory is straying from
the path toward the exact functional. Science 355:49–52
an appraisal and geochemical significance. Geochim Cosmochim Acta 66:1881–1889
Moran D, Simmonett AC, Leach FE, Allen WD, Schleyer PvR, Schaefer HF (2006) Popular theoretical methods
predict benzene and arenes to be nonplanar. J Am Chem Soc 128:9342–9343
Mukherjee A, Smirnov VV, Lanci MP, Brown DE, Shepard EM, Dooley DM, Roth JP (2008) Inner-sphere mechanism
for molecular oxygen reduction catalyzed by copper amine oxidases. J Am Chem Soc 130:9459–9473
Müller IA, Violay MES, Storck J-C, Fernandez A, van Dijk J, Madonna C, Bernasconi SM (2017) Clumped isotope
fractionation during phosphoric acid digestion of carbonates at 70 °C. Chem Geol 449:1–14
Mulliken RS (1928) Interpretation of the atmospheric oxygen bands; electronic levels of the oxygen molecule. Nature
122:505–505
Murray ST, Arienzo MM, Swart PK (2016) Determining the Δ47 acid fractionation in dolomites. Geochim Cosmochim
Acta 174:42–53
Nicholson D, Stanley RHR, Doney SC (2014) The triple oxygen isotope tracer of primary productivity in a dynamic
ocean model. Global Biogeochem Cycles 28.5:538–552
Nier AO (1947) A mass spectrometer for isotope and gas analysis. Rev Sci Instrum 18:398–411
NIST Mass Spectrometry Data Center WEW, director (2020) Infrared spectra. In: NIST Chemistry WebBook, NIST
Standard Reference Database Number 69. Linstrom PJ, Mallard WG, (eds). National Institute of Standards and
Technology, Gaithersburg, MD
O’Leary MH (1981) Carbon isotope fractionation in plants. Phytochemistry 20:553–567
as paleo-CO2-barometer. Geochim Cosmochim Acta 102:306–317
Pack A, Höweling A, Hezel DC, Stefanak MT, Beck A-K, Peters STM, Sengupta S, Herwartz D, Folco L (2017) Tracing
the oxygen isotopic composition of the upper Earth’s atmosphere using cosmic spherules. Nat Commun 8:15702
Paier J, Marsman M, Kresse G (2007) Why does the B3LYP hybrid functional fail for metals? J Chem Phys
127:024103
Paizs B, Suhai S (1998) Comparative study of BSSE correction methods at DFT and MP2 levels of theory. J Comput
Chem 19:575–584
Passey BH, Hu H, Ji H, Montanari S, Li S, Henkes GA, Levin NE (2014) Triple oxygen isotopes in biogenic and
sedimentary carbonates. Geochim Cosmochim Acta 141:1–25
Peng C, Calvin JA, Pavošević F, Zhang J, Valeev EF (2016) Massively parallel implementation of explicitly correlated
coupled-cluster singles and doubles using tiledarray framework. J Phys Chem A 120:10231–10244
Perdew JP, Zunger A (1981) Self-interaction correction to density-functional approximations for many-electron
systems. Phys Rev B 23:5048–5079
Perdew JP, Schmidt K (2001) Jacob’s ladder of density functional approximations for the exchange-correlation energy.
In: Density Functional Theory and Its Application to Materials. Vol 577. Doren VV, Alsenoy CV, Geerlings P,
(eds). American Institute of Physics, Melville, NY
Perdew JP, Burke K, Ernzerhof M (1996) Generalized Gradient Approximation made simple. Phys Rev Lett 77:3865–3868
Perdew JP, Ruzsinszky A, Csonka GI, Constantin LA, Sun J (2009) Workhorse semilocal density functional for
condensed matter physics and quantum chemistry. Phys Rev Lett 103:026403
Petersen SV, Defliese WF, Saenger C, Daëron M, Huntington KW, John CM, Kelson JR, Bernasconi SM, Colman AS,
Kluge T, Olack GA (2019) Effects of improved 17O correction on interlaboratory agreement in clumped isotope
calibrations, estimates of mineral-specific offsets, and temperature dependence of acid digestion fractionation.
Geochem Geophys Geosy 20:3495–3519
Peverati R, Truhlar DG (2011) Improving the accuracy of hybrid Meta-GGA density functionals by range separation.
J Phys Chem Lett 2:2810–2817
Peverati R, Truhlar DG (2012) M11-L: A local density functional that provides improved accuracy for electronic
structure calculations in chemistry and physics. J Phys Chem Lett 3:117–124
Peverati R, Truhlar DG (2014) Quest for a universal density functional: the accuracy of density functionals across a
broad spectrum of databases in chemistry and physics. Philos Trans R Soc London, Ser A 372:20120476
Piasecki A, Sessions A, Peterson B, Eiler J (2016) Prediction of equilibrium distributions of isotopologues for
methane, ethane and propane using density functional theory. Geochim Cosmochim Acta 190:1–12
Piasecki A, Sessions A, Lawson M, Ferreira AA, Santos Neto EV, Ellis GS, Lewan MD, Eiler JM (2018) Position-specific
13
C distributions within propane from experiments and natural gas samples. Geochim Cosmochim Acta 220:110–124
Prokhorov I, Kluge T, Janssen C (2019) Laser absorption spectroscopy of rare and doubly substituted carbon dioxide
isotopologues. Anal Chem 91:15491–15499
Redlich O (1935) Eine allgemeine Beziehung zwischen den Schwingungsfrequenzen isotoper Molekeln. 28B:371
Richter FM, Mendybaev RA, Christensen JN, Hutcheon ID, Williams RW, Sturchio NC, Beloso AD (2006) Kinetic
isotopic fractionation during diffusion of ionic species in water. Geochim Cosmochim Acta 70:277–289
Röckmann T, Brenninkmeijer CAM, Saueressig G, Bergamaschi P, Crowley JN, Fischer H, Crutzen PJ (1998) Mass-
independent oxygen isotope fractionation in atmospheric CO as a result of the reaction CO+OH. Science 281:544–546
Roth JP, Wincek R, Nodet G, Edmondson DE, McIntire WS, Klinman JP (2004) Oxygen isotope effects on electron transfer
to O2 probed using chemically modified flavins bound to glucose oxidase. J Am Chem Soc 126:15120–15131
Rustad JR, Nelmes SL, Jackson VE, Dixon DA (2008) Quantum-chemical calculations of carbon-isotope fractionation
in CO2(g), aqueous carbonate species, and carbonate minerals. J Phys Chem A 112:542–555
Rustad JR, Casey WH, Yin Q-Z, Bylaska EJ, Felmy AR, Bogatko SA, Jackson VE, Dixon DA (2010) Isotopic fractionation
of Mg2+(aq), Ca2+(aq), and Fe2+(aq) with carbonate minerals. Geochim Cosmochim Acta 74:6301–6323
Rutherford E (1919) LIV. Collision of α particles with light atoms. IV. An anomalous effect in nitrogen. The London,
Edinburgh, and Dublin Philosophical Magazine and Journal of Science 37:581–587
Schauble EA (2004) Applying stable isotope fractionation theory to new systems. Rev Mineral Geochem 55:65–111
Schauble EA, Ghosh P, Eiler JM (2006) Preferential formation of 13C–18O bonds in carbonate minerals, estimated
using first-principles lattice dynamics Geochim Cosmochim Acta 70:2510–2529
Schinke R, Grebenshchikov SY, Ivanov MV, Fleurat-Lessard P (2006) Dynamical studies of the ozone isotope effect:
a status report. Annu Rev Phys Chem 57:625–661
Schmidt MW, Baldridge KK, Boatz JA, Elbert ST, Gordon MS, Jensen JH, Koseki S, Matsunaga N, Nguyen KA, Su
S, Windus TL (1993) General atomic and molecular electronic structure system. J Comput Chem 14:1347–1363
Insights from an intermediate complexity isotope model. J Geophys Res-Atmos 121:11,215–211,247
Schwerdtfeger P (2011) The pseudopotential approximation in electronic structure theory. ChemPhysChem 12:3143–3155
Seltzer AM, Ng J, Severinghaus JP (2019) Precise determination of Ar, Kr and Xe isotopic fractionation due to
diffusion and dissolution in fresh water. Earth Planet Sci Lett 514:156–165
Sengupta S, Pack A (2018) Triple oxygen isotope mass balance for the Earth’s oceans with application to Archean
cherts. Chem Geol 495:18–26
134 Yeung & Hayles
Shao Y, Gan Z, Epifanovsky E, et al. (2015) Advances in molecular quantum chemistry contained in the Q-Chem 4
program package. Mol Phys 113:184–215
Sharp ZD (1990) A laser-based microanalytical method for the in situ determination of oxygen isotope ratios of
silicates and oxides. Geochim Cosmochim Acta 54:1353–1357
Sharp ZD, Gibbons JA, Maltsev O, Atudorei V, Pack A, Sengupta S, Shock EL, Knauth LP (2016) A calibration of
the triple oxygen isotope fractionation in the SiO2–H2O system and applications to natural samples. Geochim
Sharp ZD, Wostbrock JAG, Pack A (2018) Mass-dependent triple oxygen isotope variations in terrestrial materials.
Skaron S, Wolfsberg M (1980) Anomalies in the fractionation by chemical equilibrium of 18O/16O relative to 17O/16O.
J Chem Phys 72:6810–6811
Staroverov VN, Scuseria GE, Tao J, Perdew JP (2003) Comparative assessment of a new nonempirical density
functional: Molecules and hydrogen-bonded complexes. J Chem Phys 119:12129–12137
Stefurak EJT, Fischer WW, Lowe DR (2015) Texture-specific Si isotope variations in Barberton Greenstone Belt
cherts record low temperature fractionations in early Archean seawater. Geochim Cosmochim Acta 150:26–52
Steig EJ, Gkinis V, Schauer AJ, Schoenemann SW, Samek K, Hoffnagle J, Dennis KJ, Tan SM (2014) Calibrated
high-precision 17O-excess measurements using cavity ring-down spectroscopy with laser-current-tuned cavity
Stipp SLS, Eggleston CM, Nielsen BS (1994) Calcite surface structure observed at microtopographic and molecular
scales with atomic force microscopy (AFM). Geochim Cosmochim Acta 58:3023–3033
Stolper DA, Fischer WW, Bender ML (2018) Effects of temperature and carbon source on the isotopic fractionations
associated with O2 respiration for 17O/16O and 18O/16O ratios in E. coli. Geochim Cosmochim Acta 240:152–172
Su NQ, Zhu Z, Xu X (2018) Doubly hybrid density functionals that correctly describe both density and energy for
atoms. PNAS 115:2287–2292
Sun T, Bao H (2011) Thermal-gradient-induced non-mass-dependent isotope fractionation. Rapid Commun Mass
Sutherland KM, Wostbrock JAG, Hansel CM, Sharp ZD, Hein JR, Wankel SD (2020) Ferromanganese crusts as
recorders of marine dissolved oxygen. Earth Planet Sci Lett 533:116057
Swart PK, Murray ST, Staudigel PT, Hodell DA (2019) Oxygen isotopic exchange between CO2 and phosphoric acid:
implications for the measurement of clumped isotopes in carbonates. Geochem Geophys Geosy 20:3730–3750
Tanaka R, Nakamura E (2012) Determination of 17O-excess of terrestrial silicate/oxide minerals with respect to Vienna
Standard Mean Ocean Water (VSMOW). Rapid Commun Mass Spectrom 27:285–297
Tao J, Perdew JP, Staroverov VN, Scuseria GE (2003) Climbing the density functional ladder: nonempirical meta-
generalized gradient approximation designed for molecules and solids. Phys Rev Lett 91:146401
Thiemens MH (2006) History and applications of mass-independent isotope effects. Annu Rev Earth Planet Sci 34:217–262
Thiemens MH, Heidenreich JE (1983) The mass-independent fractionation of oxygen: A novel isotope effect and its
Tyroller L, Brennwald MS, Mächler L, Livingstone DM, Kipfer R (2014) Fractionation of Ne and Ar isotopes by
molecular diffusion in water. Geochim Cosmochim Acta 136:60–66
Tyroller L, Brennwald MS, Busemann H, Maden C, Baur H, Kipfer R (2018) Negligible fractionation of Kr and Xe
isotopes by molecular diffusion in water. Earth Planet Sci Lett 492:73–78
Urey HC (1947) The thermodynamic properties of isotopic substances. J Chem Soc 47:562–581
Valiev M, Bylaska EJ, Govind N, Kowalski K, Straatsma TP, Van Dam HJ, Wang D, Nieplocha J, Apra E, Windus TL,
De Jong WA (2010) NWChem: A comprehensive and scalable open-source solution for large scale molecular
simulations. Comput Phys Commun 181:1477–1489
Voarintsoa NRG, Barkan E, Bergel S, Vieten R, Affek HP (2020) Triple oxygen isotope fractionation between CaCO3
and H2O in inorganically precipitated calcite and aragonite. Chem Geol 539:119500
Vogiatzis KD, Ma D, Olsen J, Gagliardi L, Jong WAd (2017) Pushing configuration-interaction to the limit: Towards
massively parallel MCSCF calculations. J Chem Phys 147:184111
Waldeck AR, Cowie BR, Bertran E, Wing BA, Halevy I, Johnston DT (2019) Deciphering the atmospheric signal in
marine sulfate oxygen isotope composition. Earth Planet Sci Lett 522:12–19
Wang Z, Schauble EA, Eiler JM (2004) Equilibrium thermodynamics of multiply substituted isotopologues of
molecular gases. Geochim Cosmochim Acta 68:4779–4797
Watkins JM, DePaolo DJ, Watson EB (2017) Kinetic fractionation of non-traditional stable isotopes by diffusion and
crystal growth reactions. Rev Mineral Geochem 82:85–125
Watkins JM, Hunt JD, Ryerson FJ, DePaolo DJ (2014) The influence of temperature, pH, and growth rate on the δ18O
composition of inorganically precipitated calcite. Earth Planet Sci Lett 404:332–343
Weigend F, Ahlrichs R (2005) Balanced basis sets of split valence, triple zeta valence and quadruple zeta valence
quality for H to Rn: Design and assessment of accuracy. Phys Chem Chem Phys 7:3297–3305
Wigner E (1932) On the quantum correction for thermodynamic equilibrium. Phys Rev 40:749
Winta CJ, Gewinner S, Schöllkopf W, Wolf M, Paarmann A (2018) Second-harmonic phonon spectroscopy of
α-quartz. Phys Rev B 97:094108
Witte J, Neaton JB, Head-Gordon M (2016) Push it to the limit: Characterizing the convergence of common sequences
of basis sets for intermolecular interactions as described by density functional theory. J Chem Phys 144:194306
Wostbrock JAG, Sharp ZD, Sanchez-Yanez C, Reich M, van den Heuvel DB, Benning LG (2018) Calibration and
application of silica-water triple oxygen isotope thermometry to geothermal systems in Iceland and Chile.
Wostbrock JAG, Cano EJ, Sharp ZD (2020) An internally consistent triple oxygen isotope calibration of standards for
Xu X, Alecu IM, Truhlar DG (2011) How well can modern density functionals predict internuclear distances at
transition states? J Chem Theory Comput 7:1667–1676
Yeung LY, Young ED, Schauble EA (2012) Measurements of 18O18O and 17O18O in the atmosphere and the influence
of isotope-exchange reactions. J Geophys Res 117:D18306
Yeung LY, Li S, Kohl IE, Haslun JA, Ostrom NE, Hu H, Fischer TP, Schauble EA, Young ED (2017) Extreme
enrichment in atmospheric 15N15N. Sci Adv 3:eaao6741
in triple-isotope measurements. Rapid Commun Mass Spectrom 32:1811–1821
and ther geochemical and cosmochemical significance. Geochim Cosmochim Acta 66:1095–1104
Zhang S, Liu Q, Tang M, Liu Y (2020) Molecular-level mechanism of phosphoric acid digestion of carbonates and
recalibration of the 13C–18O clumped isotope thermometer. ACS Earth Space Chem 4:420–433
Vol. 86 pp. 137–178, 2021 4
Mass Dependence of Equilibrium Oxygen Isotope

Fractionation in Carbonate, Nitrate, Oxide, Perchlorate,
Phosphate, Silicate, and Sulfate Minerals
Edwin A. Schauble and Edward D. Young
Department of Earth, Planetary, and Space Sciences
UCLA
Los Angeles, CA 90095-1567
USA
schauble@g.ucla.edu
eyoung@epss.ucla.edu
INTRODUCTION
Variation in both O/ O and O/16O ratios in natural materials can now be measured
18 16 17
with unprecedented precision, with a broad range of potential geochemical applications. In this
chapter, equilibrium 18O/16O and 17O/16O fractionation factors are calculated for a selection of
minerals and molecules, using first-principles density functional theory models to estimate
vibrational frequencies, with a particular focus on investigating the potential for detectable
signatures of high-temperature equilibrium processes. Reduced partition function ratios as well
as mass-fractionation exponents are tabulated versus temperature. The results are compared
with previous theoretical studies, laboratory experiments, and field-based calibrations. Effects
of nuclear field shift isotope fractionation and double-well potential anharmonicity on the
relationship between 18O/16O and 17O/16O are also investigated. The estimated field shift effect is
much smaller than mass-dependent fractionation, yielding no more than 1 per meg in measured
∆′17O at 25 ºC, and correspondingly less at higher temperatures. Anharmonic vibration in a
double-well potential, such as might be found in a Si–O–Si linkage in polymerized silicates,
also does not seem to generate dramatic ∆′’17O signatures for plausible potential shapes, and
non-Born–Oppenheimer effects on ∆′17O signatures also appear to be limited. None of the
studied effects appear likely to generate the negative ∆′17O anomalies observed in polymerized
silicate mineral samples from high-temperature rocks on the Earth & Moon.
D17O signatures of equilibrium processes
Large mass-independent fractionation signatures in oxygen, sulfur, and mercury isotopes
in natural samples are well known, and are generally thought to result from a restricted set of
disequilibrium reaction processes including photochemistry and molecular recombination (e.g.
Heidenreich and Thiemens 1986; Mauersberger 1987 for ozone; Clayton et al. 1973; Mc Keegan
et al. 2011 for early solar system materials; Farquhar et al. 2000, for the Archean sulfur cycle;
Bergquist and Blum 2007, for mercury). Over the last two decades, possible uses of subtle
variability in the mass-dependence of isotope fractionation caused by more mundane processes
have become a topic of broad interest in geochemistry (e.g., Young et al. 2002; Deines 2003;
Farquhar et al. 2003; Barkan and Luz 2005; Rumble et al. 2007; Luz and Barkhan 2010; Cao
and Liu 2011 ; Levin et al. 2014; Pack and Herwartz 2014; Passey et al. 2014; Bao et al. 2015;
Dauphas and Schauble 2016). Variation in the mass dependence of equilibrium stable isotope
fractionation has long been predicted (e.g., Grilly 1951; Hulston and Thode 1965; Matsuhisa et al.
1978), but this phenomenon has not been widely exploited in geochemical studies until recently.
1943-2666/21/0086-0004$05.00 (online)
138 Schauble & Young
It is now recognized that measurements of variations in the mass dependence of equilibrium

oxygen isotope fractionation may help in geochemical determinations of the formation
temperatures of mineral precipitates as well as characterization of parent solutions (Cao and Liu
2011 ; Levin et al. 2014; Pack and Herwartz 2014; Passey et al. 2014; Sharp et al. 2016).
So far, most work on the mass dependence of equilibrium oxygen isotope fractionation has
focused on the hydrological cycle, and the precipitation of carbonates and silica from aqueous
solution. Cao and Liu (2011) made the first detailed theoretical study, using molecules and
molecular analogues of carbonate and silicate structures in crystals and solution. This work
focused on estimating the scaling relationships between 18O/16O and 17O/16O fractionations for
different species depending on temperature, i.e.,
qa–b(T ) = ln17/16aa–b/ln18/16aa–b (1)
∆′17Oa–b (‰) = 103 (ln17/16aa–b—qref ln18/16aa–b) (2)
where qa–b is the mass-fractionation exponent for substance “a” relative to substance “b”,
aa–b is the equilibrium 17O/16O fractionation factor, 18/16aa–b is the equilibrium 18O/16O
17/16
fractionation factor, and qref is a reference exponent (typically but not always chosen to be
0.528). ∆′17Oa–b (‰) is used here to mean a differential measure of the 17O/16O ratios in the two
substances, relative to what would be predicted by the reference exponent—equivalent to what
is sometimes called ∆∆′17Oa–b—and it is not referred to a particular isotope standard material.
Bao et al. (2015) made a generalized statistical study of mass dependence in hypothetical
systems for both equilibrium and kinetic processes, showing the expected limits of mass-
dependence for chemically plausible materials. Most recently, Hayles et al. (2018) modeled
18
O/16O and qa–b for quartz, carbonates, apatite, magnetite, and water using molecular-cluster
methods, and Liu et al. (2019) modeled 18O/16O and qa–b in the sulfate minerals gypsum and
basanite. Guo and Zhou (2019) and Hill et al. (2020) calculated 18O/16O and qa–b for water,
CO2, Dissolved Inorganic Carbon (DIC), and carbonate minerals. Guo (2020) also examined
potential signatures of kinetic processes on ∆′17OCaCO3-H2O(l).
There is tantalizing evidence of seemingly unexplained ∆′17O values for some high-
temperature minerals on the Moon and on Earth (Fig. 1) at the ~10 to 20 per meg level.
In particular, some tectosilicates exhibit lower ∆′17O values than their more mafic counterparts.
A striking example is the low ∆′17O values for both terrestrial and lunar anorthosites (Fig. 1).
The low ∆′17O values for lunar anorthosites, ferroan anorthosites, and troctolites is a robust
measurement made by numerous workers spanning two decades (e.g, Wiechert et al.
2001; Young et al. 2016, Cano et al. 2020). Relationships like those in Figure 1 are seen
in other terrestrial data (e.g., Pack and Herwartz 2014). The correlation hints at a crystal
chemical control on ∆′17O, with tectosilicates generally exhibiting lower ∆′17O values than
less polymerized silicate minerals. For example, a rough correlation between ∆′17O and
bulk modulus is suggested; tectosilicates have both low ∆′17O and lower bulk moduli than
orthosilicates, and the analyses summarized in Figure 1 suggest that quartz may have lower
∆′17O values than less compressible plagioclase feldspars. Similar trends are obtained using
thermal expansivity, and the distinctive effect for tectosilicates could be related to the nature
of the 3-D framework of corner-sharing SiO4 tetrahedra.
Analyses of this type at sufficient analytical precision are not yet plentiful enough to
ascertain their veracity by interlaboratory comparisons. As an example of potential analytical
pitfalls, Pack et al. (2016) found a non-linearity in ∆′17O vs. δ′18O of ~ 1 per meg/per mil specific
to their mass spectrometer. However, correcting for this effect is not sufficient to remove the
correlation between ∆′17O and crystal chemistry. As new technologies become available (larger
gas-source mass spectrometers, better collection systems), this trend will be scrutinized. In the
meantime, we consider it prudent to investigate possible explanations for what would seem to
be unexpected disparities in ∆′17O values among high temperature minerals.
Mass Dependence of Equilibrium Oxygen Isotope Fractionation 139
Figure 1. Triple oxygen isotope ratio analyses of lunar and terrestrial samples compared with the estimates
obtained in this study. Each datum is expressed as ∆′17O relative to San Carlos olivine vs. δ′18O relative to
VSMOW. For comparison, the calculations for a-quartz, low albite, anorthite, diopside and normal spinel
from this study are normalized to the ∆′17O coexisting forsterite and the δ′18O of San Carlos olivine. Lunar
and terrestrial anorthosites have low ∆′17O values compared with basalts and other mafic lunar and terrestrial
mineral separates. Terrestrial samples include feldspar and spinel separated from a diabase from the Bushveld
complex, anorthosite from the Bushveld, Lake County plagioclase, a commonly used plagioclase standard
material, and a whole-rock and olivine sample from a Mauna Loa basalt. Also shown is an analysis of the
Oliver Quarry Quartz (OQQ), a quartz oxygen isotope standard, with a similar precision showing that it too
has low ∆′17O values relative to terrestrial mantle and to lunar and terrestrial basalt. Two mass fractionation
lines defined by the exponent for triple oxygen isotope fractionation are shown for reference. One is for the
theoretical high-temperature maximum exponent of 0.5305 and the other is for a value of 0.524. All analyses
performed at the laboratory at UCLA using CO2 laser heating assisted fluorination (e.g., Young et al. 2016).
One possibility, explored here, is that there are anomalous and significant mass-
dependent effects based on crystal chemistry, and anharmonicity in particular, even in
systems equilibrated at igneous or high-grade metamorphic temperatures. However, other
explanations have been put forward. Cano et al. (2020) suggested that variable ∆′17O values
for lunar samples are the result of a mass-independent effect carried by vapor early in the
Moon’s history, followed by exchange of this vapor reservoir with the precursors to the lunar
crust, including the anorthosites and troctolites. The suggestion is that identical ∆′17O values
for lunar and terrestrial basalts is serendipitous. Additional possibilities include pervasive,
incipient alteration of feldspars in the terrestrial environment and analytical artifacts related
to fluorination. The latter would be consistent with an historical difficulty in analyzing alkali
feldspars by laser-assisted fluorination methods.
Motivated by these observations and their potential importance, this chapter aims to
assemble a broad survey of equilibrium isotope fractionation factors and mass fractionation
exponents for geochemically significant minerals using a consistent computational method.
We attempt to estimate both 18O/16O fractionations and θ exponents with reasonable accuracy.
In order to avoid extraneous terminology, we generally adopt the notation scheme of Cao and
Liu (2011) with one minor simplification. We describe the mass fractionation exponent between
a substance a and an ideal gas of O-atoms as qa–atoms, analogous to qa–b, rather than using the
Greek letter κ. We have also chosen to place some additional emphasis on the estimation of
changes in ∆′17O resulting from equilibrium fractionation. The relationship defined by Cao
and Liu (2011) between qa–atoms, qb–atoms, and qa–b is then written:
qa–b = qa–atoms + (qa–atoms − qb–atoms) ln18/16bb/ln18/16aa–b (3)
here bb is the reduced partition function ratio for b and their relationship with ∆′17Oa–b (‰)
becomes (Dauphas and Schauble 2016):
∆′17Oa–b (‰) = 103 (qa–atoms ln18/16ba − qb–atoms ln18/16bb − qref ln18/16aa–b) (4)
∆′ O has the advantage that it remains bounded (and small) near crossovers; i.e., where
17
ln18/16aa–b approaches zero, and is more intuitively relatable to measurable deviations in mass
dependence where 18/16aa–b is close to unity. Unless otherwise specified, we assume qref = 0.528,
in keeping with most prior studies. Because of the overlap with Hayles et al. (2018) in materials
studied, we will make a particular effort to compare their results with the present models.
Most theoretical studies of equilibrium stable isotope fractionation are based on the
assumption that molecular and crystalline vibrational frequencies may be treated harmonically;
though some recent studies have investigated anharmonic effects (Liu et al. 2010; Webb and
Miller 2014; Pinella et al. 2015; Dupuis et al. 2017). Cao and Liu (2011) concluded that
anharmonicity was not likely to have a major effect on the mass dependence of equilibrium
fractionation in typical molecules.
However, not all potentially relevant types of anharmonic potential have been examined.
A secondary goal of the present study is to examine the effects of anharmonicity involving
double-well potentials on the mass dependence of fractionation. This type of potential was
not specifically tested in the studies cited above, but double-well potentials may be relevant to
understanding systems with hydrogen bonding (such as in liquid water; e.g., Tachikawa and
Shiga 2005) as well as oscillations involving Si–O–Si and Si–O–Al angles in polymerized
silicate structures (e.g., Dove et al. 1995). These examples share the common property that
there are two “stable” asymmetric configurations with associated potential minima that are
separated by a potential barrier. For Si–O–Si the asymmetry, at its simplest, is caused by the
greater stability of a crooked bond angle relative to a linear configuration with a 180º Si–O–Si
bond angle. For hydrogen bonding, asymmetry comes from the preference for hydrogen to
be nearer to one oxygen than to the other. Double-well potentials have distinct quantum-
mechanical properties, including splitting of energies and delocalization. Whether these might
influence the mass dependence of fractionation is an interesting question.
Other equilibrium phenomena that may influence observed signatures are also discussed,
including nuclear field shifts and non-Born–Oppenheimer effects. These effects are discussed
in more detail in the following section.
PREDICTING 17O/16O SIGNATURES

Estimating mass-dependent fractionations in the harmonic approximation
In this chapter, mass fractionation laws for crystals and molecules are calculated theoretically
using vibrational (phonon) spectra modeled with density functional theory. The harmonic
oscillator approximation of the vibrational contribution to the Helmholtz free energy (i.e.,
Bigeleisen and Mayer 1947) is assumed, unless otherwise noted. Isotopic effects on the free
energy are determined by comparing a series of models with isotopic substitution of 16O, 17O,
and 18O on each crystallographically distinct oxygen site. Each heavy-isotope substituted model
in the series contains an 17O or 18O substitution on a single atom per unit cell. In some cases, the
perturbation associated with a phonon wave vector leads to the loss of symmetry. In these cases,
single atom isotope substitutions are performed on multiple positions corresponding to the same
crystallographic site in order to obtain a complete sample over the unit cell. The total fractionation
factor for each isotope in the crystal is then determined by the weighted arithmetic average
of the single-atom substitutions involving that isotope. We report the result as a fractionation
factor relative to an ideal gas of oxygen atoms, analogous to the reduced partition function ratio.
Note, however, that this definition of β is slightly different than is commonly found in older
literature, where β is determined from the vibrational frequencies of a substance with complete
substitution of 18O or 17O for 16O, rather than as the weighted average of many independent single-
atom substitutions. Cao and Liu (2011) have shown that mass law estimations are inaccurate when
the rule of the mean is applied to β determined in completely substituted isotopologues.
Here we focus on models with periodic boundary conditions, for crystalline mineral phases,
constructed using density functional theory. Molecular and cluster-based models have also been
used to assess mass dependence in crystalline phases (e.g., Cao and Liu 2011; Hayles et al.
2018). An important advantage of molecular and cluster-based methods is that highly accurate
electronic structure methods are more easily applied. However, an important disadvantage is
the need to truncate such models at some finite number of atoms, introducing edge effects and
complicating comparisons with relevant experimental data such as Raman spectra. Comparing
results from both model types against each other is one way to test their relative merits.
Ultimately it appears that both approaches yield similar results for most phases. The new model
results reported here are based on the Perdew et al. (1996) gradient corrected density functional,
abbreviated PBE. PBE is a widely used functional, including for previous studies of equilibrium
isotope fractionation and related effects (e.g., Schauble et al. 2006; Meheut et al. 2007).
Pseudopotentials and Projector Augmented Wave (PAW) datasets are used in this study,
greatly reducing the computational complexity of periodic boundary condition models by
simplifying the treatment of both inner-shell and valence electrons; in recent years several
public repositories of well-designed pseudopotentials and PAW datasets have become
available, covering most elements in the periodic table. However, the application of these
repositories to isotope fractionation calculations is still in need of testing. Here, anhydrous,
carbon-free crystals are modeled primarily using ultrasoft pseudopotentials from the GBRV
library (Garrity et al. 2014; including updates through version 1.5). An important parameter in
pseudopotential-based calculations is the cutoff energy of the plane-wave basis set—in general
a higher cutoff energy allows for more accurate representation of chemical bonds, while also
increasing the computation time, and a range of cutoff energies should be tested to achieve
a balance between convergence and efficiency in the system of interest. As a general matter,
the choice of cutoff will be dependent on the pseudopotential, the electronic structure of the
material, and the desired accuracy. The GBRV pseudopotential family is notable for tolerating
relatively low cutoff energies across a broad range of elements (Garrity et al. 2014). Our testing
with GBRV pseudopotentials indicated that an energy cutoff of 40 Rydberg (544 eV) was
sufficient for the crystals studied here (Table 1). Models of LaPO4 in the monazite structure and
LuPO4 in the xenotime structure used rare-earth element PAW data sets developed by Topsakal
and Wentzcovitch (2014); these models had somewhat higher energy cutoffs of 55 Rydberg
and 45 Rydberg, respectively, as recommended in testing done by the Standard Solid State
Pseudopotential project (http://materialscloud.org/sssp). Hydrous and carbon-bearing crystals
and molecules are modeled using a mixture of pseudopotentials drawn mainly from the
GBRV and PSLibrary pseudopotential libraries, crudely following the recommendations of
the Standard Solid State Pseudopotentials project (Dal Corso 2014; Lejaeghere et al. 2016;
Prandini et al. 2018, with a higher energy cutoff of 80 Rydberg (1088 eV). All DFT calculations
were made using version 5.4 of the Quantum Espresso software package (Giannozzi et al.
2009; http://www.quantum-espresso.org).
Having chosen a functional, pseudopotential, and energy cutoff, the next important
parameter is the grid over which electronic wave vectors are sampled. A finer grid will give
a more accurate depiction of the actual electronic structure, but also slow calculations down
and limit the complexity of crystals that can be modeled. Test calculations indicate that a
range of one to 20 discrete electronic wave vectors can make a sufficiently accurate grid,
Table 1. Pseudopotentials used in the present study.

Element GBRV-based models Ref. PSLibrary/SSSP-based models Ref.
H h_pbe_v1.4.uspp.F.UPF 1 H.pbe-rrkjus_psl.0.1.UPF 3
C c_pbe_v1.2.uspp.F.UPF 1 C.pbe-n-kjpaw_psl.1.0.0.UPF 3
N n_pbe_v1.2.uspp.F.UPF 1 N.pbe.theos.UPF 4
O o_pbe_v1.2.uspp.F.UPF 1 O.pbe-n-kjpaw_psl.0.1.UPF 3
F f_pbe_v1.4.uspp.F.UPF 1
Na na_pbe_v1.5.uspp.F.UPF 1 unchanged 1
Mg mg_pbe_v1.4.uspp.F.UPF 1 unchanged 1
Al al_pbe_v1.uspp.F.UPF 1 Al.pbe-n-kjpaw_psl.1.0.0.UPF 3
Si si_pbe_v1.uspp.F.UPF 1 Si.pbe-n-rrkjus_psl.1.0.0.UPF 3
P p_pbe_v1.5.uspp.F.UPF 1 P.pbe-n-rrkjus_psl.1.0.0.UPF 3
S s_pbe_v1.4.uspp.F.UPF 1 unchanged 1
Cl cl_pbe_v1.4.uspp.F.UPF 1 Cl.pbe-n-rrkjus_psl.1.0.0.UPF 3
K k_pbe_v1.4.uspp.F.UPF 1 K.pbe-spn-rrkjus_psl.1.0.0.UPF 3
Ca ca_pbe_v1.uspp.F.UPF 1 unchanged 1
Y y_pbe_v1.4.uspp.F.UPF 1 unchanged 1
Zr zr_pbe_v1.uspp.F.UPF 1 Zr.pbe-spn-kjpaw_psl.1.0.0.UPF 3
Ba ba_pbe_v1.uspp.F.UPF 1
La La.GGA-PBE-paw-v1.0.UPF 2
Lu Lu.GGA-PBE-paw-v1.0.UPF 2
Pseudopotential References:
1. GBRV: Garrity et al. (2014); https://www.physics.rutgers.edu/gbrv/
2. Wentzcovitch et al. (2014); http://www.vlab.msi.umn.edu/resources/repaw/index.shtml
3. PSLibrary: Dal Corso et al. (2014); http://people.sissa.it/~dalcorso/pslibrary/index.html.
4. THEOS: http://materialscloud.org/sssp
depending on the size and symmetry of a particular crystal’s unit cell. A detailed discussion
of the choice and testing of wave-vector grids is beyond the scope of this review, and different
researchers commonly make different choices for the same crystal structure. In principle, any
sufficiently dense grid should yield a similar result. Because of the large range of materials
studied here, it was decided to adopt a standard procedure for choosing grids: they are chosen
so that the shortest real-space vector that belongs to the reciprocal of the wave-vector lattice
formed by each grid is approximately 35 Bohr radii (~19 Å) or longer. The idea of a vector-
length metric for evaluating grid accuracy was proposed by the developers of the ABINIT DFT
software package (www.abinit.org), though they do not specify the 35 Bohr threshold—this
choice would indeed not be appropriate other types of materials: electronically conductive
metals require finer grids, for instance. Our own parameter testing suggests that the 35 Bohr
criterion is adequate to ensure convergence of unit-cell lengths at the ~0.1% level for the
materials studied here, with reasonable consistency in calculated phonon frequencies. For
each crystal, a thermodynamically representative sample of phonon frequencies is calculated
using a somewhat coarser grid of phonon wave vectors, typically with a vector-length metric
of approximately 19 Bohr radii (~10 Å). Both phonon and electronic wave vector grids are
chosen to give good convergence in calculated isotopic fractionation and crystal structures at
relevant temperatures. Similar tests have been described in previous studies (e.g., Elcombe and
Hulston 1975; Schauble et al. 2006; Meheut et al. 2007, and others).
Phonon frequencies—the crystal analogue of molecular vibrational frequencies—are

typically underestimated by several percent by the Perdew et al. (1996) functional. This is
important because the calculation of equilibrium stable isotope fractionation depends on the
differences in frequencies between isotopically substituted materials (Biegeleisen and Mayer
1947). In order to partially correct for this systematic error, a frequency scale factor of 1.043 has
been applied in all fractionation calculations; the scale factor is determined by a proportional
regression of calculated phonon frequencies correlated with frequencies measured with Raman,
infrared, and neutron spectroscopy (Fig. 2). This scale factor is similar to previous determinations
for PBE-based models of periodic models of silicates, oxides and carbonates (e.g., Schauble et
al. 2006, 2011; Meheut and Schauble 2014; Widanagamage et al. 2014). It should be noted that
scale factors determined by general-purpose benchmark studies of the vibrational frequencies of
small molecules (e.g., Merrick et al. 2007 and references therein) are typically closer to unity.
However, in these studies larger scale factors (similar to ours) are found for low-frequency
vibrational modes (< ~1000cm–1, depending on the study). This lower frequency range is much
more common in the crystals studied here than in molecular data sets, and frequencies less than
1200 cm–1 dominate our overall fit. Larger low-frequency scale factors appear to be typical
for gradient-corrected functionals (such as PBE) as well as for hybrid methods that include a
gradient-corrected functional component (such as B3LYP; Becke 1993); e.g., Merrick et al.
(2007). Further details of the scale factor determination are presented in the following section.
4000
Measured = 1.043 x Model
R² = 0.999
Measured frequency (cm –1)
3000
2000
1000
0
0 1000 2000 3000 4000
Model frequency (cm –1)
Figure 2. Vibrational frequency correlations with Raman, infrared, and inelastic neutron scattering mea-
surements for estimating the frequency scale factor. In general, OH stretching and bending frequencies
are not included in the correlation because of the larger anharmonic effects expected for these modes.
Frequencies in H2O, CO2, and CO vapor are compared with empirically estimated harmonic frequencies,
all other correlations are with measured fundamental frequencies. References for measured frequencies:
a-quartz – Castex and Maddon (1995); albite – Aliatis et al.(2015); microcline – Zhang et al. (1996);
diopside – Prencipe et al. (2012); jadeite – Prencipe et al. (2014); forsterite – Kolesov and Geiger (2004);
zircon – Chaplot et al. (2006); grossular – Maschio et al. (2014); spinel – Chopelas and Hofmeister (1991);
lizardite – Hofmeister and Bowey (2006); xenotime-Y – Giarola et al. (2011); xenotime-Lu – Mittal et
al. (2007); monazite-La – Begun et al. (1981) and Geisler et al. (2016); fluorapatite – Leroy et al. (2000);
anhydrite – Berenblut et al. (1973) and Iishi (1979); barite – Dawson et al. (1977) and Jayasooriya et al.
(1996); gypsum – Iishi (1979); NaClO4 – Lutz et al. (1979) and Toupry-Krauzman et al. (1978); calcite –
Pavese et al. (1992); aragonite – Carteret et al. (2013); dolomite – Pilati et al. (1998) and Gillet et al. (1993);
magnesite – Rutt and Nicola (1974) and Hellwege et al. (1970); nahcolite – Bertoluzza et al. (1981);
nitratine – Yamamoto et al. (1976), Brehat and Wyncke (1985), Eckhardt et al. (1970), and Lefebvre et al.
(1980); H2O vapor – Benedict et al. (1956); CO2 – Zuñiga et al. (2001); CO – Mantz and Maillard (1975).
During model testing, it became clear that pseudopotentials from the GBRV library are
reliable, fast, and accurate for calculations on anhydrous, carbon-free crystals. However,
hydrous materials and molecules with short, covalent bonds are more problematic—this group
includes water vapor, carbon monoxide, and carbon dioxide. GBRV models overestimate
O–H stretching frequencies in water vapor, the water vapor dimer (H2O)2, silicic acid vapor
(H4SiO4) and disilicic acid vapor (H6Si2O7) by ~90–100 cm–1, and the asymmetric C–O
stretching frequencies in carbon dioxide and carbon monoxide by ~40–50 cm–1, relative to
benchmark calculations with the PBE functional and large atom-centered Gaussian basis sets
(e.g, aug-cc-pVTZ). Large Gaussian basis sets such as these should approach the greatest
possible accuracy achievable with the PBE functional, when applied to molecules. When
tested in periodic boundary condition calculations of hydroxyl-bearing and hydrate minerals,
we find that PBE-GBRV models show offsets of 10’s of cm–1 relative to the alternative
pseudopotentials, suggesting that the artifacts are not limited to molecules modeled in vacuo.
In contrast, other pseudopotential libraries show much closer matches to the Gaussian basis
set results, suggesting that this is a problem isolated to the O- and H- (and to some extent C-)
pseudopotentials in the GBRV library, which were designed to optimize computational speed
as well as accuracy. We have generated alternative models using a mixture of pseudopotentials
drawn mainly from the GBRV and PSLibrary pseudopotential libraries, because they match
benchmark calculations for molecules more closely (rms errors of < 10 cm–1 for CO, CO2,
H2O and the H2O dimer, and ~13 cm–1 for silicic and disilicic acid vapor). Both GBRV and
alternative pseudopotentials yield very similar phonon frequencies in anhydrous, carbon-
free crystalline phases, with indistinguishable scale factors and residual root-mean-square
(rms) errors after scaling. Models using the alternative pseudopotential set are considerably
more computationally demanding than GBRV-based models, mainly because they require a
higher energy cutoff. This limits their practical application to phases with relatively simple
and compact unit cells. In general, we observe that the choice of GBRV vs. alternative
pseudopotentials has a negligible effect on both fractionation factors and mass law exponents
for anhydrous crystals. For hydrous species and molecules, the choice of pseudopotentials
has only a modest effect on the mass fractionation exponent qa–atoms, but often substantial
effects on estimated 18O/16O fractionation factors—up to several per mil at room temperature.
We therefore report results using the alternative pseudopotentials for molecules, carbonates,
and hydrous crystals, but GBRV results for anhydrous, carbon-free phases, even though we
originally intended to focus on GBRV results for all materials studied.
Molecules are treated in much the same way as crystals, with the exception of H2O, CO2,
and CO vapor. Harmonic frequencies and anharmonicity parameters have been determined for
each of these molecules on the basis of detailed spectroscopic measurements, making more
sophisticated (and presumably more accurate) fractionation factor estimates possible (e.g.,
Urey 1947; Richet et al. 1977; Rosenbaum 1997). Although there is disagreement about the
best compromise between practicality and accuracy in incorporating anharmonic effects into
isotope fractionation factor calculations, Urey (1947), Richet et al. (1977), Liu et al. (2010), and
others have confirmed that the anharmonic effect on vibrational zero-point energy is typically
the most significant correction. Therefore, we determined molecule-specific scale factors
for each of these molecules by fitting model frequencies to previously determined harmonic
frequencies for the dominant isotopologue, and applied a correction for anharmonicity in the
calculation of ln18/16β by adjusting the zero-point energy term in the equation for the reduced
partition function ratio (Urey 1947; Rosenbaum 1997). Anharmonic zero-point energies are
estimated with the perturbational VPT2 method (Barone 2005) using the PBE functional and
aug-cc-pVQZ basis sets ( Kendall et al. 1992), as implemented in the Gaussian 09 software
package (Frisch et al. 2013). An alternative anharmonic method, CC-VSCF in the quartic
force field approximation (Yagi et al. 2004) was also tested. Both methods give similar results
for CO and CO2, but CC-VSCF appears to somewhat underestimate the anhmarmonicity of
H2O vibrations, leading to ~0.7‰ higher estimated ln18/16β at 298 K. VPT2 anharmonicity
parameters for H2O, in contrast, come close to spectroscopic determinations (Benedict et
al. 1956, and subsequent studies). We find substantially improved agreement with previous
estimates of 18O/16O fractionation factors and ln18/16β using this correction method.
In all cases, qa–atoms is calculated using the harmonic oscillator assumption (Cao and Liu
2011). In agreement with the earlier study, we find limited effects of anharmonicity on the
mass dependence exponent of fractionation. Compared to the anharmonicity-corrected zero-
point energy method described above, a purely harmonic determination of θCO2–atoms is within
~2 × 10–5 from 273–1573 K; for H2O(v) the difference is less than 1 × 10–5 below 1273 K.
The simplified anharmonic correction procedure is not thermodynamically self-consistent
because rotational and excited vibrational state energies do not arise from the same potential
as the ground state (zero point) energy, and this introduces a (likely very small) artifact in θ
that we prefer to avoid. For the sake of consistency, a harmonic H2O(v) model using the crystal
frequency scale factor is used for estimating fractionations relative to ice and the hydration
water of gypsum. The harmonic model predicts 2–3‰ higher 18O/16O than the anharmonicity-
corrected model over the temperature range 273–373 K (0–100 ºC).
Our model for liquid and supercritical water is based on the model for water vapor, modified
by adding 103 ln18/16aliquid–vapor from Rosenbaum (1997) to 103 ln18/16bwater vapor. qliquid water–atoms is
estimated three different ways:
1. by applying qliquid water–water vapor = 0.529, as measured by Barkan and Luz (2005);
2. by taking the lower of qliquid water–atoms as calculated by (1) and θ water vapor–atoms, in the
expectation that condensed water will never have θ below the vapor, and
3. by taking a simple average of qwater vapor–atoms and qice Ih–atoms, based on the observation
that predicted ice–vapor 18O/16O fractionation at 273 K is roughly twice as large as
the observed liquid–vapor fractionation, and fractionations for both phases relative
to water vapor are driven by similar H-bonding interactions.
Methods (1) and (3) are crudely analogous to the “semi-empirical” and “ab initio”
approaches taken by Hayles et al. (2018) in that they incorporate empirical qliquid water–water vapor
estimates in the first case, and ab initio model results for clusters of water molecules in the
second case. In practice, the choice of estimation method does not matter very much at low to
moderate temperature—the various phases of H2O stay close to the limiting harmonic value of
qwater–atoms ≈ 0.530 regardless of the details of the theoretical treatment, and the three methods
outlined above agree to within 1 × 10–4 from 263–673 K (–10 ºC to 400 ºC). Above 400 ºC,
a naïve application of method (1) (i.e., assuming a supercritical fluid–vapor exponent of 0.529
even though the temperatures are far above those studied by Barkan and Luz) yields rapidly
increasing qliquid water–atoms, eventually reaching unrealistically high values. For this reason,
results from methods (2) or (3) are probably more reliable for higher temperatures; (2) and (3)
are also in good agreement with each other in this temperature range. Note that an alternative
parameterization of the liquid–vapor fractionation parameterization has been proposed by
Horita and Wesolowski (1994). Their fitted fractionation curve is generally quite close to that
of Rosenbaum (1997) up to the critical temperature of water, and the choice of one model over
the other has little effect on our results at most temperatures.
At this time, it is simply not possible to routinely determine anharmonic parameters for
crystals with significant structural complexity. However, the practical impact of this restriction
may be limited because the vibrational thermodynamics of H2O are more strongly affected
by anharmonicity than typical materials, and because the general scale factor determination
for crystals is fitted to fundamental frequencies. Fundamental frequencies will typically
deviate from harmonic frequencies in the same direction as the anharmonic zero-point energy
correction, albeit by a slightly larger magnitude, and so the crystal models will already tend to
overestimate this component of the anharmonic effect.
Crystal structures calculated with the PBE functional and pseudopotentials are compared
with structures measured by X-ray and neutron-diffraction in Table S1 in the Supplemental
Information. In general, unit cell volumes are overestimated by approximately 3%, as is typical
for the PBE functional. Calculated phonon frequencies are compared with Raman, infrared,
and inelastic neutron-scattering measurements in Table S2 and Figure 2. In general, model and
measured frequencies are in excellent agreement after applying the fitted scale factor, with
a root-mean-square misfit of 13 cm–1, compared to a root-mean-square frequency of about
689 cm–1, indicating an average scatter of roughly 2%. We have included the molecules H2O,
CO2 and CO in this scale factor fit, so as to sample as wide a range of bond environments
as possible, but have correlated harmonic frequencies rather than fundamentals for these
molecules. In practice, these three molecules represent relatively few modes, so their impact
on the calculated scale factor is modest (1.043 with molecules, vs. 1.045 without). It should
be noted that almost all correlated phonons correspond to a near-zero wave vector, and that
other parts of the Brilluoin Zone might conceivably show larger misfits or different scaling.
However, there are relatively few studies that probe non-zero wave vectors, particularly in the
higher-frequency range that mainly controls O-isotope fractionation ( Kieffer 1982). Overall,
vibrational frequency correlations suggest that the models will be able to make useful estimates
of mass-dependent equilibrium isotope fractionations and mass dependence exponents. Likely
uncertainties in these estimates arising will be discussed in more detail below.
Effects beyond the harmonic approximation
Anharmonicity in a double-well potential. To extend earlier work on the effects of
anharmonicity on the mass dependence of fractionation (e.g., Cao and Liu 2011), a highly
simplified model system consisting of a one-dimensional quantum oscillator in a double-well
potential is investigated. The potential is assumed to take the form2of a harmonic potential that
is perturbed by a central Gaussian peak (V = V0 + kfc x2/2 + b e–cx ) where kfc is the harmonic
spring constant, b determines the height of the Gaussian perturbation, and c controls the width
of the Gaussian perturbation (c here should not be confused with the speed of light). This is
only one of many possible types of double-well potentials, but it has two important advantages
for the present purpose:
1. the ground state and many low-lying excited state energies can be calculated quickly
with high precision, and
2. the potential and its solutions converge to the well-understood harmonic oscillator in
both the limit of small perturbations (b/kfc → 0, and/or c → ∞) and in the limit of
high energy: En → hnharmonic (n + 1/2) as n→ ∞, where nharmonic = (kfc/m)1/2/(2π) is the
characteristic vibrational frequency of the harmonic part of the potential.
Solution of the Schrödinger equation for this potential is simplified by substituting parameters
scaled to the characteristic frequency nharmonic, i.e. B = b/(hnharmonic) for the height of the
Gaussian above the minimum of the harmonic potential, γ = ch/(mnharmonic) for the sharpness
of the Gaussian, and y = x(mnnharmonic / h)1/2 for the positional coordinate. In the present
work, the Schrödinger equation for this potential is solved via the matrix method of Earl
(2008), using Scilab (v. 6.0.0). Accurate eigenvalues corresponding to energy states up
to n = 73 are obtained in roughly one CPU-minute per potential on a laptop computer by
solving separate 49 × 49 matrices for even and odd quantum numbers. Such matrices actually
yield estimates of energies for states up to n = 97, but the highest energy states are poorly
converged because of the truncation of higher-order terms that would require larger matrices
(i.e., the energies typically varied by ~10–5 hnharmonic or more in test calculations with B =
1–5 between 49 × 49, 48 × 48, and 47 × 47 matrix solutions). Representative examples of
a.
g=1 B = 10
B=5
B=2
12
B=1
10 B = 0.5
B=0
Potential Energy
8
6
4
2
0
-4 -2 0 2 4
Distance
b. B=5 g=5
g=2
10 g=1
g = 0.5
g = 0.2
8
g = 0.1
Potential Energy
0
-4 -2 0 2 4
Distance
c.
B = 10, g = 1
12
10
8
Energy
0
-4 -2 0 2 4
Distance
Figure 3. Examples of double-well potentials consisting of a central Gaussian peak within a harmonic
potential. The potential and distance units are scaled with respect to the quanta of the harmonic component
of the potential, as described in Earl (2008). a) Variation in peak height (B). b) Variation in the sharpness
of the central Gaussian (γ). c) Quantum energy levels in an example potential. Solid horizontal lines
represent the lowest even quantum numbers n = 0, 2, 4, 6, and 8; dashed horizontal lines represent odd
quantum numbers n = 1, 3, 5, 7, and 9.
potentials and calculated vibrational energies corresponding to different values of B, γ, and

n are shown in Figure 3. Attempts to construct larger matrices generated software errors,
and it was not judged necessary to try to pursue more complex calculations. Beyond n = 73,
energies are approximated by assuming a power-law convergence towards the harmonic limit,
i.e., En − En − 2 − 2hnharmonic ∝ n–3/2 (Fig. S1). Separate convergence power-laws were used for
even and odd n. To explore the effects of the size and width of the central potential peak, reduced
partition function ratios and mass fractionation exponents are calculated on a grid of varying
Β, γ, and hnharmonic/kBT, where kB is Boltzmann’s constant and T is the absolute temperature.
Numerical precision was generally worst at the highest temperatures (hnharmonic/kBT ≪ 1), and
with very localized Gaussian perturbations (γ ≫ 1). However, the impact of these imprecisions
is likely rather limited for two reasons.
1. Fractionations become very small at high temperatures and all mass fractionation
laws should approach the harmonic high-T limit at these conditions, providing an
analytic benchmark for determining numerical accuracy.
2. Highly localized Gaussians barriers are probably not physically realistic, as they
are much sharper than the minima they separate. For example, the inversion mode
in ammonia (NH3), one of the simplest and best studied examples of a double-well
potential, is fit well with γΒ ≈ 0.9–2.2 (Lin et al. 2007: γ ≈ 0.05, B ≈ 17; Earl 2008:
γ ≈ 0.3, B ≈ 8), corresponding to a broad, large Gaussian perturbation.
Reconnaissance vibrational frequency calculations on the gas-phase H6Si2O7 molecule in
its ground-state bent conformation, compared to a strained configuration with the Si–O–Si angle
fixed at 180º, suggest that γB for the Si–O–Si bending coordinate is also of order unity, with
B ≫ 1 and γ ≪ 1. In particular, the similar absolute value of the force constants associated with
the imaginary modes in “straight” Si–O–Si (nbend ≈ 50i cm–1), relative to the force constants of
Si–O–Si bending in the relaxed, bent molecule (nbend ≈ 60–100 cm–1), constrains γB ≤ ~2, while
the large energy barrier (> 1000 cm–1) between the relaxed and straight configurations indicates
B ≥ ~10, and thus γ ≤ ~0.2. For the sake of simplicity, the calculations reported here assume
nharmonic(16O) = 200 cm–1. This corresponds roughly to the frequency of the soft-mode Si–O–Si
vibrational frequency in α-quartz measured by Raman spectroscopy (e.g., Castex and Maddon
1995) but it should be noted that the characteristic frequency of the energy minima in the perturbed
potential will not be the same as the harmonic frequency of the unperturbed potential—they will
also depend on γB. Harmonic frequencies of isotopically-substituted oscillators in the model are
assumed to scale in proportion to the reciprocal of the square root of the oxygen isotopic mass,
as is observed for the soft-mode Si–O–Si vibration in completely 18O-substituted quartz (Sato
and McMillan 1987). This model is crude in important respects, for instance it treats the bending
vibrational mode of one bridging oxygen as completely isolated from all other vibrations in the
structure. However, this model does explore a scenario of anharmonic behavior that is likely to
affect many important silicate and aluminosilicate minerals to a greater or lesser degree.
The nuclear field shift effect. The nuclear field shift effect has long been studied in high-
resolution atomic spectroscopy (e.g., King 1984). It results from interactions between electrons
and the finite volume and sometimes distorted shapes of atomic nuclei. All else equal, an electron
inside the volume occupied by the nuclear charge “feels” a weaker Coulomb attraction than it
would at the same distance from an infinitesimal point nucleus, and the resulting binding energy
and orbital shape change slightly. Typically, nuclei with more neutrons (i.e., heavier isotopes)
take up a greater volume of space, and overlap with a larger proportion of the electron cloud, but
the correlation between mass and volume is irregular for many elements, including Oxygen (e.g.,
Miska et al. 1979). The nuclear field shift has been proposed to explain non-mass dependent
isotope effects in some elements with high atomic number (e.g., Bigeleisen 1996; Nomura et
al. 1996; Schauble 2007, for U and Tl, respectively) as well as a number of moderate-atomic
number elements (e.g., Fujii et al. 2009). Theoretical calculations generally indicate that this
effect is much smaller for low atomic number elements such as sulfur and (presumably) oxygen
(Knyaezev and Myasoedov 2001; Schauble et al. 2007), but D17O measurements at the per meg
level of precision suggest that quantitative estimates are now needed to evaluate hypotheses
about the relationship between the Earth and Moon (e.g., Young et al. 2016; Herwartz et al. 2017)
and other subtle oxygen isotope signatures. We estimated the volume component of the nuclear
field shift effect on oxygen isotope fractionation in silicates using the method of Schauble
(2013). In this method, the effect of varying the nuclear charge radii of oxygen atoms on the
electronic energies of a calibration set of small molecules is calculated using relativistic all-
electron quantum mechanics. These results are compared to electron densities at oxygen nuclei
in the same molecules determined by Density Functional Theory models based on the Projector
Augmented Wave (DFT-PAW) method. The calibrated energy/electron density relationship
can then be used to estimate nuclear volume isotope effects with DFT-PAW models of crystal
structures that are too complex to be directly modeled by the relativistic all-electron method.
Relativistic all-electron models use a high-accuracy electronic structure method, the
coupled-cluster technique CCSD(T), including single and double excitations and a perturbative
estimate of triple excitations, combined with the cc-pVTZ family of basis sets (Dunning 1989).
Test calculations show little difference between field shift effects determined with this basis set,
relative to the larger (and likely more accurate) aug-cc-pVTZ and cc-pVQZ basis sets, indicating
that cc-pVTZ is accurate enough for the present purpose. In fact, the CCSD(T) energy differences
are closely approximated by computationally simpler Dirac–Fock models, as was found previously
in studies of higher atomic number elements (Cd, Sn, and Hg; Schauble 2013). Relativistic all-
electron calculations were performed using the DIRAC15 code (Jensen et al. 2015).
There appears to be substantial uncertainty in experimental determinations of nuclear
charge radii of the stable oxygen isotopes. For the present purpose, 18O/16O and 17O/16O field
shift effects on relativistic CCSD(T) energies are calculated using the radii tabulated by Angeli
and Marinova (2013) (2.6991 fm, 2.6932 fm, and 2.7726 fm for 16O, 17O, and 18O respectively).
Alternative radii determined by Singhal et al. (1970) and Miska et al. (1979) have also been
considered. The Miska et al. (1979) results closely agree with Angeli and Marinova (2013),
but Singhal et al. (1970) find a ~25–30% smaller difference in mean squared radii between 18O
and 16O. However, all of these sets of radii yield very similar deviations from typical mass-
fractionation exponents (i.e., disagreements in radii largely cancel so that any calculated D17O
is less sensitive to the choice of a radius reference).
DFT-PAW models are constructed mainly using version 1.0 of the JTH set of PAW data
(Jollet et al. 2014), which is a publicly available library covering most of the periodic table
that is similar in many respects to the GBRV and PSLibrary repositories described earlier.
The JTH set has the advantage of being easier to set up for automatic computation of electron
densities at the nuclei of oxygen atoms. A slightly modified PAW data set for O is used, with
a shortened cut-off radius of 1.3 a0 (0.69Å) and a finite-radius nucleus, as well as short cut-off
radii PAW datasets for hydrogen and carbon (1.0 a0 ≈ 0.53 Å) from an older (pre-2013) public
library associated with the AtomPAW project (see http://users.wfu.edu/natalie/papers/pwpaw/
periodictable/oldperiodictable.html; Holzwarth et al. 2001). These short-cutoff PAW datasets
are chosen to provide realistic results for molecular species with short bonds, including H2O,
CO, and CO2, because PAW calculations generally work best when interatomic distances are
at least as large as the sum of PAW cutoff radii. All DFT-PAW calculations used plane-wave
basis sets with a kinetic energy cutoff of 60 Hartree (1633 eV). The models also all have
periodic boundary conditions: crystal structures are modeled in experimentally determined
periodic unit cells, and molecules are isolated individually in cubic or rectangular-prismatic
cells that are at least 20 Bohr radii (10.3 Å) on a side to approximate the in vacuo condition of
the corresponding relativisitic all-electron models.
Deviations from the Born–Oppenheimer appromixation of electronic structure. Typically,

electronic structure models of molecules and crystals adopt the Born–Oppenheimer approximation,
which assumes that the motions of electrons and nuclei can be treated separately because nuclei
are much more massive (and slower-moving) than electrons (Born and Oppenheimer 1927).
A consequence of this assumption is that electronic energies are insensitive to the masses of
isotopes, and equilibrium mass-dependent isotope effects arise only from the quantization of
nuclear motion. However, in more accurate quantum-mechanical treatments nuclear masses do
influence the coupled nuclear/electronic structure and energy in a way that is not captured by the
Born–Oppenheimer approximation (e.g., Kleinman and Wolfsberg 1973; Bigeleisen 1996).
Zhang and Liu (2018) made a comprehensive examination of effects beyond the harmonic,
Born–Oppenheimer approximation on equilibrium isotope fractionation, including an adaptation
of the Born & Huang (1956) perturbative Diagonal Born–Oppenheimer Correction (DBOC) to
account for nuclear/electronic coupling. They find that non-Born–Oppenheimer corrections are
important for H/D substitution in molecules, and can even be significant (if much smaller) for
isotope substitutions of heavier elements such as 13C/12C and 18O/16O. For instance, in the 18O/16O
fractionation between CO2 and H2O(v), they find a DBOC correction of ~1‰ at 300 K, very
similar in magnitude to the ~1‰ anharmonic ZPE correction described in the preceding section.
Although Zhang and Liu (2018) do not directly address possible ∆′17O signatures arising from
Born–Oppenheimer corrections, subsequent work (Cao et al. 2019; personal communication)
indicates that the effects are very small (~1 ppm) at igneous temperatures, but may be more
significant (several ppm) at ambient conditions. We have not included DBOC corrections into the
results tabulated in this chapter, and this is an area where uncertainty remains.
How important are effects beyond the harmonic approximation?
Nuclear field shift fractionation. Calculated nuclear field shift fractionations are shown
in Table 2. Because the nuclear charge radius of 17O is similar to (or even smaller than) the
charge radius of 16O, while the radius of 18O is larger, field shift effects on ∆′17O are anti-
correlated with (and somewhat smaller than) field shift 18O/16O fractionation. The range of
estimated field shift fractionations is quite small, among O2− species the largest fractionations
are approximately 2 per meg in 18/16α and 1 per meg in ∆′17O relative to a reference 0.528
mass fractionation exponent at 25 ºC. The field shift will tend to increase 18O/16O in oxides,
particularly silicates, because these species have the smallest electron densities at the oxygen
nucleus. Higher oxidation states, including O– in hydrogen peroxide, O0 in dioxygen, and
O2+ in F2O, are predicted to be progressively depleted in 18O, with positive field-shift induced
∆′17O. Nuclear field shift contributions to apparent ∆′17O in inter-mineral fractionation among
silicates (or for silicates relative to H2O) are predicted to be very small (< 1 per meg) at all
temperatures above 0 ºC, relative to a θ = 0.528 reference exponent. F2O, containing oxygen
in the 2+ oxidation state, is predicted to somewhat more strongly fractionated by the field shift
effect relative to oxide and silicate species. These fractionations are expected to scale as 1/T,
and thus to be only ~1/3 to 1/6 as large (~0.1 per meg or less) in igneous and metamorphic
rocks at near-solidus temperatures as they are at room temperature. Overall, field shift effects
on both 18O/16O and 17O/16O fractionation appear to be so small that they are not considered
further in this chapter. However, they may need to be taken into account if the accuracy of
future analytical studies improves significantly beyond ±1 per meg level.
Anharmonic double-well potential. Representative results for the reduced partition
function ratio 103 ln18/16β and the exponent qa–atoms are given in Table 3. In general, this type
of potential does not appear to generate large deviations from typical mass-dependence for
plausible values of the potential parameters. Models with B = 10–20 and γ = 0.1, which may
roughly approximate Si–O–Si bending modes in polymerized silicates such as H6Si2O7(v) and
Table 2. Calculated 18O/16O and 17O/16O nuclear field shift fractionations, based on DFT-PAW
electronic structure models calibrated against all-electron relativistic CCSD(T)/cc-pVTZ estimates
of the nuclear volume effect on electronic energies.
Nuclear field shift effect in oxygen 106 ln18/16a(X−H2O) D′17O vs. 0.528 (per meg)
Oxidation |Ψ(0)|2
298 K 1000 K 298 K 1000 K
State (e–/a03)
Molecules
OF2 2 16.9 −4.1 −1.2 2.5 0.7
O2 0 16.2 −2.6 −0.8 1.6 0.5
H2O −2 15.1 − − − −
CO2 −2 15.7 −1.3 −0.4 0.8 0.2
CO −2 15.9 −1.9 −0.6 1.2 0.3
Crystal Structures
SiO2
−2 14.7 0.8 0.2 −0.5 −0.2
a-quartz
KAlSi3O8
−2 14.8 0.7 0.2 −0.4 −0.1
microcline
CaAl2Si2O8
−2 14.8 0.6 0.2 −0.4 −0.1
anorthite
CaMgSi2O6
−2 14.9 0.4 0.1 −0.3 −0.1
diopside
Mg2SiO4
−2 14.9 0.3 0.1 −0.2 −0.1
forsterite
H4SiO4 −2 14.9 0.3 0.1 −0.2 −0.1
CaCO3
−2 15.1 0.0 0.0 0.0 0.0
calcite
Table 3. Mass dependence of oxygen isotope fractionation in a double-well potential in the form of a
harmonic potential with a central Gaussian peak. Each entry gives D′17O (in per meg) resulting from
a single vibrational mode, consisting only of movement of an oxygen atom, relative to a q = 0.528
reference exponent. The shape of the central peak is described by its height (B) and sharpness (g).
B = 0 corresponds to an unperturbed harmonic potential. For a harmonic potential with a characteristic
frequency of 200 cm–1, hnharm/kT = 0.2, 0.5, and 1 corresponds to T = 288 K, 576 K, and 1439 K,
respectively. B ≈ 10, g ≈ 0.1 may be a reasonable approximation for the Si–O–Si bending mode in
H6Si2O7. B ≈ 5, g≈1 (italics) may not be chemically reasonable for normal chemical bonding systems.
hnharmonic/kT
B g 1 0.5 0.2
0
10.5 2.8 0.5
(harmonic)
1 1 6.7 2.5 0.5
2 1 0.9 2.3 0.5
5 1 −2.8 3.0 0.6
5 0.1 5.3 1.8 0.4
10 0.1 6.3 2.1 0.4
20 0.1 19.1 5.7 0.7
room-temperature α-quartz, show qa–atoms between 0.529 and 0.531 at temperatures above
273 K, and are within 0.001 of the pure harmonic 200 cm–1 oscillator. Because this is above the
reference exponent of 0.528, the effect of these vibrational modes will be to slightly increase
∆′17O in polymerized species, relative to other substances. 103 ln18/16β is more variable within
this range of potential parameters, from ~4‰ at 288 K for B = 10 to ~11‰ at B = 20. This can
be rationalized as reflecting the increase in the curvature of the potential at the energy minima
as B increases, leading to a larger effective force constant, and would be accompanied by a
larger measured fundamental vibrational frequency. Relative to a 0.528 reference exponent,
103 ln18/16β and 103 ln18/16β for the B = 10, γ = 0.1 potential would yield ∆′17O = +6 per meg at
288 K and +0.4 per meg at 1439 K; the B = 20, γ = 0.1 potential would yield ∆′17O = +19 per meg
at 288 K and +0.7 per meg at 1439 K. For comparison, a pure harmonic oscillator at 200 cm–1
yields ∆′17O = +10 per meg at 288 K, and +0.5 per meg at 1439 K. In fact, no combination of
parameters with Bγ ≤ 2, B ≤ 20 appears to generate ∆′17O signatures ≥ 10 per mg at ambient
temperatures (~300 K), or ≥ 1–2 per meg at temperatures above 1000 K. Potentials with larger
Bγ can (Fig. 4), especially at low temperatures and at large B, but this parameter range may
not be relevant to normal natural materials. Note that the parameter sampling used to generate
this contour plot is rather coarse and certainly is not exhaustive, so interpolated variations
should be taken as a rough guide. It should also be noted that a perturbed harmonic potential
is only one of many possible, chemically plausible double-well and multiple-well potentials.
Other realistic variants could conceivably generate larger signatures at temperatures of interest.
In summary, these reconnaissance calculations appear to show that at least one form of
double-well anharmonicity can generate ∆′17O signatures of order ±10 per meg at near-ambient
or hydrothermal temperatures, and thus may be useful to consider as a correction to harmonic
estimates in future high-precision theoretical studies of polymerized silicates, hydrogen bonded
materials such as liquid water, ice, and organic matter. However, it is unlikely to explain the
negative ∆′17O observed in polymerized silicate minerals fractionated at high temperatures
(Sharp et al. 2016; Young et al. 2016) for two reasons. First, the effect appears to be too small
(< 1 per meg at T > 1200 K), and second, the calculated effects have the opposite sign to observed
signatures (> 0 for the model potential, vs. < 0 in measured granite and feldspars). Anharmonic
effects on the mass fractionation exponent will be ignored in the discussion that follows, excepting
the anharmonic zero-point energy correction previously described for H2O(v), CO2, and CO.
∆′17O signatures of equilibrium
Based on the discussion and findings above, it appears likely that mass-dependent effects
within the Born–Oppenheimer approximation are usually the predominant driver of ∆′17O
at equilibrium, including at high (metamorphic and igneous) temperatures, and that even
simplified calculations based on these assumptions can generate reasonable estimates of
signatures found in natural systems. Newly calculated mass dependent 103 ln18/16β and qa–atoms
results from the present work are reported in Table 4. Polynomials in 1/T have been fit to each,
in order to facilitate interpolation (Tables 5 and 6) for temperatures above 243.15 K. Fits for
103 ln18/16β are accurate to within 0.03‰, and fits for qa–atoms are accurate within 1 × 10–4 over
this temperature range. Model properties for liquid and supercritical water are also tabulated
(273.15–647 K for liquid water and 647+ K for supercritical water), by adding 103 ln18/16α for
liquid–vapor or supercritical fluid–vapor (Horita and Wesolowski 1994; Rosenbaum 1997) to
the present H2O vapor results. qliquid water–atoms is estimated as described above.
Uncertainties in estimated mass-dependent fractionations. Leaving aside anharmonicity,
which has been addressed in a special case above, more generally by Richet et al. (1977), Schauble
et al. (2006) Méheut et al. (2007), and others for the calculation of stable isotope fractionation
factors, and in greater detail by Cao and Liu (2011) for estimating qa–b exponents, the main
contributors to uncertainty in estimated 18O/16O and qa–b are likely to be errors in calculated
Figure 4. Contour plots showing variation in the calculated ∆′17O (in per meg), relative to a θ=0.528 refer-
ence exponent, for an anharmonic oscillator in a double-well potential, versus the height (B) and sharp-
ness (γ) of the central Gaussian. Temperatures are given for each panel, based on the assumption that the
harmonic component of the potential has a characteristic frequency of 200 cm–1. The corresponding ∆′17O
for the unperturbed harmonic potential is shown with a white line labelled “B = 0” in each color scalebar.
The dashed black line indicates Bγ = 2; it is likely that silicates and other molecules with double-well po-
tentials are best fit by values of B and γ that fall below this line.
vibrational frequencies, errors in the frequency shifts upon isotope substitution, errors from
incomplete sampling of the phonon density of states, and errors stemming from the interaction of
18
O or 17O substitutions in adjacent unit cells. We will attempt to address each of these sources of
error in turn, though it is still not possible to provide a rigorous quantitative analysis.
Scale factor effects. Application of a frequency scale factor is intended to correct for the
typical systematic error in the PBE gradient-corrected approximation of the density functional.
This work and many previous studies have found consistent underestimation of both measured
and harmonic frequencies, by ~3–6% in materials most closely resembling the crystals
studied here. However, there are still potentially significant mismatches between measured
and calculated frequencies even when the scale factor is applied. One approach to estimating
this source of uncertainty is to compare our results with a general fitted PBE scale factor
Table 4. Calculated 1000 ln18/16b, qa–atoms, and D′17O (vs. H2O vapor) for crystals and molecules at 0 ºC (273.15 K), 25ºC (298.15 K), and 100 ºC (373.15 K). 154
Anharmonic zero-point energy corrections are included in estimated ln18/16b for H2O vapor (anharm.), CO, and CO2. qa–atoms is based on harmonic oscillator
models for all species. Liquid water results are based on both anharmonic and harmonic H2O vapor models, and assume the liquid–vapor fractionation of
Rosenbaum (1997; R-model) and Horita and Wesolowski (1994; HW-model). qa–atoms for the liquid water results shown here are calculated as the average of
H2O vapor and ice-Ih.
1000 ln18/16b qa–atoms D′17O vs. H2O vapor (per meg)

Substance 0 ºC 25 ºC 100 ºC 0 ºC 25 ºC 100 ºC 0 ºC 25 ºC 100 ºC
a-quartz 124.83 108.70 75.20 0.527906 0.528118 0.528653 −149.6 −112.4 −48.4
Albite (low) 117.11 101.69 69.85 0.528103 0.528315 0.528835 −125.9 −93.2 −39.2
Microcline (low) 116.71 101.36 69.67 0.528091 0.528304 0.528827 −127.2 −94.4 −39.9
Anorthite (ordered) 109.91 95.18 64.97 0.528301 0.528509 0.529005 −104.8 −76.8 −32.2
Diopside 104.55 90.30 61.26 0.528508 0.528708 0.529177 −84.8 −61.3 −25.4
Jadeite 112.83 97.57 66.37 0.528331 0.528540 0.529037 −100.6 −72.5 −28.6
Forsterite 99.47 85.60 57.59 0.528764 0.528955 0.529387 −61.9 −43.5 −17.6
Zircon 103.09 88.98 60.26 0.528501 0.528705 0.529180 −86.3 −62.5 −26.4
Schauble & Young
Grossular 100.67 86.55 58.10 0.528798 0.528991 0.529421 −57.5 −39.5 −14.9
Lizardite 99.51 86.28 59.37 0.528679 0.528875 0.529320 −70.3 −49.8 −19.1
Kaolinite 111.96 97.25 67.12 0.528556 0.528750 0.529201 −75.7 −52.3 −16.9
Phlogopite 103.73 89.72 61.19 0.528512 0.528718 0.529194 −84.8 −60.9 −24.4
Tremolite 107.42 92.96 63.41 0.528349 0.528561 0.529062 −100.4 −73.1 −30.1
Glaucophane 111.51 96.57 65.98 0.528261 0.528476 0.528989 −108.8 −79.3 −32.2
Spinel (normal) 96.70 82.79 55.02 0.529026 0.529214 0.529615 −38.6 −24.7 −8.6
Diaspore 105.10 90.65 61.53 0.529025 0.529193 0.529552 −30.1 −17.1 −2.0
Xenotime YPO4 108.35 93.74 63.85 0.528427 0.528627 0.529103 −91.6 −66.4 −27.1
Xenotime LuPO4 110.20 95.35 64.97 0.528401 0.528603 0.529083 −93.7 −67.8 −27.1
Monazite LaPO4 103.09 89.19 60.75 0.528464 0.528660 0.529127 −90.0 −66.4 −29.0
Fluorapatite 104.31 90.29 61.57 0.528453 0.528649 0.529115 −90.6 −66.7 −28.8
Anhydrite 111.93 97.16 66.71 0.528319 0.528515 0.528992 −102.1 −75.2 −31.3
Barite 107.41 93.19 63.89 0.528387 0.528578 0.529044 −96.3 −71.3 −30.8
Gypsum (all O) 102.18 89.34 62.71 0.528635 0.528833 0.529277 −73.0 −50.8 −17.4
Gypsum (sulfate O) 109.41 94.94 65.12 0.528366 0.528559 0.529029 −97.9 −72.2 −30.5
Gypsum (H2O) 87.58 78.04 57.88 0.529769 0.529813 0.529912 10.3 9.9 8.1
NaClO4 106.54 92.47 63.43 0.528392 0.528584 0.529052 −96.1 −71.2 −30.8
Calcite 114.44 99.75 69.22 0.528436 0.528609 0.529036 −88.0 −64.5 −25.7
Aragonite 113.69 99.16 68.91 0.528421 0.528592 0.529017 −90.0 −66.5 −27.4
Dolomite 119.53 104.15 72.22 0.528414 0.528590 0.529022 −88.4 −63.8 −23.7
Magnesite 124.82 108.68 75.26 0.528409 0.528587 0.529023 −86.8 −61.4 −20.5
Nahcolite 121.52 106.37 74.67 0.528435 0.528600 0.529009 −85.1 −61.5 −22.1
Nitratine 111.71 97.52 67.85 0.528406 0.528579 0.529013 −92.5 −68.7 −28.8
H2O vapor (anharm.) 68.29 61.70 47.26 0.530019 0.530030 0.530063 0 0 0
H2O vapor (harm.) 71.68 64.81 49.74 0.530018 0.530029 0.530062 0 0 0
Liquid water (R-model, anharm.) 79.81 70.65 52.11 0.529889 0.529918 0.529988 12.9 10.3 6.1
Liquid water (R-model, harm.) 83.20 73.75 54.59 0.529889 0.529918 0.529988 − − −
Liquid water (HW-model, anharm.) 80.04 71.01 52.34 0.529889 0.529918 0.529988 13.3 11.0 6.6
Liquid water HW-model, harm.) 83.43 74.11 54.82 0.529889 0.529918 0.529988 − −
Water Ice Ih (harm.) 91.31 81.04 59.42 0.529760 0.529807 0.529915 16.1 14.9 11.2
Water Ice XI (harm.) 91.39 81.11 59.48 0.529760 0.529806 0.529914 16.1 15.0 11.3
Mass Dependence of Equilibrium Oxygen Isotope Fractionation
CO2 vapor (anharm.) 123.57 109.54 79.39 0.527852 0.527985 0.528366 −156.2 −126.8 −68.4
CO vapor (anharm.) 112.84 101.33 76.10 0.528128 0.528201 0.528429 −123.4 −104.8 −64.9
H4SiO4 vapor 100.53 88.19 62.54 0.529053 0.529192 0.529509 −32.0 −20.2 −3.1
155
Table 5. Polynomial fits of the form 1000 ln18/16b = A/T6 + B/T5 + C/T4 + D/T3 + E/T2 + F/T for temperatures above 243.15 K (273.15 K for liquid water).
156
A temperature independent term, “G”, is included in the fits for liquid water, taken directly from earlier liquid–vapor and supercritical fluid–vapor studies
(Rosenbaum 1997; Horita and Wesolowski 1994).

H6Si2O7 vapor bridging O 126.57 109.86 75.38 0.528221 0.528429 0.528935 −109.9 −78.1 −27.0
Substance A B C D E F G
a-quartz −8.57662E+15 1.29277E+14 −6.47264E+11 1.13149E+08 1.27764E+07 −1.58089E+01 −
Albite (low) −6.38987E+15 1.00579E+14 −5.33809E+11 1.54282E+08 1.15385E+07 6.24956E+00 −
Microcline (low) −6.46928E+15 1.01670E+14 −5.38345E+11 1.56396E+08 1.15223E+07 6.38887E+00 −
Anorthite (ordered) −4.67984E+15 7.71919E+13 −4.34289E+11 1.57751E+08 1.04922E+07 1.60620E+01 −
Diopside −3.62416E+15 6.16464E+13 −3.60713E+11 1.44442E+08 9.72754E+06 1.80991E+01 −
Jadeite −4.40482E+15 7.31080E+13 −4.15958E+11 1.52614E+08 1.06345E+07 1.60881E+01 −
Forsterite −2.19103E+15 4.08989E+13 −2.65461E+11 1.19351E+08 8.92416E+06 1.89238E+01 −
Zircon −3.23144E+15 5.63538E+13 −3.38586E+11 1.45392E+08 9.50742E+06 2.04146E+01 −
Schauble & Young
Grossular −1.88297E+15 3.66292E+13 −2.47596E+11 1.14579E+08 8.94561E+06 1.92367E+01 −

Lizardite 7.73661E+15 −1.12366E+14 6.86921E+11 −2.81599E+09 1.26740E+07 −8.49823E+01 −
Kaolinite 6.51056E+15 −9.35842E+13 5.89200E+11 −2.76100E+09 1.40130E+07 −9.49342E+01 −
Phlogopite 7.27276E+14 −4.55563E+12 3.47214E+10 −9.73368E+08 1.09344E+07 −1.45729E+01 −
Tremolite −2.12065E+15 3.79732E+13 −2.04889E+11 −4.08901E+08 1.07765E+07 −2.36835E+00 −
Glaucophane −2.63255E+15 4.51509E+13 −2.37314E+11 −4.08262E+08 1.12546E+07 −6.21183E+00 −
Spinel (normal) −6.18560E+14 1.71233E+13 −1.51621E+11 6.83972E+07 8.26379E+06 1.35275E+01 −
Diaspore 4.57904E+15 −7.03550E+13 4.54600E+11 −2.10919E+09 1.21732E+07 −2.67567E+02 −
Xenotime YPO4 −4.39155E+15 7.26398E+13 −4.11008E+11 1.51631E+08 1.02582E+07 1.59587E+01 −
Xenotime LuPO4 −4.56891E+15 7.51877E+13 −4.22893E+11 1.52834E+08 1.04575E+07 1.53390E+01 −
Monazite LaPO4 −4.13682E+15 6.87433E+13 −3.90816E+11 1.47530E+08 9.75493E+06 1.62466E+01 −
Fluorapatite −4.37161E+15 7.20071E+13 −4.05288E+11 1.47629E+08 9.92196E+06 1.50223E+01 −
Anhydrite −5.96338E+15 9.37217E+13 −4.97617E+11 1.29805E+08 1.10175E+07 2.01074E+00 −
Barite −5.39636E+15 8.58034E+13 −4.62533E+11 1.30922E+08 1.04920E+07 5.27267E+00 −
Gypsum (all O) 8.03873E+15 −1.25014E+14 8.53232E+11 −3.81657E+09 1.49278E+07 −2.81301E+02 −
Gypsum (sulfate O) −5.50349E+15 8.73493E+13 −4.68492E+11 1.13100E+08 1.07298E+07 3.79932E+00 −
Gypsum (H2O) 3.50923E+16 −5.48078E+14 3.47959E+12 −1.16216E+10 2.32601E+07 −8.30661E+02 −
NaClO4 −5.30560E+15 8.48246E+13 −4.60322E+11 1.34434E+08 1.04159E+07 6.62368E+00 −
Calcite −8.13464E+15 1.16592E+14 −5.39168E+11 −1.93699E+08 1.22391E+07 −8.98300E+01 −
Aragonite −8.31575E+15 1.18153E+14 −5.37818E+11 −2.40930E+08 1.22970E+07 −1.02204E+02 −
Dolomite −8.39772E+15 1.19880E+14 −5.50664E+11 −2.25445E+08 1.27776E+07 −9.88440E+01 −
Magnesite −8.52650E+15 1.21467E+14 −5.56075E+11 −2.53019E+08 1.32925E+07 −1.06063E+02 −
Nahcolite −6.36640E+15 7.90625E+13 −2.20150E+11 −1.56554E+09 1.51029E+07 −3.28073E+02 −
Nitratine −8.18521E+15 1.19422E+14 −5.67548E+11 −1.04293E+08 1.19529E+07 −6.82541E+01 −
H2O vapor (anharm.) 4.08073E+16 −6.26044E+14 3.88795E+12 −1.25885E+10 2.28002E+07 −1.24437E+03 −
H2O vapor (harm.) 4.15943E+16 −6.37102E+14 3.94849E+12 −1.27503E+10 2.30136E+07 −4.32931E+02 −
Water Ice Ih 2.55981E+16 −4.14442E+14 2.75605E+12 −9.81791E+09 2.18496E+07 −1.07955E+03 −
Water Ice XI 2.56214E+16 −4.14778E+14 2.75816E+12 −9.82552E+09 2.18670E+07 −1.07882E+03 −
CO2 vapor −6.27638E+15 4.56379E+13 2.68853E+11 −4.27840E+09 2.05286E+07 −1.32159E+03 −
CO vapor −1.93531E+15 −6.95760E+13 1.31394E+12 −8.38839E+09 2.58267E+07 −2.00113E+03 −
H4SiO4 vapor 2.09683E+16 −3.10907E+14 1.88036E+12 −6.45554E+09 1.74651E+07 −1.60791E+02 −
H6Si2O7 vapor bridging O −6.23677E+15 1.00933E+14 −5.54067E+11 1.87949E+08 1.23447E+07 1.63985E+01 −
Sums of 103 lnb H2O vapor and 103 ln a H2O liquid-vapor or 103 ln a H2O supercritical–vapor from Rosenbaum (1997)
Liquid Water (anharm.) (273–403 K) 2.86501E+15 −6.66957E+12 −8.59748E+10 −1.15346E+08 6.40332E+06 −6.46500E+01 12.815
Liquid Water (anharm.) (403–647 K) 1.20346E+17 −6.66957E+12 −1.98057E+12 −1.15346E+08 1.67523E+07 −6.46500E+01 −6.482
157
158
Supercritical Water (anharm.) (≥ 647 K) 5.58246E+17 −6.66957E+12 −2.77257E+12 −1.15346E+08 1.29963E+07 −6.46500E+01 −
Liquid Water (harm.) (273–403 K) 3.65207E+15 −1.77280E+13 −2.54358E+10 −2.77134E+08 6.61672E+06 7.46789E+02 12.815
Liquid Water (harm.) (403–647 K) 1.21133E+17 −1.77280E+13 −1.92004E+12 −2.77134E+08 1.69657E+07 7.46789E+02 −6.482
Supercritical Water (harm.) (≥ 647 K) 5.59033E+17 −1.77280E+13 −2.71204E+12 −2.77134E+08 1.32097E+07 7.46789E+02 −
Sums of 103 ln b H2O vapor and 103 ln a H2O liquid–vapor from Horita and Wesolowski (1994)
Liquid Water (anharm.) (273–647 K) 4.08073E+16 −6.26044E+14 3.88795E+12 −1.22381E+10 2.11338E+07 5.46793E+03 −7.685
Liquid Water (harm.) (273–647 K) 4.15943E+16 −6.37102E+14 3.94849E+12 −1.23999E+10 2.13472E+07 6.27937E+03 −7.685
Fits to results of Rosenbaum (1997), used to estimate properties of liquid and supercritical water
H2O vapor 4.03612E+16 −6.19374E+14 3.84852E+12 −1.24732E+10 2.26339E+07 −1.17972E+03 −
103 ln a H2O liquid–vapor (273-403K) −3.79422E+16 6.19374E+14 −3.97392E+12 1.24732E+10 −1.63969E+07 1.17972E+03 12.815
103 ln a H2O liquid–vapor (403-647K) 7.95388E+16 6.19374E+14 −5.86852E+12 1.24732E+10 −6.04789E+06 1.17972E+03 −6.482
103 ln a H2O supercritical–vapor (≥ 647K) 5.17439E+17 6.19374E+14 −6.66052E+12 1.24732E+10 −9.80389E+06 1.17972E+03 −
Schauble & Young
Table 6. Polynomial fits of the form qa–atoms = 0.530520 + L/T4 + M/T3 + N/T2 + P/T for temperaturesabove 243.15 K (273.15 K for liquid water).
Substance L M N P
a−quartz −9.84260E+06 1.34471E+05 −5.72152E+02 6.08257E−02
Albite (low) −7.15131E+06 1.05390E+05 −4.84222E+02 5.02774E−02
Microcline (low) −7.07921E+06 1.05174E+05 −4.85387E+02 5.07437E−02
Anorthite (ordered) −5.22563E+06 8.32484E+04 −4.11634E+02 4.10861E−02
Diopside −3.43390E+06 6.23164E+04 −3.41450E+02 3.30828E−02
Jadeite −4.79241E+06 7.78984E+04 −3.95169E+02 3.92213E−02
Forsterite −1.70288E+06 4.00637E+04 −2.61182E+02 2.26548E−02
Zircon −3.01991E+06 5.85699E+04 −3.33492E+02 3.19019E−02
Substance L M N P
Grossular −1.25214E+06 3.47731E+04 −2.44601E+02 2.01696E−02
Lizardite 6.78135E+05 1.79214E+04 −1.91993E+02 −7.41326E−02
Kaolinite −1.82963E+06 4.21814E+04 −2.57960E+02 −6.81728E−02
Phlogopite −1.31818E+06 4.19366E+04 −2.82492E+02 −1.24930E−02
Tremolite −3.26905E+06 6.39611E+04 −3.58080E+02 2.02277E−02
Glaucophane −4.22525E+06 7.44971E+04 −3.92412E+02 2.75325E−02
Spinel (normal) 2.11190E+05 1.43521E+04 −1.69555E+02 9.68832E−03
Diaspore −3.20897E+06 3.77741E+04 −1.87320E+02 −7.02837E−02
Xenotime YPO4 −4.36909E+06 7.30846E+04 −3.75570E+02 3.76112E−02
Xenotime LuPO4 −4.56535E+06 7.53224E+04 −3.83345E+02 3.85453E−02
Monazite LaPO4 −4.25029E+06 7.16694E+04 −3.69102E+02 3.70972E−02
Fluorapatite −4.45849E+06 7.38756E+04 −3.75391E+02 3.78383E−02
Anhydrite −6.29166E+06 9.36614E+04 −4.35133E+02 4.48193E−02
Barite −5.82495E+06 8.81142E+04 −4.15417E+02 4.24829E−02
Gypsum (all O) −1.71512E+05 1.82458E+04 −1.62159E+02 −1.57186E−01
Gypsum (sulfate O) −5.87792E+06 8.89986E+04 −4.19591E+02 4.27316E−02
Gypsum (H2O) −6.87656E+06 5.60349E+04 −1.18337E+02 −1.82943E−01
NaClO4 −5.49862E+06 8.53229E+04 −4.09047E+02 4.19470E−02
Calcite −8.74242E+06 1.11775E+05 −4.58694E+02 4.10091E−02
Aragonite −9.24375E+06 1.16665E+05 −4.71117E+02 4.15681E−02
Dolomite −8.93017E+06 1.13330E+05 −4.63424E+02 4.06824E−02
Magnesite −8.88665E+06 1.12414E+05 −4.60718E+02 3.95939E−02
Nahcolite −1.05391E+07 1.24829E+05 −4.75553E+02 1.61644E−02

Nitratine −8.48622E+06 1.11387E+05 −4.63752E+02 4.37314E−02
H2O vapor (anharm.) −5.55222E+06 4.16832E+04 −5.59908E+01 −2.15903E−01
159
Substance L M N P
160
H2O vapor (harm.) −5.44629E+06 4.05846E+04 −5.21169E+01 −2.20711E−01

Water Ice Ih −7.55686E+06 6.38621E+04 −1.50800E+02 −1.38100E−01
Water Ice XI −7.57522E+06 6.40300E+04 −1.51169E+02 −1.38290E−01
CO2 vapor −2.75388E+07 2.78083E+05 −8.40698E+02 −2.28591E−02
CO vapor −3.93502E+07 3.58154E+05 −9.25062E+02 −1.27385E−01
H4SiO4 vapor −2.33424E+06 3.28910E+04 −1.59882E+02 −1.40973E−01
H6Si2O7 vapor bridging O −5.84665E+06 9.14111E+04 −4.40417E+02 4.54413E−02
3
Liquid and Supercritical Water (anharm.; ≥ 273K) using 10 ln a H2O liquid–vapor from Rosenbaum (1997)
qliquid–vapor = 0.529 (Barkan et al. 2005) −5.49858E+07 4.63855E+05 −1.22371E+03 7.87587E−01
Lesser of vapor vs. previous −7.28054E+06 5.94857E+04 −1.22058E+02 −1.69771E−01
50% ice model −8.79767E+06 6.73660E+04 −1.31127E+02 −1.62891E−01
Liquid and Supercritical Water (anharm.; ≥ 273K) using 103 ln a H2O liquid–vapor from Rosenbaum (1997)
qliquid–vapor = 0.529 (Barkan et al. 2005) −4.99105E+07 4.21352E+05 −1.10973E+03 6.92261E−01
50% ice model −8.79767E+06 6.73660E+04 −1.31127E+02 −1.62891E−01
Schauble & Young
3
Liquid Water (anharm.; 273K ≤ T ≤ 647K) using 10 ln a H2O liquid–vapor from Horita and Wesolowski (1994)
qliquid–vapor = 0.529 (Barkan et al. 2005) −1.25778E+07 1.08486E+05 −2.70736E+02 −2.28243E−02
50% ice model −1.24040E+06 6.01304E+03 2.66231E+01 −2.89380E−01
Liquid Water (harm.; 273K ≤ T ≤ 647K) using 103 ln a H2O liquid–vapor from Horita and Wesolowski (1994)
qliquid–vapor = 0.529 (Barkan et al. 2005) −1.18765E+07 1.02119E+05 −2.51906E+02 −3.92103E−02
50% ice model −1.24040E+06 6.01304E+03 2.66231E+01 −2.89380E−01
to alternative calculations using separate, mineral specific or mineral-type specific scale

factors, for instance applying a ~3% correction for carbonates, 4–5% for silicates, and ~5–6%
for sulfates, phosphates, and perchlorates, based on correlations of subgroups of crystals
with measured spectra reported in this study. As detailed by Meheut et al. (2007), the effects
of such scale factor adjustments can be estimated using an effective power law scaling, e.g.
ln β(SF1)/ln β(SF2) = (SF1/SF2) p, where β are predicted 18O/16O reduced partition function
ratios calculated by applying two different scale factors (SF1 and SF2) to a given set of modeled
vibrational frequencies for some substance. The exponent p is expected to be slightly less than 2
for typical oxygen-bearing compounds at geochemically relevant temperatures. Alternatively,
one can simply scale the temperature, because frequency and temperature always appear as
a ratio in the calculation of a reduced partition function ratio; i.e., ln β(SF1) at T = T1 is equal
to ln β(SF2) at T = T1(SF2/SF1). Either way, one would expect that ~1% uncertainty or scatter
in frequency scale factors will lead to ~2‰ scatter in 1000 ln18/16β, given typical values of
~80–110‰ for 1000 ln18/16β at 300 K for many of the crystals in this study. Errors will be
correspondingly smaller at high temperatures. This uncertainty is much larger than typical
18
O/16O measurement precisions, and is likely to be a major source of potential error in the
present calculations. However, the effect of scale factor uncertainty on calculated qa–atoms is
much smaller. Test calculations with varying scale factors for α-quartz, diaspore, anhydrite,
and calcite indicate a typical perturbation of ~2 × 10–5 on qa–atoms when the scale factor is varied
by 1%, which is a precision well beyond current measurement capabilities. This result agrees
with the scaling arguments presented by Cao and Liu (2011), suggesting that the mass law
exponent will be a robust product of theoretical models in the absence of large errors in the
electronic structure or major anharmonic effects. Uncertainties from residual random scatter in
frequencies after scaling are expected to be even smaller due to partial cancellations of errors.
Another approach to estimating uncertainties stemming from using the PBE functional to
calculate vibrational frequencies is to compare models using different electronic structure methods,
in the expectation that both systematic and random errors will change as well. Comparisons
with molecule-based and cluster models using hybrid density functional theory (e.g., B3LYP;
Becke 1993) and higher-order methods such as Møller-Plesset theory (MP2; Møller and Plesset
1934) provide the most robust tests of this sort. The qa–atoms results of Cao and Liu (2011) and
Haynes et al. (2018) using such methods are thus of particular value, and are discussed in detail
in a later section. As an additional check, we created models of a subset of crystals (α-quartz,
diaspore, anhydrite, calcite) and molecules (water vapor and carbon dioxide) using two different
density functionals: the Local Density Approximation functional (as parameterized by Perdew
and Zunger 1981), here abbreviated LDA, and the PBEsol gradient corrected density functional
optimized for accurately reproducing lattice constants of solids (Perdew et al. 2007). Frequency
scale factors are determined independently for the models using each functional, these are 3.8%
for PBE (slightly different from the 4.3% scaling fitted to the whole PBE model suite), 1.6%
for the LDA models, and 2.7% for the PBEsol models. The residual scatter after fitting the scale
factors for the PBE and PBEsol models for these six substances is similar (rms misfit of 17 cm–1
vs. 16 cm–1 respectively). The rms frequency over the correlated modes is 962 cm–1, suggesting
slightly less than 2% relative scatter, which is similar to the larger set of correlated PBE models.
The residual scatter for the LDA models is slightly larger (23 cm–1). Calculated reduced partition
function ratios and qa–atoms are shown in Table 7. The two gradient-corrected functionals show
typical mismatches of ~1–2‰ at 300 K, consistent with the crude scale-factor based estimate
above, but this test may be overly optimistic given the close theoretical relationship between
PBE and PBEsol. The LDA models are more variable, with mismatches of up to 4‰. qa–atoms is
much less sensitive to the choice of functional, varying by less than 5 × 10–5 between PBE and
PBEsol models, and by less than 1x10–4 between PBE and LDA models.
Effects of sampling the phonon density of states. The potential effects of incomplete
phonon sampling on reduced partition function ratios and isotopic clumping have been
discussed previously (e.g., Elcombe and Hulston 1975; Schauble et al. 2003, 2006; Méheut
et al. 2007). As in prior work, these effects are estimated in the present study by comparing
models of a subset of crystals (α-quartz, diaspore, anhydrite, calcite, and aragonite) using a
coarser phonon wave vector sampling (with roughly half as many distinct wave vectors). The
test results are shown in Table 8. They show a limited dependence of 18O/16O fractionation,
with less than 0.2‰ variation at 298.15 K in all cases. qa–atoms varies by no more than 5 × 10–5
in all cases. The lack of sensitivity of qa–atoms to phonon wave vector sampling suggests that
adequate results can often be obtained on a sampling grid containing only one distinct wave
vector, so long as it is chosen with care.
The variation in calculated reduced partition function ratios with fine vs. coarse phonon
wave vector grids is notably smaller than the ~1‰ variation found in previous work on
rhombohedral carbonates (Schauble et al. 2006) using phonon wave vector samples of similar
size. An important difference in the earlier work is that in general only one atomic position per
unit cell in each crystallographic site was included in the thermodynamic calculation, whereas
the present study accounts for all oxygen atoms on each unit cell through a series of one-atom
isotopic substitutions. This improves convergence because the crystallographic symmetries of a
lattice are commonly split along the phonon wave vector, so that (for instance) 18O substitution
on one of the O(2) atomic positions in aragonite will not necessarily generate the same frequency
shifts for a particular phonon wave vector as 18O substitution on another O(2) atom. As it turns
out, this asymmetric response has very little effect on qa–atoms. It also has little effect on the Keq of
multiply substituted isotopologues in carbonate minerals, which is the main focus of the Schauble
et al. (2006) study. A re-calculation of Keq[3866] in calcite using a grid with two distinct wave
vectors from the present test models indicates that the asymmetry effect might be responsible
for an up to 8 × 10–6 (i.e., 8 per meg) deviation for a particular C–O bond from the average over
all bonds at 25 ºC, and the failure to account for this effect may be responsible for most of the
difference in the Keq values calculated with 2 wave-vector vs. 5 wave-vector grids in that study
(Table 9). In fact, the re-calculated result using a 2 wave-vector sample of phonon frequencies
from this study, combined with the frequency scale factor of 1.0331 from Schauble et al. (2006)
and averaging over all 6 C–O bonds in the primitive calcite unit cell, yields a Keq that is within 1
per meg of the Schauble et al. (2006) 5 wave-vector sample result, despite being based on models
constructed using different pseudopotentials, electronic wave vector grids, and cutoff energies.
This asymmetric substitution effect is smaller than other likely sources of error considered in
the Schauble et al. (2006) study (such as anharmonicity and uncertainty in the frequency scale
factor), and it does not change any of the significant conclusions reached in that work.
Comparison with previous determinations of 18O/16O fractionation. Although the present
work is primarily concerned with the estimation of the mass dependence of oxygen isotope
fractionation qa–b, comparisons with measured or previously estimated 18O/16O fractionation
factors are important for two reasons.
1. First-principles models that accurately reproduce 18O/16O fractionation factors can
be reasonably be inferred to predict 17O/16O with similar accuracy, and thus yield
realistic qa–b. They can also be inferred to predict 18O/16O fractionation factors for
materials that have not yet been experimentally or empirically calibrated.
2. aa–b is itself an important term in the conversion between qa–atoms, qb–atoms, qa–b,
18/16
and ∆′17Oa–b.
Calibration of 18O/16O fractionation factors has been an ongoing project of stable isotope
geochemistry for more than a half-century, and there is a large literature to provide a basis
for comparisons. It is beyond the scope of the present chapter to provide a comprehensive
evaluation of these calibrations, and excellent compilations are available in earlier volumes
of this series, e.g., Chacko et al. (2001). However, it is worthwhile to examine a subset of
materials spanning as much variability in crystal structure types and chemistries as is feasible.
Table 7. Comparison of reduced partition function ratios and exponents calculated with different
density functionals, using frequency scale factors determined just for these six species. The PBE
results are thus slightly different from the results listed above.
298.15 K 1000 ln18/16b qa–atoms
Substance PBE PBEsol LDA PBE PBEsol LDA
a-quartz 107.87 108.52 109.99 0.528129 0.528154 0.528176
Diaspore 89.91 91.42 93.59 0.529202 0.529158 0.529118
Anhydrite 96.40 98.27 100.20 0.528525 0.528487 0.528457
Calcite 98.83 99.96 101.67 0.528623 0.528604 0.528584
H2O vapor 64.45 63.57 62.73 0.530030 0.530033 0.530035
CO2 114.84 113.89 114.07 0.527949 0.527951 0.527945
RMS Misfit vs. PBE: 1.23 2.71 0.000027 0.000051
Table 8. Comparison of reduced partition function ratios and exponents calculated with different phonon
wave vector grids. The number of distinct phonon wave vectors for each species is given in parentheses.
298.15 K 1000 ln18/16b qa–atoms
Substance Fine grid Coarse grid Fine grid Coarse grid
a-quartz (4,2) 108.70 108.69 0.528118 0.528118
Diaspore (4,2) 90.65 90.65 0.529193 0.529193
Anhydrite (2,1) 97.16 97.32 0.528515 0.528511
Calcite (5,2) 99.75 99.59 0.528609 0.528614
Aragonite (2,1) 99.16 99.20 0.528592 0.528591
RMS Misfit 0.10 0.000002
Table 9. Comparison of calculated 13C–18O–16O–16O clumping equilibrium constants in calcite.

Results are given using both the carbonate-specific 1.0331 frequency scale factor used in
Schauble et al. (2006), and the general 1.043 scale factor from the present study.
Average Keq(3866) with a 2 q-point phonon grid
273.15 K 298.15 K 373.15 K 573.15 K 1273.15K
Schauble et al. (2006) 1.000482 1.000403 1.000241 1.000073 1.000004
Present Study (SF 1.0331) 1.000491 1.000411 1.000247 1.000075 1.000005
Present Study (SF 1.043) 1.000500 1.000419 1.000253 1.000078 1.000005
Average Keq(3866) with a 5 q-point phonon grid
Schauble et al. (2006) 1.000490 1.000410 1.000247 1.000075 1.000005
Here we focus on species with experimental calibrations and (ideally) ongoing interest for stable
isotope studies, including calcite, the silicates quartz, albite, anorthite, diopside, forsterite, and
zircon, and the rare-earth element phosphate mineral monazite. These are examined in terms
of 18/16aquartz–x, the fractionation relative to quartz. The sulfate minerals barite and gypsum, the
phosphate mineral fluorapatite, and the carbonate minerals calcite and dolomite are compared
with liquid water, using the liquid–vapor model of Rosenbaum (1997) and our anharmonicity-
corrected model for H2O vapor to estimate the properties of liquid H2O. This procedure likely
introduces some error—and is clearly no longer ab initio. However, the systematics of model
crystal fractionations vs. liquid water will still depend on the accuracies of the crystal models
relative to each other. These comparisons are shown in Figure 5.
5 Prese nt study 2 Prese nt study 3 Prese nt study
Va lle y e t a l. (2003) Chiba e t a l. (1989) Chiba e t a l. (1989)
4 King et a l. (2001)
Cla yton and Kie ffe r (1991)
Cla yton and Kie ffe r (1991)
T ra il et a l. (2009) Zhe ng (1991)
1000 ln aqz-an
1000 ln aqz-zr
1000 ln a qz-ab
Zhe ng (1991) 2
3 Zhe ng e t a l. (1991)
M ehe ut a nd Sc ha ubl e (2014)
Kie ffe r (1982) 1
2
1
a-quartz - zircon a-quartz - albite a-quartz - K-feldspar

0 0 0
800 100 0 120 0 140 0 800 100 0 120 0 140 0 800 100 0 120 0 140 0
T (K) T (K) T (K)

Sha rp and Kirschner (2003) Chiba e t a l. (1989)
7 6 Ble ec ke r and S ha rp (2007)
5 Cla yton and Kie ffe r (1991)
Chiba e t a l. (1989)
6 Zhe ng (1991)
Cla yton and Kie ffe r (1991) 5
1000 ln aqz-cc
1000 ln a qz-fo
1000 ln a qz-mo
4
5 M ehe ut a nd Sc ha ubl e (2014)
Zhe ng (1991)
4
4 3
3
3
a-quartz - calcite 2
2
2
1
1 1
a-quartz - forsterite a-quartz - monazite
0 0 0
323 823 132 3 800 100 0 120 0 140 0 600 800 100 0 120 0 140 0
T (K) T (K) T (K)
Ve nne ma n et a l. (2001) Chiba e t a l. (1981)

30 30 Kusak abe and Robinson (1977)
Longi ne lli and N uti (1973) Ha la s a nd Pluta (2000) 20
25 25
Barite - liquid water
Pucéa t e t al. (2010) Ze e be (2010)
1000 ln a fap-liquid water
1000 ln a anh-liquid water
1000 ln a ba-liquid water
20 Kolodny et a l. (1983)
20
Anhydrite - liquid water or 15
Chang and Bla ke (2015) sulfate(aq) - liquid water
15 15
10
10 10
5
5 5
Apatite - liquid water
0 0 0
263 283 303 323 343 363 273 473 673 873 273 373 473 573 673
T (K) T (K) T (K)
40 Prese nt study 4 Prese nt study

26 Prese nt study Ice I h
35 O 'Ne il e t a l. (1969) 3.5 24 Prese nt study Ice XI

Northrop a nd Clay ton (1966)
Kim a nd O' Ne il (1997) E lle hoj (2013)
30 3 22
She ppard and S chwarcz (1977)
M ajoube (1970)
1000 ln a cc-liquid water
1000 ln a ice-vap
1000 ln adol-cc
25 2.5 20
Horita (2014)
20 Calcite - liquid water 2 18
15 1.5 Dolomite - calcite 16
10 1 14
5 0.5 12
Ice - water vapor
0 0 10
273 473 673 873 273 773 127 3 177 3 240 250 260 270 280
T (K) T (K) T (K)
50
Prese nt study
45
Sha rp and Kirschner (1994)
40
1000 ln a qz-liquid/supercrit. water
35
30
25
a-quartz - liquid water
20
15
10
5
0
200 400 600 800 100 0
T (K)
Figure 5. Comparison of calculated 18O/16O fractionation factors with previous experimental and theoretical
studies. The models for anhydrite and fluorapatite are also compared with previous calibrations of the dis-
solved sulfate–water fractionation (Zeebe et al. 2010; Halas and Pluta 2000) and the dissolved phosphate–
water fractionation (Chang and Blake 2015).
Experiments and empirical (field-based) calibrations of oxygen isotope fractionation

between quartz and calcite give varying results—the empirical study of Sharp and Kirschner
(1994) finds fractionation twice as large as a laboratory calibration (Clayton et al., 1989).
The present models closely track the empirical calibration of Sharp and Kirschner (1994).
It should be noted, however, that at least some other silicate–calcite fractionations show
good agreement with direct exchange experiments, for instance phlogopite–calcite (−1.5‰
vs. −1.7‰ at 750 ºC; Hu and Clayton 2003), and it should also be noted that the quartz–
calcite fractionation predicted by Hayles et al. (2018) is intermediate between the Sharp
and Kirschner (1994) and Clayton et al. (1989) calibrations. Fractionations among silicate
minerals show reasonably good consistency with experimental and empirical calibrations,
particularly for quartz–albite, quartz–anorthite, quartz–forsterite, and quartz–diopside, which
all remain within 0.2–0.3‰ of tabulated experimental calibrations (Chiba et al. 1989, Clayton
and Kieffer 1991) above 600 ºC. Previous calibrations of the quartz–zircon fractionation are
more variable, with theoretical estimates tending to predict larger fractionations than are
observed. The present model behaves similarly, apparently overestimating the fractionation
by 0.4–0.8‰ at 600 ºC—and is in fact remarkably similar to a much earlier theoretical
prediction of Kieffer (1982). Qualitatively, the present model is reasonable even for this
mineral pair, but it is conceivable that a slightly higher frequency scale factor would be
appropriate to consider for zircon, as suggested by both the overestimated quartz–zircon
fractionation and the Raman and IR correlations for zircon. Quartz–monazite fractionations
measured by Breecker and Sharp (2007) are in good agreement with the present models.
Mineral–water fractionations for phosphates, sulfates, and carbonates likewise generally
show reasonable agreement with experiment, as has been found previously for similar types
of models (e.g., Schauble et al. 2006; Meheut et al. 2007). The present models appear to
underestimate sulfate–water and phosphate–water fractionations by up to 4‰ from 273–423 K.
Interestingly, a model of meridianiite, Mg(H2O)6∙SO4∙5H2O, which contains fully solvated
SO42– molecules, runs very close to previous calibrations of the aqueous sulfate–water
fractionation (Halas and Pluta 2000; Zeebe 2010), falling within 1‰ of the both calibrations
from 273–423 K. Our models closely match experimentally determined calcite–water and
dolomite–water calibrations (e.g., O’Neil et al. 1969; Kim and O’Neil 1997; Horita 2014). The
present results also match some previous dolomite–calcite fractionations well (e.g., Northrop
and Clayton 1966; Sheppard and Schwarcz 1970) with greater disagreement with Horita (2014).
Calculated fractionations between water ice and vapor (using the harmonic vapor model)
are considerably larger than measurements (e.g., Majoube 1970; Ellehoj et al. 2013), by ~5‰ at
273 K. Note that using anharmonically corrected values for vapor would worsen the mismatch (by
~3–4‰ at 273 K). The present estimates are in reasonable agreement with a previous model using
similar DFT parameters, however (Meheut et al. 2007), suggesting that this is a systematic error
for the theoretical method. This misfit may in part be consistent with the underestimation of lattice
constants for ice crystals by the PBE functional, in that both suggest that the hydrogen bonding
network of ice is not reproduced as accurately as other bond types. Relatively poor descriptions
of hydrogen-bonding interactions in water vapor clusters, liquid water, and ice are a well-known
defect of DFT methods (e.g., Gillan et al. 2016), and it should perhaps not be surprising that
this is one case where the methods used in the present chapter are less accurate. In contrast, the
estimated fractionation between liquid water (using the harmonic model of water vapor and the
liquid–vapor fractionation from either Rosenbaum 1997 or Horita and Wesolowski 1994) and the
water of hydration of gypsum is within 1‰ of experiments and empirical calibrations (Gazquez
et al. 2017; Herwartz et al. 2017), with ~4‰ higher 18O/16O in hydration water at ~20 ºC, and a
temperature sensitivity of roughly −0.01‰/ ºC. This agrees with another recent theoretical result
(Liu et al. 2019) using a similar type of model based on the PBEsol functional.
Comparison with previous determinations of qa–b and qa–atoms. Much less is known about
equilibrium deviations from the canonical mass-fractionation relationship than about 18O/16O
fractionation factors. In this section the available theoretical, experimental, and empirical
calibration data are compared to our present results. In general, comparison of qa–b with
experimental data depends on qa–atoms, qb–atoms, and 103 ln18/16aa–b = 103 ln18/16ba − 103 ln18/16β b.
The last term has just been discussed, with generally good agreement between model results
and calibration, but suggesting some reason for concern that fractionations involving water will
have modest systematic error. High-precision calibrations of θ have now been published for
silica vs. liquid water, gypsum water vs. liquid water, and liquid water vs water vapor (Barkan
and Luz 2005; Sharp et al. 2016; Gazquez et al. 2017; Herwartz et al. 2017). Theoretical
calibrations for these and other systems have also been calculated recently by other authors
(e.g., Cao and Liu 2011; Hayles et al. 2018). We compare these in turn (Fig. 6). In general,
there is excellent agreement between the present results and those of Hayles et al. 2018, and
reasonably good agreement with Cao and Liu (2011). This general agreement is consistent
with the error analysis of Cao and Liu (2011), which demonstrated that theoretical estimates
of qa–atoms, in particular, are relatively robust even for simplified models.
The most extensive comparison sets come from Cao and Liu (2011) and Hayles et al.
(2018). In the earlier work, qa–atoms (reported as κ, in their notation) is estimated for CO2,
CO, H2O, and H6Si2O7, as well as for gas-phase CO32– and molecular clusters mimicking the
structure of carbonate and silicate mineral structures. Cao and Liu (2011) mainly use hybrid
density functional theory (B3LYP; Becke 1993) to estimate vibrational frequencies in their
calculations. This method is closely related to the PBE functional used the present work—both
are substantially based on gradient-corrected density functionals—but B3LYP incorporates a
component of exact electron exchange. The exact exchange component of B3LYP is not as
easily adapted to systems with periodic boundary conditions as pure density functionals such
as PBE or BLYP, and so the calculations in Cao and Liu (2011) are limited to isolated atoms,
molecules, and clusters, using atom-centered basis sets rather than the pseudopotential+plane
wave basis set method applied here. However, a significant potential advantage for B3LYP
is that it is typically observed to reproduce vibrational frequencies, molecular structures, and
thermodynamic properties somewhat more accurately than PBE; for harmonic vibrational
frequencies and zero point energies the scale factor for B3LYP is close to unity for typical
molecular benchmark comparisons (e.g., Kesharwani et al. 2014; Alecu et al. 2010—updated
at https://t1.chem.umn.edu/freqscale/index.html), and no scale factor is applied by Cao and
Liu (2011) to their model calculations. However, the residual misfit after scaling, compared to
harmonic frequencies inferred from spectroscopic measurements, is similar for both methods
(e.g., Kesharwani et al. 2014). So cross-comparing results between the two studies is likely to
give useful information about the reliability of both methods. Typical mismatches are ~5 × 10−5
or less for the small gas-phase molecules CO, CO2, and H2O, with somewhat larger mismatch
of up to 2 × 10–4 for the bridging oxygen in vapor-phase disilicic acid H6Si2O7, and similar
disagreements between calcite with CO32–(v), and anorthite with H6SiAlO7–. The relatively
large mismatch between the present calcite model and the “calcite” result from Cao and Liu
(2011) comes mainly from their proposed crystal–vapor correction to the gas-phase carbonate
model. The close overall correspondence suggests that both model approaches are sufficiently
accurate to be useful. The good match between molecules and crystals with similar bonding
configurations around oxygen also suggests that a “building block” approach analogous to the
methods developed for predicting 18O/16O fractionation by e.g. Garlick (1966) and Zheng (1991)
will be even more well suited to predicting or rationalizing the mass dependence exponents, so
that signatures in complex, incompletely characterized, and amorphous materials can likely be
anticipated on the basis of simpler, better studied crystals and molecules.
a. a-quartz – liquid water

Bark an l -v
0.530 0.526 0.527
50% i ce model
Le sse r of Ba rka n l-v v s. v apor
0.529 Sha rp et a l. (2016) fit
Sha rp et a l. (2016) data
0.526
0.528 Cao a nd Liu (2011)
0.525 Ha yl es e t a l. (2018)
0.527
qqz-liquid water
qqz-liquid water
qqz-liquid water
0.525
0.526 Bark an l -v Bark an l -v
50% i ce model 50% i ce model
Le sse r of Ba rka n l-v v s. v apor 0.524 Le sse r of Ba rka n l-v v s. v apor
0.525
Sha rp et a l. (2016) fit Sha rp et a l. (2016) fit 0.524
Sha rp et a l. ((2016) da ta Sha rp et a l. (2016) data
0.524 Cao a nd Liu (2011) Cao a nd Liu (2011)
Ha yl es e t a l. (2018) Ha yl es e t a l. (2018)
0.523 0.523 0.523
200 700 250 300 350 400 0 20 40 60
T (K) T (K) 1000 ln 18/16a qz-liquid water
50 0 50 Bark an l -v
50% i ce model
-20 Le sse r of Ba rka n l-v v s. v apor
0 -40 0 Sha rp et a l. (2016) data
Cao a nd Liu (2011)
-60 Ha yl es e t a l. (2018)
-50 -80 -50

D'17qz-liquid water
D'17qz-liquid water
D'17qz-liquid water
-100
-100 -120 -100
Bark an l -v Bark an l -v
50% i ce model -140 50% i ce model
Le sse r of Ba rka n l-v. vs. va por Le sse r of Ba rka n l-v. vs. va por
-150 Sha rp et a l. (2016) data
-160 Sha rp et a l. (2016) data
-150
Cao a nd Liu (2011) -180 Cao a nd Liu (2011)
-200 -200 -200
200 700 250 300 350 400 0 20 40 60
b. Calcite– liquid water

0.540 0.526 0.526
Bark an l -v
0.538 50% i ce model
0.536
Cao a nd Liu (2011)
0.534 Ha yl es e t a l. (2018) Bark an l -v
0.525 50% i ce model 0.525
0.532
qcc-liquid water
qcc-liquid water
qcc-liquid water
Bark an l -v
0.530 Cao a nd Liu (2011) 50% i ce model
0.528 Ha yl es e t a l. (2018) Le sse r of Ba rka n l-v v s. v apor
0.526 Cao a nd Liu (2011)
0.524 0.524 Ha yl es e t a l. (2018)
0.524
0.522
0.520
0.518 0.523 0.523
200 700 250 300 350 400 0 10 20 30 40
T (K) T (K) 1000 ln 18/16a cc-liquid water
Bark an l -v
20 0 20
50% i ce model
0 -20 0 Le sse r of Ba rka n l-v v s. v apor
-20 -20 Cao a nd Liu (2011)
-40
Ha yl es e t a l. (2018)
-40 -40
-60
-60 -60
D'17cc-liquid water
D'17cc-liquid water
D'17cc-liquid water
-80
-80 -80
-100
-100 -100
Bark an l -v Bark an l -v
-120
-120 50% i ce model 50% i ce model -120
-140 Le sse r of Ba rka n l-v. vs. va por -140 Le sse r of Ba rka n l-v. vs. va por -140
Cao a nd Liu (2011) -160 Cao a nd Liu (2011)
-160 -160
-180 -180 -180
200 700 250 300 350 400 0 10 20 30 40
Figure 6. Comparison of calculated qa–b and ∆′17O with previous experimental and theoretical studies in the
quartz–water (a) and calcite–water (b) systems. For each system, the first row of three panels shows the expo-
nent as a function of temperature over a wide and narrow (low) range, and also a function of the equilibrium
O/ O fractionation. The irregular (hyperbolic) behavior of qcalcite–liquid water at ~700 K reflects a change in
18 16
sign of the calcite–water fractionation factor. Three variant models for qliquid water–atoms from the present study
are used, based on either qliquid water–water vapor=0.529 (Barkan l − v; Barkan and Luz 2005), the average of qwater
vapor–atoms and qice-Ih–atoms (50% ice model), or the lesser of qwater vapor–atoms and the Barkan l − v exponent. Calibra-
tion references: Sharp et al. (2016), Cao and Liu (2011), and Hayles et al. (2018). The y-axis units are per meg.
Hayles et al. (2018) focus on crystals and liquid water, mainly using a cluster-based method
and hybrid density functional theory to estimate 18O/16O fractionation factors and qa–atoms.
Many of the same crystals are modeled here, and in general the agreement is remarkably
good, with a deviation in qquartz–atoms of about 1 × 10−4 (at 273–373 K, then converging at
higher temperatures), and deviations < 2 × 10–5 for qcalcite–atoms, qdolomite–atoms, qfluorapatite–atoms,
and θH2O vapor–atoms at 273 K and above. Calculated 18O/16O fractionations also generally agree,
with the largest differences found for calcite and quartz at low temperatures. The present
models predict ~1–1.5‰ higher ln18/16β for α-quartz over the 273–373 K temperature range,
and 1.5–2.8‰ lower ln18/16β for calcite. Dolomite, fluorapatite, and water vapor models all
agree within ~1‰ at temperatures ≥ 273 K. The overall magnitude of disagreement is similar
to that observed between PBE and PBEsol-based models. There is at most a small systematic
offset for the present PBE, periodic boundary condition models versus the Hayles et al. (2018)
B3LYP, cluster models; the mean (signed) deviation in ln18/16β for the set of quartz, calcite,
dolomite, fluorapatite, and water vapor is 0.1‰ or less above 273 K.
The silica–water system has been studied empirically (Sharp et al. 2016), as well as with
theoretical models, showing negative ∆′17O in SiO2 precipitates that is most pronounced in low
temperature silica samples. All of the published theoretical models agree with the general trend
of decreasing ∆′17OSiO2 and qSiO2–liquid water in low temperature samples. In detail, the models
appear to underestimate the deviation from the 0.528 exponent somewhat, particularly at the
lowest temperatures, only reaching as low as qSiO2–liquid water ≈ 0.524 vs. 0.523 and D′17OSiO2
≈ –150 to –160 per meg vs.−180 to –190 per meg near 273 K. Note that the deviations in
qSiO2–liquid water at high temperatures in some models may be misleading, because they are highly
sensitive to small changes in the 18O/16O fractionation factor and θ liquid water–atoms that are difficult
to resolve in measurements. The overall comparison suggests that these types of models will be
useful guides to the behavior of ∆′17O and qmineral water as a function of temperature, potentially
even in low temperature samples where it is not obvious that exchange equilibrium is obtained.
The 17O systematics of the water of hydration of gypsum has become a focus of interest
in hydrological studies in arid climates. Gazquez et al. (2017) recommend qgypsum water–parent brine
= 0.5297 ± 0.0012. In a set of re-hydration experiments equilibrated at 21 ºC, Herwartz et
al. (2017) find qgypsum water–parent brine = 0.5272 ± 0.0019, and they recommend a compromise
value of 0.5286. Our models predict qgypsum water–liquid water ≈ 0.528 at 25 ºC, in good agreement
with measurements. Our calculated qgypsum water–liquid water is consistently about 0.001 lower than
the model calculations of Liu et al. (2019), though it is difficult to pinpoint the cause of the
difference, and both results are consistent with measurements, given their uncertainties.
A system of great potential interest in future work is CaCO3–liquid water, including both
calcite and aragonite polymorphs of calcium carbonate. Although a number of measurements
have been reported (e.g., Passey et al. 2014), it is difficult to convert these into a calibration vs.
temperature. Initial results for calibration studies are beginning to appear in the literature (e.g.,
Wostbrock et al. in press; Voarintsoa et al. 2020; Bergel et al. 2020). The theoretical models
of Hayles et al. (2018) and the present study are show excellent agreement; the adjusted,
CO32–(v)-based model of Cao and Liu predicts lower qcarbonate–liquid water than the other two
models, but their unadjusted CO32–-based model is closer to the more recent theoretical results.
The overall trend vs. temperature in qcarbonate–liquid water is similar in all models. The models are
also generally in agreement with low ∆′17O observed in carbonate minerals (e.g., Passey et
al. 2014), and in reasonable agreement with the qcarbonate–liquid water results of Wostbrock et al.
(2020) (0.525–0.526 at 0 ºC vs. 0.5251 in the present study, 0.5250 in Hayles et al. 2018, and
0.5253 in Guo and Zhou 2019). However, Voarinstsoa et al. (2020) and Bergel et al. (2020)
measured a somewhat lower exponent (qcarbonate–liquid water ≈ 0.520–0.523 at 283–308 K).
A final set of comparisons comes measurements of θ in higher-temperature systems,

including tabulations aimed at finding a universally applicable mass fractionation exponent for
silicate rocks (e.g., Rumble et al. 2007), including lunar samples (e.g., Young et al. 2016). As
pointed out by Cao and Liu (2011), typical values in the range of 0.525–0.528 found in these
compilations are broadly consistent with silicate–water fractionations at moderate–elevated
temperatures, coinciding for instance with qquartz–water at temperatures from ~350–850 K.
However, they are markedly at odds with the discovery of measurable ∆′17O of several per
meg in feldspars and quartz from high-temperature rocks, including lunar and terrestrial
anorthosites (Young et al. 2016). The characteristically low ∆′17O of feldspar-rich rocks, by
~10 per meg relative to related olivine and/or pyroxene-rich rocks, cannot be easily explained
in the framework of the Urey (1947) harmonic oscillator theory of stable isotope fractionation,
by nuclear field shift effects, or even by more exotic models such as the double-well potential.
Taken together, these comparisons indicate that theoretical models, even highly simplified
ones, provide a useful guide to actual 18O/16O fractionations and θ, though direct testing is still
limited to a few materials.
General properties of ∆′17O in crystals relative to water vapor. Cao and Liu (2011),
Dauphas and Schauble (2016), and Hayles et al. (2018) have shown that the variation in qa–atoms
with changing chemical structure is highly systematic, moving below the high-temperature
harmonic limit of ~0.5305 most strongly at low temperatures in materials with high force
constants and relatively high-mass bond partners for oxygen. High force constants and relatively
high-mass bond partners also correlate strongly with preferential incorporation of oxygen
with high 18O/16O, so it is not surprising to find correlation between 18O/16O fractionation
factors and deviations in ∆′17O in systems approaching equilibrium. The present results are
also consistent with these principles. For instance, the low mass of hydrogen means that the
mass fractionation exponent qwater vapor–atoms stays close to the high-temperature limit even at
Earth surface conditions, and this property largely carries over to liquid water and ice, which
will have 0–20 per meg higher ∆′17O than vapor at temperatures from 243 K (−30 ºC) up to the
critical point. This observation is important because many of the most promising geochemical
applications of high-precision ∆′17O measurements are in mineral–water systems.
Theoretical studies (e.g., Cao and Liu 2011; Hayles et al. 2018; and the present work) also find
that the range of ∆′17O between different mineral/mineral and mineral/molecule pairs is predicted
to decrease quickly at equilibrium as temperature increases towards hydrothermal, metamorphic,
and igneous conditions. However, the choice of reference exponent becomes relatively important at
elevated temperatures. At 773 K (500 ºC), the largest inter-phase ∆′17O will be less than 10 per meg
(e.g., 9 per meg for the forsterite–water vapor pair), relative to a 0.528 reference exponent, but
could be greater than 20 per meg if the high-temperature equilibrium limit exponent of 0.53052
is the reference (where CO is a fractionating phase). Using the 0.528 reference, the range only
slightly decreases to ~8 per meg at 1273 K (1000 ºC), but the range is less than 5 per meg at this
temperature if the 0.53052 high-temperature limiting exponent is used instead.
To illustrate these systematic behaviors, we have plotted the difference in ∆′17O vs.
103 ln18/16aa–water vapor for a sample of the crystal types studied. Although there is substantial
scatter, the systematic correlation in behavior of these two fractionation properties is clear
(Fig. 7). Within the silicate mineral class (and especially for anhydrous silicates), strong
18
O/16O fractionation is strongly correlated with negative ∆′17O, becoming most pronounced
in structures with the highest Si:O ratios, such as quartz and alkali feldspars. Interestingly,
hydrous Mg–Si–O–H silicates follow a similar trend of ∆′17O vs. α. Among the anhydrous
silicates studied, high Si:O corresponds to higher polymerization, whereas among the hydrous
silicates high Si:O corresponds to lower polymerization because of Al:Si substitution into
tetrahedral sites and increased OH concentrations in sheet silicates. The similar predicted
∆′17O of kaolinite and lizardite indicates that tetrahedrally coordinated cations in silicates mainly
a. b.
c. d.
e. f.
Figure 7. Calculated deviations of ∆′17O from water vapor at equilibrium from 243–1573 K. a) anhydrous
silicate minerals and molecules. b) hydrous silicate minerals, diaspore, and spinel. c) carbonates and nitratine.
d) phosphates. e) sulfates. f) CO2, CO, and sodium perchlorate. g) ice, liquid, and supercritical water. Unless
otherwise specified, intermediate tick marks are given at 273 K, 373 K, 473 K, 573 K and 673 K on panels
b–g. The same alpha-quartz data is shown in panels a–f, as a guide to the eye. The y-axis units are per meg.
control deviations from the reference fractionation exponent even though sites coordinated to
octahedral Al3+ have a significantly higher affinity for 18O (and 17O) than sites coordinated
to Mg2+. Divalent metal carbonates show the importance of strong, low-coordination
cation-oxygen bonds in controlling ∆′17O even more strikingly, and it is possible to draw
isothermal tie-lines at nearly constant ∆′17O from aragonite and calcite through magnesite,
spanning 10‰ in 18O/16O but only a few per meg in ∆′17O at 273–373 K. In these crystals the
strongest bonds are always internal to the carbonate group, and relative 18O/16O fractionation
is controlled by weaker X2+–O bonds characterized by low vibrational frequencies that do
not affect the exponent as strongly. This behavior suggests that details of cation chemistry in
mixed-composition carbonates may not be very important for interpreting ∆′17O signatures.
Nitratine, with a different trigonal oxyanion (NO3–) and even weaker inter-molecular Na+–O
bonds, closely resembles its crystallographic cousin calcite. Phosphates show a similar near-
invariance of ∆′17O versus crystal chemistry, as do sulfates (excepting the water of hydration
of gypsum). Silicate, phosphate, sulfate, and perchlorate crystals with the ZTetO4 structural
formula, including zircon, monazite, xenotime, barite, anhydrite, and NaClO4, all show similar
∆′17O vs. water vapor at a given temperature, relative to the 0.528 reference line, echoing the
similarity noted above for ZTriO3 carbonate and nitrate minerals. ∆′17O vs. 18O/16O fractionation
relationships for silicates, sulfates, phosphates, carbonates, nitrates, and perchlorates relative
to liquid (or supercritical) water and water vapor follow a characteristically concave-down
trajectory as temperature decreases. This characteristic relationship suggests that ∆′17O relative
to water could be crudely predicted even for crystals that have not be explicitly modeled, via
comparison to other minerals of similar type. For silicates, in cases where ln18/16amineral–water can
be reasonably well constrained, simply applying the ∆′17O vs 18/16aquartz–water relationship may
be sufficient to give a useful approximation to the actual system of interest.
There is no indication that the apparent ~10 per meg deviations between different silicate
phases found in lunar and terrestrial igneous rocks can be explained by equilibrium inter-mineral
fractionation (Fig. 8). As an illustration, we compare the results obtained here with the high-
temperature rock data in Figure 1. While the trend of lower ∆′17O for quartz and feldspar relative
to olivine and spinel is common to the data and the calculations, the magnitudes are entirely
different. The calculations predict quartz and feldspar ~1 to 3 per meg lower in ∆′17O than
olivine, at most, at magmatic temperatures while the data exhibit differences ten times larger.
7
973 K
6
1173 K
5
1373 K
4
1573 K
D'17 X-forsterite
2
a-quartz
1 973 K
Albite (low)
1573 K Anorthite (ordered)
0
Diopside
1573 K Spinel (normal)
-1
973 K Forsterite
-2
-1 0 1 2 3 4
1000 lna18 vs. forsterite
Figure 8. Calculated deviations of ∆′17O in quartz, albite, anorthite, diopside and spinel relative to fors-
terite (Mg2SiO4) at equilibrium at temperatures from 973.15 K (700 ºC) to 1573.15 K (1300 ºC), plotted
against each mineral’s equilibrium 18O/16O fractionation relative to forsterite. The y-axis units are per meg.
SUMMARY
ln β and θ (and by extension ∆′17O relationships at equilibrium) have been
18/16
estimated for a variety of silicate, phosphate, sulfate, and carbonate minerals, as well as for
representative nitrate, perchlorate, oxide, hydroxide, and ice crystals using first-principles
electronic structure models. The results are generally in good agreement with previous studies
of fractionation factors and mass-fractionation exponents, including both theoretical work and
measurements. The nuclear volume component of the field shift effect is shown have a minor
or insignificant influence on fractionation and θ. A reconnaissance exploration of fractionation

in an anharmonic, double-well potential does not find evidence for the generation of large
∆′17O effects, at least for the most chemically plausible potential shapes. None of the results
provide a convincing explanation for ~10 per meg ∆′17O signatures observed in polymerized
silicates in high-temperature terrestrial and lunar rock samples.
REFERENCES
Alecu IM, Zheng J, Zhao Y, Truhlar DG (2010) Computational thermochemistry: scale factor databases, scale factors
for vibrational frequencies obtained from electronic model chemistries. J Chem Theory Comput 6:2872–2887.
Note that the database generated for this reference has been subsequently updated (https://t1.chem.umn.edu/
freqscale/index.html), although the updates do not affect the conclusions of the present work
Aliatis I, Lambruschi E, Mantovani L, Bersani D, Andò S, Diego Gatta G, Gentile P, Salvioli-Mariani E, Prencipe
M, Tribaudino M, Lottici PP (2015) A comparison between ab initio calculated, measured Raman spectrum of
triclinic albite (NaAlSi3O8). J Raman Spectrosc 46:501–508
Allan DR, Angel RJ (1997) A high-pressure structural study of microcline (KAlSi3O8) to 7 GPa. Eur J Mineral 9:263–275
Angeli I, Marinova KP (2013) Table of experimental nuclear ground state charge radii: An update. At Data Nucl Data
Tables 99:69–95
Barone V (2005) Anharmonic vibrational properties by a fully automated second-order perturbative approach. J Chem
Phys 122:014108
Becke AD (1993) Density-functional thermochemistry: III. The role of exact exchange. J Chem Phys 98:5648–5652
Begun GM, Beall GW, Boatner LA, Gregor WJ (1981) Raman spectra of rare earth orthophosphates. J Raman
Spectrosc 11:273–278
Benedict WS, Gailar N, Plyler E K (1956) Rotation–vibration spectra of deuterated water vapor. J Chem Phys
24:1139–1165
Berenblut BJ, Dawson P, Wilkinson GR (1973) A comparison of the Raman spectra of anhydrite (CaSO4), gypsum
(CaSO4.2H2O). Spectrochim Acta 29A:29–36
Acta 275:36–47
Bergquist BA, Blum JD (2007) Mass-dependent and -independent fractionation of Hg isotopes by photoreduction in
aquatic systems. Science 318:417–420
Bertoluzza A, Monti P, Morelli MA, Battaglia MA (1981) A Raman and infrared spectroscopic study of compounds
characterized by strong hydrogen-bonds J Mol Struct 73:19–29
Bigeleisen J (1996) Nuclear size and shape effects in chemical reactions. Isotope chemistry of the heavy elements. J
Am Chem Soc 118:3676–3680
Bigeleisen J, Mayer MG (1947) Calculation of equilibrium constants for isotopic exchange reactions. J Chem Phys
15:261–267
Bish DL (1993) Rietveld refinement of the kaolinite structure at 1.5 K. Clays and Clay Minerals 41:738–744
Born M, Oppenheimer JR (1927) Zur Quantentheorie der Molekeln [translated: On the quantum theory of molecules]
Ann Phys 389:457–484
Born K, Huang K (1956) Dynamical Theory of Crystal Lattices. Oxford University, New York
Bostrom D (1987) Single-crystal X-ray diffraction studies of synthetic Ni–Mg olivine solid solutions. Am Mineral
72:965–972
Breecker DO, Sharp ZD (2007) A monazite oxygen isotope thermometer. Am Mineral 1561–1572
Brehat F, Wyncke B (1985) Analysis of the temperature-dependent infrared active lattice modes in the ordered phase
of sodium nitrate. J Phys C: Solid St Phys 18:4247–4259
Bunker PR (1970) The effect of the breakdown of the Born–Oppenheimer approximation on the determination of Be
and we for a diatomic molecule. J Mol Spectrosc 35:306–313
Busing W, Levy H (1958) A single crystal neutron diffraction study of diaspore, AlO(OH). Acta Crystallogr 11:798–803
Cano EJ, Sharp ZD, Shearer C K (2020) Distinct oxygen isotope compositions of the Earth and Moon. Nat Geosci
13:270–274
ocean island basalts. Acta Geochim 38:327–334
Carteret C, De La Pierre M, Dossot M, Pascale F, Erba A, Dovesi R (2013) The vibrational spectrum of CaCO3
aragonite: A combined experimental and quantum-mechanical investigation. J Chem Phys 138:014201
Castex J, Madon M (1995) Test of the vibrational modelling for the λ-type transitions: Application to the α–β quartz
transition. Phys Chem Miner 22:1–10
Catti M, Ferraris G, Hull S, Pavese A (1995) Static compression, H disorder in brucite, Mg(OH)2, to 11 GPa: a powder
neutron diffraction study. Phys Chem Miner 22:200–206
Chacko T, Cole DR, Horita J (2001) Equilibrium oxygen, hydrogen and carbon isotope fractionation factors applicable
to geologic systems. Rev Mineral Geochem 43:1–81
Chang SJ, Blake RE (2015) Precise calibration of equilibrium oxygen isotope fractionations between dissolved
phosphate and water from 3 to 37 ºC. Geochim Cosmochim Acta 150:314–329
Chaplot SL, Pintschovius L, Choudhury N, Mittal R (2006) Phonon dispersion relations, phase transitions, and
thermodynamic properties of ZrSiO4: Inelastic neutron scattering experiments, shell model, and first-principles
calculations. Phys Rev B 73:094308
Chiba H, Kusakabe M, Hirano S-I, Matsuo S, Somiya S (1981) Oxygen isotope fractionation factors between
anhydrite and water from 100 ºC to 550 ºC. Earth Planet Sci Lett 53:55–62
Chiba H, Chacko T, Clayton RN, Goldsmith JR (1989) Oxygen isotope fractionations involving diopside, forsterite,
magnetite, and calcite: application to geothermometry. Geochim Cosmochim Acta 53:2985–2995
Chopelas A, Hofmeister AM (1991) Vibrational spectroscopy of aluminate spinels at 1 atm and of MgAl2O4 to over
200kbar. Phys Chem Miner 18:279–293
Clayton RN, Kieffer SW (1991) Oxygen isotopic thermometer calibrations. In: Stable Isotope Geochemistry: A
Tribute to Samuel Epstein (eds. H.P. Taylor Jr. et al.). Geochem Soc Spec Publ vol. 3, p 3–10
Clayton RN, Grossman L, Mayeda T K (1973) A Component of Primitive Nuclear Composition in Carbonaceous
Meteorites. Science 182:485–488
Clayton RN, Goldsmith JR, Mayeda T K (1989) Oxygen isotope fractionation in quartz, albite anorthite and calcite.
Comodi P, Mellini M, Ungaretti L, Zanazzi PF (1991) Compressibility and high pressure structure refinement of
tremolite, pargasite and glaucophane. Eur J Mineral 3:485–499
Comodi P, Liu Y, Zanazzi P F, Montagnoli M (2001) Structural and vibrational behaviour of fluorapatite with pressure.
Part 1: in situ single-crystal X-ray diffraction investigation. Phys Chem Miner 28:219–224
Comodi P, Nazzareni S, Zanazzi PF, Speziale S (2008) High-pressure behavior of gypsum: A single-crystal X-ray
study. Am Mineral 93:1530–1537
Császár AG, Czakó G, Furtenbacher T, Tennyson J, Szalay V, Shirin SV, Zobov NF, Polyansky OL (2005) Water
vapor: On equilibrium structures of the water molecule. J Chem Phys 122:214305
Dal Corso A (2014) Pseudopotentials periodic table: From H to Pu. Comput Mater Sci 95:337–350
Dawson P, Hargreave MM, Wilkinson GR (1977) Polarized i.r. reflection, absorption and laser Raman studies on a
single crystal of BaSO4. Spectrochim Acta 33A:83–93
de Villiers JPR (1971) Crystal structures of aragonite, strontianite, and witherite. Am Mineral 56:758–767
Deines P (2003) A note on intra-elemental isotope effects and the interpretation of non-mass-dependent isotope
variations. Chem Geol 199:179–182
Dove MT, Heine V, Hammonds KD (1995) Rigid unit modes in framework silicates. Mineral Mag 59:629–639
Dunning TH Jr. (1989) Gaussian basis sets for use in correlated molecular calculations. I. The atoms boron through
neon and hydrogen. J Chem Phys 90:1007–1023
Dupuis R, Benoit M, Tuckerman ME, Méheut M (2017) Importance of a fully anharmonic treatment of equilibrium
isotope fractionation properties of dissolved ionic species as evidenced by Li+(aq). Acc Chem Res 50:1597–1605
Earl BL (2008) The harmonic oscillator with a gaussian perturbation: evaluation of the integrals and example
applications. J Chem Educ 85:453–457
Eckhardt R, Eggers D, Slutsky LJ (1970) Infrared spectrum of NaNO3-I. Longitudinal and transverse internal modes.
Spectrochim Acta 26A:2033–2049
Elcombe MM, Hulston JR (1975) Calculation of sulphur isotope fractionation between sphalerite and galena using
lattice dynamics. Earth Planet Sci Lett 28:172–180
Ellehoj MD, Steen-Larsen HC, Johnsen SJ, Madsen MB (2013) Ice–vapor equilibrium fractionation factor of
hydrogen and oxygen isotopes: Experimental investigations and implications for stable water isotope studies.
Farquhar J, Bao H, Thiemens M (2000) Atmospheric influence of Earth’s earliest sulfur cycle. Science 289:756–758
Farquhar J, Johnston DT, Wing BA, Habicht KS, Canfield DE, Airieau S, Thiemens MH (2003) Multiple sulphur
isotopic interpretations of biosynthetic pathways: implications for biological signatures in the sulphur isotope
record. Geobiology 1:27–36
Frisch MJ, Trucks GW, Schlegel HB, Scuseria GE, Robb MA, Cheeseman JR, Scalmani G, Barone V, Mennucci B,
Petersson GA, Nakatsuji H, Caricato M, Li X, Hratchian HP, Izmaylov AF, Bloino J, Zheng G, Sonnenberg
JL, Hada M, Ehara M, Toyota K, Fukuda R, Hasegawa J, Ishida M, Nakajima T, Honda Y, Kitao O, Nakai H,
Vreven T, Montgomery JA Jr, Peralta JE, Ogliaro F, Bearpark M, Heyd JJ, Brothers E, Kudin KN, Staroverov
VN, Keith T, Kobayashi R, Normand J, Raghavachari K, Rendell A, Burant JC, Iyengar SS, Tomasi J, Cossi M,
Rega N, Millam JM, Klene M, Knox JE, Cross JB, Bakken V, Adamo C, Jaramillo J, Gomperts R, Stratmann
RE, Yazyev O, Austin AJ, Cammi R, Pomelli C, Ochterski JW, Martin RL, Morokuma K, Zakrzewski VG, Voth
GA, Salvador P, Dannenberg JJ, Dapprich S, Daniels AD, Farkas O, Foresman JB, Ortiz JV, Cioslowski J, Fox
DJ (2013) Gaussian 09, Revision D.0,1 Gaussian, Inc., Wallingford CT
Fujii T, Moynier F, Albarède F (2009) The nuclear field shift effect in chemical exchange reactions. Chem Geol
267:139–156
Garlick GD (1966) Oxygen isotope fractionation in igneous rocks. Earth Planet Sci Lett 1:361–368
Garrity KF, Bennett JW, Rabe KM, Vanderbilt D (2014) Pseudopotentials for high-throughput DFT calculations.
Comput Mater Sci 81:446–452
Gázquez F, Evans NP, Hodell DA (2017) Precise, accurate isotope fractionation factors (a17O a18O, αD) for water and
CaSO4∙2H2O (gypsum). Geochim Cosmochim Acta 198:259–270
Geisler T, Popa K, Konings RJM (2016) Evidence for lattice strain, non-ideal behavior in the (La1−xEux)PO4. Solid
solution from X-ray diffraction and vibrational spectroscopy. Front Earth Sci 4:1–12
Giannozzi P, Baroni S, Bonini N, Calandra M, Car R, Cavazzoni C, Ceresoli D, Chiarotti GL, Cococcioni M, Dabo I, Dal
Corso A, Fabris S, Fratesi G, de Gironcoli S, Gebauer R, Gerstmann U, Gougoussis C, Kokalj A, Lazzeri M, Martin-
Samos L, Marzari N, Mauri F, Mazzarello R, Paolini S, Pasquarello A, Paulatto L, Sbraccia C, Scandolo S, Sclauzero
G, Seitsonen AP, Smogunov A, Umari P, Wentzcovitch RM (2009) QUANTUM ESPRESSO: a modular and open-
source software project for quantum simulations of materials. J Phys: Condens Matter 21:395502
Giarola M, Sanson A, Rahman A, Mariotto G (2011) Vibrational dynamics of YPO4 and ScPO4 single crystals: An
integrated study by polarized Raman spectroscopy and first-principles calculations. Phys Rev B 83:224302
Gillan MG, Alfé D, Michaelides (2016) Perspective: How good is DFT for water? J Chem Phys 144:130901
Gillet P, Biellmann C, Reynard B, McMillan PF (1993) Raman spectroscopic studies of carbonates. Part I: High-pressure
and high-temperature behaviour of calcite, magnesite, dolomite, and aragonite. Phys Chem Miner 20:1–18
Graf DL (1961) Crystallographic tables for the rhombohedral carbonates. Am Mineral 46:1283–1316
Gregorkiewitz M, Lebech B, Mellini M, Viti C (1996) Hydrogen positions and thermal expansion in lizardite-1T from
Elba: A low-temperature study using Rietveld refinement of neutron diffraction data. Am Mineral 81:1111–1116
Grilly ER (1951) The vapor pressures of hydrogen, deuterium and tritium up to three atmospheres. J Am Chem Soc
73:843–846
Guo (2020) Kinetic clumped isotope fractionation in the DIC–H2O–CO2 system: Patterns, controls, and implications.
Halas P (2000) Empirical calibration of isotope thermometer d18O (SO42–)-d18O (H2O) for low temperature brines. V
Isotope Workshop. European Society for Isotope Research, Kraków, Poland, pp. 68–71
Hawthorne FC, Ferguson RB (1975) Anhydrous sulphates. II. Refinement of the crystal structure of anhydrite. Can
Mineral 13:289–292
Hazen RM, Finger LW (1979) Crystal structure and compressibility of zircon at high pressure. Am Mineral 64:196–201
Hazen RM, Finger LW, Hemley RJ, Mao H K (1989) High-pressure crystal chemistry, amorphization of alpha-quartz.
Solid State Commun 72 (1989) 507–511
molecular oxygen: The role of molecular symmetry in isotope chemistry. J Chem Phys 84:2129–2136
Hellwege KH, Lesch W, Plihal M, Schaack G (1970) Zwei-Phononen-Absorptionsspektren end dispersion der
Schwingungszweige in Kristallen der Kalkspatstruktur (Translated title: ‘‘Two phonon absorption spectra and
dispersion of phonon branches in crystals of calcite structure’’). Z Phys 232:61–86
Herwartz D, Surma J, Voigt C, Assonov S, Staubwasser M (2017) Triple oxygen isotope systematics of structurally
bonded water in gypsum. Geochim Cosmochim Acta 209:254–266
Herzberg, G (1966) Molecular Spectra and Molecular Structure III. Electronic Spectra and Electronic Structure of
Polyatomic Molecules; Van Nostrand: New York. p 598
Hill RJ, Craig JR, Gibbs GV (1979) Systematics of the spinel structure type. Phys Chem Miner 4:317–319
Hill PS, Schauble EA, Tripati A (2020) Theoretical constraints on the effects of added cations on clumped, oxygen, and carbon
isotope signatures of dissolved inorganic carbon species and minerals. Geochim Cosmochim Acta 269:496–539
Hofmeister AM, Bowey JE (2006) Quantitative infrared spectra of hydro-silicates and related minerals. Monthly
Notices of the Royal Astronomical Society 367:577–591
Holzwarth NAW, Tackett AR, Matthews GE (2001) A Projector Augmented Wave (PAW) code for electronic structure
calculations, Part I: atompaw for generating atom-centered functions. Computer Phys Commun 135:329–347
Horita J (2014) Oxygen and carbon isotope fractionation in the system dolomite–water–CO2 to elevated temperatures.
Horita J, Wesolowski DJ (1994) Liquid–vapor fractionation of oxygen and hydrogen isotopes of water from the
Hu G, Clayton RN (2003) Oxygen isotope salt effects at high pressure and high temperature and the calibration of
oxygen isotope geothermometers. Geochim Cosmochim Acta 67:3227–3246
Hulston JR, Thode HG (1965) Variations in the S33, S34, and S36 contents of meteorites and their relation to chemical
and nuclear effects, J Geophys Res 70:3475–3484
Iishi K (1979) Phonon spectroscopy and lattice dynamical calculations of anhydrite and gypsum. Phys Chem Miner
4:341–359
Jacobsen SD, Smyth JR, Swope RJ, Downs RT (1998) Rigid-body character of the SO4 groups in celestine, anglesite
and barite. Can Mineral 36:1053–1060
Jayasooriya UA, Kettle SFA, Mahasuverachai S, White RP (1996) Lattice vibrations in compounds crystallising with
the baryte structure. J Chem Soc, Faraday Trans 92:3625–3628
Jensen HJA et al. (2015) DIRAC, a relativistic ab initio electronic structure program, Release DIRAC15 (see http://
www.diracprogram.org)
Jollet F, Torrent M, Holzwarth N (2014) Generation of projector augmented-wave atomic data: A 71 element validated
table in the XML format. Comput Phys Commun 185:1246–1254
Kendall RA, Dunning TH, Harrison RJ (1992) Electron affinities of the first-row atoms revisited. Systematic basis sets
and wave functions. J Chem Phys 96:6976–6806
Kesharwani M K, Brauer B, Martin JML (2014) Frequency, zero-point vibrational energy scale factors for double-
hybrid density functionals (and other selected methods): Can anharmonic force fields be avoided? J Phys Chem
A 119:1701–1714
Kieffer SW (1982) Thermodynamics and lattice vibrations of minerals: 5. Applications to phase equilibria, isotopic
fractionation, and high-pressure thermodynamic properties. Rev Geophys Space Phys 20:827–849
Kim S-T, O’Neil JR (1997) Equilibrium and nonequilibrium oxygen isotope effects in synthetic carbonates. Geochim
King WH (1984) Isotope Shifts in Atomic Spectra (Physics of Atoms and Molecules). Plenum, New York
King EM, Valley JW, Davis DW, Kowallis BJ (2001) Empirical determination of oxygen isotope fractionation factors
for titanite with respect to zircon and quartz. Geochim Cosmochim Acta 65:3165–3175
Kleinman LI, Wolfsberg M (1973) Corrections to the Born–Oppenheimer approximation and electronic effects on
isotopic exchange equilibria. J Chem Phys. 59:2043–2053
Knight KS, Price GD (2008) Powder neutron-diffraction studies of clinopyroxenes. I. The crystal structure and
thermoelastic properties of jadeite between 1.5 and 270 K. Can Mineral 46:1593–1622
Knyazev DA, Myasoedov NF (2001) Specific effects of heavy nuclei in chemical equilibirum. Sep Sci Technol
36:1677–1696
Kolesov BA, Geiger CA (2004) A Raman spectroscopic study of Fe-Mg olivines. Phys Chem Miner 31:142–154
Kolodny Y, Luz B, Navon O (1983) Oxygen isotope variations in phosphate of biogenic apatites. I. Fish bone apatite
—rechecking the rules of the game. Earth Planet Sci Lett 64:398–404
Kusakabe M, Robinson BW (1977) Oxygen and sulfur isotope equilibria in the BaSO4-HSO4–-H2O system from 110
to 350 ºC and applications. Geochim Cosmochim Acta 41:1033–1040
Leadbetter AJ, Ward RC, Clark JW, Tucker PA, Matsuo T, Suga H (1985) Ice XI: The equilibrium low-temperature
structure of ice. J Chem Phys 82:424
Lefebvre J, Currat R, Fouret R, More M (1980) Etude par diffusion neutronique des vibrations de reseau dans le nitrate
de sodium. J Phys C: Solid Stat Phys. 13:4449–4461
Lejaeghere K et al. (2016) Reproducibility in density functional theory calculations of solids. Science 351:1394–1395
Leroy G, Leroy N, Penel G, Rey C, Lafforgue P, Bres E (2000) Polarized micro-Raman study of fluorapatite single
crystals. Appl Spectrosc 54:1521–1527
Lin C- K, Chang H-C, Lin SH (2007) Symmetric double-well potential model and its application to vibronic spectra:
Studies of inversion modes of ammonia and nitrogen-vacancy defect centers in diamond. J Phys Chem A
111:9347–9354
Liu Q, Tossell JA, Liu Y (2010) On the proper use of the Bigeleisen–Mayer equation and corrections to it in the
calculation of isotopic fractionation equilibrium constants. Geochim Cosmochim Acta 74:6965–6983
Liu T, Artacho E, Gázquez F, Walters G, Hodell D (2019) Prediction of equilibrium isotopic fractionation of the
gypsum/bassanite/water system using first-principles calculations. Geochim Cosmochim Acta 244:1–11
Longinelli A, Nuti S (1973) Revised phosphate–water isotopic temperature scale. Earth Planet Sci Lett 19:373–376
Lutz HD, Becker RA, Kruska BG, Berthold HJ (1979) Raman-, IR- und FIR-Messungen an wasserfreiem
Natriumperchlorat NaClO4 im Temperaturbereich zwischen 90 und 600 K. Spectrochim Acta 35A:797–806
Majoube M (1970) Fractionation factor of 18O between water vapour and ice. Nature 226:1242
Mantz AW, Maillard JP (1975) Ground state molecular constants of 12C16O, J Mol Spectrosc 57:155–159
Maschio L, Demichelis R, Orlando R, De La Pierre M, Mahmoud A, Dovesi R (2014) The Raman spectrum of grossular
garnet: a quantum mechanical simulation of wavenumbers and intensities. J Raman Spectrosc 45:710–715
Matsuhisa Y, Goldsmith JR, Clayton RN (1978) Mechanisms of hydrothermal crystallization of quartz at 250 ºC and
Mauersberger K (1987) Ozone isotope measurements in the stratosphere. Geophys Res Lett 14:80–83
McKeegan KD, Kallio APA, Heber VS, Jarzebinski G, Mao PH, Coath CD, Kunihiro T, Wiens RC, Nordholt JE,
Moses RW Jr., Reisenfeld DB, Jurewicz AJG, Burnett DS (2011) The oxygen isotopic composition of the sun
inferred from captured solar wind. Science 332:1528–1532
Méheut M, Schauble EA (2014) Silicon isotope fractionation in silicate minerals: Insights from first-principles models
of phyllosilicates, albite and pyrope. Geochim Cosmochim Acta 134:137–154
Méheut M, Lazzeri M, Balan E, Mauri F (2007) Equilibrium isotopic fractionation in the kaolinite, quartz, water
system: prediction from first-principles density-functional theory. Geochim Cosmochim Acta 71:3170–3181
Merli M, Ungaretti L, Oberti R (2000) Leverage analysis and structure refinement of minerals. Am Mineral 85:532–542
Merrick JP, Moran D, Radom L (2007) An evaluation of harmonic vibrational frequency scale factors. J Phys Chem
A 111:11683–11700
Miska H, Norum B, Hynes MV, Bertozzi W, Kowalski S, Rad FN, Sargent CP, Sasanuma T (1979) Precise
measurement of the charge-distribution differences of the oxygen isotopes. Phys Lett B 83:165–168
Mittal R, Chaplot SL, Choudhury N, Loong C-K (2007) Inelastic neutron scattering, lattice dynamics and high-
pressure phase stability in LuPO4 and YbPO4. J Phys Condens Matter 19:446202
Møller C, Plesset, MS (1934) Note on an Approximation Treatment for Many-Electron Systems. Phys Rev 46:618–622
Ni Y, Hughes JM, Mariano AN (1995) Crystal chemistry of the monazite and xenotime structures. Am Mineral 80:21–26
Nomura M, Higuchi N, Fujii Y (1996) Mass dependence of uranium isotope effects in the U(IV)–U(VI) exchange
reaction. J Am Chem Soc 118:9127–9130
Northrop DA, Clayton RN (1966) Oxygen-isotope fractionation in systems containing dolomite. J Geol 74:174–196
O’Neil JR, Clayton RN, Mayeda T K (1969) Oxygen isotope fractionation in divalent metal carbonates. J Chem Phys
51:5547–5558
Pack A, Herwartz D (2014) The triple oxygen isotope composition of the Earth mantle and understanding δ17O
Carlos olivine on the VSMOW2-SLAP2 scale. Rapid Commun Mass Spectrom 30, 1495–1504
Paul GL, Pryor AW (1972) The study of sodium nitrate by neutron diffraction. Acta Crystallographica B 28:2700–2702
Pavese A, Catti M, Price GD, Jackson RA (1992) Interatomic potentials for CaCO3 polymorphs (calcite and aragonite),
fitted to elastic and vibrational data. Phys Chem Miner 19:80–87
Perdew JP, Zunger A (1981) Self-interaction correction to density-functional approximations for many-electron
systems. Phys Rev B 23:5048–5079
Perdew JP, Burke K, Ernzerhof M (1996) Generalized gradient approximation made simple. Phys Rev Lett 77:3865–3868
Perdew JP, Ruzsinszky A, Csonka GI, Vydrov OA, Scuseria GE, Constantin LA, Zhou X, Burke K (2007) Restoring
the density-gradient expansion for exchange in solids and surfaces. Phys Rev Lett 100:136406
Pilati T, Demartin F, Gramaccioloi CM (1998) Lattice-dynamical estimation of atomic displacement parameters in carbonates:
calcite, aragonite CaCO3, dolomite CaMg(CO3)2 and magnesite MgCO3. Acta Crystallogr B 54:515–523
Pinella C, Blanchard M, Balan E, Natarajan S K, Vuilleumier R, Mauri F (2015) Equilibrium magnesium isotope
fractionation between aqueous Mg2+ and carbonate minerals: Insights from path integral molecular dynamics.
Geochim Cosmochim. Acta 163:126–139
Prandini G, Marrazzo A, Castelli IE, Mounet N, Marzari N (2018) Precision and efficiency in solid-state
pseudopotential calculations. npj Comput Mater 4:72
Prencipe M, Mantovani L, Tribaudino M, Bersani D, Lottici PP (2012) The Raman spectrum of diopside: a comparison
between ab initio calculated and experimentally measured frequencies. Eur J Mineral 24:457–464
Prencipe M, Maschio L, Kirtman B, Salustro S, Erba A, Dovesi R (2014) Raman spectrum of NaAlSi2O6 jadeite. A
quantum mechanical simulation. J Raman Spectrosc 45:703–709
Pucéat E, Joachimski MM, Bouilloux A, Monna F, Bonin A, Motreuil S, Moriniere P, Hénard S, Mourin J, Dera G,
Quesne D (2010) Revised phosphate–water fractionation equation reassessing paleotemperatures derived from
biogenic apatite. Earth Planet Sci Lett 298:135–142
Redhammer GJ (1998) Mössbauer spectroscopy, Rietveld refinement on synthetic ferri-Tschermak’s molecule
CaFe3+(Fe3+Si)O6 substituted diopside. Eur J Mineral 10:439–452
Rodehorst U, Geiger CA, Armbruster T (2002) The crystal structures of grossular, spessartine between 100 and 600 K
and the crystal chemistry of grossular–spessartine solid solutions. Am Mineral 87 (2002) 542–549
Rosenbaum JM (1997) Gaseous, liquid, and supercritical fluid H2O and CO2: oxygen isotope fractionation behavior.
Ross NL (1997) The equation of state and high-pressure behaviour of magnesite. Am Mineral 82:682–688
Ross NL, Reeder RJ (1992) High-pressure structural study of dolomite and ankerite. Am Mineral 77:412–421
Rutt HN, Nicola JH (1974) Raman spectra of carbonates of calcite structure. J Phys C 7:4522–4528
Sass RL, Scheuerman RF (1962) The crystal structure of sodium bicarbonate. Acta Crystallogr 15:77–81
Sato R K, McMillan PF (1987) An infrared and Raman study of the isotopic species of alpha-quartz. J Phys Chem
91:3494–3498
Schauble EA (2007) Role of nuclear volume in driving equilibrium stable isotope fractionation of mercury, thallium,
and other very heavy elements. Geochim Cosmochim Acta 71:2170–2189
Schauble EA (2011) First-principles estimates of equilibrium magnesium isotope fractionation in silicate, oxide,
carbonate, hexaaquamagnesium(2+) crystals. Geochim Cosmochim Acta 75:844–869
Schauble EA (2013) Modeling nuclear volume isotope effects in crystals. PNAS 110:17714–17719
Schauble EA, Rossman GR, Taylor HP Jr. (2003) Theoretical estimates of equilibrium chlorine-isotope fractionations.
Schauble EA, Ghosh P, Eiler JM (2006) Preferential formation of 13C–18O bonds in carbonate minerals, estimated
using first-principles lattice dynamics. Geochim Cosmochim Acta 70:2510–2529
Sharma BD (1965) Sodium bicarbonate and its hydrogen atom. Acta Crystallogr 18:818–819
Sharp ZD, Kirschner DL (1994) Quartz–calcite oxygen isotope thermometry: a calibration based on natural isotopic
variations. Geochim Cosmochim Acta 58:4491–4501
Sheppard SMF, Schwarcz HP (1970) Fractionation of carbon and oxygen isotopes and magnesium between coexisting
metamorphic calcite and dolomite. Contrib Mineral Petrol 26:161–198
Singhal RP, Moreira JR, Caplan HS (1970) RMS charge radii of 16,17,18O by elastic electron scattering. Phys Rev Lett 24:73–75
Smith JV, Artioli G (1986) Low albite, NaAlSi3O8: Neutron diffraction study of crystal structure at 13 K. Am Mineral
71:727–733
Syme RWG, Lockwood DJ, Kerr HJ (1977) Raman spectrum of synthetic zircon (ZrSiO4), thorite (ThSiO4). J Phys
C: Solid State Phys 10:1335–1348
Tachikawa M, Shiga M (2005) Geometrical H/D isotope effect on hydrogen bonds in charged water clusters. J Am
Chem Soc 127:11908–11909
Topsakal M, Wentzcovitch RM (2014) Accurate projected augmented wave (PAW) datasets for rare-earth elements
(RE = La–Lu). Comput Mater Sci 95:263–270
Toupry-Krauzman N, Poulet H (1978) Temperature dependence of the Raman spectra of NaClO4, in relation to the
581 K phase transition. J Raman Spectrosc 7:1–6
Trail D, Bindeman IN, Watson EB, Schmitt A K (2009). Experimental calibration of oxygen isotope fractionation
between quartz and zircon. Geochim Cosmochim Acta 73:7110–7126
Urey HC (1947) The thermodynamic properties of isotopic substances. J Chem Soc (London) 562–581
Valley JW, Bindeman IN, Peck WH (2003). Empirical calibration of oxygen isotope fractionation in zircon. Geochim
Venneman TW, Hegner E, Cliff G, Benz GW (2001) Isotopic composition of recent shark teeth as a proxy for
environmental conditions. Geochim Cosmochim Acta 65:1583–1599
Wainwright J, Starkey J (1971) A refinement of the structure of anorthite. Z Kristallogr 133:75–84
Wartchow R, Berthold HJ (1978) Refinement of the crystal structure of sodium perchlorate (NaClO4). Z Krystallogr
147:307–17
Webb MA, Miller TF III (2014) Position-specific and clumped stable isotope studies: Comparison of the urey and
path-integral approaches for carbon dioxide, nitrous oxide, methane, and propane. J Phys Chem A 118:467–474
Widanagamage IH, Schauble EA, Scher HD, Griffith EM (2014) Stable strontium isotope fractionation in synthetic
barite. Geochim Cosmochim Acta 147:58–75
Wiechert U, Halliday AN, Lee D-C, Snyder GA, Taylor LA, Rumble D (2001) Oxygen Isotopes and the Moon-
Forming Giant Impact. Science 294:345–348
Wostbrock JAG, Brand U, Coplen TB, Swart P K, Carlson SJ, Brearley A, Sharp ZD (2020) Calibration of carbonate–
water triple oxygen isotope fractionation: seeing through diagenesis in ancient carbonates. Geochim Cosmochim
Acta 288:369-388
Yagi K, Hirao K, Taketsugu T, Schmidt M, Gordon MS (2004) Ab initio vibrational state calculations with a quartic
force field: applications to H2CO, C2H4, CH3OH, CH3CCH, and C6H6. J Chem Phys 121:1383–1389
Yamamoto A, Utida T, Murata H (1976) Coulomb interactions and optically-active vibrations of ionic crystals–I.
Theory and applications to NaNO3. J Phys Chem Solids 37:693–698
Zeebe RE (2010) A new value for the stable oxygen isotope fractionation between dissolved sulfate ion and water.
Zhang M, Wruck B, Graeme Barber A, Salje E KH, Carpenter MA (1996) Phonon spectra of alkali feldspars: Phase
transitions and solid solutions. Am Mineral 81:92–104
Zhang Y, Liu Y (2018) The theory of equilibrium isotope fractionations for gaseous molecules under super-cold
conditions. Geochim Cosmochim Acta 238:123–149
Zheng Y-F (1991) Calculation of oxygen isotope fractionation in anhydrous silicate minerals. Geochim Cosmochim
Acta 57:1079–1091
Zúñiga J, Bastida A, Alacid M, Requena A (2001) Variational calculations of rovibrational energies for CO2. J Mol
Spectrosc 205:62–72
Vol. 86 pp. 179–196, 2021 5
Standardization for the Triple Oxygen Isotope System:

Waters, Silicates, Carbonates, Air, and Sulfates
Zachary D. Sharp, Jordan A.G. Wostbrock
Department of Earth and Planetary Sciences
University of New Mexico
Albuquerque, NM, 87131
USA
zsharp@unm.edu
INTRODUCTION
Stable isotope analyses are a relative measurement. The precision is far higher than the
accuracy, so that subtle isotopic differences must be made relative to a reference. Modern mass
spectrometers can routinely measure the δ18O value of a gas with a precision of 0.01‰. This is
20 times more precise than the accuracy of the 18O/16O ratio of VSMOW (Baertschi 1976). It
is for this reason that isotope analyses, like most analytical measurements, are reported relative
to standards. The problem faced by the stable isotope community is that different materials
are measured using different techniques, and direct comparison between them is difficult.
Heroic efforts have been made to align the different types of analyses to the same scale, so
that data collected on different materials in different laboratories can be directly compared.
For traditional δ18O analyses, coalescence around common standards took decades. Triple
oxygen isotope studies (δ18O and d17O) of terrestrial materials is a relatively new discipline,
and agreement on standardization is only recently achieving a high level of conformity. In this
chapter, we first consider the historical path towards standardization for the well-established
18
O/16O ratios. Then the extension to standardization to 17O/16O is discussed with a goal of
presenting a uniform set of standard values for commonly used reference materials.
There was a time when stable isotope standards didn’t exist. Consider the classic paper
by Dole (1936), in which he determined the oxygen isotope composition of air relative to
water. “The atomic weight of oxygen in air is 0.000108 atomic weight units heavier than Lake
Michigan water”. This was a high precision analysis, considering that it was determined by
density measurements, but it was far lower precision than what is achievable using modern
extraction methods and modern mass spectrometers. Precise determinations of relative isotopic
differences only began in earnest in the 1950s following the development of a dual inlet mass
spectrometer (McKinney et al. 1950) under the direction of Harold Urey. Urey’s desire to
develop a carbonate paleothermometer necessitated highly precise oxygen isotope analyses in
order to obtain temperature estimates in the 1 °C range. This led to the University of Chicago
group creating what is essentially the modern high precision mass spectrometer (McKinney et
al. 1950). McKinney’s mass spectrometer measured the ratios of the voltages of 18O/16O of a
sample gas relative to a reference gas using a set of changeover valves that allowed for rapid
switching between the two gases. This reduced errors associated with the inevitable drift of
the sensitive electronics. The 18O/16O ratios were presented by McKinney in what is now the
ubiquitous delta notation in per mil units as
R 

18
O  sa  1 1000 (1)
R
 std 

1943-2666/21/0086-0005$05.00 (online)
180 Sharp & Wostbrock
In more general terms, delta is given by
Rsa
 18
O 1 (2)
Rstd
and indeed, early hydrogen isotope measurements were given in % (e.g., Friedman 1953)
rather than per mil and recent publications have presented Δ′17O values in parts per million
(ppm or per meg). In this contribution, all additional equations (e.g., Eqn. 7) and delta values
are given in terms of Equation (1).
While it is generally assumed that the first measurements of oxygen isotope ratios
were made on CO2 gas, interestingly, McKinney’s original mass spectrometer was used for
measuring both the isotope ratios of CO2 (the ratio C16O18O/C16O2) and O2 (16O18O/16O2) gas.
Today, O2 gas is the primary analyte for triple oxygen isotope analyses in order to be able to
measure the d17O value. It is perhaps surprising to see that, from the very beginning, O2 gas
was considered as a viable gas for the classic dual-inlet mass spectrometer.
Some of the first high-precision measurements were made by Epstein et al. (1951,
1953) on natural waters and carbonates, which led to the carbonate–water oxygen isotope
thermometer. Their paleotemperature equation is given as
t ( °C) = 16.5 − 4.3 δ + 0.14 δ2 (3)
In this equation δ is the difference in the δ18O value of CO2 (in per mil) produced by reaction of
carbonates with 100% phosphoric acid and the working standard CO2 gas. The reference gas
was CO2 produced by the reaction of Belemnitella americana from the Peedee formation of
South Carolina (Urey et al. 1951). They did not measure the δ18O values of either the water or
the carbonate—only the small difference between the CO2 gases obtained by decarbonating the
calcite and equilibrating CO2 with water, respectively. As long as their temperature equation
was calibrated to their own internal reference gas, they would encounter no problems.
Silverman (1951) measured the oxygen isotope composition of silicate rocks.
The reference gas he used in all analyses was oxygen derived from quartz from the Randville
pegmatite (Randville, Michigan, USA). In setting the standard for future measurements, he
adjusted all of his silicate analyses to Hawaii seawater, with an arbitrary δ18O value of 0‰.
(Hawaii seawater is 0.17‰ heavier than SMOW, Epstein and Mayeda 1953). Presumably
Silverman fluorinated his water sample to quantitatively extract oxygen (Baertschi 1950).
In those early years the different laboratories had good communication, and the standards were
commonly exchanged for cross-comparison. At the Caltech laboratory, for example, Potsdam
sandstone was the unofficial internal standard for years (Clayton 1959; Taylor and Epstein
1962) and was calibrated relative to Hawaiian water and, later, mean ocean water.
As the number of stable isotope laboratories grew, it became increasingly imperative to
develop a coordinated intercalibration so that the data from all labs could be compared with one
another. Harmon Craig published a detailed presentation of stable isotope standards, combining
existing standards from the University of Chicago laboratory with samples from the newly
created National Bureau of Standards, NBS (Craig 1957). These included the NBS Solnhofen
limestone and NBS 1 (water from the Potomac River, Maryland, Fig. 1). The oxygen isotope
values of carbonates were reported relative to the PDB carbonate standard, whereas waters were
reported relative to the NBS 1 water standard. Craig later defined the average ocean water (mean
ocean water of Epstein and Mayeda 1953) in terms of NBS 1, giving the relationship
 18 O   18 O 
 16   1.008   16  (4)
 O SMOW Standard Mean Ocean Water   O  NBS1
Standardization: Waters, Silicates, Carbonates, Air, and Sulfates 181
NBS 1 was distributed widely by the National Bureau of Standards and could be used for
interlaboratory calibration. Later, Harmon Craig collected, distilled and adjusted ocean water
at the request of the International Atomic Energy Agency (IAEA) to match the original SMOW
so that it conformed to the definition in terms of NBS 1 (Craig 1961). The new reference
material was called VSMOW (shortened from Vienna SMOW—Fig. 2) to distinguish it from
the original SMOW (Gonfiantini 1978). At the same time, a second light water standard,
SLAP (for Standard Light Antarctic Precipitation) was developed, allowing for laboratories
to have two different water standards with δ18O and δD values that were far apart. The δ18O
and δD values of VSMOW, by definition, are 0‰. Although not envisioned at the time, by
analogy the δ17O value is also 0‰. SLAP has a δ18O value of −55.5‰ and a δD value of
−428‰. There is some indication that the actual value of SMOW may be closer to −56.18‰
(Verkouteren and Klinedinst 2004). We will stick with the IAEA convention of −55.5‰ for
SLAP and SLAP2 (IAEA 2006, 2017). An intermediate standard GISP (Greenland Ice Sheet
Precipitation) has a δ18O value of −24.76‰ and a δD value of −189.5‰ (IAEA 2006). GISP is
now discontinued and has been replaced with GRESP with a δ18O value of −33.39‰ and δD
value of −257.8‰ (IAEA 2020) The absolute values for the 18O/16O and D/H ratios of VSMOW
are 2005.20 ± 0.45×106 and 155.76 ± 0.05×10−6 (see compilation in Table 2.5 of Sharp 2013).

)LJXUH Figure 2.
Figure 1. Unopened sealed vial of NBS 1. Figure 2. The SMOW-1 standard (VSMOW).
Why, if it is not to be moved, does it have handles?
[USGS Public Domain].
Two formally adopted carbonate standards followed (Friedman et al. 2007). The first,
NBS 19, essentially replaced PDB as the benchmark marine carbonate value, while the second,
NBS 18, is a carbonate of igneous origin which has a lower δ18O value, much closer to that of
the mantle. The δ18O and δ13C of NBS 19 are defined as δ18O ≡ −2.20‰ and δ13C ≡ +1.95‰
(where ≡ indicates that these values are exact by definition to the VPDB scale). NBS 19 is no
longer available and has been replaced by IAEA 603 with δ18O ≡ −2.37‰ and δ13C ≡ +2.46‰
(Assonov et al. 2020).
The relationship between VPDB and VSMOW is given by the equation (Kim et al. 2015)
δ18OVSMOW = 1.03092 δ18OVPDB + 30.92 (5)
updated from the long-used equation (Coplen et al. 1983)
δ18OVSMOW = 1.03091 δ18OVPDB + 30.91 (6)
As we will see, the difference between these two equations is significantly less than the total
uncertainties in the fractionation factors necessary to construct them.
Equation (5) allows a comparison of the δ18O values of waters reported to the VSMOW
scale with carbonates reported to the VPDB scale. Silicates are also anchored to the VSMOW
scale by the quartz sand reference material NBS 28, a sand donated by Corning Glass Co.
Friedman and Gleason (1973) introduced this standard and determined its δ18O value relative
to VSMOW by fluorinating both the quartz and VSMOW to quantitatively extract O2, which
was then converted to CO2 for isotopic analysis. Their measured δ18O value for NBS 28 was
10.0‰ ± 0.12 (2σ). Friedman and O’Neil (1977) presented a range of published δ18O values
for NBS 28 of 9.5 to 9.9‰ from different laboratories (the IAEA reference sheet for NBS 28
presents a literature range of 8.8 to 10.0‰). There is clearly uncertainty in the δ18O value of
NBS 28 relative to VSMOW. This illustrates one of the most fundamental problems facing the
triple oxygen isotope community with regards to standardization—intercalibrating reference
samples that are traditionally measured using different methods tied to different standards.
INTERCALIBRATING WATERS, CARBONATES AND SILICATES FOR d18O

In order to report the δ18O values of different materials relative to each other, it is critical
to have all types of phases calibrated to the same standard. The problem is that the methods
for analyzing carbonates, water, and silicates are completely different. Waters are generally
analyzed by equilibration with CO2 gas which is then measured on the mass spectrometer
(although laser spectroscopy methods are now commonly employed, e.g., Steig et al, 2014);
carbonates are reacted with phosphoric acid to produce CO2; silicates are fluorinated to
quantitatively produce O2. The results from these three methods are not directly comparable
and require a correction factor to bring them into agreement. As explained below, the errors in
these correction factors are far larger than the uncertainty of the analyses themselves.
Water
The standard method for analyzing the δ18O value of water was developed by the biochemist
and National Medal of Science recipient Mildred Cohn and Nobel laureate coauthor Harold
Urey over 80 years ago (1938), and is used today virtually unchanged. A trace amount of CO2
is added to the water sample. The mixture is held at a constant temperature (generally 25 °C)
for hours to days in order to allow the two phases to come into oxygen isotope equilibrium.
The CO2 is removed from the reaction vessel, purified and measured on the mass spectrometer.
Although there is a large fractionation between the water and CO2 gas, as long as there is a
large excess of H2O and the equilibration process is performed using identical protocol at
constant temperature, the isotope fractionation between the equilibrated water and CO2 gas
should be constant. Epstein and Mayeda (1953) compared the δ18O values of many natural
waters relative to the average of deep marine waters from the different oceans. All of their data
were therefore relative to MOW (mean ocean water). This ultimately led to the formation of
the SMOW and later VSMOW and still later VSMOW2 standards.
The isotopic fractionation between CO2 and water is given by
1000  18OCO2 (7)

 CO2  water 
1000  18O water
This is thought to be an equilibrium fractionation (unlike the acid digestion of carbonates—see

below). It therefore can be determined theoretically as well as experimentally. Theoretical
determinations give aCO2–water values of 1.042 (Urey and Greiff 1935), 1.039 (Urey 1947) and
1.0411 (Bottinga and Craig 1969). The difficulty in making this calculation is in the complexity
of structure of liquid water in comparison to simple diatomic gaseous species from a statistical
mechanical perspective (Cao and Liu 2011; Hayles et al. 2018).
Experimental determinations of the alpha value for Equation (7) require a quantitative
conversion of H2O to CO2 so that the δ18O value of the CO2 derived from the H2O can be
compared directly to the CO2 equilibrated with the water. At least nine experimental
determinations of the aCO2–water fractionation at 25 °C have been made using a variety of
different methods based on carbon reduction or fluorination (see Brenninkmeijer et al. 1983
for a summary). The aCO2–water values range from 1.0407 to 1.0424 (Fig. 3) and have a standard
deviation in the α values that corresponds to a 1σ uncertainty in the δ18O value of ~0.4‰.
The general consensus is that aCO2–water = 1.0412 (Friedman and O’Neil 1977; Kim et al. 2015).
1.6‰
3
2
n
1
0
1.0407
1.0409
1.0411
1.0413
1.0415
1.0417
1.0419
1.0421
1.0423
1.0425
1000lna (CO2-water)
Figure 3. Published estimates of experimental αCO2–water. From (Friedman and O’Neil 1977; Brenninkmei-
jer et al. 1983).
Carbonates
The method for analyzing carbonate, like water, is also an indirect measure. Developed
by John McCrea (1950) as part of Urey’s project to develop a paleotemperature scale, calcite
(and later other carbonates) is reacted with 100% phosphoric acid at a controlled temperature.
The reaction liberates CO2 gas, which is purified and analyzed on the mass spectrometer. As is
the case for CO2–water equilibration, the δ18O value of the evolved CO2 gas is not equal to that
of the actual carbonate. It is offset by the acid-liberated CO2–calcite fractionation. By reacting
all samples in controlled, repeatable conditions, the δ18O values of the CO2 from different
samples can be compared relative to each other.
Acid digestion of carbonates liberates only 2/3 of the total oxygen. The isotope
fractionation during this process is the aCO2(ACID)–carbonate value. It is worth pointing out that the
aCO2(ACID)–carbonate value is not the same as the aCO2–carbonate value. The latter is the equilibrium
oxygen isotope fractionation between CO2 gas and the carbonate, while the former is not
(Guo et al. 2009). The equilibrium α(18O/16O)CO2–calcite value is between 1.0014 and 1.0119
(O’Neil and Epstein 1966b; Bottinga 1968), over 1‰ greater than the average aCO2(ACID)–calcite
value. The aCO2(ACID)–carbonate value is therefore determined experimentally, and can only be
quantified and tied to the VSMOW scale if the δ18O value of the total carbonate is known.
This is determined by quantitatively extracting 100% of the oxygen from a sample by high
temperature fluorination (e.g., Sharma and Clayton 1965). The extracted O2 is then converted
to CO2 by high temperature reaction with a graphite rod. Comparison of this CO2 gas with the
CO2 liberated by reaction of the same carbonate by phosphoric acid gives the aCO2(ACID)–carbonate
value. There have been five independent determinations of this value (see Kim et al. 2015
Table 2 for compilation). They range from aCO2(ACID)–calcite = 1.01015 to 1.01058 (average is
1.01036 ± 0.00017), corresponding to an uncertainty of 0.18‰ (1σ). The generally accepted
aCO2(ACID)–calcite value at 25 °C is 1.01025 as determined by Sharma and Clayton (1965) and
amended by Friedman and O’Neil (1977).
Silicates
Silicates were first analyzed with high precision using the method of fluorination
(Baertschi and Silverman 1951; Silverman 1951). Unlike the techniques for carbonates and
water, the fluorination method is quantitative in the sense that 100% of the oxygen in the
silicate is released by the fluorination reaction. The early studies measured the 18O/16O ratios
of the O2 gas produced by the fluorination (Baertschi and Silverman 1951). Most laboratories
subsequently added a combustion step where O2 was quantitatively converted to CO2 by high
temperature reaction with spectroscopic graphite in the presence of a Pt catalyst (Taylor and
Epstein 1962). CO2 could be purified more easily and moved throughout a vacuum line by
freezing with liquid nitrogen. It is ironic that, after more than a half a century of analyzing
silicate oxygen in the mass spectrometer using CO2 as the analyte, we have returned to
analyzing O2 gas in order to also obtain the 17O/16O ratios of samples. There are two IAEA
silicate standards: NBS 28 (quartz) with a defined δ18O value of 9.57‰ and NBS 30 (biotite)
with a δ18O value of 5.12‰ (https://nucleus.iaea.org/rpst/Documents/NBS28_NBS30.pdf).
There are 16 tabulated δ18O values for NBS 28 in the abovementioned IAEA report with
a standard deviation of 0.3‰ (1σ). Two recent studies have measured the δ18O values of
both NBS 28 and VSMOW. The results are 9.56‰ (Tanaka and Nakamura 2013) and 9.58‰
(Wostbrock et al. 2020a) on the VSMOW scale, in excellent agreement with the recommended
IAEA value. Note: NBS 30 is no longer distributed from IAEA.
Comparison of all data
The crucial fractionation factors that are required to place waters, carbonates and silicates
on the same scale are the equilibrium aCO2–water and kinetic aCO2(ACID)–calcite fractionation factors
(Fig. 4). From the above discussion it is clear that there is a significant degree of uncertainty
when comparing waters to carbonates to silicates. Only silicates (and air O2) are routinely
determined quantitatively. The accuracy of carbonates and silicates on the VSMOW scale
(relative to water) is dependent on the accuracy of the alpha factors and overall translates to
an uncertainty as high as 0.5‰. This number may be slightly lower if we cull some of the
experimental fractionation values that have been published. The most widely accepted values
are aCO2–water = 1.0412 and for aCO2(ACID)–calcite =1.01025 (at 25 °C). These values may be refined
by future researchers, but at present, the techniques available for analysis are essentially
the same that were used in previous studies. Thus the ~0.5‰ uncertainty when comparing
carbonates to waters to silicates remains.
Acid liberated
CO2 equil. CO2 from PDB
with SMOW (41.49±0.2)
(41.2±0.4)
a=1.01025
PDB
(30.92)
a=1.0412
a=1.03092
NBS 28
(9.6±>0.1)
SMOW (≡0)
Figure 4. Fractionation factors and relative effects for comparing waters to carbonates to silicates. Uncer-
tainties discussed in text.
STANDARDIZATION FOR d17O VALUES OF

SELECTED REFERENCE MATERIALS
All of the same problems and obstacles for calibrating standards for 18O/16O exist for
17
O/16O as well, with several additional complications. Proper standardization for 17O/16O
ratios is particularly critical because the differences that are measured in natural materials are
so small. While the δ18O value of a material can be measured on either CO2, CO gas or O2 gas,
δ17O values generally must be measured on O2 gas. Converting samples quantitatively to O2
gas can be extremely challenging, as in the case of carbonates.
The d17O and δ18O values of most terrestrial (Earth) materials co-vary, such that they
follow the relationship d17O ≈ 0.52x × δ18O (see Miller and Pack 2021, this volume for a more
detailed discussion). Plotting all Earth materials on a d17O–δ18O plot results in a linear trend
that is often called the ‘Terrestrial Fractionation Line’ or TFL, especially in the meteoritic
literature (e.g., McKeegan and Leshin 2001). The importance of measuring the d17O values
of terrestrial materials becomes relevant only when we determine small deviations from some
reference line that is representative of natural materials. This offset is the Δ′17O value, where
Δ′17O = δ′17O – lRL δ′18O – γRL (8)
Here lRL is the assigned slope of a reference line (hence the RL subscript) and γRL is the
y intercept. l is commonly chosen to be 0.528 with γ = 0. The δ′ notation refers to the
linearized form of δ, such that d′ = 1000 × ln(δ/1000+1) in per mil notation (Hulston and Thode
1965; Miller 2002). The advantage of the linearized notation is that the d′17O versus δ′18O
relationship is close to linear, unlike d17O versus δ18O. The precision of Δ′17O measurements is
on the order of ±0.005‰, when appropriate protocols are followed. These include extreme gas
purification and long counting times on the mass spectrometer. This precision is significantly
higher than what can be obtained for either the δ18O or δ17O value. The higher precision is
obtained because the errors in the δ17O and δ18O values covary and therefore tend to cancel
each other out (see appendix of Wostbrock et al. 2018 for details).
Method of measurement
In conventional mass spectrometry, δ18O values can be measured on O2, CO2 or even CO
gas. The δ17O value, however, must be determined using O2 as an analyte, where the [33]/[32]
ratio (17O16O/16O16O) is measured. CO2 cannot be used because of the interference from 13C.
Consider that 45/44 is both 12C17O16O/12C16O16O and 13C16O16O/12C16O16O. The mass resolution
necessary to separate 12C17O16O from 13C16O16O is over 50,000, far above that attainable with
any commercially available gas-source mass spectrometer. There are several other analytical
options in place of the conventional mass spectrometer. The δ17O value of H2Ovapor is now
routinely measured using cavity ringdown spectroscopy (Steig et al. 2014) with precision that
rivals conventional mass spectrometry measurements. Promising preliminary data have also been
made measuring the d17O value of CO2 gas in a laser spectroscopy system (Sakai et al. 2017;
Stoltmann et al. 2017). A different approach is to measure the O+ fragment of CO2 produced in
the source of an electron impact isotope ratio mass spectrometer (Adnew et al. 2019). Extremely
high mass resolution (4700) is required to separate the 17O+ fragment (16.9991 amu) from the
ubiquitous OH+ ion (17.0027 amu). This high mass resolution requires an expensive doubly
focusing mass spectrometer (e.g. Thermo Fisher 253 Ultra) and has significantly reduced
transmission, necessitating counting times in excess of 20 hours for high precision.
Water
Ultimately all samples are related to VSMOW, using a stretching factor based on the
difference between VSMOW and SLAP2. Therefore, it is critical to calibrate the working
gas of a mass spectrometer and ultimately all materials to the VSMOW and SLAP2 scale.
Water was first analyzed for the d17O value using electrolysis (Meijer and Li 1998) to produce
the O2 from the water samples, which was then analyzed by conventional mass spectrometry.
These authors recognized the relationship d17O = (1 + δ18O)l − 1 (l = 0.5281 ± 0.015) by
analyzing waters of very different isotope compositions, but their precision for Δ′17O was far
lower than what would be attainable with fluorination systems.
Precise measurements for triple isotope analyses of water are made by either fluorination
of water to O2, followed by analysis on a conventional dual inlet mass spectrometer, or, in
recent years, by direct analysis of water vapor using either laser absorption cavity ringdown
spectroscopy (Steig et al. 2014) or off-axis integrated cavity output spectroscopy (Tian et
al. 2016). The latter gives precise analyses, but is not quantitative, and generally requires a
correction due to the machine-related non-linearities. Fluorination methods are, in principle,
quantitative, as 100% of the oxygen is converted to O2. Fluorination is accomplished either by
passage of water through a CoF3 heated reaction chamber in a He flow (Barkan and Luz 2005;
Schoenemann et al. 2013) or by reaction with BrF5 in a steel or nickel reaction vessel (O’Neil
and Epstein 1966a; Jabeen and Kusakabe 1997; Pack et al. 2016; Wostbrock et al. 2020a).
The data of Barkan and Luz (2005) may suffer from a slight D18O compression observed in
the CoF3 method (Schoenemann et al. 2013; Passey et al. 2014). Figure 5 shows all published
estimates for SLAP2 relative to VSMOW using the methods of fluorination. The vertical black
line represents the accepted δ18O value of SLAP2 of −55.5‰. Three analyses are close to
this value, with Δ′17O values averaging −0.011 ± 0.003‰ (1σ). All other studies report δ18O
values that are higher than the accepted value. It has been shown that analysis of light waters
suffers from a memory effect, and that the lightest (and presumably correct) values are only
obtained after multiple injections (Barkan and Luz 2005; Schoenemann et al. 2013; Wostbrock
et al. 2020a). While it has been suggested that the Δ′17O value of SLAP should be assigned a
value of 0‰ because the measured value is close to 0‰ (Schoenemann et al. 2013), we argue
that, while this may be a useful convention, accepting a 0‰ value is not necessarily correct.
The clustering of data at δ18O = −55.5‰ and −0.011‰ implies that the Δ′17O value of SLAP2
is −0.011 ± 0.005‰. The Δ′17O value of −0.006‰ from Schoenemann et al. (2013) agrees with
our suggested estimate for SLAP2 within error.
One of the subtle complexities of accurately measuring Δ′17O values of samples with very
different δ18O values is the possibility of scale distortion resulting from a pressure baseline
effect (Yeung et al. 2018). Because the effect can be non-mass-dependent, it may lead to
non-trivial errors in the measured Δ′17O value of SLAP2 relative to VSMOW. The magnitude
0.14
0.12
D'17O (‰ vs. VSMOW2, l=0.528 )
0.10
0.08
0.06
0.04
0.02
0.00
-0.02
-0.04
-56.5 -56.0 -55.5 -55.0 -54.5 -54.0 -53.5 -53.0
d18O (‰ vs. VSMOW2)
Figure 5. Published values for SLAP2 on the VSMOW2 scale. The accepted δ18O value is −55.5‰ (verti-
cal line). We suggest a Δ′17O value of −0.011 ± 0.005‰. See Wostbrock et al. (2020a, Table 2) for sources.
of this effect was evaluated by Wostbrock et al. (2020a), who reported a SLAP2 value of
δ18OSLAP2/VSMOW2 = −55.55‰ and Δ′17O = −0.015 ± 0.005‰. They found no pressure-baseline
effect in their mass spectrometer (Thermo Scientific™ 253 Plus). Two gases with δ18O values
that differed by 25‰ were analyzed against each other at operating pressures of 3, 5 and 10 V
[mass 32]. The measured Δ′17O values varied by no more than 0.002‰ over the entire pressure
range. We therefore determine that the measured values are accurate and that the Δ′17O value of
SLAP2 is between −0.015 and −0.010‰. The number is not very different from 0‰, and whether
a Δ′17O value of −0.011 or 0.0‰ is used will have a negligible effect for most samples when
adjusted to the VSMOW-SLAP2 scale. Nevertheless, it appears that the −0.011‰ is a closer fit
to the true value of SLAP2 on the VSMOW scale relative to l = 0.528. Our best estimate for
VSMOW and SLAP2 (and presumably the original SLAP, Lin et al. 2010) is given in Table 1.
Table 1. Triple oxygen isotope values of various reference standards.

Standard Material d17O (‰ relative to d18O (‰ relative to Δ′17O (l = 0.528)
VSMOW-SLAP2) VSMOW-SLAP2)
VSMOW water ≡0 ≡0 ≡0
SLAP & SLAP2 water −29.709 ± 0.006 = −55.5a −0.011 ± 0.006
San Carlos
olivine 2.75 ± 0.08 5.32 ± 0.16 −0.052 ± 0.014
Olivine (SCO)
UWG garnet garnet 2.94b 5.7c −0.064b
NBS 28b quartz 4.991b 9.57a −0.050b
d
KRS garnet −13.34 ± 0.11 −24.95 ± 0.21 −0.091 ± 0.003
SKFSd chert 17.57 ± 0.13 33.81 ± 0.26 −0.137 ± 0.004
air O2 gas 12.09 ± 0.11 23.88 ± 0.21 −0.447 ± 0.034
a
NBS 18 calcite 3.636 ± 0.009 6.99 −0.048 ± 0.009
NBS 19, 19A calcite 14.92 ± 0.010 28.65a −0.102 ± 0.010
IAEA603 calcite 14.83 ± 0.007 28.47a −0.100 ± 0.007
a
NBS 127 sulfate 4.57 8.67 ± 0.02 0.003 ± 0.02
IAEA−SO−5 sulfate 6.14 12.07 ± 0.02a −0.217 ± 0.02
IAEA−SO−6 sulfate −6.21 −11.36 ± 0.02a −0.204 ± 0.02
acid fractionation
axOCO2(ACID)–calcite 1.00535 ± (3.55 × 10 )−6
1.01025 a
factor @25 °C
Note: aIAEA defined value; bdetermined relative to SCO (see Fig. 6); cin relation to VSMOW-SLAP (see text);
d
developed by Miller et al. (2020), reporting new average of four labs (Table 3).
Silicates
Determination of the d17O values of silicates is straightforward because O2 is produced
during fluorination and the conversion from the silicate solid to O2 gas is quantitative (i.e.,
100% of the silicate oxygen is converted to O2). The problem, of course, is calibration to
VSMOW, and to a lesser extent, to the stretching factor VSMOW-SLAP2. That requires
fluorination of water and silicate standards in the same extraction line and mass spectrometer,
something that, until recently, has rarely been done.
Early analyses calibrated their d17O value to SMOW by assuming that the d17O value of
NBS-28 was equal to 0.52 × δ18O (Clayton and Mayeda 1983). Franchi et al. (1999) were
perhaps the first to fluorinate both the NBS 28 standard and SMOW. Unfortunately, the errors
in the analyses were high in comparison to more recent results. A number of more recent
studies have been made in which the IAEA standard NBS 28 and two commonly used informal
standards San Carlos olivine (SCO) and UWG garnet (Valley et al. 1995) have been analyzed
relative to VSMOW and SLAP. One problem is that all researchers get their SCO from any one
of a number of sources. It has been shown that there are at least two different populations of
SCO with different d18O values (Macpherson et al. 2005; Starkey et al. 2016). Nevertheless, in
general, the d17O and δ18O values from recent calibration papers give consistent values for SCO
data, as shown in Table 2. The data in Table 2 are all corrected to the SLAP2 value in Table 1.
The correction procedure involves the following standardization of each laboratories data:
1) The delta values for VSMOW, SLAP (or SLAP2) and SCO are converted to prime
values where δ′ = 1000 × ln(δ/1000 + 1).
2) A constant given by δ′xOVSMOW-accepted − δ′xOVSMOW-measured is added to all data
(δ′xOVSMOW-accepted = 0).
3) A stretching factor is applied such that all data are multiplied by
 x OSMOW accepted   x OSLAP accepted , which reduces to 57.100
 x OSMOW measured   x OSLLAP measured  OSMOW measured  18OSLAP measured
18
30.159
and 17O .
SMOW  measured   O SLAP  measured
17
The small differences seen in Table 2 may be due to slight differences in the oxygen isotope
compositions of the olivine samples or slight miscalibrations to VSMOW-SLAP2. The San
Carlos olivine measured in Wostbrock et al. (2020a) comes from a large aliquot of pure olivine
and is available for others to calibrate their system. Samples of this standard can be obtained
from the UNM CSI laboratory (csi.unm.edu) allowing for precise interlaboratory calibration.
Widescale use of this particular batch of SCO could potentially reduce the interlaboratory
discrepancies and result in further refinements of this value in the future.
NBS 28 quartz and UWG garnet (Valley et al. 1995) are two other commonly measured
reference silicates. There are only two direct comparisons of these two standards to VSMOW
and SLAP (Tanaka and Nakamura 2013; Wostbrock et al. 2020a). Both obtain a δ18O value
of 5.70 ( ± 0.01)‰ for UWG, in good agreement with the original published values of 5.74
to 5.8‰ (Valley et al. 1995). The reported Δ′17O values are −0.078‰ (Tanaka and Nakamura
2013) and −0.071‰ (Wostbrock et al. 2020a). For both studies, the δ18O values for NBS 28 are
within 0.01‰ of the accepted IAEA value of 9.57‰, with Δ′17O values of −0.063‰ (Tanaka
and Nakamura 2013) and −0.059‰ (Wostbrock et al. 2020a).
Table 2. Triple isotope data for San Carlos Olivine. Data are corrected for δ18O and Δ′17O
SLAP2 = −55.5 and −0.011‰.
Laboratory/reference d17O δ18O Δ′17O
1
GZG, Goettingen 2.682 5.153 −0.036
ISIE, Okayama1 2.750 5.287 −0.038
2
ISIE, Okayama 2.717 5.294 −0.075
CSI, UNM3 2.717 5.263 −0.058
4
CSI, UNM 2.892 5.588 −0.054
Average 2.75 ± 0.08 5.32 ± 0.16 −0.052 ± 0.014
Note: 1 Pack et al. (2016), 2Tanaka and Nakamura (2013), 3Wostbrock et al. (2020a), 4Sharp et al. (2016).
The Δ′17O values of UWG and NBS 28 can also be further refined by using additional
published data. There are a large number of studies where UWG and NBS 28 are compared
directly to SCO (see Fig. 6 caption for references). From Figure 6, we obtain a DΔ′17OUWG-SCO
value of −0.012‰ and a DΔ′17ONBS 28-SCO value of 0.002‰ We use the IAEA-accepted δ18O
value of 9.57‰ for NBS 28 and a δ18O value of 5.7‰ for UWG and the difference in the
Δ′17O values between SCO and NBS 28 (Fig. 6) to determine the Δ′17O value and ultimately
the d17O value of these two standards (Table 1).
Two additional standards with extremely different δ18O values have recently been developed
(Miller et al. 2020). The first is an extremely light garnet from Karelia, Russia (KRS) and the
second is a heavy chert from Denmark (SKFS). The purpose of developing standards with such
different δ18O values is to allow laboratories that are not set up to fluorinate waters to be able
to determine a stretching factor based on two silicate samples with very different δ18O values.
The standards were originally analyzed at Georg-August-Universität, Goettingen, Germany,
and The Open University, UK and have subsequently been analyzed at The Center for Stable
Isotopes, University of New Mexico, USA and The University of Oregon, USA (Table 3).
All data were corrected to the common value of UWG given in Table 1. (This is because the
SCO standard at Open University appears to have a different δ18O value from the other labs and
what is seen in Table 2). Correcting all data to a common value of UWG (Table 1) allows for a
common standardized value for these two light and heavy standards (Table 1).
0.02
DD'17ONBS 28-SCO = 0.002

DD'17O (‰ vs. SCO)
0.00
DD'17OUWG-SCO = -0.012
-0.02 12
1 2 5 7 9 10
3 11
6 8
-0.04
NBS 28
UWG
-0.06
4
-0.08
Figure 6. Published ΔΔ′17O values for UWG garnet and NBS 28 relative to San Carlos olivine values from
Table 1. The average Δ′17O value of UWG is −0.012‰ less than SCO while the average Δ′17O value for
NBS 28 is 0.002‰ heavier than SCO (pale horizontal lines). Data sources: 1 (Wostbrock et al. 2020a); 2
(Tanaka and Nakamura 2013); 3 (Pack and Herwartz 2014); 4 (Franchi et al. 1999); 5 (Starkey et al. 2016);
6 (Cowie and Johnston 2016); 7 (Miller et al. 2020); 8 (Kim et al. 2019); 9 (Ghoshmaulik et al. 2020); 10
(I. Bindeman, pers. comm.); 11 (Young et al. 2016); 12 (Yeung et al. 2018).
Air
There are three direct comparisons between atmospheric oxygen (O2) and VSMOW-
SLAP (Barkan and Luz 2005; 2011; Wostbrock et al. 2020a) and three additional comparisons
of O2 with secondary standards, either UWG or SCO (Young et al. 2014; Pack et al. 2017;
Yeung et al. 2018). Accurate Δ′17O analyses of air require the separation of N2 and especially
Ar from O2, which is done using a long gas chromatographic column chilled to ~ −80 °C.
All studies removed Ar from the O2 gas except for two (Barkan and Luz 2005, 2011) in which
an empirical correction was applied (Luz and Barkan 2005). All studies results were corrected
to VSMOW-SLAP2 using the data for standards UWG, SCO and VSMOW-SLAP2 in Table 1.
The newly normalized results are shown in Table 4.
Table 3. Triple isotope data for light and heavy standards KRS and SKFS from four different
laboratories. Data are corrected to UWG values given in Table 1. OU–Open University, UK;
Goet–Georg-August-Universität Göttingen, Germany; CSI–Center for Stable Isotopes, U. New
Mexico, USA; Or–University of Oregon, USA.
Laboratory/
d17O d18O Δ′17O
reference
KRS
OU −13.50 ± 0.05 −25.25 ± 0.09 −0.088 ± 0.010
Goet −13.34 ± 0.14 −24.95 ± 0.27 −0.091 ± 0.010
CSI −13.31 ± 0.06 −24.87 ± 0.12 −0.096 ± 0.010
Or −13.23 ± 0.05 −24.74 ± 0.03 −0.090 ± 0.015
Average −13.34 ± 0.11 −24.95 ± 0.21 −0.091 ± 0.003
SKFS
OU 17.61 ± 0.13 33.88 ± 0.24 −0.135 ± 0.009
Goet 17.38 ± 0.22 33.43 ± 0.41 −0.134 ± 0.009
CSI 17.68 ± 0.08 34.03 ± 0.11 −0.142 ± 0.005
Or 17.62 ± 0.12 33.90 ± 0.38 −0.136 ± 0.007
Average 17.57 ± 0.13 33.81 ± 0.26 −0.137 ± 0.004
Table 4. Compilation of published triple oxygen isotope

analyses of Air corrected to VSMOW-SLAP2.
Reference d17O d18O Δ′17O
Barkan and Luz 20051,2 12.08 ± 0.01 23.88 ± 0.02 −0.453
1,2
Barkan and Luz 2011 12.03 23.88 −0.507
Young et al. 20143 11.94 ± 0.02 23.57 ± 0.05 −0.429 ± 0.01
Pack et al. 20171,5 12.25 ± 0.02 24.15 ± 0.05 −0.432 ± 0.019
3,4
Yeung et al. 2018 12.03 ± 0.01 23.73 ± 0.01 −0.429 ± 0.007
Wostbrock et al. 2020a1 12.18 ± 0.07 24.05 ± 0.11 −0.441 ± 0.012
Average 12.09 ± 0.11 23.88 ± 0.21 −0.447 ± 0.034
Note: 1Measured directly to VSMOW-SLAP; 2Did not remove Ar from O2 sample; 3 Corrected to SCO; 4Includes pressure
baseline correction; 5Corrected to SCO, A. Pack, pers. comm., (2020)
Carbonates
The only way in which the δ17O of carbonates can be directly determined is by complete
conversion to O2 by the method of fluorination. 100% oxygen retrieval requires high
temperature fluorination in nickel reaction vessels. The difficulty with this method is that other
intermediate oxygen-bearing compounds, such as CO, CO2 and COFx, are produced in the
fluorination process and they are difficult to react completely to O2 (and CF4). In the one high-
precision study of this type, Wostbrock et al. (2020a) fluorinated carbonates at 750 °C for four
days with a large excess of BrF5. Yields approached 100%, but the δ18O values were often less
than the accepted values determined by phosphoric acid digestion. It was found, however,
that the lower δ18O also had proportionally lower d17O values, such that the measured l was
0.528, identical to our reference slope l. The measured Δ′17O values were therefore constant
within ± 0.01‰ (Fig. 7). Wostbrock et al. (2020a) therefore used the conventionally accepted
values for the δ18O values of IAEA carbonate standards and the measured Δ′17O values to
Figure 7. Plot of Δ′17OFigure

value7.of an individual analysis vs the difference in the accepted δ18O value and the mea-
sured δ18O value for carbonate samples. Extrapolation of 2‰ for the δ18O value will change the Δ′17O value by
less than 0.006‰. The average of these three samples give a l = 0.528. Data from Wostbrock et al. (2020b).
back-calculate the d17O values of the samples. This same correction to the endmember Δ′17O
values has been used for sulfates (Cowie and Johnston 2016) and CO2 gas (Farquhar et al.
1998). Results are given in Table 1.
Wostbrock et al. (2020a) also fluorinated the CO2 released by phosphoric acid digestion
from calcite at 25 °C. Their calculated θCO2(ACID)–calcite value is 0.5230 ± 0.0003, where
θ = ln a17OCO2(ACID)–calcite / ln a18OCO2(ACID)–calcite. This corresponds to an a17OCO2(ACID)–calcite value of
1.00535 ± 3.55 × 10−6 relative to an a18OCO2(ACID)–calcite value of 1.01025. Using this fractionation
factor, laboratories that measure the Δ′17O value of CO2 released from carbonates by phosphoric
acid digestion (at 25 °C) can back-calculate to the original Δ′17O value of the carbonate.
There are a number of indirect measurements of the δ18O and d17O values of carbonates
obtained by measuring the CO2 gas liberated by phosphoric acid digestion. The CO2 is
analyzed for the triple oxygen isotope composition using one of several methods: conversion
to H2O followed by fluorination (Passey et al. 2014); equilibration with O2 in the presence of
a Pt catalyst (Mahata et al. 2013; Barkan et al. 2015; Fosu et al. 2020); or by direct analysis
of the O+ fragment of CO2 using a high resolution mass spectrometer (Adnew et al. 2019).
With accurate determinations of the triple oxygen isotope composition of IAEA calcite
standards, researchers who measure the CO2 released by phosphoric acid digestion can easily
use the αCO2(ACID)–calcite values in Table 1 or determine an appropriate α value that brings their
measurements into agreement with the δ17O and δ18O values of the carbonate standards.
A compilation of Δ′17O values of IAEA standards and their relative values from different
studies is given in Table 5. Although the measured Δ′17O values may be different depending on the
analytical procedure used, the difference in the Δ′17O values between any two standards should be
the same for all laboratories. The community is coming to consensus on the ΔΔ′17O of standards,
although there are still large outliers, potentially due to issues with exchanging CO2–O2.
Sulfates
The triple oxygen isotope composition of sulfates is measured by laser fluorination with
either BrF5 (Bao and Thiemens 2000) or F2 with additional gas purification using a GC column
(Cowie and Johnston 2016). The laser fluorination methods result in molar yields of 25–35%
when using BrF5 and ~50% when using F2 as reaction reagents. Measured δ18O values can be
as much as 15–20‰ below the accepted values. Due to the low yields, triple oxygen isotope
compositions are corrected based on the measured Δ′17O value from fluorination and the δ18O
value obtained by thermal conversion elemental analyzer (TC/EA) method (Kornexl et al. 1999;
Table 5. Published Δ′17O values of carbonate and CO2 liberated from phosphoric acid digestion and
DΔ′17ONBS 19– NBS 18 and DΔ′17OIAEA 603– NBS 18 values.
Study CO2 calcite ∆∆′17O
NBS 19 IAEA 603 NBS 18 NBS 19 IAEA 603 NBS 18 NBS 19 IAEA 603
-NBS 18 -NBS 18
Wostbrock −0.055b −0.047b
−0.155 −0.147 −0.100 −0.102 −0.100 −0.048
et al. 2020aa −0.054c −0.052c
Barkan et
−0.182 −0.194 −0.163 −0.019b −0.023b
al. 2019
Passey et al.
−0.135d −0.098d −0.037b
2014
Fosu et al.
−0.169 −0.119 −0.050b
2020
Passey and
−0.143d −0.088d −0.055b
Ji 2019
Barkan et
−0.227e +0.003e −0.230b,e
al. 2015
Sha et al.
−0.267 −0.225 −0.042b
2020
Notes: afluorination; bCO2 gas extracted from carbonate; cdirect carbonate fluorination; dextracted at 90 °C. All others extracted at
25 °C; eBarkan et al. (2019) suggest that the analyzed samples may have been contaminated.
Savarino et al. 2001). Using a calculated θLF value (the fractionation associated with incomplete
fluorination) of 0.5301, Cowie and Johnston (2016) were able to extrapolate their measured
δ17O and δ18O to obtain the Δ′17O value of their sulfate standards NBS 127, IAEA-SO-5, and
IAEA-SO-6. The δ18O values of these standards are tied to the VSMOW-SLAP scale using
waters analyzed by CO2–H2O equilibration (Brand et al. 2009). The reported Δ′17O values
in Cowie and Johnston (2016) are relative to their d17O values for the UWG standard. We
recalculate the d17O and Δ′17O values of these standards using the difference between Cowie
and Johnston’s Δ′17O values for UWG, San Carlos olivine and NBS 28 and those reported in
the present communication (Table 1). The reported δ18O values are those given in Johnston et
al. (2014). Results are presented in Table 1 calibrated to the VSMOW-SLAP scale based on the
fluorination of VSMOW and SLAP2 .
CONCLUSION
Mass dependent triple oxygen isotope geochemistry is a relatively new field. Until recently,
laboratories generally measured variations in the Δ′17O values of materials without a proper
interlaboratory calibration scheme. In some cases, the d17O values of their samples were
estimated, with no direct ties to VSMOW and SLAP (SLAP2). While this did not affect the
results from an ‘isolated’ publication, it made intercomparisons between laboratories extremely
difficult. The problems in standardizing different materials to the same scale is related to the
different methods employed for analysis. Silicates and waters are both fluorinated, but relatively
few laboratories are set up to do both. More challenging is the total fluorination of carbonates,
which requires high temperatures and long reaction times in order to approach 100% recovery.
In this communication, we have attempted to calibrate all standards relative to VSMOW

and SLAP2 as presented in Table 1. Standards include waters, silicates, carbonates, sulfates and
air. We also provide the acid fractionation factor data for CO2–calcite as reported in Wostbrock
et al. (2020a), so that laboratories that determine the triple isotope values of CO2 liberated by
phosphoric acid digestion of calcite can back calculate the isotope values of the calcite relative
to VSMOW-SLAP2. With this internally consistent standards table, practitioners can directly
compare their data for different materials with different laboratories. Invariably, adjustments will
be made to these suggested standard values. Nevertheless, the triple oxygen isotope geochemical
community is reaching consensus on standards that makes possible direct comparison of the very
subtle differences that are seen in natural materials.
ACKNOWLEDGEMENTS
We would like to thank S.W. Schoenemann and M. Miller for their constructive reviews,
and the editors I.N. Bindeman and A. Pack for their comments and support during this work.
David Johnston provided us with unpublished sulfate data. Support for this study comes from
the following grants: NSF-EAR 1551226 and NSF-EAR 1903852 to ZDS and NSF GRFP
Grant DGE-1418062 to J.A.G.W.
REFERENCES
Adnew GA, Hofmann ME, Paul D, Laskar A, Surma J, Albrecht N, Pack A, Schwieters J, Koren G, Peters W,
Assonov S, Groening M, Fajgelj A, Hélie J-F, Hillaire-Marcel C (2020) Preparation and characterization of IAEA-603,
a new primary reference material aimed at the VPDB scale realisation for δ13C and δ18O determination. Rapid
Commun Mass Spectrom 34:e8867
Baertschi P (1950) Isotopic composition of the oxygen in silicate rocks. Nature 166:112–113
Baertschi P (1976) Absolute 18O content of Standard Mean Ocean Water. Earth Planet Sci Lett 31:341–344
Baertschi P, Silverman SR (1951) The determination of relative abundances of the oxygen isotopes in silicate rocks.
Bao H, Thiemens MH (2000) Generation of O2 from BaSO4 using a CO2-laser fluorination system for simultaneous
analysis of δ18O and δ17O. Anal Chem 72:4029–4032
Spectrom 19:3737–3742
Barkan E, Luz B (2011) The relationships among the three stable isotopes of oxygen in air, seawater and marine
photosynthesis. Rapid Commun Mass Spectrom 25:2367–2369
Barkan E, Affek HP, Luz B, Bergel SJ, Voarintsoa NRG, Musan I (2019) Calibration of δ17O and 17Oexcess values
of three international standards: IAEA-603, NBS19 and NBS18. Rapid Commun Mass Spectrom 33:737–740
carbon dioxide–water. J Phys Chem 72:800–808
Bottinga Y, Craig H (1969) Oxygen isotope fractionation between CO2 and water, and the isotopic composition of
marine atmospheric CO2. Earth Planet Sci Letters 5:285–295
Brand WA, Coplen TB, Aerts-Bijma AT, Böhlke JK, Gehre M, Geilmann H, Gröning M, Jansen HG, Meijer HA,
Mroczkowski SJ, Qi H(2009) Comprehensive inter-laboratory calibration of reference materials for δ18O versus
VSMOW using various on-line high-temperature conversion techniques. Rapid Commun Mass Spectrom
23:999–1019
Brenninkmeijer CAM, Kraft P, Mook WG (1983) Oxygen isotope fractionation between CO2 and H2O. Chem Geol
41:181–190
Clayton RN (1959) Oxygen isotope fractionation in the system calcium carbonate–water. J Chem Phys 30:1246–1250
Clayton RN, Mayeda TK (1983) Oxygen isotopes in eucrites, shergottites, nakhlites, and chassignites. Earth Planet
Sci Lett 62:1–6
Cohn M, Urey HC (1938) Oxygen exchange reactions of organic compounds and water. J Am Chem Soc 60:679–687
Coplen TB, Kendall C, Hopple J (1983) Comparison of stable isotope reference samples. Nature 302:236–238
Cowie BR, Johnston DT (2016) High-precision measurement and standard calibration of triple oxygen isotopic
compositions (δ18O, Δ′17O) of sulfate by F2 laser fluorination. Chem Geol 440:50–59
Craig H (1957) Isotopic standards for carbon and oxygen and correction factors for mass-spectrometric analysis of
carbon dioxide. Geochim Cosmochim Acta 12:133–149
Craig H (1961) Standard for reporting concentrations of deuterium and oxygen-18 in natural waters. Science
133:1833–1834
Dole M (1936) The relative atomic weight of oxygen in water and air. A discussion of the atmospheric distribution of
the oxygen isotopes and of the chemical standard of atomic weights. J Chem Phys 4:268–275
Epstein S, Mayeda T (1953) Variation of O18 content of waters from natural sources. Geochim Cosmochim Acta
4:213–224
Epstein S, Buchsbaum R, Lowenstam HA, Urey HC (1951) Carbonate–water isotopic temperature scale. Geol Soc
Am Bull 62:417–426
Epstein S, Buchsbaum R, Lowenstam HA, Urey HC (1953) Revised carbonate–water isotopic temperature scale.
Geol Soc Am Bull 64:1315–1326
Farquhar J, Thiemens MH, Jackson TL (1998) Atmosphere–surface interactions on Mars: Δ17O measurements of
Fosu BR, Subba R, Peethambaran R, Bhattacharya SK, Ghosh P (2020) Technical Note: Developments and
applications in triple oxygen isotope analysis of carbonates. ACS Earth Space Chem doi.org/10.1021/
acsearthspacechem.9b00330:9 p.
Franchi IA, Wright IP, Sexton AS, Pillinger CT (1999) The oxygen-isotopic composition of Earth and Mars. Meteorit
Friedman I (1953) Deuterium content of natural waters and other substances. Geochim Cosmochim Acta 4:89–103
Friedman I, Gleason JD (1973) A new silicate intercomparison standard for 18O analysis. Earth Planet Sci Lett 18:124
Friedman I, O’Neil JR (eds) (1977) Compilation of stable isotope fractionation factors of geochemical interest.
Washington, D.C.
Friedman I, O’Neil JR, Cebula G (2007) Two new carbonate stable-isotope standards. Geostand Geoanal Res 6:11–12
Ghoshmaulik S, Bhattacharya SK, Roy P, Sarkar A (2020) A simple cryogenic method for efficient measurement of
triple oxygen isotopes in silicates. Rapid Commun Mass Spectrom 34:e8833
Gonfiantini R (1978) Standards for stable isotope measurements in natural compounds. Nature 271:534–536
Guo W, Mosenfelder JL, Goddard WA, Eiler JM (2009) Isotopic fractionations associated with phosphoric acid
digestion of carbonate minerals: Insights from first-principles theoretical modeling and clumped isotope
measurements. Geochim Cosmochim Acta 73:7203–7225
Hayles JA, Gao C, Cao X, Liu Y, Bao H (2018) Theoretical calibration of the triple oxygen isotope thermometer.
and nuclear effects. J Geophys Res 70:3475–3484
IAEA (2006) Reference values for the relative difference in hydrogen and oxygen stable isotope-amount ratio for
the international measurement standards [1] IAEA Reference Sheet https://nucleus.iaea.org/rpst/documents/
VSMOW_SLAP.pdf:5
IAEA (2017) Reference Sheet for International Measurement Standards VSMOW2 Vienna Standard Mean Ocean
Water 2, water, SLAP2 Standard Light Antarctic Precipitation 2, water. IAEA Reference Sheet https://nucleus.
iaea.org/rpst/Documents/VSMOW2_SLAP2.pdf:8
IAEA (2020) Reference Sheet for Certified Reference Material GRESP Greenland Summit Precipitation, water. IAEA
Reference Sheet https://nucleus.iaea.org/sites/ReferenceMaterials/Shared%20Documents/ReferenceMaterials/
StableIsotopes/Gresp/RS_GRESP-2020.pdf:5
Jabeen I, Kusakabe M (1997) Determination of d17O values of reference water samples VSMOW and VSLAP. Chem
Geol 143:115–119
Johnston DT, Gill BC, Masterson A, Beirne E, Casciotti KL, Knapp AN, Berelson W (2014) Placing an upper limit
on cryptic marine sulphur cycling. Nature 513:530–533
Kim NK, Kusakabe M, Park, Lee JI, Nagao K, Enokido Y, Yamashita S, Park SY (2019) An automated laser
fluorination technique for high-precision analysis of three oxygen isotopes in silicates. Rapid Commun Mass
Kim S-T, Coplen TB, Horita J (2015) Normalization of stable isotope data for carbonate minerals: Implementation of
IUPAC guidelines. Geochim Cosmochim Acta 158:276–289
Kornexl BE, Gehre M, Höfling R, Werner RA (1999) On-line δ18O measurement of organic and inorganic substances.
Lin Y, Clayton RN, Gröning M (2010) Calibration of δ17O and δ18O of international measurement standards—
VSMOW, VSMOW2, SLAP, and SLAP2. Rapid Commun Mass Spectrom 24:773–776
biogeochemistry. Geochim Cosmochim Acta 69:1099–1110
Macpherson CG, Hilton DR, Day JMD, Lowry D, Grönvold K (2005) High-3He/4He, depleted mantle and low-δ18O,
recycled oceanic lithosphere in the source of central Iceland magmatism. Earth Planet Sci Lett 233:411–427
platinum: An innovative technique for measuring Δ17O in CO2. Anal Chem 85:6894–6901
McKeegan KD, Leshin LA (2001) Stable isotope variations in extraterrestrial materials. Rev Mineral Geochem
43:279–318
Meijer HAJ, Li WJ (1998) The use of electrolysis for accurate δ17O and δ18O isotope measurements in water. Isot
Environ Health Stud 34:349–369
O’Neil JR, Epstein S (1966a) A method for oxygen isotope analysis of milligram quantities of water and some of its
applications. J Geophys Res 71:4955–4961
O’Neil JR, Epstein S (1966b) Oxygen isotope fractionation in the system dolomite–calcite–carbon dioxide. Science
152:198–200
Pack A, Herwartz D (2014) The triple oxygen isotope composition of the Earth mantle and Δ17O variations in terrestrial
rocks. Earth Planet Sci Lett 390:138–145
Pack A, Höweling A, Hezel DC, Stefanak MT, Beck A-K, Peters STM, Sengupta S, Herwartz D, Folco L (2017)
Tracing the oxygen isotope composition of the upper Earth’s atmosphere using cosmic spherules. Nature
Communications 8:15702
Passey BH, Ji H (2019) Triple oxygen isotope signatures of evaporation in lake waters and carbonates: A case study
from the western United States. Earth Planet Sci Lett 518:1–12
Sakai S, Matsuda S, Hikida T, Shimono A, McManus JB, Zahniser M, Nelson DD, Dettman DL, Yang D, Ohkouchi N
(2017) High-precision simultaneous 18O/16O, 13C/12C, and 17O/16O analyses for microgram quantities of CaCO3
by tunable infrared laser absorption spectroscopy. Anal Chem 89:11846–11852
Savarino J, Alexander B, Darmohusodo V, Thiemens MH (2001) Sulfur and oxygen analysis of sulfate at micromole
levels using a pyrolysis technique in a continuous flow system. Anal Chem 73:4457–4462
Sha L, Mahata S, Duan P, Luz B, Zhang P, Baker J, Zong B, Ning Y, Brahim YA, Zhang H, Edwards RL (2020) A
novel application of triple oxygen isotope ratios of speleothems. Geochim Cosmochim Acta 270:360–378
Sharma T, Clayton RN (1965) Measurement of O18/O16 ratios of total oxygen of carbonates. Geochim Cosmochim
Acta 29:1347–1353
Sharp ZD (2013) Principles of Stable Isotope Geochemistry, 2nd Edition. Open Educational Resources, Albuquerque, NM
Silverman SR (1951) The isotope geology of oxygen. Geochim Cosmochim Acta 2:26–42
Starkey NA, Jackson CRM, Greenwood RC, Parman S, Franchi IA, Jackson M, Fitton JG, Stuart FM, Kurz M, Larsen LM
resonance. Atmos Measur Tech 7:2421–2435
Stoltmann T, Casado M, Daëron M, Landais A, Kassi S (2017) Direct, precise measurements of isotopologue abundance
ratios in CO2 using molecular absorption spectroscopy: Application to Δ17O. Anal Chem 89:10129–10132
Tanaka R, Nakamura E (2013) Determination of 17O-excess of terrestrial silicate/oxide minerals with respect to
Vienna Standard Mean Ocean Water (VSMOW). Rapid Commun Mass Spectrom 27:285–297
Taylor HP, Jr., Epstein S (1962) Relationship between O18/O16 ratios in coexisting minerals of igneous and metamorphic
rocks: Part 1: Principles and experimental results. Geol Soc Am Bull 73:461–480
Urey HC, Greiff LJ (1935) Isotopic exchange equilibria. J Am Chem Soc 57:321–327
Urey HC, Epstein S, McKinney CR (1951) Measurement of paleotemperatures and temperatures of the Upper Cretaceous
of England, Denmark, and the southeastern United States. Geol Soc Am Bull 62:399–416
Valley JW, Kitchen N, Kohn MJ, Niendorf CR, Spicuzza MJ (1995) UWG-2, a garnet standard for oxygen isotope
ratios; strategies for high precision and accuracy with laser heating. Geochim Cosmochim Acta 59:5223–5231
Verkouteren RM, Klinedinst DB (2004) Value assignment and uncertainty estimation of selected light stable isotope
reference materials: RMs 8543–8545, RMs 8562–8564, and RM 8566. NIST Spec Publ 260–149 2004 ED:58
application of silica–water triple oxygen isotope thermometry to geothermal systems in Iceland and Chile.
Wostbrock JAG, Cano EJ, Sharp ZD (2020a) An internally consistent triple oxygen isotope calibration of standards for
Wostbrock JAG, Brand U, Coplen T, Swart PK, Carlson SJ, Brearley AJ, Sharp ZD (2020b) Calibration of carbonate–
water triple oxygen isotope fractionation: seeing through diagenesis in ancient carbonates. Geochim Cosmochim
Acta 288:369–388
Young ED, Yeung LY, Kohl IE (2014) On the D17O budget of atmospheric O2. Geochim Cosmochim Acta 135:102–125
Vol. 86 pp. 197–216, 2021 6
Mass-Independent Fractionation of
Oxygen Isotopes in the Atmosphere
Marah Brinjikji and James R. Lyons
School of Earth & Space Exploration
Arizona State University
550 East Tyler Mall, PSF 686
Tempe, AZ 85287
USA
mbrinjik@asu.edu
jimlyons@asu.edu
INTRODUCTION
We review the O isotope modeling in the atmosphere below 100 km, which is a large and
active area of research. Our review will not be exhaustive but instead will highlight some of
the key papers on the topic over the past nearly 30 years. We focus our review on modeling
the mass-independent fractionation signatures associated with O3 formation and other species
that photochemically interact with O3. After a brief discussion of isotopic O3 formation, we
present results from models of D17O in CO2, OH and H2O, nitrates, sulfates, and in pollution
in urban environments.
We then present new O isotope results for the atmosphere above 100 km. There are presently
no oxygen isotope measurements for Earth’s atmosphere above about 60 km (Thiemens 1995).
Here, we present model results for the oxygen isotope composition of Earth’s thermosphere
and ionosphere using a 1-D photochemical model. This work was motivated by the NASA
MAVEN mission, which is studying ion and neutral composition, and escape fluxes from,
the thermosphere of Mars, and includes isotopic species (Jakosky 1994; Jakosky et al. 2017,
2018). The realization that comparable isotopic measurements for Earth’s thermosphere are
lacking prompted our study.
PREVIOUS WORK IN THE LOWER ATMOSPHERE

Early measurements
Because our main interest here is in mass-independent fractionation (MIF) of O isotopes,
or, as will be seen for the upper atmosphere, apparent MIF, we will not review the large literature
on O and H isotopes of atmospheric water vapor, and the influence of exchange with soils, which
has been a topic of ongoing research for many decades. We begin with the stratospheric work
of Konrad Mauersberger. Using a balloon platform, Mauersberger (1981) performed in situ
mass spectrometry of O3 in the stratosphere, discovering an enormous enrichment of ~ 400‰ in
16 16 18
O O O (mass 50). We now know that this value is too high, but, qualitatively, the discovery
was made. In the laboratory the key discovery of mass-independent fractionation was made by
Thiemens and Heidenreich (1983) by passing an electric discharge through O2 in a flask, and
measuring the O isotope composition of the O3 produced and the residual O2. Instead of the
usual mass-dependent isotope behavior (adhering to the slope of ~0.5 in 17O/16O vs. 18O/16O
coordinates), they found that the O2 and O3 defined a line of nearly slope 1 on 3-isotope space,
leading to the term ‘mass-independent’ fractionation. It is difficult to overstate the significance
1943-2666/21/0086-0006$05.00 (online)
198 Brinjikji & Lyons
of this experiment on our understanding of atmospheric O isotopes. Ozone participates in

a wide variety of oxygen chemical reactions, either directly as O3 or as a source of O(1D),
which forms OH upon reaction with water vapor. In this manner, MIF inherited from O3
manifests itself in nearly all O-containing compounds in the atmosphere (Lyons 2001).
Continued isotopic measurements in the laboratory and in the atmosphere confirmed the
original discovery (Thiemens and Heidenreich 1983), and sought to clarify the mechanism
responsible for the unusual isotopic behavior of O3. Analyses of O3 were made in both the
troposphere and stratosphere (e.g., Krankowsky et al. 1995, 2000; Johnston et al. 1997).
Laboratory studies focused on O3 formation reactions under a variety of conditions (e.g.,
Morton et al. 1990, Janssen et al. 1999), culminating in measured rate coefficients for
formation of isotopic O3 (Mauersberger et al. 1999). Other studies discovered oxygen MIF
effects in other relevant atmospheric reactions (Röckmann et al. 1998; Savarino and Thiemens
1999). Although much smaller in magnitude than the O3 MIF signature, these latter reactions
are still informative from a chemical physics perspective.
Theory and photochemical modeling of O3 MIF signatures
Photochemical models began exploring the implications of heavy ozone in the atmosphere,
even though there was essentially no isotopic rate coefficient data available, and even the
reaction responsible for O3 was debated. An early successful model looked at stratospheric CO2,
and recognized that O(1D), produced from O3 photolysis, will form the metastable CO3 (Yung
et al. 1991). Breakup of the CO3 imparts an O3-derived MIF signature to the CO2, diminished
by the probability of 17O or 18O remaining in the CO2. This mechanism-based analysis of the
transfer of O3 MIF to another molecule, namely stratospheric CO2, demonstrated both the
significance and subtlety with which O3 can impart isotopic signatures to other molecules. With
the measurement of isotopic rate coefficients by Mauersberger et al. (1999), it became possible
to more quantitatively model O isotope exchange processes in the atmosphere. Lyons (2001)
investigated transfer of O3 MIF signatures to about 10 other molecules, and provided some
unification for the expected oxygen isotope composition of the troposphere and stratosphere.
The measurement of the isotopic rate coefficients by Mauersberger et al. (1999) also made it
possible for the physical chemists to quantitatively investigate the mechanism responsible for O3
MIF. A series of papers from Rudy Marcus’s group (Hathorn and Marcus 2000; Gao and Marcus
2002) provided a semi-empirical, and extremely useful, theory for understanding the role of zero-
point energy and symmetry in the wide range of isotopic rate coefficient values for O3 formation.
The MIF signature derived from non-statistical behavior in the asymmetric O3 isotopomers,
the exact physical origin of which is still being explored. Importantly for the photochemical
modelers, the Marcus formulation allowed prediction of unmeasured rate coefficients. This is
particularly important for the mixed-isotope O2 species, e.g., 16O17O and 16O18O, whose reactions
with 16O are essential in the atmosphere but difficult to measure in the laboratory.
Applications of oxygen MIF
Ozone MIF signatures (i.e., enhancements or depletions in 17O) have had a tremendous
range of application in geochemistry, both for the modern and ancient Earth. We will not
attempt to review all of these applications, but will mention a few that we think are especially
notable, with the recognition that we are not without bias. Many of the most significant
applications of oxygen MIF have been to measures of global primary productivity (GPP),
as first pointed out by Bender et al. (1994). Accurate measurements of the MIF signature of
O2 allowed Luz et al. (1999) to relate modern primary productivity to the cross-tropopause
transport timescales for air. The small negative MIF signature in bulk tropospheric O2 with
D17O ~ −0.2‰, results from the storage of a positive MIF signature in stratospheric CO2 via
transfer from O3 by the excited state atom O(1D) (Yung et al. 1991). This powerful technique
has also been applied to O2 collected from Antarctic ice and snow, extending productivity
estimates back to the middle ages (Savarino et al. 2003).
Mass-Independent Fractionation of Oxygen Isotopes in the Atmosphere 199
One of the most remarkable MIF-related discoveries is that of negative MIF signatures,
D17O ~ −0.8 to −1.6‰ in sulfates from the Cryogenian period (Bao et al. 2008). Reconstructing
the implied isotopic composition for O2 yields D17O ~ −2 to −6‰. By accounting for the
efficiency of subaerial pyrite oxidation, the exact same mechanism invoked by Luz et al. (1999)
may be applied to this much earlier geological period to place a constraint on atmospheric CO2.
Given plausible assumptions about productivity at that time, greatly elevated levels of CO2 may
be inferred (up to 30 times the modern value), just as predicted for the aftermath ‘hothouse’
following the end of a Snowball Earth glaciation (Bao et al. 2008). Most recently, Hodgkiss et al.
(2019) used D17O values of sulfate (also ~ −0.8‰) to argue for the collapse of productivity at the
end of the great oxidation event (GOE). Because the sulfate D17O values depend on pO2, pCO2,
and GPP, different interpretations are possible for the Snowball Earth and the GOE.
There are numerous other applications of O isotope MIF signatures, some of which are
discussed in more detail below. These include atmospherically-derived nitrates such as those in
the Atacama desert in Chile (Michalski et al. 2004), the origin of desert varnish on rocks (Bao
et al. 2001), and an improved quantification of stratospheric volcanism (Gautier et al. 2019).
Additionally, oxygen MIF signatures have been investigated for polar water vapor (Lin et al.
2013), stratospheric CO2 (Liang et al. 2007), and the sulfate formation mechanism in marine
sea-salt aerosols (Alexander et al. 2005). The breadth of applications of O3 MIF signatures are
a testament to the importance of the phenomenon.
MATERIALS AND METHODS

VULCAN photochemical model
VULCAN is a 1-D chemical kinetics model developed for high-temperature exoplanet
atmospheres (Tsai et al. 2017). The original version of VULCAN utilized a reduced C–H–O
chemical network consisting of about 300 gas-phase reactions valid in a 500−2500 K temperature
range. However, the code is highly general, and is readily adapted to low-temperature atmospheres,
as well as atmospheres with N and S-containing molecular species. We have adapted VULCAN
to the Earth thermosphere and ionosphere, and with the inclusion of O isotopes. VULCAN
solves a system of vertical 1-D continuity equations (Eqn. 1) using a Rosenbrock solver (Tsai
et al. 2017). The Rosenbrock solver has been shown to be highly effective for systems of ‘stiff’
ordinary differential equations, such as occur in atmospheric chemical kinetics. Vertical transport
due to both eddy and molecular diffusion is represented by a vertical diffusion equation.
For high-temperature atmospheres, reverse reaction rate coefficients are computed from a
measured forward rate, combined with the equilibrium constant for the reaction. The equilibrium
constant is computed from the Gibbs free energy of the reaction, adjusted for pressures (Tsai et al.
2017). To model Earth’s atmosphere, we have turned off the reverse reactions. We have done this
even for the thermosphere because the kinetics of the relevant neutral–neutral and ion–molecule
reactions are well known from decades of laboratory work (e.g., Brasseur and Solomon 1984).
The isotopic reactions are less well studied, but we assume that they do not differ greatly from
the reactions of the dominant 16O isotope. For reactions involving O3 in the lower atmosphere,
such an assumption is incorrect due to the non-statistical nature of O3 recombination reactions
(Gao and Marcus 2002). In the thermosphere, O3 formation is unimportant, and we assume that
the isotopic reaction rate coefficients are independent of isotope, apart from a mass-dependent
reduced-mass factor discussed further below.
The continuity equations solved in VULCAN are given by
dni di (1)

 Pi  Li
dt dt
where ni and fi are the number density (cm−3) and vertical flux (cm−2 s−1) for species i in the
model. Pi and Li are the production and loss rates (cm−3 s−1) for species i due to chemical and
photodissociation reactions. The vertical flux satisfies a 1st-order diffusion equation,
 dn n   dn n 
i 
 Di  i  i   K  i  i  (2)
 dz Hi   dz H a 
where Di is the molecular diffusion coefficient for species i diffusing through the background
atmosphere, and K is the vertical eddy diffusion coefficient applicable to all species. The scale
height of species i is
kT (3)
Hi 
mi g
and for the well-mixed atmosphere is
kT (3b)
Ha 
ma g
for mean molecular mass, ma. The homopause is defined as the height at which Di ~ K for most
species. Above this height, which is about 100 km for Earth, molecular diffusion dominates over
eddy diffusion. Here, eddy diffusion means mixing by eddies on the full range of size scales from
the Kolmogorov microscale to the scale height of the bulk atmosphere, Ha. The occurrence of mi
in Equation (3a), and not in Equation (3b), is the basis of diffusive separation in the thermosphere.
RESULTS FOR THE LOWER ATMOSPHERE

We present a short synopsis of O isotopes in the lower atmosphere based on the
photochemical equilibrium calculations in Lyons (2001), and then discuss more recent work
on modeling MIF signatures in O3, CO2, H2O, nitrates, sulfates, and O-containing components
of air pollution. The focus of Lyons (2001) was the transfer of O3 MIF signatures to other
oxygen-containing molecules in the atmosphere. We summarize this work below, and present
updated results for D17O of several lower atmosphere species.
The relevant isotopic O3 formation reactions are the following:
O + O2 M O3 k1 (R1)
Q + O2 M QOO k2 (R2)
O + QO M QOO k3  Q (R3a)
O + QO M OQO k3 (1   Q ) (R3b)
P + O2 M POO k4 (R4)
O + PO M POO k5  P (R5a)
O + PO M OPO k5 (1   P ) (R5b)
In these reactions k is the 3-body rate coefficient, γ is the branching ratio, Q = 18O and P = 17O, and
M is a 3rd-body collision partner, usually O2 or N2. The primary loss for ozone is photodissociation.
There are multiple photodissociation channels, but, for the purpose of illustration, we only
consider the channel forming O(1D). We also assume the photodissocation rate constants are
identical for all isotopomers of O3. This is not exactly correct, but it simplifies the resulting
equations. The resulting photodissociation reactions are shown in reactions (R6–R10).
O3  hv  O(1D)  O2 J1 (R6)
QOO + hv  O(1D)  QO 1
J1 (R7a)
2
QOO + hv  Q(1D)  O2 1
J1 (R7b)
2
OPO + hv  O(1D)  PO J1 (R8)
POO + hv  O(1D)  PO 1
J1 (R9a)
2
POO + hv  P(1D)  O2 1
J1 (R9b)
2
OPO + hv  O(1D)  PO J1 (R10)
We must also include O exchange with O2:
Q + O2  QO + O k6 f (R11a)
O + QO  Q + O2 k6 r (R11b)
P + O2  PO + O k7 f (R12a)
O + PO  P + O2 k7r (R12b)
Finally, we include quenching of the electronically excited O(1D) by collision with O2

and N2 to produce a ground state O atom. For the presentation here we will not include this
reaction, but will simply assume that all O(1D) is rapidly quenched to O. This is, of course, not
actually the case in the atmosphere, as O(1D) is involved in essential reactions with H2O (to
form OH), CO2 (isotope exchange), and many other molecules. However, this assumption does
not significantly affect the isotopic composition of O3.
From the above set of reactions, and assuming O(1D) → O proceeds rapidly, we may derive
expressions for the O3 isotopomer number densities and delta-values. We do this by setting
production and loss rates equal to each other for a given species, and then solving for the species
number densities. This will only be valid for short-lived species that are essentially not affected
by transport, i.e., species in photochemical steady state. From reaction (R11) and (R12), we find
[Q] 1 [QO]
 Q (4)
[O] K eq [O2 ]
[P] 1 [ PO] (5)

 P
[O] K eq [O2 ]
In Equations (4) and (5)  K eqQ =k 6kf 6f/ /kk66r

r = e31.631.6/T
1.9463e
1.9463 /T

and K eqP = kk f //kk7r
77f .9728 e1515.8/T
7 r = 11.9728e
.8 / T
,
where T is atmospheric temperature. The equilibrium constants have been evaluated from
statistical mechanics and are given in Yung et al. (1997). The forward rate constants are
k6
f k7f 3.4  10 12 (300 / T )1.1 , neglecting reduced factors in the rate of reaction.
Again, assuming photochemical equilibrium, we derive approximate expressions for the
O3 isotopomer number densities:
[OQO] k3 [QO]
 (1   Q ) (6)
[ O3 ] k1 [O2 ]
[QOO]  k2 1 k  [QO]
  Q
 3  Q  (7)
[ O3 ] k K
 1 eq k1  [O2 ]
[OPO] k5 [ PO] (8)
 (1   P )
[ O3 ] k1 [O2 ]
[ POO]  k4 1 k  [ PO]
  P
 5  P  (9)
[ O3 ]  k1 K eq k1  [O2 ]
The total isotopomer number densities are independent of branching ratio,
[QOO]  [OQO]  k2 1 k  [QO]

  Q
 3  (10)
[ O3 ]  k1 K eq k1  [O2 ]
[ POO]  [OPO]  k4 1 k  [ PO]

  P
 5  (11)
[ O3 ]  1 eq k1  [O2 ]
k K
Using the experimental results of Mauersberger et al. (1999) and Janssen et al. (1999),
together with the semi-empirical model results of Hathorn and Marcus (2000) and Gao and Marcus
(2002), we compile a table of relative rate coefficients is compiled for O3 isotopomer formation
(Table 1). The values in the ‘theory’ column for η = 1.15 imply gQ = 0.594 and gP = 0.575.
Lyons (2001) used gQ = gP = 0.57 based on the incomplete experimental values. These small
differences in branching ratios produce dramatically different D17O values for the symmetric
and asymmetric O3 isotopomers, i.e., OQO and QOO, respectively. We compute delta-values as
[O x OO]  O 
 x O(O3s )  1 (12)
[O3 ]  x O SMOW
[ x OOO]  O 
 x O(O3a )  1 (13)
2[O3 ]  x O SMOW
[ x OOO]  [O x OO]  O 
 x O(O3 )   xO  1 (14)
3[O3 ]  SMOW
Lyons (2001) assumed photochemical steady state for a system of reactions in the mid-
latitude atmosphere from 0 to 60 km, analogous to Equations (6−11), but yielding a coupled
non-linear system. As noted by Zahn et al. (2006), Lyons assumed gQ = gP = 0.57 based on the
experimental data available at the time, which resulted in D17O ~ −40 to −50‰ for symmetric
O3 and too high a D17O value for asymmetric O3. Here, and below we define D17O as
 (17O  1)  (18O  1)

17O (15)
where λ =0.50–0.53, and where the reference is VSMOW. Updated results from Lyons (2001)
are shown in Figure 1. The branching ratios used in Figure 1 are gQ = 0.594 and gP = 0.564,
which yields a vertically-averaged symmetric O3 ~ 0‰. Temperature and pressure affect
the formation rate of isotopic O3, and although these effects were included in Lyons (2001),
they have not been updated here. A more precise treatment was given in Liang et al. (2006),
which we discuss further below. The key point of Figure 1 is the illustration of how O3-derived
MIF signatures are transferred to other photochemically active species.
Table 1. Relative rate coefficients for O3 isotopomer formation.

Theory2 Theory2 Theory3
Rate coefficient Experiments1
(η = 1.15) (η = 1.00) (η = 1.18)
k2/k1 0.93 0.95 0.83 0.93
k3/k1 1.27 1.28 1.18 1.25
k3gQ/k1 0.73 0.76 0.66 0.74
k3(1 − gQ)/k1 0.54 0.52 0.52 0.52
k4/k1 1.03 1.02 0.89 1.03
k5/k1 1.17 1.20 1.11 1.19
k5gP/k1 not measured 0.69 0.60 0.68
k5(1 − gP)/k1 not measured 0.51 0.51 0.51
Notes:
1
Mauersberger et al. (1999); Janssen et al. (1999).
2
Hathorn and Marcus (2000). The parameter η characterizes the magnitude of non-statistical behavior in the asymmetric ozone isotopomers. Non-
statistical behavior in a chemical reaction means a deviation from the usual Boltzmann distribution of energy levels in a microcanonical statistical
mechanics description of a molecule (e.g., Gao and Marcus 2002). These values are for 140 K.
3
Gao and Marcus (2002).
60
50 NO2 O3
O(1D)
NO
40 Oa3
Altitude (km)
ClO
30 Os3
OH
20
HO2
10
0
-10 0 10 20 30 40 50 60 70 80
∆17O (permil)
Figure 1. Photochemical equilibrium model results. Calculations assume an O atom exchange rate coef-
ficient of 1×10−19 cm3 s−1 between OH and O2, well below the upper limit value (Table 2). The hatched
regions show the range of D17O measurements for stratospheric (Schueler et al. 1990; Krankowsky et al.
2000) and tropospheric (Krankowsky et al. 1995; Johnston and Thiemens 1997) O3. The narrow rectangle
near the origin is the range of measurements in rainwater H2O2 (Savarino and Thiemens 1999). This figure
closely follows Lyons (2001) but with an updated isotopic branching ratios (see text), with a non-zero
exchange rate for OH and O2, and with the addition of symmetric ozone. Slope = 0.520
Figure 1 includes exchange reactions (X1–X7), and (X9) (Table 2), with the rate coefficient for
X6 set arbitrarily to 1×10−19 cm3 s−1. If the rate coefficient for X6 is set to zero (as in Fig. 1 of Lyons
2001), the maximum D17O values for OH are about 27‰ from 28−32 km, several permil higher than
shown here, but still about 20‰ lower than Figure 1 in Lyons (2001). Using the improved isotopic
branching ratios has a quantitative impact on atmospheric D17O values, but does not change the
qualitative picture of the prediction of an enrichment of 17O in many stratospheric species (Lyons
2001). It is also possible that additional O exchange reactions are present, which will further modify
the D17O values shown in Figure 1. Exchange reactions for ClO have not been included here.
Table 2. O atom exchange rate coefficients (Lyons 2001).

Rate
Reaction coefficient Ref.
cm3s-1
X1 O + O ↔ OO + O 2.9(‒12) a
X2 Q + NO ↔ O + NQ 3.7 (‒11) a
X3 QH + H2O ↔ OH + H2Q 2.3 (‒13)e -2100/T b
X4 QH + NO ↔ OH + NQ 1.8 (‒11) c
X5 QH + NO2 ↔ OH + NOQ 1.0 (‒11) c
X6 QH + O2 ↔ OH + OQ <1 (‒17) c
X7 QH + HO2 ↔ OH + HQO 1.7 (‒11) e400/T d
X8 HOQ + O2 ↔ HO2 + OQ <3 (‒17) e
X9 NQ + NO2 ↔ NOQ + NO 3.6 (‒14) f
X10 NOQ + O2 ↔ NO2 + OQ <1 (‒24) g
X11 NOQ + H2O ↔ NO2 + H2Q rapid (?) h
X12 QH + CO ↔ OH + CQ <1 (‒15) c
X13 QH + CO2 ↔ OH + COQ <1 (‒17) c
References: a. Anderson et al. (1985); b. Dubey et al. (1997); c. Greenblatt and Howard (1989); d. Dransfield
and Wagner (1987); e Sinha et al. (1987); f Klein et al. (1963); g Sharma et al. (!970); h Jaffe and Klein (1966)
Ab initio calculations from Francisco (1998) predict a 3.1 kcal mol−1 barrier above the bond
dissociation energy of the HOClO adduct. At stratospheric temperatures this barrier (1560 K)
implies a relatively slow rate of exchange with OH, but other exchange reactions could be
important, although exchange with O2 is unlikely.
Figure 2 shows D17O values from the Lyons (2001) model, but that have not been previously
published. The CO2 are not computed here, but instead are specified from Lämmerzahl et al. (2002).
60
50
ClNO3
40
Altitude (km)
NO3
30 HNO2
CO2 MPN
20 HNO3
10
0
0 10 20 30 40 50
∆17O (permil)
Figure 2. D17O values for several other species included in the Lyons (2001) model, but not previously
shown. The CO2 values are specified (Lämmerzahl et al. 2002), whereas the others are computed. MPN is
methyl peroxy nitrate (CH3O2NO2). These values are more qualitative than quantitative. For example, we
have not considered exchange reactions for any of the computed species. Slope = 0.520.
‘MPN’ is methyl peroxy nitrate, CH3O2NO2, and is meant to be representative of the class of
peroxy nitrate compounds. A more complete analysis of the D17O of nitric acid and precursor
nitrates is given in Michalski et al. (2004). Exchange reactions have not been included for any
of these species. Far more complete models now exist, especially for the nitrates, as will be
discussed further below.
Turning next to O3 isotopic models, detailed modeling of stratospheric ozone isotopes was
presented in Liang et al. (2006). Isotope fractionation due to ozone formation reactions were
carefully modeled using the results of the semi-empirical model of Gao and Marcus (2002)
and the laboratory measurements of Morton et al. (1990). Fractionation associated with ozone
photolysis was also accounted for. The model reproduced balloon-borne mass spectrometer
measurements from 20‒35 km quite well (Liang et al. 2006). More recently Young et al. (2014)
have argued that including the mass-dependence of kinetic rate coefficients in ozone formation
reactions has a non-negligible effect on the MIF signatures of the resulting O3. As has been
discussed previously in Lyons (2001), in a hard-sphere approximation, a gas-phase bimolecular
rate coefficient may be expressed in the form k ~ c vth e  Ea / kT where sc is the collision cross
section, vth is the thermal velocity, and Ea is the activation energy. The thermal velocity is
vth  1 / u where µ is the reduced mass of the reactants. Young et al. (2014) showed that
inclusion of even this simple reduced mass factor reduces the MIF signatures of O3 products,
producing better agreement with measurements. Most of the models results described here
(Lyons 2001 included) have not accounted for the reduced mass in the isotopic rate coefficients.
Trace but measurable D17O signatures were predicted for tropospheric CO2 (Hoag et al.
2005). In the stratosphere CO2 acquires a large MIF signature due to exchange with O(1D)(Yung
al. 1991). In the troposphere CO2 exchanges O primarily with biospheric water. Using a two-
box model, and accounting for the rates of exchange in the biosphere and the flux of CO2 from
the stratosphere, Hoag et al. (2005) predict D17O = 0.13 and 0.15‰ (VSMOW, slope λ = 0.516)
for the northern and southern hemispheres, respectively (Fig. 3). These results mean that D17O
can be used as an additional tracer of biospheric productivity, and of the rate of exchange of
stratospheric and tropospheric air, as has been done for O2 (e.g., Bender et al. 1994).
Figure 3. Box model prediction of D17O for tro-

pospheric CO2 for the global average (solid line),
the northern hemisphere (dashed), and the south-
ern hemisphere (dotted). (Figure from Hoag et al.
(2005), Geophys. Res. Lett. 32, L02802.)
The predictions of Hoag et al. (2005) have been confirmed by measurements in Taiwan
(Mahata et al. 2016), and by measurements and modeling in the vicinity of Göttingen, Germany
(Hofmann et al. 2017). 3D modeling also demonstrates this MIF signature in atmospheric CO2
on a fairly well-resolved grid (Fig. 4), although with MIF signatures lower than the Hoag et
al. (2005) model. All of these results confirm that D17O of CO2 can be used as a tracer of C in
the atmosphere. As an example of this, Liang and Mahata (2015) used an enhanced D17O in
surface CO2 to infer intrusion of stratospheric air, possibly associated with tropopause folding.
Figure 4. Monthly average of simulated D17O using a 3D model (TM5) with a 6° × 4° horizontal resolution
and 25 vertical levels. (a) Heat diagram for D17O, with values are expressed in per meg rather than per mil.
(b) Comparison with previously published box models. (Figure from Koren et al. (2019) J Geophys Res
Atmos 124:8808−8836.)
Stratospheric CO2 exhibits a slope ~ 1.6‒1.7 on a 3-oxygen isotope plot (d17O vs. d18O)
discovered by balloon measurements from Lämmerzahl et al. (2002). Such a high slope in excess
of 1 arises from exchange of CO2 with O(1D) released by O3 photolysis (Yung et al. 1991;
Liang et al. 2007). Clumped isotope work on stratospheric CO2, in particular, measurement of
16 13 18
O C O (mass 47) from stratospheric air samples, has revealed a large meridional variation in
mass 47 (Yeung et al. 2009). The mid-latitude mass 47 abundances are consistent with exchange
of CO2 with O(1D), but polar vortex abundances indicate another fractionation process at work,
possible temperature dependent exchange on polar stratospheric clouds particles.
Next, we consider the MIF signature of stratospheric water. A prediction from Lyons (2001)
was that stratospheric water could have a positive D17O value, because O(1D) produced by
photolysis of one of the terminal O atoms on O3 produces OH upon reaction with stratospheric H2O.
Formation of stratospheric H2O occurs by the oxidation of CH4 by OH. A more detailed
analysis of this possibility by Zahn et al. (2006) using a 1-D photochemical model showed that
stratospheric OH could have D17O ~ +30‰ near 30 km (Fig. 5a), leading to stratospheric H2O
with D17O ~ +10‰. These values are slightly in error because Zahn et al. (2006) assumed that
symmetric O3 has D17O ∼ +25‰, rather than closer to 0‰, as discussed above. Again, the results
are dependent on several uncertain rate coefficients for O atom exchange. Zahn et al. (2006)
also present d18O values (Fig. 5b), but these results do not include the reduced mass scaling
of the isotopic rate coefficients. Measurements of water vapor samples from the Arctic (Alert,
Canada) and Antarctica (Vostok) by Lin et al. (2013) and Winkler et al. (2013), respectively, have
demonstrated a component of 17O-enriched water from the stratosphere as a result of the overall
Brewer-Dobson meridional circulation. Direct sampling of water vapor in the upper troposphere
and lower stratosphere yielded D17O ~ 0‰ with uncertainty of about 2‰, but it was not entirely
clear from where the sampled air was originating (Franz and Röckmann 2005).
Turning next to nitrates, Michalski et al. (2003, 2004) identified the key nitrate D17O
formation pathways and used D17O signatures to definitely show that the Atacama nitrates
are mostly atmospherically derived. In a time-dependent study, diurnal variation of D17O of
nitrate was simulated using an atmospheric chemistry box model for HOx, NOx, and NOy
species (Morin et al. 2011). Because of the strong diurnal variation in the chemistry of nitrate
formation, and in the NOx and NOy precursors, quite large diurnal variations in D17O were
predicted (Fig. 6). For NO2, the minimum in D17O ~ 28‰ and occurred near local noon,
while the maximum ~ 38‒40‰ and occurred at night (Morin et al. 2011). Similar variations,
Figure 5. 1-D photochemical simulation of the oxygen isotopes of several radical species in the atmo-
sphere. These results assume D17O = +25‰ for symmetric O3, which implies D17O values for asymmetric
O3 that are too low. A positive MIF signature for stratospheric OH is predicted. (Figure from Zahn et al.
(2006), Atmos. Chem. Phys. 6: 2073−2090).
Figure 6. Diurnal variation of D17O of NO2

for two cases. Case 1 assumes photochemi-
cal steady state (PSS), and Case 2 is an ex-
plicit kinetic calculation. The NO2 photo-
lytic lifetime is shown by the overlain color.
(Figure from Morin et al. (2011) Atmos.
Chem. Phys. 11: 3653−3671)
although for smaller mean D17O values, were found for HO2 and H2O2. Global simulations
of nitrate with a 1 km cloud model, with highly detailed chemical pathways, reveal large
geographical variations in D17O (Alexander et al. 2020). Over Amazonia, the dominance of
RONO2 hydrolysis yields mean values of just 3−4‰, whereas over the mid-latitude oceans,
the dominance of XNO3 hydrolysis (X = halogen atom) yields mean values over 30‰ (Fig. 7).
On the topic of MIF signatures in sulfates, a key driver has been the transfer of D17O to
sulfate during the oxidation of volcanic plume SO2. There are many oxidation pathways for
SO2 in the modern atmosphere including by O3, by H2O2, and by O2, depending on the pH
of the participating aerosol. Although our emphasis thus far has been on the modern Earth
atmosphere (Martin et al. 2014), several major eruptions have occurred since the Pleistocene
that have left well known tuff formations, including the Lara Creek Tuff (LCT, 0.64 Ma),
the Huckleberry Ridge Tuff (HRT, 2.04 Ma), and the Bishop Tuff (0.76 Ma). Analysis of the
O isotope composition of sulfate in these tuffs yielded constraints on total O3 consumed by the
corresponding volcanic eruption (Fig. 8) (Martin and Bindeman 2009).
Another key application of sulfate MIF signatures has been to elucidate oxidation
mechanisms. Alexander et al. (2011) used observations of D17O in non-sea-salt sulfate
to quantify the formation pathways in marine boundary layer sulfate. For their northeast
Atlantic ocean collection site, they determined that oxidation of S(IV) by O3 accounted
for about 36% of in-cloud sulfate formation. Similarly, but in a different context,
D17O of Antarctic ice core sulfates and nitrates have been used to infer changes in oxidation
mechanisms in the southern hemisphere since the 19th century. The observed increase in
D17O of sulfates is consistent with greater oxidation by O3, and the observed decrease in
D17O of nitrates is interpreted to be due an increase in RO2 over O3 in NOx cycling (Sofen
et al. 2014). More recently, Lin et al. (2017) used D17O and 35S (a radiogenic isotope) in
sulfates collected in eastern China to determine that the formation pathway for tropospheric
sulfate is vertically uniform, likely due to strong convection in this region.
Finally, we also consider applications of MIF signatures to atmospheric pollution,
particularly urban environments. Li et al. (2013) analyzed a long record of sulfates collected in
Figure 7. Simulation of annual mean global nitrate D17O values for a 1 km cloud model. The wide range
in nitrate MIF signature results from the relative importance of XNO3 hydrolysis (high D17O) and RONO2
hydrolysis (low values). (Figure from Alexander et al. (2020) Atmos Chem Phys 20:3859−3877).
Figure 8. Estimates of ozone layer consumption from several large volcanic eruptions in California dry
lake beds over the past 2 Ma. BT, LCT, and HRT are defined in the text. Several important SO2 oxidation
mechanisms are shown. (Figure reprinted with permission of Elsevier from Martin E, Bindeman I (2009)
Mass-independent isotopic signatures of volcanic sulfate from three supereruption ash deposits in Lake Tecopa,
California. Earth Planet Sci Lett 282:102–114.)
Wuhan and found D17O ~ 0.6‰, similar to Baton Rouge, Louisiana. The similarity is surprising
given the difference in rainwater pH in the two locations, which leads Li et al to suggest that
S(IV) oxidation by metals is more important in the dust-rich Wuhan atmosphere. Laskar et al.
(2020) have used D17O of CO2 air in Delhi, India to distinguish between the fossil fuel and rice
straw burning contribution. Combined with a mixing model, they have determined that rice
straw burning contributes ~70% to anthropogenic CO2.
RESULTS FOR THE UPPER ATMOSPHERE

The isotopic composition of Earth’s upper atmosphere (above 100 km) has not received
much attention. The lower atmosphere is mostly well-mixed due to the motions of turbulent
eddies and breaking atmospheric waves. In photochemical models, eddy transport is
approximated as a diffusive process, and is referred to as eddy diffusion. Compositionally, the
upper atmosphere above the homopause, the upper boundary of the well-mixed atmosphere at
~ 100km, is dominated by O atoms, and transport by molecular diffusion dominates over eddy
diffusion. Considering these conditions, we present simulations of the O isotope composition
of O atoms in the thermosphere up to escape altitudes (~ 500 km).
The thermosphere represents the transition from convective and radiative energy
transport below the homopause to primarily conductive energy transport above the homopause
(Chamberlain and Hunten 1990). Generation of hot O atoms due to far-UV and extreme-UV
heating of O2 and N2 in the ionosphere creates the high temperatures (~ 1000 K) from which the
thermosphere’s name originally derived. Below the homopause, isotope fractionation results
from photochemical reactions (mostly mass-dependent), non-statistical photochemistry, such
as for O3 (mostly mass-independent), or from condensation/evaporation processes (mass-
dependent). Eddy mixing below the homopause keeps transport-derived isotope fractionation
to a minimum. Above the homopause, molecular diffusion dominates transport, and massive
mass-dependent fractionation results.
For the thermosphere, diffusive equilibrium (i.e., zero flux) conditions are informative.
Setting fi = 0 in Equation (2), and solving for ni, we find
 zD K  1 
( z ) ni ( z0 )exp     i 
ni dz  (16)
 z  Hi H a  Di  K 
 0 
where z0 is a reference height, which could be the ground, the homopause, or some other
preferred height (Banks and Kockarts 1973). We will use Equation (16) to get a sense of
the expected isotope fractionation in the thermosphere, and compare those results to a
photochemical simulation using the VULCAN model (Tsai et al. 2017).
We first ran VULCAN to simulate Earth’s atmosphere from the ground up to 500 km, the
approximate upper boundary of the atmosphere for which the mean free path of an atom is
comparable to its scale height. This height varies with the solar cycle up to a maximum of about
1000 km for peak solar activity. Our VULCAN runs included the minor isotopes of oxygen
(17O and 18O), but did not include the non-statistical rate constants of O3 formation (Hathorn
and Marcus 2000; Gao and Marcus 2002), which is the primary source of mass-independent
fractionation in the atmosphere below the homopause (e.g., Thiemens and Heidenreich 1983,
Lyons 2001). The temperature and pressure profiles are from the US standard atmosphere model
(e.g., Chamberlain and Hunten 1990). The corresponding ion and electron number density
profiles were also computed. These non-isotopic number density profiles are not shown here,
but will be presented elsewhere. The ion species do not include water cluster positive ions or
negative ions (e.g., Brasseur and Solomon 1984), corresponding to the D-layer of the ionopshere.
We focus on O atoms calculated from both diffusion theory (Eqn. 16) and from the
full VULCAN photochemical model of the upper atmosphere. Figure 9 shows the number
densities for the stable oxygen isotopes calculated by these two methods. The solid lines are the
results from the diffusion theory equations for an assumed isothermal temperature of 500 K,
and the dashed lines are the VULCAN full kinetic results. The number density (in cm−3) for
16
O is in blue, 17O is in green, and 18O is in red. We expect to see the slope of the number
density lines to be different below and above the homopause, due to diffusive separation.
The isothermal diffusion theory approximation has a slope change at around 200 km.
This height differs significantly from the true Earth homopause because Earth’s atmospheric
temperature profile is not isothermal, but instead varies from about 200 to 1000 K.
The VULCAN results diffusively separate near 100 km, closer to the true Earth homopause.
Figure 9. The number densities for the stable oxygen isotopes of O atoms, calculated from diffusion theory
for a 500 K isothermal atmosphere (solid lines) and using VULCAN (dashed lines). The true homopause
is shown with the solid black line. (Brinjikji and Lyons 2020a).
The delta-values for thermospheric O atoms are shown in Figure 10. The values for d17O and
d18O (‰) calculated are shown for diffusion theory (solid lines) and VULCAN photochemical
modeling (dashed lines). The delta values from the diffusion theory approximation are depleted by
about −100‰ more than the VULCAN results for both d17O and d18O above about 400 km. Both
models show evidence of diffusive separation above the homopause. As in Figure 9, the diffusion
theory model is for an isothermal atmosphere that does not have the correct location of the
homopause, yet the isotope fractionation is qualitatively similar to that obtained with VULCAN.
The D17O for these two cases is shown in Figure 11 with from diffusion theory shown
as solid lines and from VULCAN shown as dashed lines. However, the slope factor of
0.528 results in enrichments on the order of 40‰ near the exobase (Figure 11). Based on
the barometric law behavior of the individual isotopes undergoing molecular diffusion, we
redefined D17O to have slope λ = 0.500 which results in a depletion of −2‰ from the diffusion
theory calculation and an enrichment of 5‰ from VULCAN (Fig. 11). We have not yet
determined the source of these still very large values, but it’s clear that a fractionation law
using 0.528 or 0.52 is not correct in the thermosphere.
Figure 10. The delta values (‰) for and calculated from isothermal diffusion theory (solid lines) and using
VULCAN (dashed lines). The homopause is shown with the solid black line.
Figure 11. The values for thermospheric O atoms calculated from isothermal diffusion theory (solid lines)
and using VULCAN (dashed lines). (Brinjikji and Lyons 2020a).
DISCUSSION
Improvements in the accuracy and interpretation of D17O modeling in the well-mixed
atmosphere (below 100 km) will come from two key areas. First, a complete accounting
of mass-dependent effects in the kinetics. Young et al. (2014) make this point for O3
formation, but this will extend to most of the other species isotopically influenced by O3.
Second, the data on exchange reactions is still incomplete, especially for gas phase exchange.
Several of the exchange reactions in Table 2 have rather high upper limits to their rate
coefficients. Laboratory measurements could provide much tighter constraints on these rate
coefficients, which would significantly improve the accuracy of isotopic model predictions.
For the upper atmosphere, we focus discussion on diffusive separation of isotopes.
To illustrate why a 0.50 fractionation law better describes mass-dependent isotope fractionation
in the thermosphere (Fig. 11), we consider number density profiles for O atoms well above the
homopause. This ensures that only molecular diffusion is affecting isotope ratios. For diffusive
equilibrium in an isothermal atmosphere, the number density profile is given by
ni ( z )  ni ( zh )e  ( z  zh )/ Hi (17)
where zh is the height of the homopause. Using the ratios of number densities to compute
delta-values relative to the homopause, we have
 1 1 
nx ( z ) nx ( zh )  ( z  zh ) H x  H16  (18)
 e
n16 ( z ) n16 ( zh )
where the difference of the inverse scale heights is
1 1 g
  ( mx  m16 ) (19)
H x H16 kT
n
The corresponding delta-values are  x
O x
 1. . For small arguments in the exponential
n16
of Equation (18), the delta-values become
g (20)
 x O  ( z  zh ) ( mx  m16 )
kT
From equation 20, the linear dependence on mass difference means that d17O/d18O ≈ 0.50.
This continues to be largely true for the full exponential form of Equation (18) as well.
The barometric law behavior of gases in diffusive equilibrium in an atmosphere yields
mass-dependent behavior, but with a fractionation law exponent that differs fundamentally
from the more usual 0.52‒0.53 law that results from either zero-point energy, vibrational
partition function, or collisional mass dependence (Bigeleisen 1965; Matsuhisa et al. 1977;
Weston 1999; Young et al. 2000).
CONCLUSIONS
We have reviewed model results for the mass-independent fractionation of O isotopes in
the lower atmosphere. We presented updated results from Lyons (2001), together with later
modeling of D17O of stratospheric H2O from Zahn et al. (2006). A definitive measurement
confirming a MIF signature (possibly as high as 10‰) in stratospheric water has yet to be
made, but there are strong suggestions of MIF stratospheric water in Arctic and Antarctic
tropospheric water samples (Lin et al. 2013; Winkler et al. 2013). The small MIF signature
(D17O ~ 0.1‰) of tropospheric CO2 has been shown to be an effective tracer of C in the lower
atmosphere in a 2-box model (Hoag et al. 2005) and in a global 3D model (Koren et al. 2019).
Recent work on global 3D modeling of D17O of nitrates demonstrates the importance of the MIF
carrier, with XNO3 hydrolysis producing large signatures (D17O ~30‰) versus RONO2 hydrolysis
(e.g., over Amazonia) yielding dramatically smaller MIF signatures (D17O ~3‰) (Alexander et
al. 2020). MIF signatures of sulfate formation have been used to elucidate the complex oxidation
pathways for S(IV) in various environments (e.g., Alexander et al. 2005), and to estimate ozone
layer consumption from large volcanic eruptions (Martin and Bindeman 2009). A theme that
emerges from these studies is the impressive diversity of applications of O MIF signatures in
the atmosphere. It is important to remember that the models may be limited by the available
data, particularly with respect to poorly constrained rate coefficients for exchange reactions.
In addition, the models need to properly account for mass-dependent fraction of O isotopes, by,
at the very least, including the reduced mass in isotopic rate coefficients.
We have also presented new results of a model analysis of O isotope fractionation in
the upper atmosphere, specifically above the homopause (~ 100 km). Diffusive separation of
isotopes is the predominant feature, with massive (several 100’s of permil) depletion of 17O
and 18O relative to 16O for atomic oxygen, the primary component of the upper atmosphere.
Preliminary supporting evidence for such massive fractionation comes from measurements of
Al samples from the Long Duration Exposure Facility (LDEF), which sampled the atmosphere
at altitudes from 470 to 340 km over a 6 year period (Brinjikji and Lyons 2020b). In an
additional modeling result, we found that the mass-dependent fractionation that accompanies
diffusive separation has a fractionation law exponent closer 0.50 than to the more usual values
of 0.52−0.53 (Brinjikji and Lyons 2020a). This derives from the difference in scale heights in
the barometric laws that describe atoms in the thermosphere.
In summary, our main conclusions are as follows.
1. Improved laboratory rate coefficients for exchange reactions are needed for models of
MIF signatures in the lower atmosphere (below 100 km). Also, a more careful accounting
of mass-dependent fractionation should improve the accuracy of model computed MIF
signatures.
2. In the upper atmosphere (> 100 km), diffusive separation due to molecular diffusion of
the isotopes of atomic oxygen leads to massive depletions (by 100’s ‰) of 17O and 18O
by altitudes approaching the exobase (~ 500 km). 3) Diffusive separation is a type of
mass-dependent fractionation, but it exhibits a slope (λ) closer to 0.50 than to the more
usual mass-dependent fractionation slopes of 0.52‒0.53.
ACKNOWLEDGEMENTS
The authors gratefully acknowledge numerous discussions with Shang-Min Tsai on the use of
the VULCAN photochemical code, and on python in general. JRL acknowledges support from
the NASA LARS program (grant #80NSSC18K0841 to ASU).
REFERENCES
Alexander B, Park RJ, Jacob DJ, Li QB, Yantosca RM, Savarino J, Lee CCW, Thiemens MH (2005) Sulfate formation
in sea-salt aerosols: Constraints from oxygen isotopes. J Geophys Res Atmos 110:1–12
Alexander B, Allman DJ, Amos HM, Fairlie TD, Dachs J, Hegg DA, Sletten RS (2011) Isotopic constraints
on the formation pathways of sulfate aerosol in the marine boundary layer of the subtropical northeast
Atlantic Ocean. J Geophys Res 117:D06304
Alexander B, Sherwen T, Holmes CD, Fisher JA, Chen Q, Evans MJ, Kasibhatla P (2020) Global inorganic nitrate production
mechanisms: comparison of a global model with nitrate isotope observations. Atmos Chem Phys 20:3859–3877
Anderson SM, Klein FS, Kaufman F (1985) Kinetics of the isotope exchange reaction of 18O with NO and O2 at 298 K.
J Chem Phys 83:1648–1656
Bao H, Campbell DA, Bockheim JG, Thiemens MH (2000) Origins of sulphate in Antarctic dry-valley soils as
Bao H, Michalski GM, Thiemens MH (2001) Sulfate oxygen−17 anomalies in desert varnishes. Geochim Cosmochim
Acta 65:2029–2036
Bao, H, Lyons JR, Zhou C (2008) Triple oxygen isotope evidence for elevated CO2 levels after a Neoproterozoic
Banks PM, Kockarts G (1973). Aeronomy, Part B. Academic Press, New York
Bender ML, Sowers T, Labeyrie L (1994) The Dole effect and its variations during the last 130,000 years as measured
Bigeleisen J (1965) Chemistry of isotopes. Science 147:463–471
Brasseur G, Solomon S (1984) Aeronomy of the Middle Atmosphere, D. Reidel, Dordrecht
Brinjikji M, Lyons JR (2020a) The effects of diffusion on oxygen isotopes in earth’s upper atmosphere. LPSC abstract #2317
Brinjikji M, Lyons JR (2020b) Oxygen isotope fractionation in earth’s upper atmosphere. LPSC abstract #2322
Chamberlain R, Hunten D (1990) The Theory of Planetary Atmospheres, An Introduction to their Physics and
Chemistry (2nd Ed). Academic Press, New York
Dransfield P, Wagner HG (1987) Comparative study of the reactions of 16OH and 18OH with H16O2. Z Naturforsch
42a:471–476
Dubey MK, Mohrschladt R, Donahue NM, Anderson JG (1997) Isotope specific kinetics of hydroxyl radical (OH)
with water (H2O): testing models of reactivity and atmospheric fractionation. J Phys Chem A 101:1494:1500
Francisco JS (1998) Oxygen atom exchange in reactions of OH radicals with NO and ClO. Chem Phys Lett 285:138–142
Franz P, Röckmann T (2005) High-precision isotope measurements of H216O, H217O, H218O, and the Δ17O-anomaly of
water vapor in the southern lowermost stratosphere. Atmos Chem Phys 5:2949–2959
Gao YQ, Marcus RA (2002) On the theory of the strange and unconventional isotopic effects in ozone formation.
J Chem Phys 116:137–154
Gautier E, Savarino J, Hoek J, Erbland J, Caillon N, Hattori S, Yoshida N, Albalat E, Albarede F, Farquhar J (2019)
2600-years of stratospheric volcanism through sulfate isotopes. Nat Comm 10:1–7
Greenblatt GD, Howard CJ (1989) Oxygen atom exchange in the interaction of 18OH with several small molecules.
J Phys Chem 93:1035–1042
Hathorn BC, Marcus RA (2000) An intramolecular theory of the mass-independent isotope effect for ozone:
II. Numerical implementation at low pressures using a loose transition state. J Chem Phys 113:9497–9509
Hoag KJ, Still CJ, Fung IY, Boering KA (2005) Triple oxygen isotope composition of tropospheric carbon dioxide as
a tracer of terrestrial gross carbon fluxes. Geophys Res Lett 32:L02802
Hodgskiss MSW, Crockford PW, Peng Y, Wing BA, Horner TJ (2019) A productivity collapse to end Earth’s Great
Oxidation. PNAS 116:17208–17212
Hofmann MEG, Horváth B, Schneider L, Peters W, Schützenmeister K, Pack A (2017) Atmospheric measurements of
D17O in CO2 in Göttingen, Germany reveal a seasonal cycle driven by biospheric uptake. Geochim Cosmochim
Acta 199:143–163
Jaffe S, Klein FS (1966) Isotopic exchange reactions atomic oxygen produced by the photolysis of NO2 at 3660 Å.
Far Soc Trans 62:3135:3141
Jakosky BM, Jones JH (1994) Evolution of water on Mars. Nature 370 :328–329
Jakosky BM, Slipski M, Benna M, Mahaffy P, Elrod M, Yelle R, Stone S, Alsaeed N (2017) Mars’ atmospheric
history derived from upper-atmosphere measurements of Ar−38/Ar−36. Science 355:1408
Jakosky BM, Brain D, Chaffin M, Curry S, Deighan J, Grebowsky J, Halekas J, Leblanc F, Lillis R, Luhmann JG,
Andersson L (2018) Loss of the Martian atmosphere to space: Present-day loss rates determined from MAVEN
observations and integrated loss through time. Icarus 315:146–157
Janssen C, Guenther J, Krankowsky D, Mauersberger K (1999) Relative formation rates of 50O3 and 52O3 in 16O–18O
mixtures. J Chem Phys 111:7179–7182
Johnston GC, Thiemens MH (1997) The isotopic composition of tropospheric ozone in three environments. J Geophys
Res 102:25395–25404
Klein FS, Spindel W, Stern MJ (1963) Catalysis of isotopic exchange in nitric oxide. J Chim Phys 60:148–153
Koren G, Schneider L, van der Velde IR, van Schaik E, Gromov SS, Adnew GA, Mrozek Martino DJ, Hofmann ME,
Liang MC, Mahata S, Bergamaschi P (2019) Global 3-D simulations of the triple oxygen isotope signature Δ17O
in atmospheric CO2. J Geophys Res Atmos 124:8808–8836
Krankowsky D, Bartecki F, Klees GG, Mauersberger K, Schellenbach K, Stehr J (1995) Measurement of heavy
isotope enrichment in tropospheric ozone. Geophys Res Lett 22:1713–1716
Krankowsky D, Lämmerzahl P, Mauersberger K (2000) Isotopic measurements of stratospheric ozone.
Geophys Res Lett 27:2593–2595
Lämmerzahl P, Röckmann T, Brenninkmeijer CAM, Krankowsky D, Mauersberger K (2002) Oxygen isotope
composition of stratospheric carbon dioxide. Geophys Res Lett 29:1582
Laskar AH, Maurya AS, Singh V, Gurjar BR, Liang M-C (2020) A new perspective of probing the level of pollution
in the megacity Delhi affected by crop residue burning using the triple oxygen isotope technique in atmospheric
CO2. Environ Pollution 263:114542
Li X, Bao H, Gan Y, Zhou A, Liu Y (2013) Multiple oxygen and sulfur isotope compositions of secondary atmospheric
sulfate in a mega-city in central China. Atmos Environ 81:591–599
Liang M-C, Mahata S (2015) Oxygen anomaly in near surface carbon dioxide reveals deep stratospheric intrusion.
Sci Rep 5:11352
Liang M-C, Irion FW, Weibel JD, Miller CE, Blake GA, Yung YL (2006) Isotopic composition of stratospheric ozone.
J Geophys Res 111:2302–2313
Liang M-C, Blake GA, Lewis BR, Yung YL (2007) Oxygen isotopic composition of carbon dioxide in the middle
atmosphere. PNAS 104:21–25
Lin Y, Clayton RN, Huang L, Nakamura N, Lyons JR (2013) Oxygen isotope anomaly observed in water vapor from
Alert, Canada and the implication for the stratosphere. PNAS 110:15608–15613
Lin M, Biglari S, Zhang Z, Crocker D, Tao J, Su B, Liu L, Thiemens MH (2017) Vertically uniform formation
pathways of tropospheric sulfate aerosols in East China detected from triple stable oxygen and radiogenic sulfur
isotopes. Geophys Res Lett 44:5187–5196
as a tracer of biospheric productivity. Nature 400:547–550
Lyons JR (2001) Transfer of mass-independent fractionation in ozone to other oxygen-containing radicals in
the atmosphere. Geophys Res Lett 28:3231–3234
Mahata S, Wang C-H, Bhattacharya SK, Liang M-C (2016) Near surface CO2 triple oxygen isotope composition.
Terr Atmos Ocean Sci 27:99–106
Martin E, Bindeman I (2009) Mass-independent isotopic signatures of volcanic sulfate from three supereruption ash
deposits in Lake Tecopa, California. Earth Planet Sci Lett 282:102–114
Martin E, Bekki S, Ninin C, Bindeman I (2014) Volcanic sulfate aerosol formation in the troposphere. J Geophys Res
Atmos 119:12660–12673
Matsuhisa Y, Goldsmith JR, Clayton RN (1978) Mechanisms of hydrothermal crystallization of quartz at 250 °C and
15 kbar.Geochim Cosmochim Acta 42:173–182
Mauersberger K, Erbacher B, Krankowsky D, Günther J, Nickel R (1999) Ozone isotope enrichment: Isotopomer
specific rate coefficients. Science 283:370–372
Michalski G, Böhlke JK, Thiemens M (2004) Long term atmospheric deposition as the source of nitrate and other salts
in the Atacama Desert, Chile: New evidence from mass-independent oxygen isotopic compositions. Geochim
Michalski GM, Scott Z, Kabiling M, Thiemens MH (2003) First measurements and modeling of D17O in atmospheric
nitrate. Geophys Res Lett 30:1870
Morin S, Sander R, Savarino J (2011) Simulation of the diurnal variations of the oxygen isotope anomaly (D17Ο) of
reactive atmospheric species. Atmos Chem Phys 11:3653–3671
Morton J, Barnes J, Schueler B, Mauersberger K (1990) Laboratory studies of heavy ozone. J Geophys Res 95:901–907
Röckmann T, Brenninkmeijer CA, Saueressig G, Bergamaschi P, Crowley JN, Fischer H, Crutzen PJ (1998) Mass-
independent oxygen isotope fractionation in atmospheric CO as a result of the reaction CO + OH. Science 281:544–546
Savarino J, Thiemens MH (1999) Mass-independent oxygen isotope (16O, 17O, 18O) fractionation found in Hx, Ox
reactions. J Phys Chem A103:9221–9229
Savarino J, Bekki S, Cole-Dai J, Thiemens MH (2003) Evidence from sulfate mass independent oxygen isotopic compositions
of dramatic changes in atmospheric oxidation following massive volcanic eruptions. J Geophys Res Atmos 108:4671
Schueler B, Morton J, Mauersberger K (1990) Measurement of isotopic abundances in collected stratospheric ozone
samples. Geophys Res Lett 17:1295–1298
Sharma HD, Jervis RE, Wong KY (1970) Isotopic exchange reactions in nitrogen oxides. J Phys Chem 74:923–933
Sinha A, Lovejoy ER, Howard CJ (1987) Kinetic study of the reaction of HO2 with ozone. J Chem Phys 87:2122–2128
Sofen ED, Alexander B, Steig EJ, Thiemens MH, Kunasek SA, Amos HM, Schauer AJ, Hastings MG, Bautista
J, Jackson TL, Vogel LE, McConnell JR, Pasteris DR, Saltzman ES (2014) WAIS Divide ice core suggests
sustained changes in the atmospheric formation pathways of sulfate and nitrate since the 19th century in
the extratropical Southern Hemisphere. Atmos Chem Phys 14:5749–5769
Thiemens MH, Heidenreich HE (1983) The mass-independent fractionation of oxygen: A novel effect and its possible
cosmochemical implications. Science 219:1073–1075
Thiemens MH, Jackson T, Zipf EC, Erdman PW, van Egmond C (1995) Carbon dioxide and oxygen isotope anomalies
in the mesosphere and stratosphere. Science 270:969
Tsai S-M, Lyons JR, et al (2017) VULCAN: An open-source, validated chemical kinetics python code for exoplanetary
atmospheres. Astrophys J Supp 228:2–28
Weston RE (1999) Anomalous or mass-independent isotope effects. Chem Rev 99:2115–2136
Winkler R, Landais A, Risi C, Baroni M, Ekaykin A, Jouzel J, Petite JR, Prie F, Minster B, Falourd S (2013)
Interannual variation of water isotopologues at Vostok indicates a contribution from stratospheric water vapor.
PNAS 110:17674–17679
Yeung LY, Affek HP, Hoag KJ, Guo W, Wiegel AA, Atlas EL, Schauffler SM, Okumura M, Boering KA, Eiler JM (2009)
Large and unexpected enrichment in stratospheric 16O13C18O and its meridional variation. PNAS 106:11496–11501
Young ED, Yeung LY, Kohl IE (2014) On the D17O budget of atmospheric O2, Geochim Cosmochim Acta 135:102–125
Yung YL, DeMore WB, Pinto JP (1991) Isotopic exchange between carbon dioxide and ozone via O(1D) in the
stratosphere. Geophys Res Lett 18:13–16
Yung YL, Lee AYT, Irion FW, DeMore WB, Wen J (1997) Carbon dioxide in the atmosphere: Isotopic exchange
with ozone and its use as a tracer in the middle atmosphere. J Geophys Res 102:10857–10866
Zahn A, Franz P, Bechte C, Grooß J-U, Röckmann T (2006) Modelling the budget of middle atmospheric water
vapour isotopes. Atmos Chem Phys 6:2073–2090
Vol. 86 pp. 217–240, 2021 7
Isotopic Traces of Atmospheric O2 in Rocks,

Minerals, and Melts
Andreas Pack
Georg-August-Universität Göttingen
Geowissenschaftliches Zentrum
37077 Göttingen
Germany
apack@uni-goettingen.de
INTRODUCTION
In this chapter I review some of the knowledge about the oxygen isotope exchange
between air O2, rocks, minerals, melts, including some technical products. The rise of free
atmospheric molecular oxygen since the great oxygenation event at the Archean–Proterozoic
boundary 2.3 Ga ago (Farquhar et al. 2000; Luo et al. 2016) was one of the major events of
the Earth environment. The triple oxygen isotope composition of atmospheric O2 provides
exciting insights into the Earth atmospheric composition and biosphere productivity (Bender
et al. 1994; Luz et al. 1999, 2014; Luz and Barkan 2011; Young et al. 2014; Crockford et al.
2018). The triple isotope composition of O2 of the past atmosphere can directly be obtained
from ice cores (Blunier et al. 2002, 2012), but no such data are available for the time >1 Ma.
Here, I review how one can get information about the isotope composition of atmospheric O2
(especially Δ′17O) from rocks, minerals, and melts beyond the 1 Ma limit.
THE ISOTOPE COMPOSITION OF THE ATMOSPHERE

Modern air O2
The modern atmosphere contains (by volume) 21% O2 as the main oxygen-bearing
component. With respect to molecular oxygen being the main oxygen-bearing component,
this situation has been relatively stable at least during the Phanerozoic (for variations in O2
partial pressures, see, e.g., Tappert et al. 2013). Before the rise of atmospheric O2 2.3 Ga ago
(Farquhar et al. 2000), CO2 was likely the dominant oxygen-bearing species in the atmosphere
(e.g., Catling and Zahnle 2020). In the present atmosphere, water vapor is the second most
abundant oxygen-bearing species after O2 roughly amounting ~ 1–2% at sea level and rapidly
decreasing in concentration and mixing ratio with altitude. The upper troposphere and higher
altitude atmospheric layers are very dry and contain only ≤ 5 ppmv water vapor (Leblanc et
al. 2011). The third most abundant oxygen carrier in the modern atmosphere is CO2, currently
making up 0.04% and is increasing due to anthropogenic CO2 emissions. Hence, for the upper
atmosphere, 99.8% of oxygen is carried by O2. Oxygen atoms are the dominant form above
~120 km At sea level, 96.4% of the oxygen is carried by O2, 3.4% by vapor, and 0.2% by CO2.

1943-2666/21/0086-0007$05.00 (online)
218 Pack
The isotopic compositions are expressed in form of the δ (McKinney et al. 1950) and
Δ′17O notations with:
 17
O 
 16 
O Sample
 O 
17
 1   1000
 17
O 
 16 
 O Reference 
 18
O 
 16 
O Sample
 18 O   1   1000
 18
O 
 16 
 O Reference 
  17 O    18 O 
 17 O 
1000  ln   1   0.528  ln   1
 1000   1000 
One can debate whether the factor of 1000 is not already included in the ‰ notation.
As it is referred to McKinney et al. (1950), I added the factor of 1000. Throughout this
publication I use a Δ′17O of San Carlos olivine of −0.052‰ as anchor point (average value of
the studies of Pack et al. 2016; Sharp et al. 2016; Wostbrock et al. 2020) to ensure that δ17O is
reported relative to VSMOW scale (see also Miller and Pack 2021, this volume, and Sharp and
Wostbrock 2021, this volume). Literature data were recalculated accordingly.
Early and first isotopic data of air O2 have been reported from density measurements
of waters by Dole (1936). He measured the isotope composition of air O2 relative to the
composition of Lake Michigan water (δ18O = –5.9‰). The “air” water sample was prepared by
reacting atmospheric O2 with H2. In order to eliminate the effect of variable D/H ratios in the
waters, Dole (1936) prepared two equivalent electrolysis cells and electrolyzed the “air” water
and Lake Michigan water separately down to the same residual volume (40 ml). The generated
O2 from each water was purified and reacted with the same bottle of H2 to produce water for
density determination. The observed density difference between these two waters was only
6.0 ± 0.6 ppm, which transforms to a difference in atomic weights of oxygen in these waters of
0.000108 ± 0.00001 amu (Dole 1936). Casting this number into the now common δ18O notation
(McKinney et al. 1950), Dole (1936) determined that air O2 has a δ18O of 21.0 ± 1.5‰ relative
to VSMOW. This remarkable number is, within uncertainty, identical to later and much more
precise mass spectrometric measurements. Kroopnick and Craig (1972) measured the oxygen
isotope composition of air O2 by reacting the O2 with hot (800–900°C) graphite leading to its
conversion to CO2. During the reaction, CO2 was continuously separated by freezing it out with
liquid nitrogen in order to shift the reaction quantitatively from CO towards CO2. Because of
isobaric interference of 13C16O2 with 12C17O16O, no δ17O could be measured by these authors.
The reported δ18O was 23.5 ± 0.3‰. Thiemens and Meagher (1984) analyzed the δ17O and δ18O
of air by cryogenetic separation of N2 from O2 in 13X molecular sieves and subsequent mass
spectrometric analysis of the molecular oxygen. Nitrogen can be trapped at higher temperatures
on molecular sieve than O2 and Ar. Thiemens and Meagher (1984) obtained an average value
of δ18O = 23.43 ± 0.07‰ and Δ′17O = –0.12 ± 0.13‰ (1σ SEM; Δ′17O recalculated relative to a
slope 0.528 reference line). The large uncertainty in Δ′17O may be related to the fact that Ar was
not separated from O2 (which affects mass spectrometric sensitivity, see Barkan and Luz 2003)
and/or that a dual collector instead of the now-common triple-collector mass spectrometer
was used. Later, Thiemens et al. (1995) reported values for the atmospheric O2 from high-
altitude rocket launch experiments between 30 and 60 km altitude of δ18O = 23.4 ± 0.1‰
Isotopic Traces of Atmospheric O2 in Rocks, Minerals, and Melts 219
and Δ′17O = –0.4 ± 0.1‰. These air samples were treated as described in Thiemens and Meagher
(1984). Barkan and Luz (2005) measured the composition of VSMOW water relative to the
composition of their air reference (HLA, “Holy Land Air”). For the measurements, they applied
a correction for the effect of Ar contamination in the samples. The O2 of the VSMOW water
was extracted using the CoF3 fluorination method. They obtained a composition of their HLA of
δ18O = 23.88 ± 0.02‰ and Δ′17O = –0.453 ± 0.010‰; again reported relative to 0.528 reference
line. Later, Barkan and Luz (2011) revised their Δ′17O for air O2 to –0.507 ± 0.004‰. Young et
al. (2014) analyzed air O2 and report a δ18O of 23.533 ± 0.007‰ relative to VSMOW. The δ17O
of the reference gas tank was calibrated relative to O2 extracted from San Carlos olivine with an
assigned “Δ′17O” value of the San Carlos olivine of –0.004‰. Instead, the average Δ′17O of San
Carlos Olivine with both, δ17O and δ18O on VSMOW scale from Pack et al. (2016), Sharp et al.
(2016), and Wostbrock et al. (2020), one obtains a value of –0.052‰.
Young et al. (2014) present an atmospheric mass balance model and analyzed the triple
isotope composition of O2 released from San Carlos olivine and O2 extracted from air. Casting
the δ17O and Δ′17O of Young et al. (2014) into a framework (slope 0.528 reference line) relative to
San Carlos olivine with a Δ′17O = –0.052‰, a Δ′17O of air O2 of –0.422 ± 0.001 is obtained. Pack
et al. (2017) reported a δ18O of air of 24.15‰ and a Δ′17O of –0.424 ± 0.008‰ when recalculating
their datum relative to San Carlos olivine at –0.052‰. Yeung et al. (2018) reported a composition
of air O2 (δ18O = 23.6‰) with Δ′17O relative to San Carlos olivine of −0.421 ± 0.004‰.
Including data reported relative to UWG2 garnet and assuming UWG2 garnet having a Δ′17O =
–0.061‰, gives –0.423 ± 0.002‰ for Yeung et al. (2018). Wostbrock et al. (2020) also separated
air O2 from N2 and Ar by means of low-T (–80°C) gas chromatography (Yeung et al. 2012;
Young et al. 2014; Pack et al. 2017) and reported δ18O = 24.05 ± 0.11‰ and –0.441 ± 0.012‰.
For this publication, we consider the composition of atmospheric O2 to be δ18O = 23.9 ± 0.3‰
and Δ′17O = –0.432 ± 0.015‰ (1σ SD, weighted mean of Barkan and Luz 2005, Young et al.
2014, Pack et al. 2017, Yeung et al. 2018, and and Wostbrock et al. 2020; Fig. 1).
Including the datum of –0.507‰ by Barkan and Luz (2011) would pull the mean Δ′17O value
to –0.444‰. The reference gas used by Pack and Herwartz (2014) has been calibrated relative to
VSMOW-SLAP calibrated tank O2 relative to which the value of –0.507‰ has been measured.
The reported Δ′17O of San Carlos olivine in Pack and Herwartz (2014) then became –0.089‰
Figure 1. Plot of selected δ18O (Dole 1936; Kroopnick and Craig 1972; Thiemens and Meagher 1984;
Thiemens et al. 1995; Barkan and Luz 2005, 2011; Young et al. 2014; Pack et al. 2017; Yeung et al. 2018;
Wostbrock et al. 2020) (A) and Δ′17O (Barkan and Luz 2005, 2011; Young et al. 2014; Pack et al. 2017;
Yeung et al. 2018; Wostbrock et al. 2020) (B) values of air O2 vs. publication year. For calculation of the
average composition of air O2, we used published data by Barkan and Luz (2005), Young et al. (2014), Pack
et al. (2017), Yeung et al. (2018), and Wostbrock et al. (2020). Giauque and Johnston (1929) first reported
on 17O and 18O in air.
220 Pack
(recalculated relative to a slope 0.528 reference line), which is 0.051‰ lower than what later has
been measured by direct comparison of O2 from VSMOW and SLAP fluorination with O2 from
San Carlos olivine fluorination. Also, Sharp et al. (2016) and Wostbrock et al. (2020) obtained
higher Δ′17O values for San Carlos olivine than what was initially reported by Pack and Herwartz
(2014). From this line of evidence, we favor excluding the −0.507‰ value and suggest a mean
composition of air O2 of –0.432 ± 0.015‰ (see also Sharp and Wostbrock 2021, this volume).
Notably, the studies on Δ′17O of air O2 show much larger discrepancies than can be
explained by the respective reported analytical uncertainties only (see also Pack and Herwartz
2014). This shows that there are still systematic errors that need consideration.
Why is the reconstruction of the Δ′17O of air O2 from rocks interesting?
The isotope composition of rocks, minerals, and quenched melts is potentially preserved
over millions to billions of years. Finding rocks, minerals, and glasses that have interacted
with atmospheric oxygen may allow to reconstruct the past atmospheric isotope composition.
This provides information about the atmospheric O2 and CO2 mixing ratios and global primary
productivity (GPP, Luz et al. 1999; Young et al. 2014).
AIR–ROCK INTERACTION
Meteorite fusion crust
Meteorites are fragments of celestial bodies (Moon, planets, asteroids) that arrive
on the Earth′s surface. Meteorites hit the Earth′s atmosphere with velocities in the range
of 10 to 40 km s–1. Friction with atmospheric molecules and atoms leads to heating of the
meteoroid surfaces above their respective melting temperatures (Genge 2016). The interiors
of larger meteorioids remain cool and only the outermost millimeter-thick layer is affected.
Aerodynamic erosion of the melt layer leaves only a thin remaining fusion crust (Fig. 2).
Stony meteorite fusion crusts, however, are mixtures of the indigenous meteoritic
material and melt exchanged oxygen or even equilibrated with air O2 and other atmospheric
gases like CO2, but only when its partial pressure was higher than today (Payne et al. 2020).
The indigenous meteorite material has very large variations in Δ′17O, which makes isolation of
the atmospheric signature difficult.
In contrast, iron meteorites are composed of oxygen-free Fe, Ni alloy with minor sulfides.
Oxides and silicates occur only in trace amounts (Buchwald 1977). The fusion crusts of iron
meteorites are predominantly composed of magnetite (Fe3O4). The oxygen in that magnetite
entirely sources from the atmosphere. Heinzinger et al. (1971) reported the first oxygen isotope
analyses of iron meteorite crusts. Heinzinger et al. (1971) heated the oxidic fusion crust with
graphite (t = 1350°C) under vacuum and formed CO and CO2. The CO was quantitatively
Figure 2. Photography showing the 0.3 mm thin fusion crust of LL6 ordinary chondrite NWA 5882 (cour-
tesy of Svend Buhl, https://www.meteorite-recon.com).
converted to CO2 by glow discharge between Pt plates. The δ18O was analyzed on the CO2
by means of mass spectrometry. They observed δ18O = 17.6 ± 0.4‰ for fusion crusts
of unaltered iron meteorite falls (Braunau, IIAB; N′Goureyma, ungrouped; Trysa, IIIAB;
Sikhote Alin, IIAB; classification from the Meteoritical Society Meteoritical Bulletin
database). For the falls Braunau, Treysa, and Sikhote Alin, retardation heights of 15, 16, and
4.4 km were reported (Krinov et al. 1961), i.e., the fusion crust formed through oxidation by
tropospheric O2. The retardation height is the altitude, at which the strongest deceleration
and heating occurs. Heinzinger et al. (1971) concluded that the observed –7‰ fractionation
between the fusion crust and the atmosphere (using a value of δ18O = 23.9‰) is equilibrium
fractionation between molten Fe-oxides and air O2. They attributed lower δ18O values of
some fusion crusts to be the result of terrestrial weathering. Heinzinger et al. (1971) noted
that I-type cosmic spherules, which they regarded as ablation products of iron meteorites
and which are found in sediments, should record the δ18O of the ancient atmosphere with an
offset of –7‰. The δ18O of the atmosphere, i.e., the δ18O of molecular O2 then would provide
information about variations of the Dole effect on geological timescales.
More than a decade later, Clayton et al. (1986) analyzed the triple oxygen isotope
composition of fusion crusts of six different iron meteorites, including some of those analyzed
by Heinzinger et al. (1971) (Braunau, NGoureyma, Sikhote Alin, Treysa). The δ18O values
show a wide range from −10 to 15‰ with four out of the six meteorites clustering around 15‰
(Fig. 3), about ~2‰ lower than reported by Heinzinger et al. (1971). As did Heinzinger et al.
(1971), Clayton et al. (1986) concluded that the 8‰ difference (using a value of 23.9‰ for air)
between unaltered iron meteorite fusion crust and air O2 reflects magnetite–O2 equilibrium.
The triple isotope analyses of Clayton et al. (1986) allow one to test if the fusion-crust
magnetite oxygen is of atmospheric origin and if equilibration occurred between magnetite
and air O2 (Fig. 3).
The low Δ′17O of the fusion-crust magnetite that clusters around δ18O = 15‰ supports
the hypothesis that the oxygen is at least partly of atmospheric origin. In contrast, the fusion
crusts with δ18O < 15‰ have likely exchanged with meteoric water, removing much of the
atmospheric O2 signal. The lowest δ18O is observed in fusion crusts of ALHA 76002, an
Antarctic IIAB iron meteorite. The low δ18O reflects exchange with Antarctic snow and ice.
Meteori
c water
θ=
0.5
15
θ = 0.5305
Figure 3. Plot of Δ′17O vs. δ18O of iron meteorite fusion crust (data from Clayton et al. 1986) and air O2.
(δ18O = 23.9‰, Δ′17O = −0.431‰). The composition of meteoric waters (including surface freshwater) is
from a compilation of >1000 water analyses with data from Luz and Barkan (2010), Landais et al. (2008),
Affolter et al. (2015), Li et al. (2015, 2017), Surma et al. (2015), Touzeau et al. (2016), Surma et al. (2018)
Tian et al. (2018), and Tian and Wang (2019). The lowest δ18O was observed for the fusion crust of Ant-
arctic iron meteorite ALHA 76002.
222 Pack
The triple isotope data of Clayton et al. (1986), however, do not unequivocally support an
equilibrium fractionation of 7–8‰ between magnetite fusion crust and air O2 as suggested by
Heinzinger et al. (1971) and Clayton et al. (1986). The high-temperature approximation for the
equilibrium fractionation of oxygen is θ = 0.5305 (Matsuhisa et al. 1978; Young et al. 2002).
For high-temperature silicate and oxide mineral assemblages, Pack and Herwartz (2014)
reported θ values in the range of 0.5281–0.5290. Instead, a kinetic isotope fractionation effect
with a θ value around 0.515 is more compatible with the observations (Fig. 3). A short lifetime
of the molten Fe oxide layer during the atmospheric entry is caused by intense aerodynamic
erosion and likely prevents equilibration even at temperatures 1430 °C (minimum melting
temperature in the magnetite–wüstite phase diagram; Laughlin and Hono 2015). Although
not a high-temperature process, the chemisorption of oxygen on steel and copper is associated
with a large kinetic fractionation of 26 and 61‰, respectively (Dole et al. 1954). Notably, the
61‰ for chemisorption of oxygen on Cu coincides with the fractionation assuming diffusion
of O atoms and pure Graham′s law (Graham 1863) fractionation. The corresponding pure
Graham′s law θ would be 0.515, which could explain the composition of iron meteorite fusion
crust (Fig. 3). It should be emphasized here that the Δ′17O data by Clayton et al. (1986) are not
comparable in precision and accuracy as data obtained with modern mass spectrometers. Scale
distortion effects (Yeung et al. 2018) may also have played a role. However, even if the fusion
crust does not reflect isotope equilibrium with the atmosphere, triple oxygen isotope data of
“fossil” iron meteorite fusion crust would still have the potential of providing information
about the atmosphere. A 1‰ lower Δ′17O of atmospheric O2 would certainly be reflected in an
about 1‰ lower Δ′17O in the fusion crust. Data from unaltered, ancient “fossil” iron meteorite
fusion crusts, however, have not yet been reported.
Cosmic spherules
During their atmospheric entry, meteoritic particles in the millimeter size range are heated
and melted throughout to round spheres. These are termed cosmic spherules (Genge et al.
2008). A sub-group of these spherules are termed “I-type” cosmic spherules. These, often
magnetic spherules were correctly identified as extraterrestrial in origin when they were first
observed in deep-sea sediments collected during the research cruise of the HMS Challenger
in the late 19th century. Heinzinger et al. (1971) suggested that these spherules are ablation
products of iron meteorites. Although this view on their origin has not been supported by
succeeding studies, Heinzinger et al. (1971) correctly noted “As these spherules are found in
sediments of different geological ages, their oxygen isotope ratio can give information on the
development of atmospheric oxygen.” The oldest such cosmic spherules have been reported
from Archean carbonates (Tomkins et al. 2016).
It is now established that cosmic I-type spherules form by oxidation of small Fe,Ni metal
particles during atmospheric entry. Extraterrestrial Fe-Ni alloys with 5 ≤ wt.% Ni ≤ 50 are not
stable in the Earth atmosphere. Upon heating and melting during entry, the metal becomes
oxidized. The I-type cosmic spherules are composed of Fe oxides (magnetite, wüstite) and,
in some cases, residual Ni-rich metal. The oxides are Ni-poor compared to the original metal
because Ni behaves more siderophilic than Fe during the oxidation. Clayton et al. (1986)
reported the first triple isotope analyses of I-type cosmic spherules. Clayton et al. (1986)
observed that I-type cosmic spherules (as do iron meteorite fusion crusts) roughly fall on
a “terrestrial fractionation line”, then defined as δ17O = 0.52 · δ18O. Clayton et al. (1986)
observed 40 ≤ δ18O < 47‰ for I-type spherules; a range much higher than what they observed
for iron meteorite fusion crusts (δ18O ≈ 15‰). From the fusion crust data and their supposed
equilibrium with atmospheric O2, Clayton et al. (1986) concluded that the Fe oxides in the
small spherules equilibrated with the atmosphere with ~7–8‰ equilibrium fractionation.
They hence concluded that the upper atmosphere in ~100 km altitude has a δ18O in the range
of 36‰, which is 12‰ higher than tropospheric O2.
It was later reported by Davis et al. (1991) and Davis and Brownlee (1993) that Fe and
Ni (Davis and Brownlee 1993) in I-type cosmic spherules are highly enriched in the heavy
isotopes. The Fe isotope enrichment was reported to be 10–19‰ amu–1 and the Ni isotope
enrichment was reported to be 4–32‰ amu–1. Such extreme enrichment has neither been
observed in terrestrial nor in any meteoritic metal. Davis and Brownlee (1993) concluded
that between 74 and 94% evaporation led to the observed enrichment in Fe and Ni isotopes.
Engrand et al. (2005) reported O, Fe, Ni, and Cr isotope compositions of I-type cosmic
spherules from deep-sea sediments and confirmed strong isotope enrichments. Again, data
were explained by mass fractionation during evaporation.
During mass fractionation, fractionation among the different isotope ratios is highly
correlated so that mass-independent signals may be preserved. For oxygen this means that
the Δ′17O of air O2 may still be preserved in I-type cosmic spherules although they have seen
such massive evaporation. If so, Heinzinger et al. (1971) were right in their statement that
I-type cosmic spherules can provide unique information about the atmospheric composition
throughout the geologic history. In order to provide usable information about the 17O anomaly
of atmospheric O2 it is necessary to analyze the Δ′17O with an uncertainty < ~ 0.1‰, which is
currently not achieved by ion microprobe measurements (Engrand et al. 2005).
Pack et al. (2017) reported measurements of the triple oxygen and iron isotope
composition of Antarctic I-type spherules. They used laser fluorination in combination with
gas source mass spectrometry in continuous-flow mode and analyzed a small set of large
I-type cosmic spherules collected in Antarctica. They reported δ18O values between 36 and
42‰, well within the ranges reported by Clayton et al. (1986) and Engrand et al. (2005).
The aim of this study was to test the hypothesis that the Earth′s atmosphere is isotopically
homogenous up to ~ 100 km altitude. In order to quantify the degree of evaporation, Pack et
al. (2017) measured the Fe isotope composition of the spheres. In contrast to oxygen, there is
no Fe in the atmosphere and exchange with the atmosphere does not influence the Fe isotope
composition. Therefore, δ56Fe was regarded as a solid constraint for the degree of evaporation.
Wang et al. (1994) conducted evaporation experiments for FeO melts. They observed that the
δ18O and δ56Fe both increase with increasing degree of evaporation. From the experiments
of Wang et al. (1994), Pack et al. (2017) calculated an empirical relation for the Fe and O
isotope fractionation during evaporation with ln (δ18O + 1) = 1.18 · ln (δ56Fe). This relation
allowed a reconstruction of the pre-evaporation composition of the I-type spherules, i.e., the
composition of the spherules after melting and oxidation, but before evaporation. Spherules
with such a composition, of course, do not exist during atmospheric entry because melting and
oxidation and evaporation probably occur all at the same time. For the modeling, however, it is
useful to disentangle the two processes: fractionation during oxidation and during evaporation.
The suggested pre-evaporation oxygen isotope compositions of the studied I-type spherules
were in the range 2 ≤ δ18O ≤ 14‰, i.e. considerably lighter than air O2. The discrepancy
between air O2 (δ18O = 24‰) and pre-evaporative compositions of the spherules (δ18O ≈ 8‰)
is on the same order as that observed in iron meteorite fusion crusts (Fig. 3), where evaporation
does not play a role. This fractionation was attributed to kinetics during metal oxidation.
Fractionation of isotopes during evaporation is dominated by kinetic fractionation with an
associated θ that is generally “low”. This is true for both Fe and O. Although analytical precision
was likely lower than what nowadays can be achieved, the experiments by Wang et al. (1994),
indeed, show that during evaporation, the associated θ for triple oxygen is as low as 0.510.
It is interesting to note that also the θ for Fe is low, i.e., variations in θ can not only be observed for
a light element like oxygen, but also for heavy elements. Apart from the evaporation, Pack et al.
(2017) also considered kinetic fractionation during oxidation of the metal. As mentioned above,
very large isotope effects were observed by Dole et al. (1954) for the oxidation of steel (–26‰ in
18 16
O/ O) and oxidation of copper (–61‰ in 18O/16O). The high-T oxidation experiments by Pack
et al. (2017) were associated with a –4‰ fractionation in δ18O. Along with the iron meteorite
fusion rusts as natural experiments, an associated θoxidation = 0.506 was suggested.
224 Pack
In conclusion, Pack et al. (2017) demonstrated that the isotope composition of the Earth′s
atmosphere is homogenous with respect to Δ′17O up to ~100 km and that I-type cosmic
spherules are suitable tracer for the Δ′17O of the atmosphere. because Δ′17O of atmospheric
molecular O2 is function of GPP and atmospheric CO2 mixing ratios, fossil cosmic spherules
are very interesting proxies.
From air to sulfates and beyond
The triple isotope composition of sedimentary sulfates is reviewed in detail by Bao (2015),
Bao et al. (2016), and Cao and Bao (2021, this volume) and only a short overview is given here.
It has been observed that sedimentary sulfates can carry a distinctly negative Δ′17O when
compared to other sedimentary minerals (Bao et al. 2008). The only component known on Earth
that carries a negative anomaly in Δ′17O is atmospheric O2, which counter balances the positive
anomalies of O3 and stratospheric CO2. Bao et al. (2008) suggested that the negative anomaly
down to –0.7‰ that they have observed in Neoproterozoic barites is inherited by the sulfate from
the process of subaerial sulfide (e.g., pyrite) oxidation in an overall reaction:
2 FeS2 + 7.5 O2 + 4 H2O → Fe2O3 + 4 H2SO4
As such, sedimentary sulfate can be used to obtain information on the Δ′17O of atmospheric
O2. This information provides insights into the GPP and pCO2 (Bender et al. 1994; Luz et al.
1999; Blunier et al. 2002; Young et al. 2014). The uncertainty in this approach is dominated
by the uncertainly of the portion of oxygen sourcing from O2, (see Kohl and Bao 2011) but
also depends on the assumed GPP and the stratosphere–troposphere exchange mass flux (Cao
and Bao 2013; Graham et al. 2019). The high pCO2 estimates of up to 20,000 ppmv by Bao et
al. (2008) are based on the assumption of the same GPP in the Neoproterozoic as today. A
dynamic view of the evolution of atmospheric pCO2, pO2, and GPP in a post-Snowball Earth
meltdown world can be found in Cao and Bao (2013). Crockford et al. (2018) also reported
Δ′17O of Proterozoic sulfates with minimum Δ′17O of –0.88‰. They concluded that the GPP
was only 6–41% of modern GPP, depending on the assumed atmospheric O2 mixing ratio
(0.02–2 vol.%) and pCO2 (540–8100 ppmv).
Peters et al. (2020) investigated the triple oxygen isotope composition of magnetite
from iron deposits in Iran by means of high-precision laser fluorination. They observed that
magnetite from the Yazd iron oxide–apatite deposit shows a significant negative anomaly down
to –0.2‰, which Peters et al. (2020) related to the incorporation of oxygen from older sulfate
from late Proterozoic, early Cambrian evaporite rocks reported by Crockford et al. (2018)
from that region. These evaporites, in turn, had obtained a negative anomaly inherited from
subaerial sulfide oxidation as originally suggested by Bao et al. (2008). By their triple oxygen
isotope data, Peters et al. (2020) could exclude that the magnetite from the Yazd deposit last
equilibrated with “normal” magmatic fluids.
The study of Peters et al. (2020) shows that Δ′17O anomalies can be a robust tracer for
the pathway of oxygen in complex geologic processes. On its pathway from air through pyrite
oxidation, sulfate dissolution, sulfate reduction, fluid mobilization and participation in magnetite
precipitation, much of the Δ′17O has been diluted with “normal” oxygen but some atmospheric
signature is preserved. The observed Δ′17O can therefore only be regarded as a maximum value
during the time of evaporite precipitation with limited information on GPP · pCO2. Surely, direct
analyses of the sulfate as reported by Crockford et al. (2018) from the same region provide more
information. Nevertheless, the study by Peters et al. (2020) is yet another trace of anomalous
atmospheric oxygen in rocks and can be applied in cases where no evaporite is preserved and/or
accessible.The magnetite Δ′17O provides information about the minimum anomaly in 17O and
hence provides a minimum GPP · pCO2 number. Caution must be exercised when dealing with
sulfate Δ′17O that are within ±0.25‰, as various mass-dependent processes can also impart
small, yet variable 17O anomalies for sulfate (Cao and Bao 2021, this volume).
Deep sea manganese nodules

Large portions of the deep oceans floor are covered by small, centimeter-sized
ferromanganese nodules. These nodules formed by oxidation of dissolved Mn2+ and
precipitation of Mn-IV-oxides. The oxidation of Mn2+ requires a strong oxidizer, which
is dissolved O2. The isotope composition of seawater dissolved O2 is closely linked to the
composition of air O2, which makes ferromanganese nodules a potential archive for the isotope
composition of tropospheric O2, especially the Δ′17O.
As in the case of sedimentary sulfate or cosmic spherules, oxidation of Mn2+ by O2
involves mass-transfer of O2 into the resultant Mn-IV-oxides (“MnO2”, manganates). Unlike
in the process of oxidation of ferric to ferrous iron in aqueous solution, the isotope signature
of O2 is partially preserved in the MnO2 phases. Mandernack et al. (1995) demonstrated that
experimentally precipitated manganate (abiological and biological) contains 32–50% oxygen
from dissolved O2 with the remaining oxygen in the manganate sourcing from H2O. This is in the
same range as suggested for the pyrite-derived sulfate (8–15%, Balci et al. 2007; 0–60%, Kohl
and Bao 2011). The isotope fractionation between O2 and manganate, however, was measured
to vary between ~ 0 and –22, which complicates the interpretation of ferromanganese nodule
oxygen isotope data. The fractionation between H2O and manganate precipitate was measured
to vary between –5 and 1‰ (Mandernack et al. 1995), which is similar to the fractionation
between H2O and iron oxides. For the interpretation of ferromanganese nodule data, it also
has to be considered that they are a mixture of manganate, authigenic silicates, and Fe oxides.
The authigenic silicates and the Fe oxides are not expected to carry the anomalous signature
of atmospheric O2. A further aspect that needs consideration when using manganese nodules
as a paleoatmospheric-archive is that the isotope composition of seawater dissolved O2, i.e.,
the Mn2+ oxidizer, is not identical to that of tropospheric O2, the eventual quantity of interest.
The equilibrium fractionation between dissolved O2 in water and air at 25°C is
1000 · ln(αO2,diss.–air) = 0.7‰ (Benson and Krause 1984; Li et al. 2019), i.e., the oxygen dissolved
in the ocean surface layer in immediate contact with air has δ18O ≈ 25‰. The corresponding
Δ′17O is close to the value for air O2, i.e., –0.432‰. This oxygen is mixed with photosynthetic
O2, which has an isotope composition close to that of seawater. Photosynthesis in the photic
zone produces O2 with δ18O ≈ Δ′17O ≈ 0‰. Mixing of anomalous air O2 with such O2 leads
to an increase of dissolved O2 Δ′17O with increasing O2 concentration (e.g., Luz and Barkan
2000, 2009, their Fig. 2). At depths of a few hundred meters, respiration consumes dissolved
O2 and enriches the remaining dissolved O2 in 17O and 18O relative to 16O (Dole effect, Dole et
al. 1954). In this oxygen minimum zone, Kroopnick and Craig (1976) measured δ18O values
for the dissolved O2 as high as 38‰. Respiration is a kinetic process and is accompanied by
a θresp. in the range of 0.518, i.e., respiration leads to a decrease in Δ′17O with increasing δ18O
when reporting Δ′17O relative to a reference line with slope 0.528. The precise value for θresp.
depends on a variety of factors and it is referred to recent works by Stolper et al. (2018) or Ash
et al. (2020) for details. Only a small number of studies report the triple isotope composition of
O2 dissolved in abyssal regions of world oceans (here, we consider data from depths >1000 m).
Recasting the data reported by Yeung et al. in the online data repository at https://www.bco-
dmo.org/dataset/753594 as of 2020 with δ17O and δ18O on VSMOW scale and Δ′17O relative to
a slope −0.528 reference line gives deep-Pacific data with 32 ≤ δ18O ≤ 40‰ and Δ′17O falling
along a slope ~ 0.518 trend (Fig. 4). This trend coincides with the respirators fractionation trend,
but is a result of complex relations between fractionation and mixture processes (Hendricks et
al. 2005; Nicholson et al. 2014). The fan-shape of the area covered by data (Fig. 4) indicates
that mixing between photosynthetic O2 and anomalous air O2.
226 Pack
Recently, Sharp et al. (2018) and Sutherland et al. (2020) reported the first high-precision
triple oxygen isotope data of ferromanganese nodules from the ocean floor. Sutherland et al.
(2020) clearly identified anomalous oxygen in their set of ferromanganese nodules (Fig. 4).
Their interpretation, in line with experimental work of Mandernack et al. (1995), is that these
marine Mn-IV-oxides preserved the anomaly of dissolved O2 and hence are potential archives
for dissolved O2 that had exchanged with atmospheric O2.
Here I discuss the interpretation of the Sutherland et al. (2020) data in terms of hypothetical
Mn oxide endmembers. Mn oxides that form in equilibrium with seawater would have a
composition as marked “Mn oxide I” in (Fig. 4). If Mn oxides formed by oxidation of deep sea
dissolved O2 only, their composition is expected to plot somewhere between the composition of
the dissolved O2 (“Mn oxide II”; no fractionation between O2 and Mn oxide) and a composition
marked “Mn oxide III” with 20‰ kinetic fractionation between O2 and Mn oxide (Mandernack
et al. 1995). A θkinetic of 0.515 has been assumed (Sutherland et al. 2020) and we adopt this
value here. For details on mechanisms of O2 binding and O2 reduction I refer the interested
reader to, e.g., Ash et al. (2020). Assuming that the Mn is oxidized in ocean layers that were
Global photosynthetic O2
Mn oxide (I), Clay Marine authigenic
Fe oxides silicates
Silica
Mixing
Mn-oxide (V)
Stratospheric MIF
varies with GPP and pCO2
Mn-oxide (III)
Kinetic fractionation Deep sea
dissolved O2
(>1000 m)
Mn-oxide (IV)
Mn-oxide (II)
Figure 4. Plot of Δ′17O vs. δ18O illustrating the oxygen pathways and sources of deep-sea ferromanganese
nodules (brown filled circles, Sutherland et al. 2020). I assume that photosynthetic oxygen (red filled hexa-
gon) is a mixture of marine (red filled pentagon) and terrestrial (green filled diamond; average composition
of leaf water) photosynthesis. The seawater composition (light blue filled circles) is from Barkan and Luz
(2011). The marine authigenic clay composition is from Sengupta and Pack (2018). The silica composition is
from Sharp et al. (2016). The dark grey filled area outlines oxygen mixing between authigenic minerals (clay,
silica) and seawater. The dissolved O2 isotope data (light gray filled circles, >1000 m: red filled crosses) are
taken from Yeung et al.’s dataset https://www.bco-dmo.org/dataset/753594 as of 2020. “Mn oxides I” outlines
the composition of Mn and Fe oxides in equilibrium with seawater. “Mn oxides II” have acquired the entire
oxygen from deep-sea dissolved O2 without fractionation (Mandernack et al. 1995), whereas “Mn oxide IV”
would have acquired the oxygen from air equilibrated dissolved O2. “Mn oxide III and V” would have formed
from the respective dissolved O2 under participation of a 20‰ kinetic isotope fractionation (Mandernack et al.
1995). The light gray shaded area outlines the possible field for bulk ferromanganese nodules.
well-mixed with the atmosphere, and no fractionation would occur during the oxidation,
Mn oxide would have a composition identical to that of dissolved and air equilibrated O2 (“Mn
oxide IV”). This composition has been considered as one endmember by (Sutherland et al. 2020).
If 20‰ fractionation is associated with the oxidation of Mn (Mandernack et al. 1995), one
would get a Mn oxide endmember with a composition labeled “Mn oxide V”. This is the
second endmember considered by Sutherland et al. (2020). Any of these pure endmembers are
likely to exist. The experiments by Mandernack et al. (1995) show that during the oxidation,
also equilibration with the surrounding water plays a role and only ~40% of the oxygen in
the Mn oxides sources from O2. This means that the Mn oxides in ferromanganese crusts are
expected to have compositions on the mixing trends between “Mn oxides II–V” and the water
equilibrated “Mn oxide I”. Because the ferromanaganese nodules form in the deep sea, a mixture
between “Mn oxide II” and “Mn oxide III” should be considered rather than “Mn oxide IV” and
“Mn oxide V”. Bulk ferromanganese nodules are mixtures of Mn and Fe oxides, and silicates.
The Fe oxides will have a composition close to that of seawater. Following the suggestion by
Sutherland et al. (2020), the silicate fraction can be approximated by a mixture of silica (Sharp
et al. 2016) and clay (from Sengupta and Pack 2018; orange filled region in Fig. 4).
The composition of bulk ferromanganese nodules is expected to fall in the light gray
shaded area, which is, indeed, the case (Fig. 4). Because of their deep-sea formation and
considering 40% O2 in the Mn oxides, the data by Sutherland et al. (2020) can be explained
by “Mn oxide III” as one endmember, mixed with seawater-equilibrated Fe and Mn oxides
and various portions of silicates. In such a model, those deep-sea ferromanganese nodules
with Δ′17O below the “Mn oxide II,III”–“Mn oxide I, Fe oxide” mixing trend would be
explained by a lower Δ′17O of atmospheric O2. A lower Δ′17O of atmospheric O2 could be due
to elevated pCO2 and/or decreased GPP (Bender et al. 1994; Luz et al. 1999; Young et al. 2014).
As a quantitative paleo-CO2 and paleo-GPP proxy, more data are required to better understand
the particular fractionation processes during the crust formation. Nevertheless, the deep-sea
ferromanganese nodules are an interesting target for future triple oxygen isotope studies.
Tektites
Air melt exchange. Urey (1955) discussed the origin of tektites and concluded that they
are of extraterrestrial origin. He based his argument on the observation of tektite liquidus
temperatures that are beyond temperatures known from magmatic processes and their (inter-
group) chemical homogeneity that is unrelated to the local rock chemistry. He also noted that
tektite occurrences are unrelated to the spatial distribution of terrestrial volcanism. Based on
their distribution on Earth, Urey (1955) speculated that tektites may have arrived from the
Moon. The chemical composition of tektites, however, is very similar to that of the Earth′s
crust and makes a lunar origin less likely. Two decades earlier, Spencer (1933) had put forward
the now-established idea that tektites formed during melting end ejection of target rock during
a meteorite impact. He based his suggestion on the occurrence of tektites in the vicinity
of impact craters and on the chemical similarity of tektites and the Earth surface material.
A review on the origin of tektites on basis of geochemistry is given by Koeberl (1988, 1994).
The first oxygen isotope data of tektites was published by Silverman (1951). He analyzed
tektites from the Phillippines, which belong to the Australasian strew field, and from Bohemia
(moldavites), which were ejected from the Ries crater in Southern Germany ~15 Ma ago.
For both glasses, Silverman (1951) measured δ18O = 10.4‰. He noted that their composition
is different from stony meteorites, but similar to terrestrial sedimentary rocks. He leaves the
reader with the options that tektites are either of terrestrial origin or of extraterrestrial origin and
that similar sedimentary processes that enrich rocks in 18O operate on other bodies of the Solar
System. Taylor and Epstein (1962) analyzed a wider range of tektites and found δ18O in a narrow
range between 9.6 and 10.4‰. They noted that the oxygen-isotope composition resembles that
228 Pack
of felsic igneous rocks, whereas their chemistry rather agrees with that of sedimentary rocks.
Sedimentary rocks, however, typically have δ18O > 10‰. In conclusion, Taylor and Epstein
(1962) suggested that tektites were of extraterrestrial origin. The total range of published
tektite data is 7 ≤ δ18O ≤ 15‰ (see Zák et al. 2019, and references therein), i.e., covering the
range typical of sedimentary rocks.
The first triple oxygen isotope data for tektites were published by Clayton and Mayeda
(1996). They reported data for 14 tektites of different origin. Their δ18O values range between
8.6 and 10.3‰, typical for continental surface rocks (e.g., Bindeman 2021, this volume).
All tektite data are in the δ18O range typical for sedimentary rocks and are by <10‰ lighter
than air O2. If one assumes that the equilibrium fractionation at high temperatures is small, this
suggests little exchange between atmospheric O2 and tektites. This conclusion is supported
by the Δ′17O values of Clayton and Mayeda (1996). Casting their data into the λRL = 0.528
framework, their average Δ′17O is –0.08‰ with a 1σ standard deviation of the mean 0.05‰
(Fig. 5). This puts the tektites well in the field of terrestrial rocks (Sharp et al. 2018; Bindeman
2021, this volume), but is much higher than the –0.432‰ of air O2. More triple isotope data
of tektites were published by Zák et al. (2019) on tektites from the Austalasian strew field
(Fig. 5). Their δ18O values range from 8.7 to 11.6‰, a typical range for sedimentary rocks.
The reported Δ′17O values (cast into λRL = 0.528 notation) range from –0.07 to –0.04 with
an average of –0.055 and a standard deviation of 0.010‰. As in case of the data by Clayton
and Mayeda (1996), the new data fall well within the range typical of terrestrial rocks.
No indication for exchange with air O2 is observed in the dataset by Zák et al. (2019).
Earth
Crustal rocks
mantle
end
mixing tr
ceous
Carbona
M
ix
chondrite
in
g
Equilibrated
melt
Air
Figure 5. Plot of Δ′17O vs. δ18O of triple oxygen isotope data of tektites (data are from Clayton and Mayeda
1996, Magna et al. 2017, and Zák et al. 2019). Most tektites follow a narrow trend with Δ′17O close to the
composition of the Earth mantle with δ18O typical of felsic igneous and sedimentary rocks (e.g., Pack and
Herwartz 2014; Bindeman et al. 2018, 2019; Bindeman 2021, this volume). For the Clayton and Mayeda
(1996) data an uncertainty in Δ′17O of ± 0.1‰ is assumed. The more recent data have an uncertainty of only
0.01‰ and demonstrate the progress made in the analytical protocols for triple oxygen isotope analyses.
The irghizites follow a mixing trend between target material and silicate melt that equilibrated with air O2
(red filled pentagon; see explanation below).
Magna et al. (2017) published triple oxygen isotope data on Moldavites and glasses from
the ~ 1 Ma old Zhamanshin impact crater in Kazahstan (48°24′N 60°58′E, see also Florenski
1977, Bouşka et al. 1981, and Koeberl and Fredriksson 1986; Fig. 5). The δ18O values of
the Moldavites and the basic splash forms from the Zhamanshin impact site (excluding the
irghizites) range from 7 to 12‰ and fall within the range of tektites from other localities.
The Δ′17O values for moldavites, i.e., the ejecta from the ~ 15 Ma old Ries impact structure
(48°53′N, 10°32′E), and the basic splash forms from the Zhamanshin impact range from
–0.068 to –0.033‰ and are indistinguishable from sedimentary rocks with similar δ18O (Sharp
et al. 2018; Bindeman 2021, this volume).
Only the felsic irghizites ejected from the Zhamanshin crater plot outside the field typical
for sedimentary and igneous rocks and show distinct lower Δ′17O values (Fig. 5, red filled
diamonds). For these impact glasses, lower Δ′17O could either come from mixing of target rock
with a carbonaceous chondrite impactor or from exchange with low Δ′17O atmospheric oxygen.
Magna et al. (2017) excluded that the low Δ′17O is due to admixing of a carbonaceous chondrite
component. If so, a steep vertical trend is expected (Fig. 5). Instead, the irghizites follow a trend
that can broadly be explained by exchange of the melt with atmospheric oxygen. Magna et al.
(2017) assumed that the fractionation between silicate melt and O2 is negligible at the high
temperatures of the impact. The age of the Zhamanshin crater is ~ 1 Ma. At this time, the Δ′17O
of the atmospheric O2 was similar to today, i.e. ~ –0.4‰. It is noted here that glacial–interglacial
variations in atmospheric Δ′17O amount for as much as 0.06‰ in Δ′17O (Blunier et al. 2012).
Why do the irghizites appear to have exchanged with the atmosphere whereas the
Zhamanshin basic splash forms, Moldavites and Australites do not? The sizes of the
Zhamanshin crater (source of the irghizites) and the Ries crater (source of the Moldavites) are
both 15 km, so it is not likely the intensity of the impact that makes the difference. Irghizites are
found close to the crater, whereas Moldavites and Australites have been transported hundreds
to thousands of kilometers away from the impact site. For these proximal tektites, Mizera et
al. (2012) suggested longer heating times compared to the distal tektites. Another feature that
distinguishes the irghizites from other, distal tektites is the “effective” size. The irghizites are
often composed of agglomerates of small former melt spherules (Mizera et al. 2012), which
reduces the “effective” size, i.e., the size relevant for estimating the exchange rate with the
ambient atmosphere. The surface/volume ratio of these spherules is orders of magnitude larger
than for the monolithic distal tektites. Both prolonged heating times and smaller “effective”
size of the irghizites compared to Moldavites and Australites could explain why the irghizites
exchanged oxygen with the atmosphere, whereas the latter two did not.
However, are the short heating times in the range of a few minutes at maximum sufficient
to exchange ~ 20% oxygen with the atmosphere as has been concluded by Magna et al. (2017)?
In order to test what heating times are required to exchange oxygen between O2 and silicate melts,
we conducted experiments in our laboratory. Silicate samples (49 wt.% SiO2, 23 wt.% CaO, 11
wt.% Al2O3, 17 wt.% MgO, Tliquidus ≈ 1350 °C, phase diagram from Winter 2001, p. 106) were
melted for different times (5 s–50 min) in a Gero vertical gas mixing furnace using the Pt-loop
technique (Donaldson et al. 1975). Iron was avoided in order to reduce complexity due to redox-
related mass transfer between the melt and the atmosphere and the melt and the Pt loop. The
diameter of the sample loops was roughly 2–3 mm. Temperatures were 1350 °C and 1500 °C, both
above the liquidus of the chosen composition. The furnace atmosphere was air. The oxygen isotope
composition of the samples was analyzed by laser fluorination in combination with gas-source
mass spectrometry. At time of the project, problems with the source controller board prevented use
of the Thermo MAT253 mass spectrometer and a gas chromatograph. Therefore, all analyses on
the starting materials and run products were conducted offline, i.e., by transferring the samples in
glass vials filled with 5Å molecular sieve to a Finnigan DeltaPlus mass spectrometer and without
gas chromatographic purification. Therefore, the uncertainty in Δ′17O was ± 0.03‰. San Carlos
olivine was used as standard. The results are listed in (Table 1) and illustrated in (Fig. 6).
230 Pack
The data demonstrate that 20% exchange between silicate melt and atmospheric O2
occurs within < 1 min. It is therefore conceivable that the spherules forming the irghizites have
exchanged with the atmosphere during their minutes-long residence time in the hot impact
plume (Shuvalov and Dypvik 2013). A second result of the experiments is that the equilibrium
fractionation between air O2 and silicate melt at 1350 and 1500 °C is still about −2.5‰ (Fig. 6).
Although associated with a large uncertainty, an associated high-T θ of 0.5290–0.5305 is
compatible with the observation. Also, the data suggest that the silicate analyses and the
analyses of air O2 are on the same scale for δ17O. High-T equilibration of silicate melt with
excess air O2 using the air data by Pack et al. (2017) or Wostbrock et al. (2020) could be a
means of calibrating the Δ′17O of laboratory reference gases relative to the VSMOW.
The data by Magna et al. (2017) show that tektites can have ~ 20% exchange with
atmospheric oxygen, if the individual melt droplets are in the millimeter size range. In order to
use fossil tektites as a proxy for pCO2 · GPP, it is necessary to extrapolate the tektite data toward
the locus of equilibrated melt at a δ18O of about 21.5‰ in order to account for the equilibrium
fractionation between O2 and silica melt. The use of tektites as a paleo-CO2 · GPP proxy,
however, requires that the data plot along a narrow trend instead of showing large, uncorrelated
variations in δ18O and Δ′17O as observed by Magna et al. (2017) (Fig. 5).
The experimental data (Fig. 6) suggest that at 1350–1500 °C exchange occurs within tens
of seconds to minutes, timescales that can be realized in impact plumes. Temperature in the
impact plume could well have been higher than the experimental temperatures (Macris et al.
2018), which would further accelerate the exchange. It is therefore conceivable that detailed
analyses of, e.g., only outer layers of tektites could show more exchange or even equilibration
and thus would make a suitable proxy for the Δ′17O millions of years back from now.
Table 1. List of the experimental samples and the corresponding oxygen isotope data (Stübler 2017).
The Δ′17O is reported relative to a slope − 0.528 reference line.
Sample Comment δ17O δ18O Δ′17O
CSS Starting material 3.78 7.36 −0.10
CS5−1350 t = 5 s, T = 1350 °C 5.06 9.97 −0.19
CS10−1350 t = 10 s, T = 1350 °C 6.1 11.96 −0.20
CS40−1350 t = 40 s, T = 1350 °C 6.42 12.77 −0.30
CS60−1350 t = 60 s, T = 1350 °C 6.04 11.9 −0.23
CS160−1350 t = 160 s, T = 1350 °C 7.57 14.87 −0.26
CS640−1350 t = 640 s, T = 1350 °C 10.89 21.58 −0.44
CS1280−1350 t = 1280 s, T = 1350 °C 10.68 21.21 −0.46
CS2560−1350 t = 2560 s, T = 1350 °C 10.54 21 −0.49
CS2997−1350 t = 2997 s, T = 1350 °C 10.5 20.78 −0.42
CS10−1500 t = 10 s, T = 1500 °C 7.38 14.55 −0.27
CS15−1500 t = 15 s, T = 1500 °C 8.4 16.62 −0.34
CS35−1500 t = 35 s, T = 1500 °C 9.87 19.59 −0.42
CS105−1500 t = 105 s, T = 1500 °C 10.82 21.43 −0.43
CS240−1500 t = 240 s, T = 1500 °C 10.85 21.37 −0.37
CS960−1500 t = 960 s, T = 1500 °C 11 21.75 −0.42
San Carlos olivine average (N = 7) 2.69 5.19 −0.05
10%
20%
5s
40%
High-T θ
60% (0.529-0.5305)
15 s 70%
80%
High-T equilibrium:
1000 ln αsilicate melt–O2 = –2.5‰
Figure 6. Plot of Δ′17O vs. δ18O of silicate melt (red filled diamond) and air (gray filled star) equilibrated
for different times at 1350°C (orange) and 1500°C (yellow). At prolonged equilibration times, silicate melt
and air O2 come into equilibrium with a fractionation of ~ –2.5‰. The study was conducted within a BSc
thesis project by Christian Stübler in 2017. The data are listed in (Table 1).
Air inclusions in tektite glasses

Tektites may preserve information about the atmosphere at time of their formation other
than by having isotopically exchanged. Instead, bubbles of air could potentially be preserved
in unaltered tektite glass for thousands or millions of years.
Zähringer and Gentner (1963) reported K–Ar ages for tektites from different strew fields.
In addition, they report on excess gases from bubble-rich tektites. The measured 40Ar/36Ar ratios
varied between 296 and 305, which overlaps with the value of modern air of 296.2 (Mark et al.
2011). The gas composition in bubbles of indochinites, moldavites, Libyan desert glass, and glass
from the Ries impact site in Nördlingen was further investigated by means of gas chromatography
by Müller and Gentner (1968). The discovered O2 with and observed a N2/O2 ratio varied
around 4, which is close to the modern atmospheric value of 3.7 (from the volume mixing ratio).
The measured N2/CO2 ratios for the Muong Nong tektites, which are younger than 1 Ma, vary
between 16 and 160 and are much lower than the ratio of ~3000 about 0.7 Ma ago. Jessberger
and Gentner (1972) studied the gas composition (N2, O2, CO2, CO, SO2, Ar, Kr, Xe) in bubbles
from the Muong Nong tektites and from Libyan desert glass by means of mass spectrometry.
Samples were crushed in vacuum and the released gases were analyzed in a quadrupole mass
spectrometer. They reported noble gas isotope ratios resembling that of modern air. The N2/O2
ratios all exceeded the atmospheric value by up to a factor of a thousand. Jessberger and Gentner
(1972) suggested that O2 was consumed by an unspecified oxidation process.
To date, the author is not aware of any other isotope data of the O2 from the inclusions
in tektites. Even if the δ18O would have been modified by mass-dependent fractionation,
e.g., thorough preferential uptake of 16O during oxidation processes, the Δ′17O could still
be reconstructed. Also, diffusional loss of gas and contamination by recent air needs to be
considered. Small amounts of gas, however, make such measurements an analytical challenge.
232 Pack
Skeletal apatite and eggshell carbonate

The oxidation of carbohydrates, protein, and fat by inhaled O2 in animals is accompanied
by mass transfer from anomalous O2 into body water. From that body water, skeletal apatite
(bones and teeth) and calcite (eggshells) are precipitated and can, in principle, record the triple
isotope composition of the respective body fluid (see Kohn 1996). If that animal lived in times
of elevated CO2 concentrations, Δ′17O of the inhaled O2 was lower and hence the Δ′17O of body
fluid would have been lower. The Δ′17O signature is then transferred from the body water into
bioapatite and egg shell carbonate, where its being preserved.
This approach has been introduced by Pack et al. (2013) for bioapatite and Passey et al.
(2014) for eggshell calcite. A few applications have since been published (Pack et al. 2013;
Passey et al. 2014; Gehler et al. 2011, 2014). Details about that approach are summarized by
Passey and Levin (2021, this volume).
High temperature technical products
Technical magnesia and alumina. In this section, I will present a few data on technical
products that had interacted with atmospheric O2. Sintered magnesia (periclase, MgO,
Tmelting = 2852 °C) and alumina (corundum, Al2O3, Tmelting = 2072 °C) are common materials
for high-T applications. Magnesia is produced by thermal decomposition of magnesite
(MgCO3) or brucite (Mg[OH2]; e.g., Drnek 2018; Herbrich et al. 1990). The reaction product
is so-called caustic magnesia, which is extremely fine-grained and highly reactive. For high-T
refractories, coarse-grained magnesia is required, which is obtained by sintering or melting of
caustic magnesia (Drnek 2018). Magnesia sintering requires temperatures of 1800–2200 °C.
Electric arc melting reaches temperatures exceeding 2852°C to melt the magnesia.
Sintering includes destruction and reconstruction of the crystal structure. Such process
may allow the exchange of oxygen between the furnace atmosphere (air) and the oxide
(alumina, magnesia) refractory. Pure diffusional exchange is a slow process, even at very high
temperatures and is not considered here to be important for the exchange of oxygen between
the atmosphere and refractories.
Herbrich et al. (1990) studied the isotope composition of magnesia during the decalcination
and the sintering in process. They fluorinated magnesia using F2 gas in conventional Ni reaction
vessels. The furnace atmosphere during the Czochralski growth process was oxidizing.
They observed that the δ18O of the magnesia products changes towards the composition of
the furnace atmosphere, which was measured to be identical to that of air O2. The sintered
pellets had a higher δ18O on their outer parts compared the interior, which had δ18O ≈ 13‰.
A similar observation has been made by Pack et al. (2005), who studied the oxygen isotope
composition of a 3.5 cm large sintered magnesia pellet. The magnesia was analyzed by means
of laser fluorination with F2 as oxidant in combination with continuous flow gas source mass
spectrometry (Pack 2000; Pack et al. 2005). The δ18O increased from 8‰ in the interior towards
15‰ at the surface of the pellet. Pack et al. (2005) reported bulk δ18O = 14‰ for three different
magnesia raw materials used for steelmaking, which is close to the bulk number of 13‰
reported by Herbrich et al. (1990). Herbrich et al. (1990) present a detailed photography of
the sintered magnesia that revealed newly formed cubes of periclase. The isotope composition
of this magnesia as measured to δ18O = 22‰, i.e., very close to the composition of air O2.
The idiomorphic crystals of magnesia either formed by sublimation and re-sublimation of MgO
or by localized magnesia reduction and re-precipitation by oxidation with O2 from the furnace
atmosphere (Herbrich et al. 1990). Herbrich et al. (1990) favor the reduction in combination
with evaporation and oxidation in combination with re-sublimation pathway and suggest that
carbon is the reducing agent. The difference of 1.9‰ between air O2 and magnesia could
reflect the high-T equilibrium between magnesia and O2. High-T equilibrium fractionation
of the same order (–2.5‰) has been observed for the silicate melt O2 equilibrium (Fig. 6).
No triple oxygen isotope data have been published of technical magnesia. It is expected that
the newly-formed magnesia described by Herbrich et al. (1990) has a Δ′17O that is about
0.004‰ lower than that of air O2, i.e., –0.435‰. This estimate is based on a θ of 0.53 for the
magnesia-O2 high-T equilibrium.
As in the case of magnesia, sintering of alumina may also have an effect on its triple
oxygen isotope composition. The bauxite raw materials have δ18O values in the range of
6–13‰ (Bird et al. 1989, 1993; Ellahi et al. 2016) expected to fall in the Δ′17O field typical
for low-T rocks and minerals. The Bayer process (Habashi 1995) includes dissolution of
bauxite and precipitation as Al(OH)3, which may lead to mass-dependent fractionation due
to exchange with the solvent. Alumina is then produced from the hydroxide by dehydration
at 1200–1300 °C, a step that likely also includes mass-dependent fractionation effects and,
possibly, already some exchange with the atmosphere. The major exchange step, however,
would be the sintering, which is done at >1600 °C. For two alumina materials used as refractory
in steelmaking Pack et al. (2005), determined 18‰ ≤ δ18O ≤ 20‰.
So far, no triple oxygen isotope analyses have been published on high-T refractories.
A PostDoc from my group, Dr. Nina Albrecht, conducted first triple oxygen isotope
measurements on technical alumina (Table 2). The material was a densely sintered alumina
crucible for application in high-T furnace experiments. She obtained a δ18O of 16.5‰ and a
Δ′17O as low as −0.270‰ (Fig. 7).
The δ18O of the sintered alumina is close to the data reported by Pack et al. (2005) for
alumina refractories. The Δ′17O of the sintered alumina is clearly much lower than what has
been measured for natural rocks and minerals (Fig. 7). A simple mass balance indicates ~60%
exchange with atmospheric O2 during sintering (Fig. 7). For comparison, Dr. Albrecht analyzed
a set of natural corundum, ruby, and sapphire samples (Table 2). They range in δ18O from 12 to
26‰, but have Δ′17O values typical of sediments (Fig. 7). For comparison, Giuliani et al. (2005)
reported similarly spreading δ18O values of 3 to 23‰ from various ruby and sapphire deposits.
Crustal rocks
Raw materials
Figure 7. Plot of Δ′17O vs. δ18O of natural (purple filled crosses, corundum, ruby sapphire) and sintered
technical (yellow filled triangles) alumina. The range of typical terrestrial crustal rocks and data for San
Carlos olivine (light green filled circles) and air O2 (gray filled star) are displayed (e.g., Pack and Her-
wartz 2014; Bindeman et al. 2018, 2019; Bindeman 2021, this volume). Data for synthetic corundum start-
ing materials (pentagon, diamonds), and Czochralski grown ruby and sapphire crystal (red filled square)
are displayed. The corresponding data are listed in Table 2.
234 Pack
Table 2. List with description and oxygen isotope data for the analyzed corundum samples.
The materials were provided by Dr. Klaus Dupré of the Forschungsinstitut für Mineralogische
und Metallische Werkstoffe Edelmetalle/Edelsteine (FEE) in Idar Oberstein and by Dr. Alexander
Gehler from the collection of the Geoscience Center in Göttingen (GZG). The Δ′17O is reported
relative to a slope − 0.528 reference line.
Sample Comment Source δ17O δ18O Δ′17O
Sapphire (Ceylon) Sri Lanka GZG Collection 9.336 17.873 –0.082
Ruby (Ceylon) Sri Lanka GZG Collection 13.558 25.993 –0.066
– “” – – “” – – “” – 10.633 20.361 –0.082
– “” – – “” – – “” – 11.534 22.116 –0.061
Ruby (Slatoust) Ural, Russia GZG Collection 11.874 22.767 –0.082
Technical alumina Commercial Laboratory 8.44 16.58 –0.281
– “” – – “” – – “” – 8.29 16.22 –0.239
~ 50 µm spherulitic 11.548 22.637 –0.338
Alumina microbeads powder, oxyhydrogen FEE Idar Oberstein
flame fused
Crackle corundum Coarse fraction, Verneuil FEE Idar Oberstein 14.865 29.237 –0.460
– “” – Fine fraction, Verneuil FEE Idar Oberstein 14.866 29.25 –0.466
– “” – – “” – FEE Idar Oberstein 15.652 30.777 –0.475
– “” – – “” – FEE Idar Oberstein 16.71 32.925 –0.533
Sapphire FEE Czochralski XX FEE Idar Oberstein 14.258 27.995 –0.420
Ruby FEE Czochralski XX FEE Idar Oberstein 14.115 27.754 –0.438
San Carlos olivine N = 20 (av.) Arizona, USA 2.745 5.306 –0.052
The δ18O of magnesia refractories and first δ17O and δ18O data for technical alumina clearly
show that during the high-T production processing of these materials, considerable exchange
occurs between the atmosphere and the refractory product. The studied alumina refractory
carries a triple oxygen isotope signature that suggests ~ 60% equilibration with the atmosphere.
Using Δ′17O as a new tool to distinguish synthetic and natural gems
The most common technique of producing single crystals of alumina is growing them by
continuously pulling the growing crystal out of a melt (Czochralski 1918). After his inventor,
this technique is now called the Czochralski method (see Tomaszewski 2002, for some dispute
on the name). An illustration of the process is given in (Fig. 8A). Earlier, Verneuil (1904)
described a method for the production of single crystals of corundum, ruby or sapphire;
a method now named after him as Verneuil process (Smith 1908). In that process, fine alumina
powder is rinsed through an oxyhydrogen torch, in which the powder melts and coagulates to
fine alumina droplets (Fig. 8B). The droplets of liquid alumina then fall onto a seed crystal,
where they grow to form a single crystal of corundum, ruby, or sapphire. For high-quality and
larger crystals, this technique is inferior compared to the Czochralski method. It is, however,
used for the preparation of starting materials used in the Czochralski process.
We have obtained three natural ruby and sapphire samples and five samples of synthetic
corundum (Table 2).
Sample “Alumina microbeads” consists of a granulate of ~50 µm large alumina spherules.
No details about the production process are available. The alumina spherules were likely
produced by oxyhydrogen flame fusion in an apparatus similar to that invented by Verneuil
(1904), except that the molten droplets were not grown on a single crystal but quenched and
B
Hammer
O2
Al2O3 powder
Sieve
H2
Oxyhydrogen flame
Al2O3 melt
droplets
Al2O3 crystal
Figure 8. Sketches illustrating the Czochralski method (A) and the Verneuil method (B, modified after Blum-
berg and Müller 1970). In the Czochralski method (A) a crystal is grown by continuously pulling out of an
alumina melt. In the Verneuil process (B), alumina powder is melted to small droplets that fall onto the grow-
ing crystal, where they crystallize in the same lattice orientation as the underlying crystal substrate.
collected as small spherules. Data about the composition of the alumina starting material
as well as about the oxygen isotope composition of the O2 used, however, are not known.
The second sample is “Crackle corundum”, which is a granulate made by crushing of alumina
crystals that were produced by the Verneuil process. As in the case of “Alumina microbeads”,
not further details about the production process, e.g., the composition of the O2 gas used, are
known. A mixture of “Alumina microbeads” and “Crackle” corundum was used as starting
material for the Czochralski process. The pre-fused material is used because of its higher bulk
density. Usage of powdered alumina from the Bayer process would require fusion, followed
by refilling of the crucible before the growth can start.
A Czochralski-grown ruby and a Czochralski-pulled sapphire (both FEE, Idar-Oberstein,
Table 2) were also analyzed. The ruby and sapphire were pulled from an alumina melt at a
temperature exceeding the melting point of alumina at 2072°C. The atmosphere during the
crystal growth was a mixture of N2 with 0.5 vol.% of air, i.e., 0.105 vol.% O2. The melt mass
was 6–8 kg and the crystal pulling took ~ 10 days. The apparatus was flushed at a rate of
0.5 m3 h–1. This translates to a total amount of oxygen of 10.3 mol O (in form of O2) that went
through the apparatus. This number is small compared to the 208 mol O that is contained in
7 kg of well-mixed liquid alumina. This small mass balance calculation demonstrates that,
under given growth conditions, as maximum, only 5% of the alumina can have equilibrated
with air. In reality, this number may well be an order of magnitude lower as only a small
fraction of the O2 contained in the N2 atmosphere ever will have touched the melt surface.
The oxygen isotope compositions of the oxyhydrogen flame fused and the Verneuil
alumina starting materials and the Czochralski grown ruby and sapphire are listed in Table 2 and
illustrated in Fig. 7. The δ18O of the oxyhydrogen flame fused and Verneuil starting material
(“Alumina microbeads” and “Crackle”) varies from 22 to 33‰. The variations among the
different fragments of the “Crackle” (Table 2) are attributed to the rapid and likely incomplete
exchange process between the alumina melt droplets and flame oxygen while the droplets fall
through the flame. The data fall on a tight mixing trend of the raw materials with slope λ = 0.509.
236 Pack
Although air O2 does not fall on that trend, the low Δ′17O of the raw materials and the crystals
made from them is clearly related to the influence technical O2 that was prepared by liquefaction
from air O2. Pack et al. (2007) showed that technical O2 can vary in δ18O between 0 and 30‰,
but, because of mass-dependent fractionation during the oxygen production (λ ≈ 0.524), that
the anomaly in Δ′17O is preserved.
The isotope compositions of the Czochralski grown ruby and sapphire plot on a mixing
trend between “Alumina microbeads” and “Crackle” (Fig. 7), which is expected as these
materials were used as starting materials. It was discussed that likely no isotope exchange had
affected the isotope composition of the melt during the Czochralski process. The observation
that the Δ′17O of the sapphire and ruby crystals is identical to that of air is a mere coincidence
and results from the compositions and mixing ratios of the raw materials.
The data presented here show that synthetic alumina (corundum, ruby, sapphire) can
clearly be distinguished from natural corundum, ruby and sapphire in that the synthetic
crystals have a much lower Δ′17O resembling that of air O2 and its technical derivatives. Hence,
Δ′17O is a clear and difficult-to-manipulate indicator of the gemstone formation process.
The studied Czochralski grown ruby and sapphire crystals largely reflect the Δ′17O of the modern
atmosphere along with mass-dependent variation due to the liquefaction process. As such, they
can be regarded a suitable proxy for the pCO2 · GPP. One can imagine that a centimeter to
decimeter sized synthetic alumina crystal is highly resistant to any kind of isotopic modification
by geological processes. In the far future (e.g., billions of years from now), such then fossil
Czochralski-grown ruby and sapphire crystals buried in sedimentary rocks could serve an
isotope geochemist as a unique tracer of anthropogenic activity and as an atmospheric proxy.
CONCLUSIONS
It has been stated by Bao (2015) “[…] that sulfate, to this point, is the only compound
from which direct atmospheric O2 and O3 signals from the distant past can be retrieved.”
In this chapter, I highlight that not only sulfate, but a number of other solids carry information
about the isotope anomaly of air O2. The processes are exchange at high temperatures and/or
transfer of O2 during an oxidation process. Some of the materials are potential proxies for the
Δ′17O of air O2.
ACKNOWLEDGEMENTS
The constructive reviews by Laurence Y. Yeung and Huimin Bao helped improving the
manuscript. Ilya Bindeman is thanked for initiating this volume, his comments and for the
editorial handling. Dennis Kohl, Nina Albrecht, and the entire lab team is thanked for their
various contributions. Klaus Dupré is thanked for providing alumina crystals ad background
information. Christian Stübler is thanks for allowing me to use his data from his BSc theses.
REFERENCES
Affolter S, Häuselmann AD, Fleitmann D, Häuselmann P, Leuenberger M (2015) Triple isotope (δD, δ17O, δ18O)
study on precipitation, drip water and speleothem fluid inclusions for a Western Central European cave
(NWSwitzerland). Quat Sci Rev 127:73–89
Ash JL, Hu H, Yeung LY (2020) What fractionates oxygen isotopes during respiration? insights from multiple
isotopologue measurements and theory. ACS Earth Space Chem 4:50–66
Balci N, Shanks IIIW C, Mayer B, Mandernack KW (2007) Oxygen and sulfur isotope systematics of sulfate produced
by bacterial and abiotic oxidation of pyrite. Geochim Cosmochim Acta 71:3796–3811
Bao H (2015) Sulfate: A time capsule for Earths O2, O3, and H2O. Chem Geol 395:108–118
Bao H, Lyons J, Zhou C (2008) Triple oxygen isotope evidence for elevated CO2 levels after a Neoproterozoic
Bao H, Cao X, Hayles JA (2016) Triple oxygen isotopes: fundamental relationships and applications. Ann Rev Earth
Barkan E, Luz B (2003) High-precision measurements of 17O/16O and 18O/16O of O2 and O2/Ar ratio of air. Rapid
Spectrom 19:3737–3742
Bender M, Sowers T, Labeyrie L (1994) The Dole effect and its variations during the last 130,000 years as measured
Benson BB, Krause Jr D (1984) The concentration and isotopic fractionation of oxygen dissolved in freshwater and
seawater in equilibrium with the atmosphere. Limnol Oceanogr 29, 620–632
Bindeman I, Zakharov D, Palandri J, Greber ND, Dauphas N, Retallack G, Hofmann A, Lackey J, Bekker A (2018) Rapid
emergence of subaerial landmasses and onset of a modern hydrologic cycle 2.5 billion years ago. Nature 557:545–548
Bird MI, Chivas AR, Andrew AS (1989) A stable-isotope study of lateritic bauxites. Geochim Cosmochim Acta
53:1411–1420
Bird MI, Longstaffe FJ, Fyfe WS, Kronberg BI, Kishilda A (1993) An oxygen-isotope study of weathering in the eastern
Amazon Basin, Brazil. Washington DC. American Geophysical Union Geophysical Monograph Series 78:295–307
Blumberg H, Müller R (1970) Automatische Verneuil—Apparatur zur Züchtung von Rubin-Einkristallstäben. Kristall
und Technik 5:33–36
Blunier T, Barnett B, Bender ML, Hendricks MB (2002) Biological oxygen productivity during the last 60,000 years
from triple oxygen isotope measurements. Global Biogeochem Cycles 16:1029
Blunier T, Bender M, Barnett B, von Fischer J (2012) Planetary fertility during the past 400 ka based on the triple
isotope composition of O2 in trapped gases from the Vostok ice core, Climate of the Past 8:1509–1526
Bouşka V, Povondra P, Florenskij P, Řanda Z (1981) Irghizites and zhamanshinites: Zhamanshin crater, USSR.
Meteoritics 16:171–184
Buchwald VF (1977) Mineralogy of iron meteorites. Phil Trans R Soc London Ser A 286:453–491
Cao X, Bao H (2013). Dynamic model constraints on oxygen-17 depletion in atmospheric O2 after a snowball Earth.
PNAS 110:14546–14550
Cao X, Bao H (2021) Small triple oxygen isotope variations in sulfate: Mechanisms and applications. Rev Mineral
Geochem 86: 463–488
Catling DC, Zahnle KJ (2020) The Archean atmosphere. Sci Adv 6:1420
Clayton RN, Mayeda TK (1996) Oxygen isotope studies of achondrites. Geochim Cosmochim Acta 39:569–584
Clayton RN, Mayeda TK, Brownlee DE (1986) Oxygen isotopes in deep-sea spherules. Earth Planet Sci Lett 79:235–240
Crockford PW, Hayles JA, Bao H, Planavsky NJ, Bekker A, Fralick PW, Halverson GP, Bui TH, Peng Y, Wing BA
Czochralski J (1918) Ein neues Verfahren zur Messung der Kristallisationsgeschwindigkeit der Metalle. Z für
Physikalische Chemie 92:219–221
Davis AM, Brownlee DE (1993) Iron and nickel isotopic mass fractionation in deep-sea spherules. XXIV Lunar and
Planetary Science Conference (Houston) p. 373–374
Davis AM, Clayton RN, Mayeda TK, Brownlee DE (1991) Large mass fractionation of iron isotopes in cosmic spherules
collected from deep-sea sediments. XXII. Lunar and Planetary Science Conference (Houston) p. 281–282
Dole M (1936) The relative atomic weight of oxygen in water and in air. J Chem Phys 4:268–275
Dole M, Lane GA, Rudd DP, Zaukelies DA (1954) Isotopic composition of atmospheric oxygen and nitrogen.
Donaldson CH, Williams RJ, Lofgren G (1975) A sample holding technique for study of crystal growth in silicate
melts. Am Mineral 60:324–326
Drnek T (2018) Magnesit im Überblick. res montanarum. Z des Montanhistorischen Vereins für Österreich 58:5–12
Ellahi SS, Taghipour B, Zarasvandi A, Bird MI, Somarin AK (2016) Mineralogy, geochemistry and stable isotope
studies of the Dopolan Bauxite Deposit, Zagros Mountain, Iran. Minerals 6:1–21
Engrand C, McKeegan KD, Leshin LA, Herzog GF, Schnabel C, Nyquist LE, Brownlee DE (2005) Isotopic
compositions of oxygen, iron, chromium, and nickel in cosmic spherules: Toward a better comprehension of
atmospheric entry heating effects. Geochim Cosmochim Acta 39:569–584
Farquhar J, Bao H, Thiemens M (2000) Atmospheric influence of Earths earliest sulfur cycle. Science 339:780–785
Florenski PV (1977) Der Meteoritenkrater Zhamanshin (nördliches Aralgebiet, UdSSR) und seine Tektite und
Impaktite. Chemie der Erde 36:83–95
238 Pack
Gehler A, Tütken T, Pack A (2011) Triple oxygen isotope analysis of bioapatite as tracer for diagenetic alteration of
bones and teeth, Palaeogeogr Palaeoclimatol Palaeoecol 310:84–91
PETM using triple oxygenisotope analysis of mammalian bioapatite, PNAS 113:7739–7744
Genge MJ, Engrand C, Gounelle M, Taylor S (2008) The classification of micrometeorites. Meteorit Planet Sci
35:807–816
Genge MJ (2016) The origins of I-type spherules and the atmospheric entry of iron micrometeoroids. Meteorit Planet
Sci 51:1–19
Giauque WF, Johnston HL (1929) An isotope of oxygen of mass 17 in the Earth’s atmosphere, Nature 54:349–392
Giuliani G, Fallick AE, Garnier V, France-Lanord C, Ohnenstetter D, Schwarz D (2005) Oxygen isotope composition
as a tracer for the origins of rubies and sapphires. Geology 33:249–252
Graham T (1863) On the molecular mobility of gases. Philos Trans R Soc 153:385–405
Graham RJ, Shaw TA, Abbot DS (2019). The Snowball Stratosphere. J Geophys Res: Atmospheres 124:11819–11836
Habashi F (1995) Bayers process for alumina production: a historical perspective. Bull History Chem 17/18:15–19
Heinzinger K, Iunge C, Schidlowski M (1971) Oxygen isotope ratios in the crust of iron meteorites. Z Naturforsch
26:1485–1490
Hendricks MB, Bender ML, Barnett BA, Strutton P, Chavez FP (2005) Triple oxygen isotope composition of dissolved
O2 in the equatorial Pacific: A tracer of mixing, production, and respiration, J Geophys Res: Oceans 110: C12
Herbrich P, Mörtl G, Hoernes S (1990) Der Anteil von schwerem Sauerstoff (18O) in Sintermagnesia-Produkten.
Radex-Rundschau 1990:275–283
Jessberger E, Gentner W (1972) Mass spectrometric analysis of gas inclusions in Muong Nong glass and Libyan
Desert glass. Earth Planet Sci Lett 14:221–225
Koeberl C (1988) The origin of tektites: a geochemical discussion. Proceedings of the National Institute of Polar
Research Symposium on Antarctic Meteorites 1:261–290
Koeberl C (1994) Tektite origin by hypervelocity asteroidal or cometary impact. Large meteorite impacts and
planetary evolution 293:133–151
Koeberl C, Fredriksson K (1986) Impact glasses from Zhamanshin crater (USSR): chemical composition and
discussion of origin. Earth Planet Sci Lett 78:80–88
Kohl I, Bao H (2011). Triple-oxygen-isotope determination of molecular oxygen incorporation in sulfate produced
during abiotic pyrite oxidation (pH= 2–11). Geochim Cosmochim Acta, 75:1785–1798
Kohn MJ (1996) Predicting animal δ18O: accounting for diet and physiological adaption, Geochim Cosmochim Acta
39:569-584
Krinov E, Brown H, Anders E (1961) Principles of Meteoritics. Pergamon (London), 552 pp
Kroopnick P, Craig H (1972) Atmospheric oxygen: isotopic composition and solubility fractionation. Science 175:54–55
Kroopnick P, Craig H (1976) Oxygen isotope fractionation in dissolved oxygen in the deep sea. Earth Planet Sci Lett
32:375–388
Landais A, Barkan E, Luz B (2008) Record of δ18O and 17Oexcess in ice from Vostok Antarctica during the last 150,000
years. Geophys Res Lett 35:L02709
Laughlin DE, Hono K (Eds.) (2015) Physical Metallurgy. Elsevier (Amsterdam), 2960 pp
Leblanc T, Walsh T, McDermid I, Toon G, Blavier J-F, Haines B, Read W, Herman B, Fetzer E, Sander S, Pongetti
T, Whiteman D, McGee T, Twigg L, Sumnicht G, Venable D, Calhoun M, Dirisu A, Hurst D, Hauchecorne A
(2011) Measurements of Humidity in the Atmosphere and Validation Experiments (MOHAVE)-2009: overview
of campaign operations and results. Atmos Meas Tech Discuss 4:3277–3336
Li S, Levin NE, Chesson LA (2015) Continental scale variation in 17Oexcess of meteoric waters in the United States.
Li S, Levin NE, Soderberg K, Dennis KJ, Caylor KK (2017) Triple oxygen isotope composition of leaf waters in
Mpala, central Kenya. Earth Planet Sci Lett 468:38–50
transfer and implications for tracing oxygen cycling in the ocean. Mar Chem 210:61–71
Luo G, Ono S, Beukes NJ, Wang DT, Xie S, Summons RE (2016) Rapid oxygenation of Earth’s atmosphere 2.33
billion years ago, Sci Adv 2: e1600134
Luz B, Barkan E (2009) Net and gross oxygen production from O2/Ar, 17O/16O and 18O/16O ratios. Aquatic Microbial
Ecol 56:133–145
Luz B, Barkan E (2010) Variations of 17O/16O and 18O/16Oin meteoric waters. Geochim Cosmochim Acta 74:6276–6286
Luz B, Barkan E (2011) Oxygen isotope fractionation in the ocean surface and 18O/16O of atmospheric O2. Global
Biogeochem Cycles 25: GB4006
Luz B, Barkan E, Severinghaus JP, (2014) The stable isotopic composition of atmospheric O2. In: Holland HD,
Turekian KK (Eds.) Treatise in Geochemistry Elsevier (Amsterdam), p. 363–383
Magna T, Žák K, Pack A, Moynier F, Mougel B, Skála R, Jonášová S, Řanda Z, Mizera J (2017) Zhamanshin
astrobleme provides evidence for carbonaceous chondrite and post-impact exchange between ejecta and Earth’s
atmosphere. Nat Commun 8:227
Mandernack KW, Fogel ML, Tebo BM, Usui A (1995) Oxygen isotope analyses of chemically and microbially
produced manganese oxides and manganates. Geochim Cosmochim Acta 59:4409–4425
Macris CA, Asimow PD, Badro J, Eiler JM, Zhang Y, Stolper EM (2018) Seconds after impact: Insights into the
thermal history of impact ejecta from diffusion between lechatelierite and host glass in tektites and experiments,
Mark D, Stuart F, De Podesta M (2011) New high-precision measurements of the isotopic composition of atmospheric
argon. Geochim Cosmochim Acta 75:7494–7501
Miller MF, Pack A (2021) Why Measure 17O? Historical perspective, triple-isotope systematics and selected
applications. Rev Mineral Geochemistry 86:1–34
Mizera J, Řanda Z, Tomandl I (2012) Geochemical characterization of impact glasses from the Zhamanshin crater by
various modes of activation analysis. Remarks on genesis of irghizites. J Radioanal Nucl Chem 293:359–376
Müller O, Gentner W (1968) Gas content in bubbles of tektites and other natural glasses. Earth Planet Sci Lett 4:406–410
Nicholson D, Stanley RH, Doney SC (2014) The triple oxygen isotope tracer of primary productivity in a dynamic
ocean model, Global Biogeochem Cycles 28:538–552
Pack A (2000) Tracing the origin of oxide inclusions in continuously casted steel. PhD thesis, Rheinische Friedrich
Wilhelms Universität, Bonn, 164 pp
Pack A, Hoernes S, Göbbels M, Broß R, Buhr A (2005) Stable oxygen isotopes—a new approach for tracing the origin
of oxide inclusions in steel. Euro J Mineral 17:483–493
as paleo-CO2-barometer, Geochim Cosmochim Acta 102: 306–317
Carlos olivine on VSMOW2-SLAP2 scale. Rapid Commun Mass Spectrom 30:1495–1504
Pack A, Höweling A, Hezel DC, Stefanak M, Beck AK, Peters ST M, Sengupta S, Herwartz D, Folco L (2017) Tracing
the oxygen isotope composition of the upper Earth atmosphere using cosmic spherules. Nat Commun 8:15702
Payne RC, Brownlee D, Kasting JF (2020) Oxidized micrometeorites suggest either high pCO2 or low pN2 during the
Neoarchean, PNAS 117:1360
Passey BH, Levin NE (2021) Triple oxygen isotopes in meteoric waters, carbonates, and biological apatites:
implications for continental paleoclimate reconstruction. Rev Mineral Geochemistry 86:429–462
Peters ST, Alibabaie N, Pack A, McKibbin SJ, Raeisi D, Nayebi N, Torab F, Ireland T, Lehmann B (2020) Triple
oxygen isotope variations in magnetite from iron-oxide deposits, central Iran, record magmatic fluid interaction
with evaporite and carbonate host rocks. Geology 48:211–215
Sharp ZD, Wostbrock J (2021) Standardization for the triple oxygen isotope system: Waters, silicates, carbonates, air, and
sulfates. Rev Mineral Geochem 86: 179–196
Sharp ZD, Gibbons JA, Atudorei V, Pack A, Sengupta S, Shock EL, Knauth LP (2016) A calibration of the triple oxygen isotope
fractionation in the SiO2–H2O system and applications to natural samples. Geochim Cosmochim Acta 186:105–119
Sharp Z, Wostbrock J, Pack A (2018) Mass-dependent triple oxygen isotope variations in terrestrial materials.
Geochem Perspect Lett 7:27–31
Shuvalov V, Dypvik H (2013) Distribution of ejecta from small impact craters. Meteorit Planet Sci 48:1034–1042
Smith GH (1908) Note on synthetical corundum and spinel. Mineral Mag 15:153–155
Spencer LJ (1933) Origin of tektites. Nature 131:117–118
Stolper DA, Fischer WW, Bender ML (2018) Effects of temperature and carbon source on the isotopic fractionations
associated with O2 respiration for 17O/16O and 18O/16O ratios in E coli, Geochim Cosmochim Acta 240:152–172
Stübler C (2017) Hochtemperaturaustausch von Sauerstoffisotopen zwischen Silikatschmelzen und der Atmosphäre,
Georg-August-Universität, Göttingen, 55 pp
Surma J, Assonov S, Bolourchi M, Staubwasser M (2015) Triple oxygen isotope signatures in evaporated water
bodies from the Sistan Oasis, Iran. Geophys Res Lett 42:8456–8462
240 Pack
Surma J, Assonov S, Herwartz D, Voigt C, Staubwasser M (2018) The evolution of 17Oexcess in surface water of the arid
environment during recharge and evaporation. Sci Rep 8:1–10
Sutherland KM, Wostbrock JA, Hansel CM, Sharp ZD, Hein JR, Wankel SD (2020) Ferromanganese crusts as
Taylor HP, Epstein S (1962) Oxygen isotope studies on the origin of tektites. J Geophys Res 67:4485–4490
Tappert R, McKellar RC, Wolfe AP, Tappert MC, Ortega-Blanco J, Muehlenbachs K (2013) Stable carbon isotopes of C3 plant
resins and ambers record changes in atmospheric oxygen since the Triassic. Geochim Cosmochim Acta 121:240–262
Thiemens MH, Meagher D (1984) Cryogenic separation of nitrogen and oxygen in air for determination of isotopic
ratios by mass spectrometry. Anal Chem 56:201–203
Thiemens MH, Jackson T, Zipf EC, Erdman PW, van Egmond C (1995) Carbon dioxide and oxygen isotope anomalies
in the mesosphere and stratosphere. Science 339:780–785
Tian C, Wang L (2019) Stable isotope variations of daily precipitation from 2014–2018. Scientific Data 6:190018
snowfall in the central United States. Sci Rep 8:1–15
Tomaszewski PE (2002) Jan Czochralski–father of the Czochralski method. J Cryst Growth 236:1–4
Tomkins AG, Bowlt L, Genge M, Wilson SW, Brand HE A, Wykes JL (2016) Ancient micrometeorites suggestive of
an oxygen-rich Archaean upper atmosphere. Nature 533:235–238
Touzeau A, Landais A, Stenni B, Uemura R, Fukui K, Fujita S, Guilbaud S, Ekaykin A, Casado M, Barkan E (2016)
Acquisition of isotopic composition for surface snow in East Antarctica and the links to climatic parameters.
The Cryosphere 10:837–852
Urey HC (1955) On the origin of tektites. PNAS 41:27–31
Verneuil AV L (1904) Mémoires sur la reproduction artificielle du rubis par fusion. Annales de Chimie et de Physique
Sér. 8, Vol. III: 20–45
Wang J, Davis AM, Clayton RN, Mayeda TK (1994) Kinetic Isotopic Fractionation During the Evaporation of the
Iron Oxide from Liquid State. XXV Lunar and Planetary Science Conference (Houston) abstract #1459
Winter J (2001) An introduction to igneous and metamorphic petrology. Prentice-Hall (Upper Saddle River, NJ), 697 pp
Wostbrock JA, Cano EJ, Sharp ZD (2020) An internally consistent triple oxygen isotope calibration of standards for
Yeung LY, Young ED, Schauble EA (2012) Measurements of 18O18O and 17O18O in the atmosphere and the role of
isotope-exchange reactions, J Geophys Res (Atmospheres) 117:18306
in tripleâ. Rapid Commun Mass Spectrom 32:1811–1821
Zähringer J, Gentner W (1963) Radiogenic and atmospheric argon content of tektites. Nature 199:583–583
Zák K, Skála R, Pack A, Ackerman L, Křížová S (2019) Triple oxygen isotope composition of Australasian tektites.
Meteorit Planet Sci 54:1167–1181fs
Vol. 86 pp. 241–290, 2021 8
Triple Oxygen Isotopes in Evolving Continental Crust,

Granites, and Clastic Sediments
Ilya N. Bindeman
Department of Earth Sciences
University of Oregon
Eugene OR 97403-1272
USA
bindeman@uoregon.edu
INTRODUCTION
This Chapter considers triple oxygen isotope variations and their 4 Gyr temporal evolution
in bulk siliciclastic sedimentary rocks and in granites. The d18O and ∆′17O values provide
new insights into weathering in the modern and ancient hydrosphere and coeval crustal
petrogenesis. We make use of the known geological events and processes that affect the rock
cycle: supercontinent assembly and breakup that influence continent-scale and global climate,
the fraction of the exposed crust undergoing weathering, and isotopic values of precipitation.
New data from a 5000 m Texas drillhole into the Oligocene Frio Formation demonstrate
minimal isotopic shifts from mudrocks to shales during diagenesis, mostly related to expulsion
of water from smectite-rich loosely cemented sediment and its conversion to illite-rich shale.
Inversion of triple oxygen isotope fractionations return isotopic values and temperatures
along the hole depth that are more consistent with weathering conditions in the Oligocene and
modern North America (d18O = −7 to −15‰, and T of +15 to +45 °C) rather than d18O from
8 to 10‰ diagenetic water in the drill hole at 175–195 °C. More precise T and d18Owater are
obtained where the chemical index of alteration (CIA) based detrital contribution is subtracted
from these sediments. Triple oxygen isotopes from suspended sediments in major world rivers
record conditions (T and d18Ow) of their watersheds, and not the composition of bedrock
because weathering is water-dominated.
In parallel, the Chapter presents new analyses of 100 granites, orthogneisses, migmatites,
tonalite-trondhjemite-granodiorite (TTG), and large-volume ignimbrites from around the world
that range in age from 4 Ga to modern. Most studied granites are orogenic and anatectic in origin
and represent large volume remelting/assimilation of shales and other metasediments; the most
crustal and high-d18O of these are thus reflect and record the average composition of evolving
continental crust. Granites also develop a significant progressive increase in d18O values from
6–7‰ (4–2.5 Ga) to 10–13‰ (~1.8–1.2 Ga) after which d18O stays constant or even decreases.
More importantly, we observe a moderate −0.03‰ step-wise decrease in ∆′17O between 2.1
and 2.5 Ga, which is about half of the step-wise decrease observed in shales over this time
interval. We suggest that granites, as well as shales, record the significant advent and greater
volumetric appearance of low-∆′17O, high-d18O weathering products (shales) altered by meteoric
waters upon rapid emergence of large land masses at ~2.4 Ga, although consider alternative
interpretations. These weathering products were incorporated into abundant 2.0–1.8 Ga orogens
around the world, where upon remelting, they passed their isotopic signature to the granites. We
further observe the dichotomy of high-∆′17O Archean shales, and unusually low-∆′17O Archean
granites. We attribute this to greater contribution from shallow crustal hydrothermal contribution
to shales in greenstone belts, while granites in the earliest 3.0–4.0 Ga crust and TTGs require
involvement of hydrothermal products with lower-∆′17O signatures at moderately high-d18O,
which we attribute to secondary silicification of their protoliths before partial melting.
1529-6466/21/0086-0008$500 (print) http://dx.doi.org/10.2138/rmg.2021.86.08
1943-2666/21/0086-0008$5.00 (online)
242 Bindeman
The Chapter further discusses evolution of the shale record through geologic history and
discusses the step-wise change in d18O and ∆′17O values at Archean–Proterozoic transition.
Denser coverage for shales in the past 1 billion years permits investigation of the rocks and their
weathering in the last supercontinent cycle, with observed lighter d18O values, characteristic for
the mid-Phanerozoic at the initiation of Gondwana breakup. The continuing increase in d18O
values of the shales since 4 Ga is interpreted to reflect accumulation of weathering products
via shale accretion to continents, as low-density and buoyant shales tend to not subduct back
into the mantle.
The rock cycle passes triple oxygen isotopic signatures from precipitation to sedimentary,
metasedimentary, and finally to anatectic igneous rocks. Continental crust became progressively
heavier in d18O, lighter in ∆′17O due to incremental accumulation of high-d18O sediments in
accretionary wedges. Second-order trends in d18O and ∆′17O are due to supercontinent cycles
and glacial episodes.
The evolution of continental crust since its formation in the early Archean has been a major
theme of petrologic investigation for many decades (Goldschmidt 1933; Taylor and MacLennan
1995; Wedepohl 1995; Rudnick and Gao 2003; Greber et al. 2017; Spencer et al. 2017;
Bucholz et al. 2018). The appearance of each new chemical analytical method or a new isotopic
system, as well as improving geological and geochronologic investigation and recognition of
ancient geologic formations adds greater understanding of Earth and the early processes of
its evolution. It is reasonable to assume that weathering conditions on Earth have evolved as
conditions (such as solar luminosity and CO2 and O2 concentrations) changed, and thus both
uniformitarian and non-uniformitarian approaches can apply, especially to the Archean. As the
Archean and Proterozoic eons capture ~88‰ of Earth′s history, most events occurring in Earth′s
crust are rooted there: the timing and rate of plate tectonics, the appearance of first continents,
their composition (mafic vs. silicic); their emergence from water as subaerial land masses, the
evolution of the atmosphere and the causes of the Great Oxidation Event at 2.3 Ga, Snowball
Earth glaciations in early and late Proterozoic, and the appearance of microbial and multicellular
life (Holland 1984; Buick et al. 1998; Bekker and Holland 2012; Hoffman 2013).
Shales, what do they reflect?
Mudrocks constitute 50 to 80% of total sedimentary rock mass (Pettijohn 1957, Ronov and
Yaroshevsky 1969), and shales have been traditionally used to deduce the role of continental
weathering and sediment recycling through geologic time, and to constrain the average chemical
composition of the exposed and eroding continental crust (Goldschmidt 1933; Wedepohl 1995;
Rudnick and Gao 2003). Their chemical and isotopic compositions have been used since the
early days of geochemistry to address weathering conditions and sediment provenance, and
address fundamental questions of elemental and isotopic fluxes from and into the mantle and
the hydrosphere, and element and isotope recycling in the framework of plate tectonics, after
the acceptance of that theory (Holland 1984). Unmetamorphosed clay-rich shales are known
to exist since 3.4–3.5 Ga, and they are relevant to the study of conditions under which organic
matter becomes incorporated into weathering products throughout geologic history (Buick et
al. 1998; Hayes et al. 1999; Retallack 2001; Sageman and Lyons 2004; Kump 2014).
Like many other sedimentary rocks, shales and mudrocks represent a mixture of detrital
and authigenic (equilibrated with weathering fluid at a specific temperature and water, and less
pronounced exchange during transport and post-depositional diagenesis) components (Sageman
and Lyons 2004). The authigenic component is important for recording environmental factors.
Silverman (1951), Savin and Epstein (1970a,b) and Bindeman et al. (2019) demonstrated that
oxygen isotopic values of recent sediments and authigenic shale samples reflect exclusively
isotopic values of meteoric water at respective values of temperature, since weathering
proceeds with a great excess of water over rocks.
Evolving Continental Crust, Granites, and Clastic Sediments 243
Other isotopic systems in sediments such as Sr, Nd, Hf, Mg, Li, Ti, and Si, whose dissolved
concentration in weathering water is low, reflect provenance (exposed bedrock) of the crust
undergoing weathering (Goldstein and Jacobsen 1988; Bayon et al. 2015, 2018, 2020; Li et al.
2016; Dellinger et al. 2017; Greber et al. 2017). Therefore shales and their constituent detrital and
secondary minerals are a valuable resource for the investigation of the detrital portion of crustal
maturation. In their oxygen and hydrogen isotopes they additionally record evolving weathering
conditions: d18O and temperature on the Earth′s surface (Fig. 1), which themselves are related to
watershed latitude, and global climate as well as the concurrent (and hotly debated!) d18O value of
seawater. This complexity creates both significant opportunities and challenges in the application
of isotope geochemistry to sedimentary environments. Shales remained the “most studied but
poorly understood rock” (Shaw and Weather 1965), but recent advances in isotopic analysis of
multiple elements in shales and sediments will solve many of the puzzles stored in shales.
Prior efforts to characterize stable isotopes in shales, mudrocks, soil weathering
profiles, and marine sediments date back to the early 1970s (Savin and Epstein 1970a,b)
using conventional methods and relying on 30 mg of material. The d18O and dD isotopic
investigation of shales and the details of their partitioning among their constituent minerals
(clay minerals and secondary quartz) has been performed in the past in great detail by Taylor
and Epstein (1962), Savin and Epstein (1970a,b), Yeh and Savin (1976), Yeh and Epstein
(1978), and others (Churchman et al. 1976; Savin and Lee 1988; Sheppard and Gilg 1996).
Their results showed that: 1) The clay and quartz components in shales are isotopically heavy,
and d18O is shifted during weathering by a few to +25‰ with respect to the the mantle or
the igneous protolith, with meteoric waters diverse in d18O, and at different temperatures.
2) Upon formation of secondary minerals and their accumulation in sedimentary basins, rates
of isotope exchange with ambient water are slow, and thus shales show little isotope exchange
with lacustrine or sea waters after deposition (Land and Lynch 1996). 3) Diagenesis occurs
in a narrow temperature window of 80 to 120 °C (Eberl 1993) and leads to a profound change
in clay mineralogy and morphology (e.g., smectite and kaolinite are transformed into illite
Mackenzie
Volga
Fraser Nelson
Don
Mississippi Amu Darya Yellow
4
Yangtze
2 G-B
0 Red
annual precipitation
-2 Niger
Orinoco Nile
-4 Mekong
-6 Mae
Sepik
Klong
-8 Amazon Congo
Betsiboka
-10 Fly
W Fitzroy
-12
World:
-14 Murray
-16
Bedrock: Waikato
-18 Large River, mixed bedrock
G18OMW
-20 Sedimentary/mixed
-22 22%
Volcanic bedrock Illite
-24 Igneous/Metamorphic
-26
-28
Figure 1. The isotopic composition of precipitation in the modern world, and river basins that were sam-
pled for triple oxygen isotope analysis of suspended sediments. The sampled river basins cover most of
)LJ&RORU
the modern environments of weathering from tropical to Arctic, and from wet to arid. XRD-determined
clay proportions for selected rivers and global average are shown. Predominant watershed bedrock types
are also shown, but isotopic values of clays are not influenced by the bedrock, and reflect predominantly
the temperature of weathering and d18OMW. [Used by permission of Elsevier, from Bindeman et al. (2019),
EPSL, Vol. 528:115851, Fig. 1]
244 Bindeman
and chlorite, developing shistosity (Fig. 2). However, these changes may cause only 1–2
permil d18O increases due to loss of porewaters (Wilkinson et al. 1992), and they do not erase
the multi-permil isotopic differences between shales formed in diverse environments. 4) Low
post-diagenetic Darcy permeability of fine-grained shales leads to the preservation of both O
and H isotopic signatures in the geologic record, far better than in cherts or carbonates (Land
and Lynch 1996; Sheppard and Gilg 1996). 5) After diagenetic reactions are complete, fine-
grained shales become virtually impermeable to secondary waters (Fig. 2, Pettijohn 1957).
Of the many chemical and isotopic proxies targeting these diagenetic processes, oxygen
isotopes serve as the boundary between evolving lithosphere and hydrosphere, as oxygen
makes up 49–53 wt% of rocks and 89 wt% of waters. Upon weathering, oxygen (and hydrogen)
isotopic values of weathering products reflect that of the water, due to the great quantitative
excess of water over rock (Silverman 1951; Savin and Epstein 1970a,b). Likewise, chemical
precipitates from sea or meteoric water reflect the oxygen isotopic value of that water. If
unaltered by subsequent isotopically different waters, these chemical sediments (carbonates,
cherts, phosphates, etc.) should reflect the original oxygen isotopic values. It is also important
to stress that shales record d18O and d17O values of precipitated water, while cherts, carbonates
and phosphates for the most part provide a record of seawater.
A
2 cm

/DQGHWDO %XONVKDOHWKLVZRUN
B C y = 0.0011x + 23.722
ILQHFOD\
R² = 0.8454
y = 0.0011x + 23.812
/DQG/\QFK
G2&OD\Å9602:
R² = 0.763
G18O, ‰ VSMOW
H[FOXGHG
\ [
5ð

\ [
5ð

)LJ*UH\
G2EXONVKDOH 'HSWKP
Figure 2. Shales and their isotopic values. These figures demonstrate the utility of using bulk shale as a
proxy for weathering conditions, and that more tedious analysis of fine clay fractions shows equal scatter in
the results. A) An unaltered shale with fine-grained texture, and a shale core sample that underwent second-
ary alteration, visible in disruption of thinly laminated schistosity, and recrystallization into coarser texture.
Wells 6 and 1 Pleasant Bayou, Brazoria County Texas. B) Comparison of bulk shale analysis with that of
the fine clay < 0.1 mm fraction showing offset linear relationships. Plotted data for Phanerozoic shales are
from Land and Lynch (1996) showing whole rock (WR) analyses offset by 0.33‰, and data from Land
et al. (1997) sampling the Oligocene Frio Formation (Brazoria county, coastal Texas) offset by 0.89‰
because the quartz fraction in shale is a few per mil heavier than the bulk clay by mass balance. There is
an overlap within the field of bulk shale and fine clay data, taken from Land et al. (1997) sampling the Frio
Formation, and the shallower Miocene formation. C) Comparison of bulk shale analysis herein (Table 1)
with isotope analyses for fine < 0.05 mm clay from Yeh end Savin (1977) and < 0.1 mm clay from Morton
and Land (1985) from the same depths of the Frio Formation. Notice close overlap justifying the use of
bulk shale measurements. The scatter is likely due to inherent depositional heterogeneity of the deposits
or perhaps may indicate the additional presence of high-d18O (amorphous?) quartz in fine-clay fraction.
Table 1. Oxygen isotope analyses of Pleasant Bayou #1 Brazoria county, Texas drill hole shale samples.
Depth, m T °C d′17O, ‰ d′18O, ‰ D′17O0.5305 D′17O0.528
0 21 9.142 17.545 −0.166 −0.122 surface sediment

loose smectite-rich shale,
−1326a 62 10.154 19.41 −0.143 −0.094
Miocene
−1326b 62 10.139 19.409 −0.158 −0.109 "
−2042 85 11.059 21.204 −0.19 −0.137 "
−3114 119 10.707 20.469 −0.152 −0.101 denser shale, Frio Fm Oligocene
−3117 120 10.398 19.852 −0.133 −0.083 "
laminated illite-rich shale Frio
−3840 143 10.219 19.489 −0.12 −0.071
Fm Oligocene
−3845 143 10.038 19.204 −0.15 −0.102 "
−4287 157 10.268 19.656 −0.159 −0.11 "
−4498 164 9.397 18.027 −0.166 −0.121 "
−4740 171 9.066 17.329 −0.127 −0.084 "
a
−4743 171 9.86 18.841 −0.135 −0.088 "
b
−4743 171 9.804 18.782 −0.16 −0.113 "
c
−4743 171 11.127 21.267 −0.155 −0.101 quartz
c
−4743 171 8.107 15.500 −0.116 −0.077 clay
Note: Temperature is estimated using Dutton et al. (2015). d′17O has ±0.01‰ average (1s) error based on five 8-cycles analysis (40 sample-standard
repetitions). d′18O has error of 0.1‰ based on standards; a,b means different areas of the same core sample. c refers to minerals (bulk Qz and bulk
clay) extracted from this sample, computed liambda of quartz-clay triple oxygen isotope fractionation is 0.5239 at this temperature,
In the Figure 3 we plot concurrent records for carbonates, cherts, and shales. All of these
show a general temporal increase of d18O, but with a different pattern. Note for example the
decrease in d18Oshale values in the Permo–Triassic (~0.25 Ga), and lack of a similar signal in
carbonates and cherts. There has been a bitter and contentious debate regarding these records
in the past 50 years: either they record the result of decreasing d18O value of seawater and
higher temperatures in the Archean and the Precambrian, or they record the result of alteration
by meteoric waters (Muehlenbachs and Clayton 1976; Perry et al. 1978; Holland 1984;
Gregory et al. 1991; Muehlenbachs 1998; Wallmann 2001, 2004; Veizer et al. 2002; Kasting et
al. 2006; Sengupta and Pack 2018; Kanzaki 2020a). This Chapter will touch upon these lines
of discussion by employing triple oxygen isotope values.
Mineralogically, shales contain two primary components: 60–70% clays, e.g., kaolinite,
chlorite, illite, and smectites, and 20–30% quartz, with only minor amounts of other
constituents e.g., feldspars, zircon, rutiles, etc. This mineralogical constancy characterizes
shales of different ages (Shaw and Weaver 1965). Therefore, the oxygen isotope composition
of mineralogically similar bulk shales through time, provides a consistent record of the
weathering of terrestrial surfaces exposed to atmosphere and hydrosphere.
No systematic triple oxygen investigation of shales and sediments has been performed
to our knowledge except for our recent efforts (Bindeman et al. 2018, 2019), although related
low-temperature rock types such as cherts, low-T silica, and clays have been targeted both
historically and recently (Fig. 3, Knauth and Lowe 2003; Levin et al. 2014; Hayles et al. 2017;
Sharp et al. 2016; Liljestrand et al. 2020). The advent of Secondary Ion Mass Spectrometry
(SIMS) in the past decade has resulted in tens of thousands of zircon detrital and igneous
246 Bindeman
analyses that permit linking their O and Hf isotopic values to U–Pb age (Valley et al. 2005;
Dhuime et al. 2015; Hawkesworth et al. 2017; Spencer et al. 2017, 2019). Zircons preserve
d18O values well even when there is no host rock available (especially for the oldest Hadean
earth, 4.4–4.0 Ga, Wilde et al. 2001; Valley et al. 2002; Cavosie et al. 2018), this record however
is “twice removed” from surficial environments as it reflects magmatic and metamorphic
crystallization temperatures. However, these efforts by different methods and research groups
may nicely complement each other by providing cross-checks
Figure 3. Comparison of d18O values of carbonates and cherts (A) with those for bulk shales and til-
lites (B). Data in this figure are from Bindeman et al. (2016) compilation with addition of data from Payne
et al. (2015, shales and pelites only) and Gaschnig et al. (2016, tillites). Shales and tillites both indicate
temporal d18O increase with an offset. These three sedimentary archives all show temporal increase but
exhibits significant differences in the shape of the trends. In particular, the Phanerozoic record for shales
exhibit a U-shaped trend, while the other proxies show an overall increase with time. Arrows indicate
required temperature range and bounds to maintain d18O ocean at modern ~0‰ values, or a required d18O
change of ocean to keep the temperature constant. Each arrow is pegged to specific d18O values on the y-
axis corresponding to indicated temperatures at 0‰ seawater.
Shales, granites and continental crust growth

Continental erosion generates fine-grained weathering products, which accumulate in
depositional basins, river deltas, and are less dense (2 g/cc) than the bulk of the continental
crust (2.7 g/cc). Even upon diagenesis and porosity reduction, shales and sediments remain
much less dense and thus largely do not subduct into the mantle but instead accumulate in
accretionary sedimentary prisms during plate collisions (Lackey et al. 2008, 2012). The rock
cycle turns them into metamorphic, anatectic, and then igneous rocks (Fig. 4), commonly
forming granitic stocks and batholiths. These high-grade crystalline rocks eventually
become part of consolidated cratons. The mantle-derived juvenile magmas and their
differentiates often provide heat and mass for crustal remelting and assimilation. In other
cases however, relamination of silicic crust in subduction zones (e.g., Hacker et al. 2011)
accretes metamorphosed accretionary (and high-d18O) sediments to the arc margins. Shallow
subduction and trench migration also aid in the re-accretion of sediments, and slab brake-off
lead to granite and ignimbrite flare ups (Chapman et al. 2013), remelting these sediments, such
as in wide areas of western N America and Mexico (Watts et al. 2011; Ducea and Chapman
2018). Older metavolcanics, sandstones, carbonates, and greywackes constitute other
components of arc rocks in creating the new crust, however siliciclastic sediments and shales
constitute the majority of sedimentary rocks (Wedepohl 1995; Simon and Lecuyer 2005).
Their subsequent weathering again into shale, passes their isotopic and chemical signatures to
the second generation of shale. This rock cycle serves as an interesting topic to balance and
reconcile multitude of elemental and isotopic fluxes and cycling and continental crust growth
and evolution, multiple research groups around the world are engaged into this effort using
their elements or isotopes of expertise.
Here I stress and demonstrate that transformation of sediments into metamorphic rock,
then migmatite, and then anatectic granite (e.g., Fig. 4), inherit the absolute majority of oxygen
originally derived from the hydrosphere via weathering. Bucholz and Spencer (2019) recently
presented a nice overview of strongly peraluminous granites in the Precambrian, as these
granites most closely reflect bulk sediment melt (cf Fig. 4). It is thus somewhat paradoxical that
even the anatectic granite contains most of the oxygen that is ultimately derived from ancient
rain water! The evolution of the continental crust, from the oxygen isotope perspective, thus
includes continuous build-up of weathering-derived oxygen with its triple oxygen signatures
(high-d18O, low-∆′17O), into the crust. A good analogy is our prior effort in studying low-d18O
magmas worldwide (Bindeman 2008) showing that magmatic rocks with d18O values less
than the mantle value of 5.7‰ are derived from re-melted hydrothermally altered rocks, and
thus surface-derived oxygen (e.g., Friedman et al. 1974). A good example is the Yellowstone
hotspot interacting with continental crust, and driving widespread hydrothermal alteration
which gives the crust low-d18O values (e.g., Colón et al. 2019). Then, multiple overlapping
calderas bury, remelt, and recycle this low-d18O hydrothermally altered surface of volcanic
and plutonic rocks (Watts et al. 2011). These 1–2 Ma caldera cycles at hot-spot volcanoes
represent an analogy for the entire crust evolution on perhaps 1–2 Ga timescales.
A B C
Figure 4. Rock cycle and largely isochemical and iso-isotopic transformation of shales into granite [Used
Fig. 4 Greyscale
by permission of Oxford Journals from Bucholz and Spencer (2019), Journal of Petrology, Vol. 60, p.
1299–1348, Fig. 1].
Here we extend this statement by suggesting that oxygen in most recycled continental
crust reflects inheritance and cumulative build-up of oxygen inherited via exchange reaction
from the hydrosphere. The hydrospheric effect on rocks largely excludes mantle and mantle-
derived igneous rocks, because while slabs with diverse d18O have been subducted at least
since mid-early Archean (Condie et al. 2013; Brenner et al. 2020), their mass is negligible to
that of the mantle, and thus the shift in mantle d18O is also negligible, but still is occasionally
recognizable (Turner et al. 2007).
248 Bindeman
Investigation of triple oxygen isotopes in high-grade metamorphic, in anatectic rocks, and

in granites provide a needed complementary view into the evolution of the continental crust
in each time interval. Many orogenic granites represent the result of direct anatectic melting,
or a large-degree assimilation of metasedimentary crust. Thus they average its composition
and serve as a “time interval” average of hydrosphere conditions acquired during previous
rock and attendant supercontinent cycles. The increasing d18O values of shales (Fig. 3, and
decreasing ∆′17O values) and granites (figures below) are considered to be correlated; they
both indicate the inheritance of heavy oxygen from surface weathering reactions, and their
time-integrated addition of hydrosphere-cycled oxygen into the silicate earth.
PART I: TRIPLE OXYGEN ISOTOPES IN WEATHERING PRODUCTS FROM

MODERN AND RECENT ENVIRONMENTS
Weathering reactions and stable isotopic parameters
Continental weathering of silicate rocks proceeds with depletion of alkaline elements and
calcium, followed by silica removal, leading to overall enrichment in Al in secondary neo-formed
clays (Retallack 2001). As a consequence, Al-based indices of alteration, such as the Chemical
Index of Alteration (CIA= Al2O3/(Al2O3+CaO+Na2O+K2O)mol; Nesbitt and Young 1982), make
use of Al/Si ratios, or other proposed chemical parameters, and typically serve as a measure of
weathering intensity. To illustrate, a weathering reaction for anorthite proceeds as follows:
CaAl2Si2O8 + H2O + 2H+ + CO32− = Al2Si2O5(OH)4 + CaCO3

Feldspar Kaolinite Calcite (1)
CIA=50 CIA=100
As oxygen is exchanged in these reactions, the products carry a time-integrated isotopic

record of weathering waters d18,17OWW, and temperatures:
d′18,17Oweathering product = ∆′18,17Omineral–MW f(T) + (d′18,17OWWf(T) + ∆′18,17MW–WW) (2)
Similarly,
∆′17Oweathering product = ∆∆′17O mineral–MW f (T) + (∆′17OWW f (T) + ∆′17OMW–WW) (3)
where for both meteoric waters (MW) and weathering waters (WW), fractionation factors are
functions of temperature, and ∆MW–WW is a potential offset between meteoric and weathering
waters (e.g., due to evaporation in soils and watersheds and other effects).
Here and below, we use linearized definitions of d18,17O = (d18O/1000 + 1)∙1000
(Miller 2002; Miller and Pack 2021, this volume), see Miller and Pack (2021, this volume)
for definitions, and ∆′17O is defined using a “rock” scale slope of 0.5305, but data is also
presented in Tables and some figures with respect to the common scale with the slope of
0.528. In linearized coordinates these relationships are strict and fully interchangeable:
d′18,17Omin − d′18,17Owater = ∆ ′mineral–MW = 1000 ln amineral–MW. Data presented in this work is
normalized to VSMOW scale using d18,17O values of standards obtained at the University
of Oregon: SCO = 5.24‰ ∆′17O = −0.05‰; UWG2 garnet ∆′17O = −0.06‰; SKFS chert
(∆′17O = −0.14‰) and others; see Miller et al. (2020), Miller and Pack (2021, this volume),
and Sharp and Wostbrock (2021, this volume).
Globally, isotopic values d′18,17O and dD in precipitation are strong functions of
mean annual temperature (MAT) (e.g., Fig. 1; Dansgaard 1964; Rozanski et al. 1993;
Surma et al. 2021, this volume), a relationship that has been utilized widely to target paleoclimate
based on isotopic values of surface mineral proxies. Furthermore, since the mass-dependence
global waters varies along the widely utilized meteoric water line: dD = 8 ∙ d18O + 10 (Dansgaard
1964). A multitude of studies since have demonstrated better applicability of local meteoric
water lines, which are different from global by slope and intercept. For D/H ratio, the offset, or
d-excess, is typically related to humidity of the boundary layer undergoing evaporation, and to
a lesser extent, to the amount of precipitation (Criss 1999; Surma et al. 2020).
For triple oxygen isotopes, Luz and Barkan (2010) have proposed the widely utilized
global triple oxygen meteoric water line:
d′17O = 0.528· d′18O + 0.033 (4)
where the 0.528 slope of mass dependent oxygen fractionation is largely adhered to in
most subsequent studies. Likewise, the 0.033 intercept value which we below call Y, or
“Excess_D17OMW” is related to humidity and precipitation effects, affecting offsets in
triple oxygen isotope plots (e.g., ∆MW–WW). Investigation of triple oxygen isotope effects in
precipitation is a topic of current research represented by Passey and Levin (2021, this volume)
and Surma et al. (2021, this volume).
Unlike bio(chemo)genic water precipitates for which water isotopic compositions are
better known, each clay mineral during formation and transport interacts with a variety of
waters: mean annual precipitation (MAP), evaporated soil waters, higher-altitude waters,
diagenetic waters, and finally river water, and at different environmental temperatures
including mean daily and seasonal climatic variations. Furthermore, some clays such as illite
and chlorite may be derived from recycling of older sedimentary rocks, having experienced
repeated weathering cycles that can reset initial O isotopic values. However, due to the large
excess of meteoric water involved in weathering, we argue and test below that MAT and
MAP can be used as first order proxies for world clay d′18O and ∆′17OMW signatures, and
using the simplest global relationships such as Equation (4) (with and without intercept) is
the most parsimonious approach to weathering, and that meteoric water serves as the average
weathering water (Bindeman et al. 2019, Fig. 1).
Using triple oxygen isotopes in weathering products to estimate T and d′18Ow values
Before we enroll into a process and algorithm to deconvolute and interpret paleoclimate
signatures of mudrocks and shales—the dominant but mixed sedimentary rock throughout the
Earth′s history—the reader may ask the question: why simply not use bio(chemo)genic proxies
such as cherts and carbonates that directly precipitate from waters (e.g., Fig. 3)? There are two
main answers. First, having an independent proxy using the most abundant sedimentary rock
type can provide most robust constraints into evolution of weathering through time, as we
think the uniformitarian nature of isotopic fractionation and equations that govern evaporation,
water cycling and isotope fractionations holds through time. Second, carbonates and especially
cherts have been plagued by the evidence of their post-depositional alteration (McKenzie
2005; Ryb and Eiler 2018; Liljestrand et al. 2020), which can be avoided by considering
petrographically intact shales, e.g., those without any recrystallization and coarsening features
(Fig. 2b). Additionally, Precambrian carbonates and especially cherts are polygenic (high-T
hydrothermal and low-T chemogenic, Marin et al. 2010; Zakharov et al. 2021, this volume),
and commonly represent mixed sources and processes.
Sedimentary proxies involving only d18O has always suffered from resolving effects
of temperature or the starting d18Ow value. It plagues discussion of their relative effects and
interpretation of isotopic records such as shown on Figure 3, as due to higher temperatures or
lower d18O value of the hydrosphere in the Precambrian (Muehlenbachs and Clayton 1976;
Veizer et al. 1999; Knauth and Lowe 2003). Triple oxygen isotope geochemistry is called
250 Bindeman
to resolve these question as there is an additional independent fractionation equation for

d17O. Likewise, when hydrogen isotopes are deemed primary for hydrous phases, such as
opal, smectites and kaolinites, for which both D/H and d18O fractionations factors are known
(Savin and Lee 1988; Sheppard and Gilg 1996), it is possible to independently determine three
unknowns: T, d18O and dD (Capuano 1992; Delgado and Reyes 1996; Mix and Chamberlain
2014). This involves solving two fractionation equations for d18O and dD in combination
with a dD − d18O meteoric water line equation, e.g., dD = 8 ∙ d18O + 10 of Dansgaard (1964).
Hydrogen isotopes are however not part of the same isotopic system and easily modified by
secondary processes in ancient weathering products and thus emphasis here on triple oxygen
isotope geochemistry is warranted.
The triple oxygen isotope value of a particular phase of the weathering product or, by
extension, a chemogenic precipitate (e.g., chert and carbonate) will reflect three parameters:
i) the isotope fractionation relative to the parental water (Eqns. 2–3: ∆′18Omineral–MW =
d′18OM − d′18OW), ii) the mean annual temperature of weathering, iii) the d18O and d17O values
of parental meteoric water that participated in weathering reactions, or an assumption about
their relationship, through for example a global or local MWL (e.g., Eqn. 4).
The appearance of the first independently determined triple oxygen fractionation equation
for quartz (Pack and Herwartz 2014; Sharp et al. 2016) and analogous efforts for other
minerals (Chapters by Wostbrock and Sharp 2021; Passey and Levin 2021, both this volume),
added another equation, yielding three equations (D17Omineral–water fractionation, d18Omineral–water
fractionation, and meteoric water line) for three unknowns (d18OW, d17OW, T).
For example, we proposed these two equations for bulk shale–water fractionation (e.g.,
Bindeman et al. 2018, 2019):
 3.97  10 4.9  10
6 3
 18
103 ln  Rshale 18   (5)
 Rwater  T2 T
 3.97  106 4.9  103   2.05 

17Oshale-water   2
   0.5305   (6)
 T T  T 
Here, to characterize the triple oxygen composition of a sample, the Δ′17O value is used to
show the amount of 17O excess relative to a reference line λ = 0.528 per Equation (4).
Bindeman et al. (2019) numerically solved the Equations (4–6) to obtain the temperature
of weathering and d′17O and d′18O values of meteoric waters in equilibrium with modern
suspended sediments in modern river watersheds, for which these parameters are independently
known (MAT and d′18O for river or watershed waters). They further explored parameter space:
changes in the slope of the Equation (4) and 17O excess parameters (Fig. 5a). This Chapter
additionally presents computed meteoric waters in equilibrium with river clays (Fig. 5b).
These authors observed an overall match of observed modern river sediments and weathering
products, and computed parameters based on them. They suggested a slope of 0.528 and an
intercept Y that ranges from 0.033 to 0. Furthermore, Equations (5) and (6) can be linearly
combined with the quartz–water fractionation equations of Sharp et al. (2016) to derive
fractionation for any shale with a known wt% of authigenic quartz.
Comparison of waters involved in weathering at each well-studied watershed is a worthy
task (see Fig. 5b and chapters by Passey and Levin 2021; Herwartz 2021, both this volume).
Special environments, such as extremely dry regions in Africa, have high d18O values and
local MWL with 17O excess parameter in Equation (4) of 0 or even less than 0 for lake
waters (Passey and Ji 2019). A study of triple oxygen variations of soil water, river water,
and mean annual precipitation is thus urgently needed for different regions around the world.
A B
A Evaporation
Reconstructed meteoric waters 0.1
Effects
0.08
0.06
''20:Å
0.04
y = -0.0018x + 0.0077 0.02

R² = 0.0725
x
0.5305
0
tropical
-20 -18 -16 -14 -12 -10 -8 -6 -4 -2 0
602:
temp humid -0.02
G 1820:Å
d′ OMW, ‰ Lake
arctic
Sand av. -0.04
temp dry
x Silt av.
Big 5 -0.06
rivers
Archean shales Antarctic clay
Fig. 5 Color, feel free to arrange A,B vertically

Figure 5. A) Triple-oxygen isotopes for the < 4 mm fraction suspended sediment load from world rivers
(Fig. 1), plotted for mean annual temperature (MAT) of weathering conditions in their respective water-
sheds. Clay–water fractionation lines emanate from the parental weathering waters. Note the significant
overlap of clay triple-oxygen isotopic values from different climates, broadly explained by counterbalanc-
ing relationships of isotopic fractionation and MAT–d18OMW values. Clays from the modern world overlap
the field of post-Archean shales, and this advocates for a comparable range of d18OMW and surface tempera-
tures, and overall hydrologic conditions of weathering since the post-Archean (inset showing Effects). The
concave up yellow curves denote the mixing of weathering products with their detrital crustal material (see
Fig. 4); the proportion of detritus is greater in global silt and sand. Two water lines are shown, the MWL of
Luz and Barkan (2010), and the WW, “weathering water” line explained by the extent of the evaporation
of water participating in weathering in soils (small vector under Effects). [Used by permission of Elsevier,
from Bindeman et al. (2019), EPSL, Vol. 528:115851, Fig. 7]. B) Meteoric water line reconstructed us-
ing triple oxygen isotopic values of clays in rivers and temperature (MAT) and d18OMW values of their
watersheds using formula 8 in text and data in Bindeman et al. (2019). Note that inverted this way waters
are shifted downward, towards evaporated waters. Field and a line to the lake shows data from Passey and
Levin (2021, this volume).
Bindeman et al. (2019) compiled d18O data from IAEA online resources and observed that river
water and mean annual precipitation in their watershed basins for most rivers and precipitation
around the world broadly agree (within ~1‰). We thus assume that weathering can be
approximated, at least for now, by the values of precipitation, and use global MWL Equation (4).
To illustrate the above procedures, Figure 6 plots triple oxygen isotope values of a
hypothetic weathering product in d′18O and ∆′17O space, and explains the equation-solving
approach. A system of three Equations (4–6) has three unknowns: temperature, d′18O and d′17O
of parental meteoric water. We can thus solve them analytically, given an extra fractionation
equation provided by d′17O.
The above Δ′17O−d′18O reference framework is that which we will employ in the rest of
this Chapter, and can also be conveniently thought in terms of previously used slopes of d′17O
vs. d′18O. Slopes of the fractionation lines in terrestrial environments range between < 0.516 and
0.5305 and are specific to temperature, process, and the compositions of substances. Highest
temperature equilibrium between silicates in igneous rocks show a slope close to 0.53, horizontal
line in Figure 6 (Herwartz and Pack 2014; Pack et al. 2016). In this Chapter, we adopt the
“rock” 0.5305 reference line, in which the d′17O deviation from this reference fractionation line
is always negative and expressed as Δ′17O = d′17O − 0.5305 ∙ d′18O. We also report ∆′17O values in
Tables relative to a reference line with a slope of 0.528 and zero intercept, most commonly used in
low-T and hydrologic communities. In particular this slope is characteristic for VSMOW-SLAP
(Surma et al. 2021, this volume). For minerals that formed at lower temperature in equilibrium
with water, such as amorphous silica, clays, quartz, and calcite, fractionation between water
252 Bindeman
A B
Modern World Past Hydrospheres
Modern World Past Hydrospheres
)LJColor or Greyscale, feel free to arrange vertically
Figure 6. A) An example of the application of bulk shale–water triple oxygen isotopic values to resolve the
isotopic values of weathering waters and the temperature of weathering. This involves solving a system of
3 unknowns: temperature, Δ′17O, and d′18O, of parental weathering waters using a system of three Equa-
tions (4–6) in text. Concave down fractionation lines are for bulk shale emanating from d′18Ow = −5‰, and
−15‰ parental waters. Also shown is the concave up mixing curve between weathering product and the de-
arrange verticallytrital component of shales. Weathering product is computed via a mass balance equation using independent
chemical information for sample in this case the Chemical Index of Alteration (CIA). Diagenesis (see text)
is mostly rock-dominated and is conservative of the weathering-related Δ′17O and d′18O values, but leads to
scatter in the data for each formation. B) Effect of changing d′18O values of the ocean on its Δ′17O. Sample 1
can be solved for temperature and d′18Ow, if they belong to an accepted MWL, as in A). For example Ar-
chean MWL2 gives +2 °C and d′18Ow = −25‰, and using the modern MWL gives +76 °C and −8‰ d′18Ow.
Red dots along the line denote hypothetical ancient ocean water, each with a corresponding MWL. The shift
likely proceeds along a line connecting the mantle and the modern ocean water as it predominantly involves
high-T interaction of seawater at mid ocean ridges (Sengupta and Pack, 2018), and weathering with the
comparable slopes shown from Bindeman et al. (2019). Y is an excess Δ′17O as per Equations. (4) and (8).
and minerals results in a smaller slope value within the 0.525–0.524 range (Cao and Liu 2011;
Bao et al. 2016; Sharp et al. 2016). Some lower T kinetic processes involving water and gases
fractionate isotopes yielding an even lower exponent of 0.516 (Young et al. 2002; Pack 2021,
this volume) or even 0.500 for gases in vacuum (Clayton and Mayeda 2009). These smaller
slopes lead to strongly negative Δ′17O values at the same d′18O values (Fig. 6). For example,
below we explore formation of secondary quartz under low-T conditions (silicification) as the
mechanism to explain strongly negative Δ′17O values in some shales.
Computing the weathering product

Shales and recent sediments represent a mixture of solely physically weathered to strongly
chemically weathered materials (Shaw and Weather 1965). The contribution of unweathered
detrital (feldspars, amphiboles, etc) and previously weathered detrital (muscovites, illites)
products needs to be evaluated to estimate d′18O and Δ′17O isotopic values of the weathering
product, by for example, using XRD. Figure 6 explores this. The proportions of detrital
contribution to weathering products can also be evaluated based on a chemical analysis
of a rock, and by using the simplest of weathering proxies, the CIA parameter defined by
Nesbitt and Young (1982). CIA essentially measures the degree of feldspar conversion to
clay minerals; a value of 45–50 signifies the absence of chemical weathering and the average
composition of continental crust from Rudnick and Gao (2003) yields values between 45–50.
Values between ~85 for K–Na bearing clays, and 100 for Ca–Na–K free kaolinite and chlorite,
denote a rock that had undergone nearly complete chemical weathering, and thus likely oxygen
isotope exchange with weathering waters. Since the CIA parameter does not include silica,
addition of detrital quartz would not affect the CIA value. We note however that the detrital/
authigenic proportions of quartz very likely proportional to that of clays during weathering
and sediment transport in silt- and smaller-size particles (Pettijohn 1957). For example, wind-
blown and water-transported silt-size and smaller quartz was considered entirely authigenic
as demonstrated by its high d18O values of +19.1 ± 1.4‰ worldwide (Churchman et al. 1976).
Figure 6 illustrates a graphical way of derivation of a weathering product, which
algebraically involves a solution of a simple linear relationship d1xOshale = d1xOweathering product X +
d1xOdetrital (1 − X), where X are normalized CIA values (CIA = 50 corresponds to X = 0, and
CIA = 100 corresponds to X = 1) and solve it for d1xOweathering product since the concentration of
oxygen is the same in detrital and weathered materials. One needs to know or assume d18Odetrital
and Δ17Odetrital values, note the unprimed delta values needed for mass balance calculations.
Bindeman et al. (2019) computed a global average silicate surface undergoing weathering as
+11.5‰ and −0.105‰ respectively. Different watershed areas in the modern world, especially
for any sizable rivers that sample a variety of bedrock types are shown to be close to these
values, as 60–70% of all silicate surface is represented by well mixed sedimentary rocks.
The sensitivity analysis shows that variation of CIA within ±10 units results in ±1‰
change in the final output, computed d′18Ow, and ±8 °C in computed temperature (Bindeman
et al. 2018). Such uncertainties are comparable to other sedimentary proxies, such as cherts
and carbonates. Evidence for this is that a typical CIA index for shales is 75 and does not
vary much across their geologic history (Bindeman et al. 2016, 2018; Gaschnig et al. 2016).
Thus, any procedure of correction for detrital contribution, likely offsets detrital estimates
by a comparable constant, on average. Given this, one can simply compare measured values
for shales as proxies for the environments, without computing absolute temperature and
isotopic values of weathering waters involved. Such apple-to-apple comparisons permits
broad statements regarding the change of the weathering conditions on earth through time.
If no CIA correction is applied to measured d′18O and Δ′17O values of modern river sediments,
and thus effectively assuming that they are entirely composed of the weathering product, the
computed d′18O of the water shows the comparable range, but the T of weathering is increased
by ~10 °C and the resulting values are more scattered (Bindeman et al. 2019). These results
are easily understood considering that the mixing lines are sub-parallel to the fractionation
lines in Figure 6, where an increase in detrital proportion corresponds mostly to an increase in
temperature with less effect on d′18Ow.
Weathering in different climate regions
Studies of d18O and dD in monomineralic clays are used historically to reconstruct
paleoclimate (Savin and Epstein 1970a,b; Capuano et al. 1996; Sheppard and Gilg 1996; Bauer
and Vennemann 2014; Mix and Chamberlain 2014). Monomineralic clays such as smectite and
254 Bindeman
kaolinite do exhibit scattered correlation with temperatures below ~15 °C (Sheppard and Gilg
1996; Savin and Hsieh 1998) and overlap at higher temperatures. Bulk clays and weathering
products are typically a mixture of common clay types: smectite, kaolinite, illites, and chlorite,
with secondary quartz and albite, and proportions of these depends on the climate (e.g.,
Fig. 1, Newman 1987). It is difficult to nearly impossible to separate individual clays from a
mixture and when handling common sediments derived by weathering, and so bulk analysis is
sometimes the only option available (Gilg 2004).
Increasing Mean Annual Temparature (MAT) causes a decrease in clay–water isotope
fractionation, thus d18Ow of meteoric waters decrease with decreasing MAT (Dansgaard 1964;
Rozanski et al. 1993).
d18OMW = 0.69 · MAT − 13.6 (7)
As a result of opposing effects of temperature on clay–water fractionation (Eqn. 5) and this

hydrologic relationship (Eqn. 7) between temperature and d18OMW, bulk clays from different
climatic regions do not vary much isotopically (Bindeman et al. 2019; Fig. 5a). Only subarctic
clays formed from very low-d18O waters display low-d18O values, there is little difference
among different climate zones. Although this is discouraging news for paleoclimatology
involving isotopic values of bulk clays, this effort captures the global climate conditions in the
modern world with its temperature, hydrologic cycle, and weathering conditions.
Furthermore, measured d′18O and d′17O of clay samples from the river basins modern
world (Fig. 1) with broadly known MAT of their watersheds and d′18Ow values of precipitation
constitute a “natural experiment” of weathering. It enables solution of fractionation
Equations (6) and (7), to solve the MWL Equation (4) for the intercept Y. The inverted equation
for this purpose is:
Y = d′17OS − 0.528 · d′18OS − (d′18OS − d′18OMW) · (0.5305 − 2.05/T) (8)
where Y is ∆′ O—Excess as per Equation (4) (e.g., 0.033), T is MAT for each river basin, and
17
d′18OMW is compiled water data, subscript “S” refers to measured sample data.
Data for d′17OS, d′18OS, d′18OMW and MAT are substituted to this equation to compute the Y
parameter, and results of inversion are presented on Figure 5b. One can see that the computed
MWL, using this procedure plots subparallel to the global MWL (Eqn. 4), with smaller Y.
This likely reflects more evaporated waters in soils involved in weathering and production of
clays, than in precipitation (Passey and Ji 2019: Passey and Levin 2021, this volume). This
was also suggested in Bindeman et al. (2019) by comparing computed vs. compiled MAT
and d′18OMW. The Equation (8) enables further and more detailed analysis and quantification
of the weathering products found in a particular water basin. This equation should be further
improved by experimental determination of the clay–water d17O fractionation equation, as was
done for quartz and carbonates (Wostbrock and Sharp 2021, this volume). This may result in
revision of the 1.85 or 2.05 fit parameter in the last terms of Equations (6) and (8).
Bindeman et al. (2019) presented triple-oxygen isotope analyses of clay- and silt-size
sediments from 45 rivers worldwide (Figs. 1, 5), which together cover about 25% of Earth′s
continental area, extending from the tropics to polar regions. They found that a majority of the
studied clays closely approximate weathering products, always having high-d18O signatures
regardless of the bedrock type, and in equilibrium with local meteoric waters at the respective
MAT. Globally averaged, sediment-flux weighted suspended river sediment d18O and D17O
values are +14.8‰ and −0.16‰, respectively. These values are significantly skewed toward
O isotope signatures for the southeast Asia and western Pacific regions, characterized by
very high sediment fluxes to the ocean (Milliman and Farnsworth 2011). Using both clay-
and silt-size fractions, the total flux-weighted silicate weathering d18O signature exported to
the world′s ocean is −2.59‰, almost 50% higher than the previous estimate, underscoring
greater proportional effect of modern weathering on global hydrospheric fluxes. This is also
yielding an ice-free world hydrosphere estimate of −0.78‰, very close to estimates provide
by Sengupta and Pack (2018) and −1‰ value of Shackleton and Kennett (1975). Overall, the
modern river clays represent a snapshot of modern weathering conditions on continents, and
associated first-order climatic signatures related to MAT and d18O of the hydrosphere.
Effects of diagenesis on d18O and Δ′17O in shales
Sediments weathered by low-d18O meteoric waters most commonly accumulate in
marine environments, forming kilometer-thick sedimentary sequences. Yeh and Savin (1976)
demonstrated with marine sediments that clay residence times of up to 106–107 yr in cold
~0 to −1‰ seawater does not affect their O and H isotopic values acquired during weathering.
Yeh and Esslinger (1986) demonstrated no d18O modification of original detrital clays in
Mississippi River delta or DSDP site 323 upon burial to depths as deep as 600 m.
It is indeed important to relate isotopic values of weathering products to those of
consolidated shales that went through diagenesis. For weathering products, these values reflect
surface weathering conditions (T, d18O, d17O), with consolidated sedimentary rocks, and it
is thus essential to establish the offset (if any) diagenesis induces in the original products.
The latter is important to our ability to read the climate signature in sedimentary rocks.
The most important reaction of siliciclastic sediment diagenesis is the smectite to illite
transition, that occurs between ~70 and 130 °C (Eberl 1993). This process leads to strong
reduction of porosity from > 20% to commonly < 5%. This topic has been visited previously
using d18O of bulk sediments, clays and quartz, carbonate cement, and pore fluids (Hower et
al. 1976; Wilkinson et al. 1992; Land et al. 1997; Lynch et al. 1997). An excellent locale to
test this question is in the thick piles of Oligocene–Miocene sediments along the Texas coast,
where many wells up to 5000 m in depth fully span the smectite/illite transition (Fig. 7)
Our recent measurements of ∆′17O and d18O (Fig. 7) were matched with samples from
the same Pleasant Bayou and General Crude drill holes, according to depth, from Lynch et
al. (1997), Land et al. (1997), clay and fine quartz analyses of Yeh and Savin (1977), where
XRD-determined mineralogy is presented from these studies. Furthermore, d18O of diagenetic
pore waters has also been analyzed in this Texas drillcore, as well as in DSDP, other Texas
drill holes, and the Bengal fan in India (Lawrence et al. 1979; Land et al. 1992; Wilkinson et
al. 1992). Collectively, they demonstrate a trend from ~−1‰ for ice-free seawater to “shifted”
d18O values of +8–10‰ at the greatest depths of 4–5 km, and within the illite stability field.
Significant scatter of analyzed d18O waters with depth (Fig. 7) permit different statistical fits,
from linear to parabolic. This scatter is likely not analytical, but reflects different, and not-
linear, progress of illitization reactions, and perhaps lateral dewatering via permeable sandstone
layers at different depths. Isotopically negative water in shallow depth of Mississippi delta for
example (Fig. 7) reflect volcanic glass alteration and carbonate precipitation from pore water
in limited pore space (Lawrence and Taylor 1972).
The main observation of our new and previous analyses is that the d18O value of bulk
sediment undergoing transformation to illite is shifted by 1–2‰ to slightly lighter values, in
agreement with Lynch et al. (1997). Equilibrium smectite and illite-water fractionation factors
as a function of temperature, and partial expulsion of water, are sufficient to explain the trends
on Fig. 7 (Suchecki and Land 1983; Wilkinson et al. 1992). We here observe a minimal effect
on ∆′17O (Fig. 7), showing perhaps a slightly positive trend with depth. Water is lost during
diagenesis from the smectite to illite transition. We suggest that isotope shifts related to
progressive diagenetic water loss likely work in the same direction, favoring heavier d18Ow with
decreasing total water with depth as is observed in Figure 7. No ∆′17O analyses of basinal water
is available to our knowledge; we thus concentrate on d18O values assuming that ∆′17O will
follow similar trends, by relating shales with basinal water on Figure 8. Fractionation lines are
concave up with slopes of 0.525~0.528, mixing curves are concave up with slopes of ~0.51, and
256 Bindeman
A B C D
G2Å 7& '¶2Å ,OOLWH +2ZW

0 0
bulk shale, this work
-500 35° -500 Land et al. 1997
97
Mississippi delta clay
-1000 -1000
fine clay, drillcore
62°
-1500 pore water -1500
-2000 Smectite -2000

88°
'HSWKP
-2500 -2500
-3000 120° -3000
-3500 -3500
-4000 140° -4000

Illite
-4500 -4500
170°
-5000 -5000
Lynch et al. 1997
-5500 195° -5500
Figure 7. Triple oxygen isotopes in the deep Texas drillhole crossing the Oligocene Frio Formation, and
Miocene and younger sediments in coastal Texas (A–B). Also shown are the diagenetic transformation of
smectite
)LJ
into illite, which is accompanied by dehydration. A field for < 0.01 and 0.05 mm bulk clays in A
is from Yeh and Savin (1977), and the Mississippi River delta data are for < 0.3 mm clay, and the water
Greyscale
and are from the hole DSDP 323 (Yeh and Eslinger, 1986). Deeper water analyses from the same Texas
drillhole are from Willkinson et al. (1992). Dutton et al. (2012) estimated a 24 °C/km geothermal gradient,
and aging of sediment at 25 Myr per 100 °C increase, suggesting that the studied section represents ~45 Ma
worth of sediment accumulation and burial diagenesis. These sediments sample Mississippi watershed,
with relatively constant sediment input flux since the Oligocene (Galloway et al. 2011). There is little
isotopic change with depth, with an ~2‰ decrease in d18O, and no resolvable trend in Δ′17O, despite the
pronounced transformation from smectite-rich unconsolidated sediment, to illite-bearing laminated shale
(Fig. 2a). Some variation in O isotopic composition is likely due slight variations in sediment composition.
the linear “diagenetic water line” on Figure 8 has an intermediate slope around ~0.52. Upon
transition to illite, the micas, quartz, water, and pore water-precipitated cements are broadly in
internal equilibrium with each other (Yeh and Savin 1986). Milliken et al. (1981) and Land et
al. (1997) reported stages of mineral cement evolution in the Frio Formation drillhole involving
dissolution of detrital K-feldspar and quartz and precipitation of secondary quartz and mica.
We verified this result by plotting d18O values of the < 2 µm fraction quartz reported in Land et
al. (1997), and observed (Fig. 8) adherence to quartz–bulk shale fractionation using the quartz–
water fractionations of Sharp et al. (2016) for any quartz formed below smectite–illite transition
where the temperature >120 °C. Using the 0.1–0.5 µm quartz fraction from Yeh and Savin (1977),
quartz–water equilibrium is achieved at 85 °C, in broad agreement with Milliken et al. (1981) on
growth of authigenic quartz in pore spaces. This result is a test for internal equilibrium in shale
upon precipitation of secondary silica during transition of smectite into illite.
When plotted on a ∆′17O vs. d18O diagram, there is a significant overlap of samples from
different depths. When data points are compared to empirically calibrated bulk-shale–water
curves (Fig. 9) our new triple oxygen isotope analyses are scattered, but cluster around d18O
values of −5 to −15‰ equilibrium for parental water, and at T of +10 and +40 °C. The average
d18O and ∆′17O shale of the Frio Formation belongs to 15 °C and d18O of −11.5‰ parental water.
Although the bulk-shale–water fractionation are semi-empirically calibrated by Bindeman et
al. (2018) using the quartz–water fractionation of Sharp et al. (2016), datapoints are positioned
much lower in d18O and ∆′17O than would be required for shales in the borehole temperature
range of 85–170 °C and pore waters with +10‰ d18O.This indicates that shales provide a
record of weathering conditions, rather than diagenetic transformations. Furthermore, if
corrected for detrital contribution (as per Fig. 6, using CIA = 70 in these rocks), the computed
40 Smectite Illite
Kaolinite Chlorite
Bulk Clay Quartz
35
Bulk Shale Serpentinite
∆18O(mineral-porewater)
Texas drillhole bulk Texas drillhole Qz (fine)
or Bulk Shale δ18O, ‰

30 Bulk shale, Land et al. 1997
97 Bulk Shale, this work
25
S I
20 Bulk Texas
shale δ18O
Global Average
15 sediments δ18O
10
0 50 100 150 200
Temperature, °C
Fig. 8 Color Figure 8 ∆18O (bulk shale–porewater) and ∆18O (quartz–porewater) values vs. temperature in the Texas
drillhole (Fig. 7), is compared to experimentally determined isotopic fractionation factors (Savin and Lee
1988; Sheppard and Gilg 1996; Sharp et al. 2016). Porewater d18O values are linearly approximated from
Fig. 7 to compute the ∆18O(shale–porewater) values shown. Notice that shale quartz with T > 122 °C is in
equilibrium with pore waters, while fine < 0.5 mm quartz from Yeh and Savin (1977) is in equilibrium at
~80 °C. The Texas drillhole bulk shales (grey diamonds and lines) hover around the red dashed line
representing computed bulk shale–water equilibrium fractionation. Superimposed on this graph using the
same vertical axis are bulk shale d18O data from Fig. 7 plotted against drillhole temperature. Shale d18O
values can also be read as ∆18O(shale–0‰ seawater) fractionation. Note that in the smectite (S) to illite
(I) transition temperature window (dashed box with arrows), computed bulk-shale–0‰ seawater frac-
tionation is equivalent to the original bulk shale d18O value, yielding no net flux of oxygen isotopes at
these temperatures for the Texas drillhole samples. For modern average global sediments (e.g., Fig. 5a),
the crossover temperature is ~30 °C higher. At T higher than the indicated crossovers, bulk shales will be
shifted down due to diagenesis, and waters are shifted up as is observed in Fig. 7. For lower in d18O global
average shale, this shift is likely to be less, and only for deeper diagenetic conditions shown. Notice that
plotted shale bulk d18O values do not follow the bulk shale–water dashed red curve, and thus retain their
original d18O values because the shale–porewater system is a rock-dominated.
weathering product corresponding to the Frio Formation will shift further down and to the
right, and will overlap the composition of weathering products in equilibrium with either the
modern Mississippi River watershed, or those slightly higher d18O weathering products in
equilibrium with Oligocene waters.
We further observe no definite relationship between borehole T and the measured isotopic
values. While some variation is due to analytical error (±0.2‰ d18O and 0.01‰ ∆′17O, 1σ),
other variation may reflect slight variation in provenance of sediments. In either case the data
does not show any systematic trend with temperature of diagenesis or transition of smectite
to illite. Frio Fm changes d18O from ~21‰ at smectite-rich top and T from 25 to 35 °C to
+18.5‰ at depth. However, variations in modern Mississippi river muds show a similar range,
suggesting that observed vertical isotopic trends in the Frio Formation may be partly reflecting
temporal depositional variations. Hower et al. (1976) interpreted mineralogical and chemical
changes as the response of the shale to burial metamorphism and concluded that the shale was
a closed system for all components except H2O, CaO, Na2O, and CO2. This isochemical nature
was challenged somewhat by Land et al. (1997) who observed minor uptake of K2O with
depth related to illitization. These authors suggested that 0.2–0.3 wt% extra K2O was derived
from KFsp dissolution from adjacent silt and sandstone layers. Given that oxygen is a major
element, its isotopes should be predominantly conserved.
258 Bindeman
0.5305

G2Å

VHDZDWHU
seawater
Å Borehole T, °C
SRUHZDWHU

171
+100°C 120
ǻ¶2Å
171
143 119
watershed 151 157
T>100°
164
6KDOH
85
T<70° :3
G2Å
)ULR2OLJRFHQH)P
0LRFHQH)P -15‰
$YHUDJH)ULR
ǻ¶2Å
-10‰
10

0LVVLVVLSSLPRGHUQ(T -5‰
+15°
ÅG2&
diagenetic
0LVVLVVLSSL2OLJRFHQH(T
ÅG2&

Figure 9. A triple oxygen diagram for samples from a Pleasant Bayou Texas drillhole (Fig. 7–8) superim-
RU
posed on a grid of calculated values for bulk-shale–meteoric water fractionation lines, each emanating from a
point on the meteoric water line. The evaporated meteoric water line with a slope of 0.528 and intercept Y = 0,
is shown in blue as in Figure 5a, and a hypotherical diagenetic water line connecting 0‰ seawater with bulk
shales is shown in green. Concave up shale–diagenetic water equilibrium fractionation lines are also shown
in green and are emanating from the diagenetic water (dashed) line in main figure and square inset. These
may denote evolution of triple oxygen isotopes in pore waters and in shales experiencing diagenetic shifts.
Pre-diagenetic pore waters likely evolve along the straight line resulting from a combination of concave up
mixing and concave down fractionation lines. The approximate directions for low and high-T diagenetic
shifts is shown in the square inset. The brown box denotes the likely detrital composition of the Mississippi
watershed, taken as the average of modern Earth′s surface triple oxygen isotopic values. A concave up yellow
line connects this box with the average value of the Frio Formation shale to compute its weathering product
based on whole-rock chemical analysis (e.g., Fig. 6a); the blue angular polygonal field represents computed
weathering products for the Oligocene Frio and Miocene formations. Shale samples scatter around +15 to
+45 °C and −5 to −15‰ d18O parental meteoric water values, weathering products are shifted to lower T and
higher d18O, and overlap with equilibrium weathering product estimated values for both modern and Oligo-
cene–Miocene weathering conditions. The scatter in data reflect the some depositional isotopic diversity,
and minor effects related to diagenesis as is explained in the square inset and Figure 7. The circular inset
shows the portion of the diagram with borehole temperatures included for each sample. Notice that there is
no correlation between the borehole temperatures and their isotopic temperatures that are computed based on
shale-meteoric water isotope fractionation. This is due to the fact that diagenesis is a strongly rock-dominated
process and is largely isochemical, and thus preserving the weathering conditions.
The average composition of the Frio Formation plots much closer to the temperature and
d18O of precipitation in the Mississippi River Basin. Global temperatures that were involved in
generating weathering products at the Oligocene–Miocene boundary were 5–6 °C higher than
today′s 12.8 °C, thus around ~+17 °C (Retallack 2007). Given comparable to modern orography
and latitudes of N America then, d18O of precipitation was likely comparable in d18O to that of
modern Mississippi River draining wide swaths of the N. American continent, by analogy with a
relationship between average precipitation and river water (Galloway et al. 2011). In the warmer
periods of the Oligocene and Miocene, d18O values of precipitation may have been 1‰ heavier
given the global d18O–T relationship (Eqn. 7), but Antarctica was just in the beginning of its
~34 Ma glaciation, and thus the oceans were ~1‰ lighter than today (Shackleton et al. 1975).For
example, Zanazzi et al. (2015) found no variation in d18O of precipitation in northern part of North
America based on Perissodactyl mammal tooth enamel. The Mississippi River, and this Texas
drillcore thus represent a good insight into average Earth surface conditions in Late Cenozoic.
The paragraphs below provide further insight into diagenetic processes and their effects on
isotopic values of shales. The Frio Formation was deposited in −1‰ d18O Oligocene seawater,
and the exchange reactions were inhibited for tens of millions of years (Savin and Epstein 1972b).
Burial to several kilometer depth and geothermal T increase, perhaps approaching that of the
smectite/illite transition of ~70–120 °C, (Figs. 7, 9) initiated chemical solution-reprecipitation and
thus isotopic exchange reactions (Hower et al. 1976; Eberl 1993). At this stage, the mildly shifted
d18O = 0‰ seawater, filling the remaining ~10–15% porespace, begins reacting and exchanging
oxygen with the surrounding silicates. The important observation on Figure 8, is that the sense of
isotopic fractionation is such that the original bulk d18O of +18‰ of surface sediment has zero
permil D18Oshale–water (= d18Oshale − d18Oporewaters) fractionation at 85 °C. For a world average surface
sediment of 14.1‰ (Bindeman et al. 2019), the temperature of the inversion is higher at ~100 °C.
Thus, the initial transition of smectite to illite is accompanied by zero isotopic shifts in marine
porewater, and expelled porewater may be in equilibrium with silicate. If ancient seawater were
−10‰ as is commonly proposed, then the corresponding meteoric water evaporating from that
ocean, and weathering products would have also have shifted to lower values, and the logic on
minimal diagenetic transformation holds for ancient times.
Subsequent burial, heating, and exchange leads to dissolution of smectite and formation
of illite and appearance of diagenetic quartz, (Hower et al. 1976; Land et al. 1997), that
is a predominantly isochemical reaction. The precipitating quartz fills the porespace,
and precipitating K-mica from dissolution of detrital K-feldspar (Land et al. 1997) further
reduces porosity. We plotted d18O analyses of < 2 mm quartz presented in Land et al. (1997)
along the Texas drillcore (Fig. 8) and observe that quartz is in (d18OQz − d18Oporewaters)
equilibrium starting at 122 °C, and fine < 0.5 mm quartz from Yeh and Savin (1977) is already
in equilibrium at ~80 °C. Data for porewaters at respective depths are approximated from data
in Figure 7, and are chiefly derived from Land (1992) and Wilkinson et al. (1992).
Subsequent diagenesis in a rock-dominated system proceeds with slightly decreasing
d18O values for shale, and an increase in porewater d18O (Fig. 7), as is observed in the highest
temperatures at 5 km in the Texas drillcores. Upon temperature decrease, due to for example
a hypothetical closed system retrogression of Frio formation as a result of the uplift, shale
would regain its original d18O value due to retrograde reaction with porewater. Such retrograde
reactions would likely close to isotopic exchange at temperatures not that different than the
illite-smectite transition, and yield zero overall isotopic effects. The reasoning above depends
of course on the quite important diagenetic smectite–illite transition. However, many shales
worldwide do not reach 5 km depths nor the temperature of 200 °C as in deepest holes of the
Frio Fm, and thus hover around illite–smectite boundary, and so they maintain the original
d18O value characteristic of deposition. Longstaffe (1983) reviewed several such examples.
I thus conclude that a typical shale retains isotopic values of the original sediment.
Diagenesis shifts its d18O value to within ~1‰, as is observed in our dataset (Figs. 7–9). We also
believe that d18O (and ∆′17O) values in shales are preserved in geologic record. Consolidated
shale (Fig. 2), due to its very small crystal size, has negligible filtration porosity and permeability
to secondary fluids with different in d18O (Land and Lynch 1997). If secondary interaction
with such water occurs, it will result in petrographic evidence such as crystal size coarsening,
recrystallization, and chemical alteration that are possible to recognize petrographically,
in XRD and thin sections (Fig. 2) and to be avoided during sampling and analysis.
We draw attention here that this Chapter reviews only shales and not sandstones, which
commonly undergo a complex recrystallization history by low-d18O artesian waters generating
overgrowths with secondary d18O values recognizable by SIMS (Longstaffe 1983; Pollington
et al. 2011). Likewise, lacustrine, periglacial, and syn-snowball earth low-d18O, high-∆′17O
waters and depositional environments, will undoubtedly have an alternate scenario to marine
260 Bindeman
deposition, and leading to generation of the low-d18O shales. This is observed for shales
and tillites formed at the 2.2–2.4 Ga Huronian and 0.5–0.6 Ga Cryogenian age, when shales
presumably formed from ultra-low-synglacial waters (Fig. 9, Bindeman et al. 2016), and is
also observed for the Pliocene Antarctic claystone analyzed by us in Figure 5a, and perhaps
lower overall d18O values of tillites as compared to shales (Fig. 3). Land and Lynch (1996),
who compared lacustrine and ocean-deposited shales for non-glacial time intervals argued
that these are lower by only 1–2‰ d18O, again emphasizing expulsion of porewater as the
dominant mechanism of diagenesis. Further work needs to be done on the particular synglacial
formations and their diagenetic history.
Effects of metamorphism and anatexis on d18O and Δ′17O
It is important to discuss what happens to triple oxygen isotopes during further
transformation of shales into schists, gneiss, and later anatectic granites (e.g., Fig. 4).
The Texas drillhole with its illite-dominated mineralogy and ~190 °C bottom temperature
can be viewed as lowermost greenschist metamorphic facies (Fig. 7). Illite-dominated Frio
Formation shale has 4.1 ± 0.8 wt% H2O stored in clays and 7.7 ± 1.4 wt% carbonate cement
(3.2 wt% CO2) in pore space (Land et al. 1997). Carbonate cement has the average d18O value
of +24.7 ± 0.7‰ SMOW (Land et al. 1997), largely in equilibrium with +6‰ pore waters
and bulk silicate shale at these depths at the drillhole temperatures (Land et al. 1992, 1997).
To estimate overall d18O fractionation shifts related to hypothetical transformation of this
shale into a crystalline schist, and further into amphibole gneiss, we choose the illite starting
composition of Frio Formation. Using temperature-dependent fractionation factors for illite–
H2O and quartz–CO2 (Zheng et al. 1993) enable computation of overall d18O effects related to
loss of H2O and CO2 in either Rayleigh or a batch process.
The D18O(shale–CO2) fractionation stays negative over a 200–750 °C temperature range,
and a complete loss of 3 wt% CO2 at 450 °C will result in −0.2‰ depletion of the remaining
rock. The D18O(shale–water) is initially positive (+0.2‰ at 200 °C, thus initially compensating
for the negative isotopic effect of CO2 loss, but crosses over to negligibly negative −0.05‰
values at temperatures higher than ~425 °C. The overall effect upon d18O of losing both 3 wt%
of clay-held water, and carbonate-hosted CO2 is thus close to zero to perhaps −0.2‰ for a
shale containing 3.2 wt% initial CO2 (7–8% initial carbonate). The calculations and conclusion
are in agreement with earlier work of Valley (1986) on decarbonation reactions of limestones
and calcareous sediments and their limited extent on changing bulk rock d18O values. Details
of devolatilization, whether continuous or punctuated, have little effect.
Thus, at post-diagenetic temperatures higher than 200 °C, there are minimal isotopic
effects on ∆′17O as in Figures 5–6. This observation when coupled with a small d18O effects
of devolatilization and metamorphism of shale into schist, suggests preservation of diagenetic
d18O and ∆′17O values essentially unchanged well into the metamorphic and igneous realms.
It also adds to the credence of using metamorphic schists as recorder of environmental surface
conditions, provided that these rocks have not been modified by secondary fluids with different
oxygen isotopic values. Moreover, if zircons crystallized within these schists and gneisses, these
zircons should reflect and preserve triple oxygen isotopic values of the host rock. Thus, if zircons
formed in this way and are found as detrital grains in sediments, especially where the original
host rock is now absent, especially for the oldest zircons (Valley et al. 2002; Hawkesworth et
al. 2015; Spencer et al. 2019), these indeed provide a distant glimpse into surface weathering
conditions that existed before crystallization of zircons as metamorphic mineral grains. I propose
to couple the triple oxygen investigation of zircons with their chemical and isotopic features that
proves their supracrustal origin.
PART II: SHALES ACROSS THE GEOLOGIC HISTORY

Overview of temporal trend of d18O and ∆′17O and Archean vs. post Archean shales
The earliest observation that early Precambrian shales have lower d18O than modern shales
belongs to Silverman (1951). Longstaffe and Schwatrz (1977) and Land and Lynch (1997)
presented early data of temporal evolution of d18O in clastic sedimentary rocks and shales.
Subsequent work incrementally added to the temporal records. Figure 10 shows a temporal
increase in shale d18O values with variation modulated by the supercontient cycle. Valley et
al. (2005), Dhuime et al. (2015) and others observed temporal increases in zircon d18O values
(Fig. 10). However, zircons record a history of high-temperature igneous and metamorphic
crystallization and recrystallization reactions in maturing metasedimentary crust, but diluted
by a normal-d18O, mantle-derived component, with constant d18O value across the geologic
history. The reason for the zircon d18O increase, like the record of granites that we present
below, is rooted in evolution of shales.
The reason for a d18O temporal increase is unlikely to be related to a dramatic change in the
crustal compostion from mafic to silicic. Taylor and McLennan (1993), Rudnick and Gao (2003),
Gashnig et al. (2016) provided a detailed discussion of chemical composition of shales and
tillites through time, suggesting that exposed areas of the crust undergoing weathering retained
approximately similar proportions of crustal and mantle derived compositions, on average
dioritic through time; an exception noted is of greater proportions of komatiites in the Archean
(yielding high Ni, Cr, V concentrations in shales, Li et al. 2016). The d49Ti measurements
for the same shale suite studied by us for oxygen isotopes in Figure 10, demonstrated that
proportions of silicic to mafic rocks undergoing weathering has also remained approximately
constant since at least 3.5 Ga (Greber et al. 2017). Furthermore, Ptáček et al. (2020) used
an inversion of elemental composition in fine terrigenous sediments and demonstrated a
that a continuous > 50% felsic contribution is required starting from 3.5 billion years ago.
This finding is consistent with an early onset of plate tectonics. We demonstrate above that the
composition of the crust, whether slightly more mafic or more granitic, plays only a small role
in the current d18O (and ∆′17O) values of weathering products; these are determined primarily
by temperature and d18O of the weathering waters.
Bindeman et al. (2018) presented triple oxygen data for shales across geologic history
(Fig. 10b), and I add here new data covering the Phanerozoic. There is a stepwise shift to more
negative and more diverse ∆′17O values during and after the Archean–Proterozoic transition.
The extended dataset presented here (Table 2) supports this observation, and the granites
discussed below also show this step-wise change at ~2.1–2.4 Ga. Shales older than 2.5 Ga, yield
an average ∆′17O = 0.047 ± 0.012‰ (±2σ), largely overlapping with the mantle values, while
shales younger than 2.2 Ga, yield an average ∆′17O = –0.118 ± 0.024‰. This difference in ∆′17O
between the two age groups cannot be explained by different temperatures of equilibration or
by mixing in different proportions of variably weathered detrital materials. The shale record
spans the d′18O–∆′17O trends defined by these processes (Fig. 11), and thus requiring different
initial d′18O meteoric waters. Taking clues from the modern world, where meteoric water shows
variable d18O–∆′17O compositions, the simplest explanation for some of the oxygen isotopic
variations is that they were partly inherited from the waters involved in rock alteration on the
continents. Starting roughly at the Archean-Proterozoic transition, this emerged crust interacted
with waters that had more variable and on average more negative d′18OW and more positive
∆′17Ow values than before the GOE, with ∆′17Ow shifted by approximately +0.1‰.
The d′18OW and ∆′17OW of precipitation in the modern world depends on the cumulative
history of water loss from an air parcel traveling inland away from the coasts, to higher
latitude, and higher altitude, resulting in lower d′18OW, higher ∆′17OW, and more diverse
compositions overall, as in Figures 1 and 11. This combined effect that I call “continentality”
262 Bindeman
Figure 10. The d18O (A) and Δ′17O (B, C) of average shales through time (data from Bindeman et al. 2016;
2018 with added data points for the last 1.2Ga, smaller circles, Table 2). A) average values and bounds,
the highest d18O values (upper 10%) for each time interval are exceeding the 90th percentile, closely ap-
proaching the weathering product and thus the full extent of chemical weathering. The record is correlated
to the supercontinent cycle and glaciations; the GOE is Great Oxidation Event which occurred broadly at
2.32 Ga, after the first Paleoproterozoic glaciation (Zakharov et al. 2017). The igneous zircon d18O record
is shown (yellow line, Valley at al. 2005) demonstrating the coeval maturation of the crust resulting from
increases in heavy in d18O sedimentary component. B) Notice step-wise Δ′17O change at ~2.4–2.2 Ga,
explained by the appearance of diverse, and similar to modern hydrologic and weathering cycle upon
emergence of subaerial crust after the Archean, see Fig 11 and 5 for explanation of effects of diverse in d18O
waters and temperatures. C) Same data as in B but plotted vs. a reference slope of 0.528, and demonstrating
similar relationships. [Fig 10B is used by permission and with modification from Nature, from Bindeman
et al. (2018), Nature, Vol. 557, p. 545–548, Fig. 2a].
Table 2. New triple oxygen isotope analyses of shales, computed.
Age,
Spec number d18O d′18O d′17O d17O D′17O ±D′17O n D′17O d18OWP d17OWP d′18OWP d′17OWP T °C d18OMW D′17OMW
Ma
16 F118066* 9.833 9.785 5.117 5.13 −0.074 0.01 7 −0.0495 8.166 4.284 8.133 4.275 74 −10.253 0.059
165 F114833 7.8 7.77 4.029 4.037 −0.093 0.006 5 −0.0736 4.1 2.098 4.092 2.096 27 −23.141 0.091
169 F108499 14.199 14.099 7.364 7.391 −0.116 0.004 6 −0.0808 16.898 8.806 16.757 8.767 19 −12.59 0.063
218 F106537 12.81 12.729 6.64 6.662 −0.113 0.01 5 −0.0812 14.12 7.348 14.021 7.321 20 −15.009 0.07
253 F112216 12.797 12.716 6.635 6.657 −0.111 0.006 5 −0.0792 14.094 7.338 13.996 7.311 19 −15.292 0.071
no
270 F108479 14.945 14.834 7.713 7.742 −0.157 0.009 5 −0.1199 18.39 9.509 18.223 9.464
solution
318 F110201A 22.512 22.262 11.669 11.737 −0.141 0.012 6 −0.0858 33.524 17.498 32.974 17.347 48 3.464 0.024
430 F110412 15.719 15.597 8.154 8.187 −0.12 0.007 5 −0.081 19.938 10.399 19.742 10.345 25 −10.45 0.059
453 F106636B 13.807 13.713 7.169 7.194 −0.106 0.009 5 −0.0717 16.114 8.413 15.986 8.378 37 −11.515 0.062
472 F112980 16.39 16.257 8.505 8.541 −0.12 0.003 6 −0.0794 21.281 11.106 21.057 11.045 36 −11.6 0.062
495 F109879 13.248 13.161 6.87 6.894 −0.112 0.007 5 −0.0791 14.996 7.811 14.885 7.781 26 −15.202 0.071
506 F108173 14.569 14.464 7.563 7.592 −0.11 0.012 5 −0.0738 17.638 9.208 17.484 9.166 34 −10.58 0.059
no
545 F123797A 14.48 14.376 7.553 7.581 −0.074 0.007 5 −0.0381 17.46 9.187 17.309 9.145
solution
1325 F115996A 16.318 16.186 8.485 8.522 −0.101 0.015 5 −0.0608 21.136 11.067 20.915 11.007 60 −1.834 0.037
Note: *Sample numbers, WR chemistry and description are from Retallack et al. (2019); WP – computed weathering product, T °C, MW computed values of
temperature and meteoric water values solved based on triple oxygen isotopic values. weathering product, temperature and meteoric water values.
Evolving Continental Crust, Granites, and Clastic Sediments
263
264 Bindeman
A B
)LJFRORU
Figure 11. a) Triple-oxygen isotopes in shales of different age; (Bindeman et al. 2018 with added data for
the last 1.2 Ga, smaller circles). See Figure 6 for explanations and Figure 5 for modern world values. Notice
that Archean shales tend to be lower in d18O, and higher in Δ′17O, with scatter around higher temperatures.
b) Hypsometry maps for the Archean (top) and post-Archean (post) world hypsometry, showing their
pronounced difference. This explains that the limited hydrologic cycle in the Archean is due to the small
continental size. [Fig 11 is used by permission and with modification from Nature, from Bindeman et al.
(2018), Nature, Vol. 557, p. 545–548, Figs. 1a and 3].
is invoked to explain the shale record, showing a stepwise change that coincides with the
Archean–Proterozoic boundary (Fig. 10). Most likely, the observed changes in the shale
triple-oxygen isotope record reflect the appearance of larger continents (Fig. 10) and higher
elevations starting in the Proterozoic, which is broadly contemporaneous with final stages
in the assembly of the first documented supercontinent, Kenorland (Mertanen and Pesonen
2012; Condie 2013) or alternatively, several supercratons occurring immediately before the
GOE (Bleeker 2003). Supercontinent assembly and orogenic events result in high-mountain
ranges and plateaus, e.g., the Himalayas and Tibet. Mountains even in relatively low to mid
latitudes yield precipitation with very light d′18O values correlated with elevation, about 2–3‰
drop per 1 km increase in altitude (Rowley et al. 2001). To illustrate “continentality”, we here
compiled d18OW values for the modern world, from the Waterisotopes.org website (Fig. 12);
future analyses should permit similar efforts with ∆′17OW, and these would also account for
the evaporation effects. Large continents such as Asia extended latitudes and with mountain
ranges show significantly larger variations in d18OW, as compared to smaller land masses such
as Madagascar or Malaysia for example. These effects are indeed expected to be much more
dramatic in a supercontinent-type environment (Fig. 13).
Supporting our interpretations of triple oxygen variations in shales is the Sr isotope record
of marine carbonates, that suggests the area of emerged continental crust increased significantly
and irreversibly at around the Archean–Proterozoic boundary (Flament et al. 2008). From a
geodynamic perspective, models of a cooling Earth call for a thickening of the lithosphere
and the establishment of a higher continental freeboard by ca. 2.5 Ga due to increased mantle
viscosity (Vlaar 2000; Condie 2013; Korenaga et al. 2017). Emergence of large landmasses
(Fig. 10b) would also have led to a larger weathering sink for CO2 that occurred in greater
concentrations in the Archean, resulting in a transition to moderate surface temperatures
after the GOE. This set of large-scale tectonic and near-surface changes may best explain the
observed shift in the d′18O–∆′17O composition of shales between 2.5 to 2.2 Ga (Fig. 10).
)LJ*UH\VFDOH

1±5° 5±4° 12±6° 19±6° 20±30° 24±16°
0±5°
0.54M 0.79M 7.5±1.5° 0.3M 0.59M 17.8 M 7.7 M
0.47M
2.3km 4.9km 5±5° 0.14M 2.9km 2.9km 6.9km 2.2km 37±6°
3.8km
0.18M 3.7km 0.23M
PNG 3.8km 72±11°
Kalimantan 2.1km
Australia
56±5° 2.2M
Sumatra Malaysia 3.7km
Madagascar
0.27M
Africa
4.8km
S America
Java
Philippines
90±20°
Asia
Kamchatka
14M
Honshu
4.9km
Greenland
0±35°
G20$3
30.1 M
5.9km

Antarctica

40±35°
44 M
8.8km

-65‰
-
-
Figure 12. A compilation and summary of d18O values of mean annual precipitation (MAP) for the major
modern terrestrial land mases of different size and hypsometry, scaled proportionally within each land mass
area. Shown for each land mass is its latitude ranging from equatorial to polar, its surface area in square
kilometers, and its highest elevation in km. Data for plotting is retrieved from waterisotopes.org.
3
4
2 1 weathering products
4
3 2 1
∆'17O, ‰
3 Ice
δ18O
Figure 13. An example of paleoenvironments on a Pangea-type supercontinent in which weathering prod-
ucts (shales) or chemogenic sediments (e.g., carbonates and opals) form. 1) tropical island, hot and high-
Fig. 13 color
d18O waters; 2) wet interior of the continent; 3) glaciated region;, 4) dry interior of the continent, warm
and cold currents and mountain ranges are shown. Inset is showing triple oxygen isotope the effect of tem-
perature on the isotope fractionations (concave down curves) and effects of meteoric waters. Inset shows
weathering environments shown. Weathering product is shown by ovals on fractionation curves (See Fig 6
for further explanations). Temperature of interaction (formation) is indicated by dot (blue—cold 1–10 °C
MAT, red—hot, 25–35 °C MAT). eMWL is evaporated MWL, with a zero offset Y in Δ′17O as in Figure 9.
Dashed circle indicate the hypothetical pre-early Phanerozoic low-d18O seawater, dashed ovals denote
their corresponding continental weathering products, assuming the slope of meteoric water line stays the
same at 0.528. The base image credit is from the AtlasPro project on YouTube (https://www.youtube.com/
watch?v=VKq0pr4rbRs&t=37s).
266 Bindeman
The observed shift in the triple-oxygen isotope composition of shales further requires
lower post-GOE surface temperatures (Fig. 10) if modern meteoric water is used. Higher
apparent temperature in the Archean oceans may also be related to a greater contribution
of hydrothermal clays, or to Archean shales that would result in less positive d′18O and less
negative ∆′17O values. This is broadly consistent with previous studies of d′18O in cherts
using the same duality of temperature vs. d′18O of the ocean. The herein inferred rapid
increase in Earth′s subaerial surface and overall hypsometry after ~2.5 Ga (Fig. 9) would
have also increased Earth′s albedo, the flux of nutrients to the oceans from the continents
undergoing subaerial weathering, and the extent of continental margins, resulting in a higher
burial rate of organic carbon and diminishing CO2 concentration in the air. In combination,
these changes could have contributed to global cooling, the Snowball Earth glaciations of
the early Paleoproterozoic, and the following GOE, highlighting how Earth′s interior could
have influenced surface redox conditions and chemistry. The most dramatic change in Earth′s
history was marked by a transition from generally hot and largely anoxic surface conditions to
an oxygenated atmosphere with moderate surface temperatures.
The Phanerozoic shale record
The Phanerozoic is a period of Earth′s history when plate configuration, orogenies, and
exposed continental areas are known, offering further information on the sequence of events.
The Phanerozoic also includes the last complete supercontinent cycle involving destruction of
Rodinia at 750–650 Ma, accretion of Pangea by ~330 Ma (Fig. 11), followed by its destruction
starting in Late Triassic, ~220–190 Ma and advent on plant life on continents. Hypothetical
climate and weathering conditions on Pangea are presented in Figure 11 which shows the
anticipated triple oxygen effects. Early Permian glaciation and a colder climate at ~280 Ma was
likely accompanied by a global decrease in temperatures, and thus the d18O value of precipitation
(cf Eqn. 7). Therefore, weathering products should have lower d18O and higher ∆′17O values,
which should characterize mid-Pangean history. Triple oxygen isotopes in shales can assist in
resolving the details of Pangean climate (Fig. 13) for the four environments shown. Tropical
islands and tropical near coastal type environment 1 will exhibit higher d18O and ∆′17O values.
Intracontinental type environment 2 will display lower d18O but comparable ∆′17O due to lower T
of weathering. Polar environment 3 will exhibit the lowest d18O and ∆′17O. Finally, environments
with dry, deeply intra-supercontinental conditions will yield waters so highly evaporated that
they belong to a different “strongly evaporated water line” (aka modern arid Africa; Passey and
Ji 2019), showing the lowest ∆′17O and high-d18O values. As arid intracontinental conditions are
perhaps well represented in the modern world (Fig. 1) by dry African environments, there is a
great potential to discover even lower ∆′17O values in African Permian–Triassic shales, lacustrine
diatoms, and biogenic carbonates. Targeted investigation of geologic formations with known
depositional environment and climate should test this hypothesis.
Figure 14 shows that the d18O and to a lesser extent the ∆′17O values for the shales exhibit
U- or W-shaped trends throughout the Phanerozoic (previously published and new data from
Tables 1 and 2). Sr isotopic values in coeval carbonates, are also plotted for comparison, and
it shows a similar U- shaped trend with a minimum near the Paleozoic–Mesozoic boundary.
Since Sr and O isotopes in supracrustal rocks commonly correlate, we will explain these
observed trends together. Figure 14 also shows estimated surface area of exposed continental
crust in latitudes between 30° N and 30° S based on Lawver et al. (2011). The rationale for
this estimation is that the majority of silicate chemical weathering occurs in the tropics, e.g.,
2/3 of modern global Earth′s sediment flux is from SE Asia (Milliman and Farnsworth 2011).
We observe that the area of the Earth′s surface in the tropics also exhibits W-shaped (or smile
face) pattern, coeval with O and Sr isotopic trends. We present the following hypotheses
that are in accordance with the Phanerozoic supercontinent cycle of Pangean assembly and
disassembly, with associated orogenies and rifting events:
10
land, 106 km2

Tropical area
9
8
7
)LJJUH\VFDOH
E Permian Glaciation
6
5
Pangea Rifting
carbonate
0.709
4
Siberian Trapps
87Sr/86Sr
3
0.708
2
0.707
1
0

G2Å602:

'¶2Å

-0.02

'’17O0.528 ‰
-0.06
-0.10
-0.14
0 0.1 0.2 0.3 0.4 0.5 0.6
Age, Ma
Figure 14. The correlation of the subaerial land masses between 30° N and S (derived from Atlas of Plate
Reconstruction Project, Lawver et al. 2011) with d18O and Δ′17O values of shales, and Sr isotopic values of
coeval carbonates. Data from Bindeman et al. (2016, 2018) with added data points for the last 1.2 Ga, Table 2.
(i) Orogeny during supercontinent assembly in the Devonian is accompanied by uplift,

exposure and rapid erosion of fresh, unweathered igneous and metamorphic rocks with
lower d18O. The detrital contribution of these rocks to shales (since average shale is a mixture
of weathering products and detrital components, the latter can also be previously weathered
sedimentary rocks) drives the bulk of the shale d18O values down.
(ii) Since orogenesis and erosion occurred in central Pangea and at higher altitudes during the
Caledonian collisions, waters involved in weathering are depleted in d18O and enriched in ∆′17O;
their interaction with rocks imprints these signatures on shales (cf. Fig. 5).
Secondary effects are related to the glacial episode in the Early Permian at ~280 Ma,
causing further d18O depletion of meteoric water worldwide. Glaciation additionally produces
fine-grained tillite, which upon interaction with synglacial and intracontinental low-d18O
waters transfer these signatures to shales. It is thus instructive to investigate Early Permian
formations for these effects. As an example, synglacially altered rocks formed during Snowball
Earth episodes at 2.4–2.2 Ga (Herwartz et al. 2015; Zakharov et al. 2017, 2019), generated
the lowest d18O rocks yet measured, formed in exchange with −45‰ d18O meteoric waters.
The Pliocene Antarctic clay also records < 28‰ d18O waters (Fig. 5).
268 Bindeman
The episode of supercontinent splitting and intracontinental rifting that has dominated the
world since ~200 Ma results in opposite isotope effects:
i) Lack of major orogenies, except the most prominent India–Asia collision after
~50 Ma, leads to peneplain development and formation of eroded mountain belts
such as the Appalachian and Ural Mountains. This in turn leads to formation of
more sedimentary rock, and thus higher d18O values in shales. The ongoing chemical
weathering and attendant repeated and more thorough isotope exchange of exposed
sedimentary rocks in flood plains should further increase d18O values of newly-
formed shales.
ii) Since the breakup of Pangea leads to continents that are smaller in size, especially in
tropical latitudes, the d18O value of meteoric water involved in tropical weathering is
globally higher, and D′17O is lower (cf. environment 1 in Fig. 13). For example, the
d18O in sediment-flux-weighted river water in the modern world is only −7.5‰, and
these signatures are thus passed to shales.
Likewise, the Sr isotopic record (Fig. 14) in coeval carbonates also indicates competition
between the erosional supply of high-87Sr/86Srinitial from eroding continents (including
carbonates), and the supply of mantle-derived low-87Sr/86Srinitial Sr at mid-ocean ridges during
rifting. As erosional flux reflects mountain denudation rates especially when they occur in the
tropics, the Sr isotopic curve reflects that. To the first order, the Neoproterozoic shales have the
highest d18O and Sr isotopic values, reflecting simply the weathering of a large proportion of
continental crust and mountains after accretion of a Rodinia supercontinent at tropical latitudes at
that time. The lowest 87Sr/86Sr and d18O values are coeval with the lowest amount of continental
surface area exposed in tropical areas, combined with initiation of Pangea rifting at 210 Ma, and
preceded by large quantities of erosion at the Siberian Traps, which erupted at 250 Ma.
PART III: THE TRIPLE OXYGEN ISOTOPE ANALYSIS OF GRANITES

IN COMPARISON WITH SHALES RECORD: INSIGHT INTO THE
EVOLUTION OF THE CONTINENTAL CRUST AND WEATHERING
This work presents triple oxygen isotope analyses of granites across geologic history
(100 new triple oxygen analyses and > 200 compiled d18O, Fig. 15, Table 3) with emphasis on
orogenic and S-type granites available for post-Archean history, and a broader suite of granites in
the Archean and Early Proterozoic, including orogenic, TTGs, and synmetamorphic (anatectic,
migmatitic) types for which geodynamic classification is more difficult to unambiguously
establish. For example, our oldest granites and TTGs from the Kaapvaal craton are interpreted
to be partial melts of silicified metabasalts and metasediments based on silicon isotopes and
bulk chemical compositions (André et al., 2019). In our sample selection and compilation
of literature data for d18O in granites we exercised criteria of “maximum crustal origin” and
minimized the use of I-type granites and A-type, intracontinental rift-related granites (e.g.,
Anderson and Morrison 2005) that commonly have a significant mantle-derived component.
For example, voluminous orogenic granites that form batholiths typically represent large scale
remelting of metamorphic supracrustal protoliths that span large areas/volumes of crust (e.g.,
Sylvester 1998; Schaltegger et al. 2019), and these large magma bodies provides a broad view
of crustal evolution on giga-year timescales. Orogenic granites studied were commonly referred
to as S-type, i.e., sedimentary-derived, many with peraluminous characteristics (e.g., Chappell
and White 1974). Bucholz and Spencer (2019 and references therein) recommended replacing
the term S-type with the expression “strongly peraluminous” granites. We further compiled
published d18O values in well-studied orogenic granites (see Reference list) and utilized the
recent dataset of Bucholz and Spencer (2019) for strongly peraluminous Precambrian granites
with excess of Al2O3 and SiO2 and are thus mostly of anatectic origin.
Table 3. Oxygen Isotope analyses of quartz and other minerals in granites and metaigneous rocks of different ages.
d18O
Age D′17O Qz 0.5305, d18O Country/
Sample Type ±D′17O D′17O0.528 d17O Other D′17O0.5305 Collector Reference/locality
Ga ‰ VSMOW Quartz US State
Minerals
ID3−85 ms 3.8 −0.059 0.003 −0.036 9.152 4.786 Greenland A. Kays Isua, Kays et al. (1972)
IC3−85 ms 3.8 −0.079 0.010 −0.056 9.412 4.902 " " Ashvall (1986)
2.8ma−2, vein ms 3.8 −0.067 0.011 −0.057 4.334 " " "
2.8Ma−2 ms 3.8 −0.047 0.012 −0.036 4.214* 1.464 −0.041 ± 0.011, Grt " " "
2.8 M−1−85 ms 3.8 −0.070 0.010 −0.055 6.017 3.267 −0.064 ± 0.011, Grt " " "
IC3−85 qz−1 ms 3.8 −0.063 0.012 −0.040 9.436 4.931 " " "
IF−3 quartzite ms 3.8 −0.077 0.014 −0.046 12.644 6.610 " " "
IF−1−85 ms 3.8 −0.113 0.010 −0.082 12.308 6.400 " " "
2.8MI−85 ms 3.8 −0.059 0.011 −0.050 3.261 " " "
Acasta gneiss i 4 −0.061 0.010 −0.045 6.559 6.81, 6.77, Qz Canada UO collection N Canada
Chinese Granite i 3.8 −0.086 0.013 −0.071 6.131 3.167 N China " Bindeman et al. (2018)
Morton
mi 3.5 −0.103 0.008 −0.079 9.826 5.097 MN T. Vislova Town of Morton MN
gneiss−1
Morton
mi 3.5 8.220 " " "
gneiss−2
granitic gneiss,
MN23−16 mi 3.5 −0.085 0.013 −0.062 9.258 4.815 " D. Blackwell
Montevideo, MN
grey gneiss, Republic
LS−13 p 3.3 −0.094 0.008 −0.072 8.748 4.537 MI UO collection
Mine Marquette MI (1)
LS−13 p 3.31 8.250 " UO collection "
Grey granite, S.E.
UO−605 p 3.2 7.950 Scotland UO collection
Bushveld (2)
Bar−2 p 3.46 8.270 S Africa A. Hoffman Andre et al. (2019)
Bar−5 p 3.26 9.100 " " "
Bar−9 p 3.24 −0.080 0.005 −0.058 8.857 4.608 5.092 −0.054 ± 0.006, Zrc " " "
Bar−8 p 3.23 −0.074 0.006 −0.051 9.125 4.756 " " "
Bar−10 p 3.2 −0.089 0.015 −0.070 7.854 5.354 −0.083 ± 0.015, Zrc " " "
d18O
Minerals
Bar−15 p 3.12 −0.055 0.016 −0.030 10.038 5.257 " " "
Bar−14 p 3.12 −0.065 0.021 −0.041 9.781 5.112 " " "
Bar−1 p 3.11 −0.068 0.012 −0.048 8.336 4.357 " " "
Bar−7 p 3.11 −0.077 0.012 −0.054 9.392 4.905 " " "
Bar−11 p 3.11 8.730 " " "
Bar−13 p 3.11 −0.109 0.014 −0.086 9.133 4.725 " " "
SCD−5 mi 3 −0.099 0.016 −0.074 10.171 5.284 Scotland I. Bindeman Scouri
Bar−12 p 2.74 −0.082 0.020 −0.058 9.595 4.997 S Africa A. Hoffman Andre et al. (2019)
granitic gneiss,
MN7−14 p 2.7 −0.099 0.014 −0.076 9.196 4.768 MN D. Blackwell
Ojakanngas (2009) (3)
Bar−16 p 2.69 −0.060 0.010 −0.038 8.900 S Africa A. Hoffman
Superior province
DD96−4 zrc p 2.688 −0.073 0.006 −0.053 8.167 5.667 −0.067 ± 0.006, Zrc Canada J. Valley
Valley et al. (2005)
BengW−1 p 2.65 7.130 India UO collection West Bengal
Paradiso p 2.62 8.140 " " Dharwar craton, Karnataka
Big Horn granite,
WY12−16 p 2.6 −0.047 0.007 −0.026 8.199 4.294 WY D. Blackwell
Shell Falls
Pudukottai Town,
LaDr−1 p 2.6 7.760 India UO collection
Tamil Nadu
CF8562 zrc p 2.594 −0.067 0.009 −0.043 9.799 7.299 −0.061 ± 0.009, Zrc China John Valley Valley et al. (2005)
granite, Milbank,
SD−20−18 mi 2.588 8.790 SD Dave Blackwell
Gries (1996)
Closepet Granite,
VisW−1 p 2.55 9.700 India UO collection
Dharwar Craton
granite Ramakona,
R5855 p 2.156 −0.067 0.010 −0.047 7.910 " Greg Retallack
Madhya Pradesh (3)
migmatite of 1.9 Ga
JupB−1 mi 2 −0.134 0.010 −0.107 10.691 9.640, Grt Brazil UO collection orogeny of mid-Archean
gneiss
VerB−1 mi 2 10.690 " "
Gr B−1 mi 2 9.380 " "
d18O
Minerals
FN35−72 p 1.85 −0.085 0.01 −0.055 12.172 6.405 SW Finland Allan Kays Kays et al. (1972)
FN35−72 p 1.85 12.090 " " "
DJ−1 p 1.82 −0.111 0.01 −0.086 10.169 5.270 N Baffin Land Dana Johnston Cumberland batholith
Volodarsk-Volynsky
Volga Blue p 1.78 6.760, Grt Ukraine UO Collection
7.360 Intrusion
NE Lake Hara area
K263−70 p 1.75 −0.114 0.011 −0.088 10.272 5.318 A. Kays
Saskatchewan Kays et al. (1972)
K332−70 p 1.75 9.200 " " "
K247−70 p 1.75 −0.076 0.008 −0.049 11.021 5.761 " " "
K247−70 p 1.75 9.850 " " Wollaston, Hearne craton
Lake Hara area
K388−70 mi 1.75 −0.105 0.004 −0.078 10.675 5.544 " "
Kays et al. (1972)
Vernal Mesa Qtz
CoNM 6−08 p 1.741 12.540 CO D. Blackwell Monzonite,
Scott et al. (2001)
Harney Peak Granite,
SD14−15 p 1.697 −0.134 0.011 −0.096 14.970 7.779 SD "
Lufken et al. (2009)
SD1−10 p 1.697 −0.127 0.012 −0.090 15.169 7.890 " " "
Vernal Mesa Qtz
BCNGP−2-Q8 p 1.48 −0.070 0.013 −0.044 10.583 5.530 " "
Monzonite, pegmatite
BCNGP−1−08 p 1.48 −0.080 0.013 −0.051 11.628 6.072 " " "
Grenville Granite,
MR−85 p 1.15 −0.122 0.013 −0.096 10.405 5.383 NY W. Peck
Peck et al. (2004)
PR−21 p 1.15 −0.135 0.022 −0.098 14.770 7.673 " " "
BP−16 p 1.15 −0.110 0.01 −0.076 13.860 7.218 " " "
quartzite.
IPQ−1 ms 1.1 −0.105 0.008 −0.074 12.234 6.366 " J. Valley
Grenville Province
gneissic granite.
R5856 p 0.967 −0.124 −0.092 12.790 India G. Retallack
Nadol, Rajasthan (4)
d18O
Minerals
Brazil Espirito Ribeira Belt, Macluf,
Uba−1 p 0.6 12.000 9.260, Grt UO collection
Santo Buchwaldt (2001)
Pedrosa-Soares et al.
SCC−1 p 0.65 −0.106 0.01 −0.074 12.990 10.080, Grt " " (1999), Mendes et al.
(2000)
VG−1 p 0.67 13.230 " "
KWG−1 p 0.68 12.240 " "
CrBrd−1 p 0.69 9.160 6.410, Grt " "
Zanz−1 p 0.56 13.530 " "
BiscW−1 p 0.56 10.330 " "
Brazil, Aguia
DalW−1 p 0.55 12.410 10.000, Grt "
Branca
Buryatia, Angaro-Vitim batholith,
B−420 p 0.44 −0.129 0.007 −0.095 13.841 7.189 B. Litvinovsky
Russia Wickham et al. (1996)
Australian S type Granites,
BB83 p 0.429 −0.064 0.005 −0.036 11.083 5.800 SW Australia UO collection
Berrydale batholith
Cooma 4(5) p 0.414 −0.078 0.010 −0.047 12.432 6.498 " " Chappell and White (1974)
KB32 p 0.41 −0.081 0.005 −0.047 13.580 7.100 " " "
AB40 p 0.389 9.060 4.778 " " "
Norway, Jayananda et al. (2000),
Blue Pearl p 0.35 7.330 6.730, Plag UO collection
larrvakite Moyen et al. (2001)
Central
MD−1 p 0.35 −0.068 0.005 −0.038 12.060 6.309 10.32, Plag M. Dungan Dent de Cheval, Caudjol
France
Sierra Nevada,
OISG04 zrc p 0.101 −0.084 0.014 −0.057 11.002 8.502 −0.078 ± 0.014, Zrc CA J.S. Lackey
Valley et al. (2005)
Ashland Granite p 0.05 11.550 11.3 OR UO collection Ashland, OR
Adamello Adamello batholith,
p 0.04 −0.133 0.016 −0.106 10.699 4.081 7.949 −0.127, Amph N Italy U. Schaltegger
PB782 Schaltegger et al. (2019)
VAL zrc powder p 0.04 −0.065 0.020 −0.045 8.159 5.659 −0.059 ± 0.02, Zrc " " "
d18O
Minerals
Ignimbrites
BFCQ Fish
v 0.028 −0.108 0.006 −0.089 7.470 2.270 4.477 −0.102 ± 0.006, Sphene CO O. Bachmann Fish Canyon tuff
Canyon
TM−10 Zrc v 0.0012 −0.055 0.008 −0.032 9.178 6.677 −0.049 ± 0.008, Zrc NV I. Bindeman Timber Mt
Caucasus,
Cheg−9 v 0.003 −0.085 0.006 −0.065 8.090 4.198 I. Bindeman Chegem Tuff
Russia
CG−141 v 0.003 −0.122 0.012 −0.098 9.669 4.995 Bolivia UO Collection Cerro Galan Tuff
HRT−1 v 0.002 −0.082 0.006 −0.062 7.717 4.005 WY I. Bindeman Huckleberry Ridge
HRT−1 Zrc v 0.002 −0.079 0.009 −0.059 8.110 5.610 −0.073 ± 0.009, Zrc " I. Bindeman tuff, Yellowstone
HRT−3a qz v 0.002 −0.062 0.007 −0.043 7.517 3.992 " I. Bindeman "
Krakatoa Px v 0 −0.062 0.003 −0.045 7.009 4.509 −0.056 ± 0.003, Cpx Indonesia A. Belousov Krakatoa 1883 AD
Bishop tuff Late v −0.098 0.013 −0.078 8.123 CA I. Bindeman Bishop tuff
Toba v 0.0007 −0.084 0.016 −0.061 9.135 4.751 Indonesia C. Chesner Toba tuff
LBT−4a v 0.0001 −0.059 0.013 −0.040 7.731 4.035 NM I. Bindeman Lower Bandelier Tuff
Note: p: plutonic igneous; ms-metasedimentary; mi-metaigneous, v-volcanic. d18O analyses without D17O data were done by CO2-method for d18O only; Ages in italics are approximate, based on geologic relations. *Offset italics represent
quartz equivalent (typically for quartz-undersaturated rocks) calculated using mineral analyzed and corrected to quartz values by adding (Qz-min) fractionation increment using high-T fractionation factors of Chiba et al. (1989) and
corresponding D17O along the slope 0.528; (1) Ajibic Fm., Whitley et al. (2004) Gneiss Dome book, 3.2–3.6 Ga; (2) 20 miles from Grobersdal on road to Bronkhorstspruit, S.E. Bushveld; (3) Vermillion Granitic Complex Ojakanngas,
2009, p 90, (3) India. From roadcut 10 km north of Ramakona on National Route 268, The granite is a 15 cm thick dike in schist; (4) India. From quarry dump 5 km east of Nadol on National Route 67, quarry
274 Bindeman
The d18O of quartz and other minerals in granites analyzed in this study is increasing
with time (Fig. 15), comparable to the record for shales (Fig. 3, 10) and zircons (Valley et
al. 2005; Dhuime et al. 2015; Hawkesworth et al. 2017, Fig. 10). A novel observation of this
work is that I observe a step-wise change of ∆′17O from −0.072 to −0.115‰ occurring at
2.1–2.4 Ga, broadly coinciding with the change in shale at Archean–Proterozoic transition
(Fig. 10b). This step-wise change of ∆′17O is somewhat less-well pronounced for granites than
for shales (Fig. 15, 10b). The d18O of granites also increased by 3–4‰ from 7.8‰ to 10.5‰
across the Arc hean/Proterozoic boundary, although in a more gradual fashion. The lower d18O,
and higher-∆′17O range of granites of any age group indicate greater contribution by mantle
derived magma differentiates or their remelts, and these values stay relatively constant across
geologic history. For example rift-related and some anorogenic magmas (some Proterozoic
rapakivi granites), are close in d18O to the mantle values. It is the upper d18O, lower-∆′17O
values that are characteristic of the crustal isotopic values. Spencer et al. (2019) used the
zircon d18O record, also emphasized a step-wise change at ~2.3Ga.
However, we observe a clear difference in ∆′17O between Archean shales and granites: shales
are significantly shifted and overlap with the mantle −0.05‰ value, while granites are noticeably
more negative than the mantle. Pure igneous differentiation of mantle-derived compositions (and a
complementary processes of partial melting of basalts) leads to <1‰ increase in d18O (Eiler 2001)
and thus should correspond to a very minute −0.01–0.015 decrease in ∆′17O along any high-T
slopes between 0.528 and 0.5305. This is insufficient to explain stronger negative ∆′17O values.
The second observation is that Mid-Late Proterozoic granites in our dataset have the
highest d18O and lowest ∆′17O values as compared to Phanerozoic granites. Similar relations
occur in the record of shales (Figs. 3, 10). Overall the shale record has wider range of d18O and
∆′17O variations than that of granites, and this is expected. Isotopically, granites range from
being mantle-like, represented by partial melts of basalts and their differentiates, to S-type,
derived from diverse sedimentary protoliths in anatectic environments.
Comparing coeval record of Archean granites and shales
We explore d18O and ∆′17O variations in granites and shales in Figure 16, which indicates
potential trajectories of assimilation of a mantle-derived magma with crustal protoliths
represented by shales, cherts, and hydrothermally altered rocks. Archean granites would require
overall a narrower range of assimilants, ranging from marine-derived, higher-∆′17O cherts, to
moderately low-∆′17O shales. Post-Archean granites clearly require a predominant high-d18O
shale source. Granites on the lower part of the figure show low d18O but strongly negative
∆′17O, including early the Archean Acasta orthogneisses studied by Rumble et al. (2013) and
us, are too low in ∆′17O to be explained by any realistic igneous assimilation of these crustal
sources. It is possible however that the Acasta orthogneisses with their oldest whole rock age
of 4.02 Ga (Reimink et al. 2016) are produced by wholesale anatectic remelting of sedimentary
or low-T hydrothermal protoliths without addition of mantle-like material, inheriting low-∆′17O
signatures. As examples, the Acasta, Jack Hills and the Singhbhum craton zircons have the
lowest εHf value of −5 among its EoArchean peers (Bauer et al. 2017, 2020). Reimink et al. (2016,
2020) advocate for shallow processes including hydrothermal alteration for the earliest 4.02 Ga
Acasta complex, consistent with an oceanic plateau setting, followed by deeper anataxis 3.6 Ga.
Alteration by early Archean low-∆′17O and d18O waters remain as a topological possibility
in Figure 16 (labeled “Low-∆′17O assimilant”). Involvement of evaporated low ∆′17O waters,
similar to those occurring in dry and lacustrine regions of Africa today (Passey and Ji 2019),
can also shift protolithic ∆′17O values downward. Addition of silica via protolith silicification
(e.g., Fig. 11) by either globally low-d18O, or evaporated waters, is another possibility. Note
that hydrothermal alteration by recent 0‰ ocean or predominant d18O = −7.5‰ meteoric
water will increase rather than decrease ∆′17O values, and thus measured values cannot be a
Fig. 15 Color, please publish by taking the
E
D
Acasta,
Acasta,
Isua
Isua
supercontinents
3.5
3.5
3
3
Literature orogenic
Peraluminous
2.5
2.5
This work
GOE
Age, Ga
2
2
glacial episodes
1.5
1.5
mantle
mantle
mantle
1
1
0.5
0.5
shales
0
0
-0.02
-0.04
-0.06
-0.08
-0.1
-0.12
-0.14
-0.16
-0.02
-0.04
-0.06
-0.08
-0.10
-0.12
20
18
16
14
12
10
δ18O, ‰ VSMOW
∆'17O0.5305 ‰ VSMOW ∆’17O 0.528, ‰
A
Figure 15. Triple oxygen isotopic values in quartz and other minerals in granites through time (Table 3)
showing increase in d18O (A) and decrease in Δ′17O (B) after ~2.2 Ga. Panel C plots the dataset in B vs.
0.528 reference line and shows similar relations as in B. The shale record is superimposed in the back-
ground showing coeval relationship with granites. In particular, both shales and granites exhibit the
highest-d18O values in 1.1–1.4 Ga than anywhere else in the geologic history. The + markers represent
compiled data for selected well-studied orogenic granitic batholiths from the literature that indicate that
Grenvillian orogeny (1.1–1.2 Ga) generated highest d18O (data from Peck et al. 2004), and coincident with
the highest d18O values in shales. Strongly peraluminous granites (data compiled by Bucholz and Spencer,
2019) largely follow the orogenic granite trend. Thin jagged black line represents running average of
d18O values of 9000 detrital zircons from Spencer et al. (2017). Notice comparable relations to the present
dataset. A step-function describes Δ′17O data with change at 2.2–2.4 Ga, results are statistically significant
using t-test, p values are 0.0003 (B) and 0.009 (C). Parabolic fits are also presented. Notice that Archean
granites have lower Δ′17O values as compared to shales. The oldest samples from Acasta and Isua have
metaigneous and metasedimentary origin. Panels D and E plot the same dataset using whiskers plot and
interquartile statistics as for coeval shales in Fig. 10b.
276 Bindeman
Archean SW?
-15‰
-28‰
0.01
0‰ δ18O, ‰
SW
-10 -5 -0.01 0 5 10 15 20
-0.03
hydrothermal Archean
shale cherts
-0.05
mantle
Isua
Isua, literature -0.07
Acasta Hadean
-0.09
Acasta, literature Low-∆’17O
Assimilant?
-0.11
Archean
Δ’17O0.5305, ‰
post-Archean -0.13
1.9-1 Ga
-0.15 Post Archean
shale
Figure 16. Triple oxygen isotopic variations in granites based on quartz (or quartz equivalents based on other
minerals, Table 3) as compared to and early metasedimentary crust and shales (colored fields). Quartz–water
fractionation curves (from Sharp et al. 2016) emanating from 0‰, −15 and −28‰ waters on bulk shale–water
fractionations emanatingFigfrom
16 −10‰
Colorwater are shown for reference. Maximum extent of igneous differentia-
tion of deep, wet amphibole- and pyroxene-bearing assemblages (e.g., Bucholz et al. 2017) is shown to reach
d18O value of ~7.2‰ with a hypothetical slope of 0.528. Like in Figure 14, Archean granites occupy smaller
range of d18O and Δ′17O values, while 1.9–1.1 Ga granites are the highest in d 18O and lowest in Δ′17O. The
3.8–4.0 Ga Isua and Acasta gneisses and metasedimentary rocks (this work and literature data mostly from
Rumble et al. 2013), are calibrated relative to San Carlos olivine of −0.051‰, equivalently to this Chapter
and as in Miller and Pack (2021, this volume) and Sharp and Wostbrock (2021, this volume). Brown arrows
denote mixing (assimilation) trajectories between mantle materials and shales and cherts on respective frac-
tionation lines to explain the spread in the datapoints, blue lines denote high-T hydrothermal trends explain-
ing the spread of the early Archean Isua data, requiring hydrothermal alteration and a mysterious low-Δ′17O
assimilant. Literature data for Isua scatters likely due to greater analytical errors in these earlier studies.
result of recent exchange with water. Additionally, the low-∆′17O signature is also measured
by us in Archean zircons and other minerals (Table 3) and thus secondary alteration of quartz
in petrographically intact granites and TTGs is unlikely. If the earliest, Hadean “low-∆′17O
assimilant” was indeed present, it likely has been reworked, similar to εHf and 142Nd record for
the earliest crust (e.g., Bauer et al. 2017, 2020; Carlson et al. 2019 and references therein). See
below our later speculation on the origin of this “low-∆′17O” component via cometary delivery.
Notice that if the entire hydrosphere including oceans were lower in d18O (e.g., Perry et
al. 1978; Knauth and Lowe 2003), and thus all fractionation lines were shifted upward and to
the left on Figure 16, this will assist in explaining the low-∆′17O data points. A hypothetical
scenario wherein the Archean hydrosphere was both low-d18O, and also relatively low-∆′17O,
e.g., shifted left and retained at its current ∆′17O, will explain these low-∆′17O granites, and
also Archean cherts (Sengupta and Pack 2018; Hayles et al. 2019; Liljestrand et al. 2020;
Zakharov et al. 2021, this volume), and shales. This brings us back to the above proposed
testable hypothesis on testing the Early Earth, primordial low-∆′17O crustal values.
Figure 17 proposes a cartoon depicting an Archean greenstone belt with shallow submarine
rifting, and partially flooded low elevation continents barely emerged from seawater, filled
by evaporating lakes with hydrothermal activity. Diagenesis also proceeds with participation
of such fluids, similar to evaporated African Lakes (Passey and Levin 2021, this volume).
Hot climate, ~+37°C
High CO2, weathering rates

-6‰
Shallow, pancakes-like, flooded continents
silicification Evaporating Lake,
Seawater -5‰ low ∆’17O
TTG
Hydrothermal alteration of rocks inherits low-δ18O,
Greenstone belts anatexis high ∆17O clays and passes them to shales
Figure 17. A cartoon cross section illustrating Archean continents with a relatively warm (ca. +37 °C)
Fig. 17 A Color
average climate, yielding high weathering and silicification rates on continents, aided by intracontinental
rifting and hydrothermal alteration. Inset shows proposed trajectory of hydrospheric changes moving along
a straight line connecting the mantle and modern ocean water. Each circle represents ocean isotopic values
and MWL maintains the modern slope of 0.528 but is shifted upward by the amount Y, which we use in
resolving the isotopic value of meteoric water and the ocean (Eqn. 8) if assumption about temperature are
made. See also Figure 6b and text for discussion.
This would support climate in the Archean that is comparable to the modern tropics at 35–
37 °C. The origin of secondary silicification in the Archean rocks is a subject of discussion
extending far beyond the field of triple oxygen isotopes (Condie 2013; Trail et al. 2018; André
et al. 2019). Secondary silicification may result from strongly evaporated waters on land, and
from extreme weathering on barely emerged land in a CO2-rich atmosphere.
In summary, it appears that the shale and granite record for the Archean is both coeval
and dichotomous at the same time. Shales will predominantly form in greenstone belts
and rift zones, will contain significant proportion of higher-T hydrothermal clays, and are
isotopically shifted toward seawater and perhaps near coastal, low altitude higher d18O waters.
This reflects the small size of Archean continents and Archean continental surface areas and
thus the narrow compositional range of meteoric waters available on such small continents or
island arcs (Fig. 10b). Note that if Archean seawater were −5–15‰ lower in d18O as is shown
on Figure 16, the argument still holds, requiring lower temperature of clay formation (dashed
circles on Figs. 13 and 16). Granites and TTGs would form in adjacent thickened crust of
protocontinents (Fig. 17). A hot Archean climate would favor evaporation and silicification,
aided by perhaps shallow hydrothermal processes involving meteoric and shallow marine
waters. Granites would inherit a greater proportion of silicified material and this additional
silica may contribute to making granites themselves (André et al. 2019; Borisova et al. 2020).
Post-Archean granites and shales
The overlapping post-Archean granites, when compared to coeval shales are easier to
explain using uniformitarian principles of derivation from and assimilation of coeval and
preexisting shales (Fig. 16). Granites inherit the low-∆′17O, and high-d18O signatures from
shales, siliciclastic sediments, and their metamorphic equivalents during anatexis (e.g., Fig. 4).
The rock cycle takes tens to hundreds million years so one would expect the low ∆′17O granites
to “lag behind” low-∆′17O and high-d18O values of shales. Figure 15 hints at a lag of perhaps
of up to 0.2 Ga or less. Chapman et al. (2013) observe minimal (a few million years) time lag
between subducting sediments and magmatism in Sierra Nevada, based on d18O analyses of
zircons and their rims.
278 Bindeman
More research involving geochronology and rock cycle transitions at particular locales
with well-studied and dated sedimentary to metamorphic to igneous transitions, is needed to
address this question.
Figure 15 compares three coeval records: shales, quartz in granites (this work), and detrital
and igneous zircons (Spencer et al. 2017). To the first order, on ~0.5 billion year timescales,
peaks and troughs coincide (within ~0.2 Ga) in the geologic record, which is an important
observation. Supercontinents and panglobal glacial episodes are also shown and exhibit some
connection to the record. In particular, the lowest d18O shales troughs coincide with Neo and
Paleoproterozoic and early Permian glaciations, supporting conclusions in Bindeman et al.
(2016) on the involvement of ultra-low-d18O meteoric and diagenetic waters during glacial
episodes (e.g., Bindeman and Lee 2018). Indeed, the lowest-d18O rocks with d18O values of
−27‰ SMOW are dated to two Paleoproterozoic glacial episodes at 2.4 and 2.3 Ga (Zakharov
et al. 2019). Low-d18O shales also coincide with the end of the supercontinent and increased
rifting, supplying juvenile +5–6‰ mantle oxygen to the record. There is a general lack of
granites during Huronian and Neoproterozoic panglobal ice ages, and is known as the magmatic
shutdown (Condie 2013). In this case the d18O (and ∆′17O!) in dated detrital zircons should help
fill the gap. Zircons also show the low values in the Neoproterzoic, but the minimum appears
to slightly predate the earliest Sturtian glaciation. More research is needed to identify coeval
(subglacial) granites and zircons with ultra-low-d18O values of this age and also Paleoproterozoic
age. The existing dataset of Spencer et al. (2019) with 9000 d18O analyses should be further
split into detrital and igneous subsets to get better resolution of the above possibilities of glacial
influence, supercontinent assembly, and granite-shale time lag questions. For example, the issue
of the time lag is described in Bucholz and Spencer (2019) as future directions of research.
Alternative testable hypotheses of Archean–Proterozoic ∆′17O transition
The argument here then, is that granite and shales record exhibit a step-wise change
of ∆′17O at the Archean–Proterozoic boundary. A question in opposition: what if low-∆′17O
granites appeared first, and shales simply inherited this signature via their weathering? As
studied low-∆′17O granites are metasedimentary derived, this brings the question back to older
sediments. Even with the addition of mantle material via direct mixing with mantle–magma
differentiates, to direct derivation by melting of metabasaltic (rather than siliciclastic) protoliths,
the question is redirected to simultaneous ∆′17O change in the mantle-derived magmas.
The record is further explained if, for some reason, the upper mantle ∆′17O value changed
step-wise at the Archean–Proterozoic boundary. Consider a scenario where the Archean
upper mantle is providing magmas that generated Archean crust and sediments, and was
lower in ∆′17O by up to −0.05‰. This does not require that the ∆′17O = −0.05‰ difference
is representative of the entire convecting mantle, past or present, but may reflect its upper
portion, perhaps a stagnant lid of sublithospheric mantle formed from the magma after the
giant impact at ~4.44 Ga. This early “low-∆′17O” component would have since been largely
dissolved in the mantle and disappeared since 2.4Ga, due to for example intensification of
plate tectonics. The ancient Archean continental keels, made of buoyant harzburgite and
dunite, while not participating in mantle convection may still preserve, if originally present,
this ∆′17O depleted material. This hypothesis can be tested using xenoliths of Archean age from
kimberlites. In order to create a −0.05‰ upper mantle, the Giant Impact itself can be called as
an explanation. The impactor in this case would have been HED-meteorite like planetary body
with ∆′17O = −0.3‰. A cometary material of “Late Veneer” Bombardment can also be called
as an additional source of added material.
Current discussion on the difference between the Earth and the Moon in triple oxygen
space (Herwartz et al. 2014; Young et al. 2016; Greenwood et al. 2018; Cano et al. 2020)
should be tested against the earliest rocks presently available. These authors argue in favor
of a much smaller ∆′17O range not exceeding 0.1‰, likely <0.04‰, and thus we consider
“primordial” ∆′17O of the early crust testable, but a less likely hypothesis than continental
emergence at the Archean–Proterozoic boundary.
PART IV. SEDIMENTARY PROXIES SHOWING TEMPORAL d18O INCREASE

Several paleoclimatic archives such as carbonates, cherts, and shales, show a d18O increase
throughout geologic time (Fig. 3). At face value, these provide independent lines of support
for a temporal increase in d18O of the hydrosphere, by as much as 10–15‰, and strongly
negative (−10 to −20‰) Precambrian oceans (Veizer et al. 1999; Veizer and Prokoph 2015).
Alternatively, this record assuming constant d18O ocean similar to modern (e.g., Muehlenbachs
and Clayton 1976) indicates greater temperatures in the past and thus smaller mineral–water
fractionations per Equations (2) and (3) (Fig. 3). Although diagenetic transformation and
secondary alteration processes involving meteoric waters were undoubtedly involved and
important (see discussion above), I believe that these three records still carry at least a portion
of primary depositional signatures, reflecting surface conditions. I argued above that shales
do. One should also note that an increase in three proxies on Figure 3 appears convincing of
global change of d18Oseawater or T, each of these proxies record somewhat different reservoirs and
processes, and exhibit a different pattern and magnitude of increase. Cherts represent biotically
and abiotically precipitated amorphous silica that matures and commonly become lighter in
d18O during diagenesis (Marine-Carbonne et al. 2010; Cammack et al. 2018). The origin ranges
from low-T cold seawater diatoms or chemical precipitate, to vent fluid-dominated (thus mostly
basaltic origin). Carbonates represent organic (shells, foraminifera, stromatolites), or inorganic
precipitates from carbonate oversaturated seawater or lake water. Both carbonates and cherts are
also subject to alteration, and reset after formation when exposed to the low-d18O hydrosphere
(Ryb et al. 2018; Liljestrand et al. 2020). Shales represent the products of weathering by diverse
in d18O meteoric waters, shifted down on weighted average by −7.5‰ in the dominant modern
sediment formed in the tropics to mid-latitudes (Bindeman et al. 2019).
The long debate about constant vs. increasing d18O of oceans and hence hydrosphere is
related to debate about paleotemperature on early earth and especially in the Archean. Given
mineral proxy–water isotope fractionations as function of temperature, one solution is zero
permil d18O oceans and high temperatures of 45 to 85 °C (reaching clearly unrealistic 130 °C!
for some cherts, Perry et al. 1978). This is viewed as unsustainable for any life except extreme
theromphilic, and violating long term “wet-bulb” sea-land hydrological climate feedback
not allowing the warmest temperature to exceed globally averaged 37 °C (Pierrehumbert
1994; Huber 2012; Buzan and Huber 2020). Hydrothermal alteration products in ophiolites
and altered oceanic crust on the other hand demonstrate relative (±2‰) constancy of d18O
of seawater through time (Muehlenbachs and Clayton 1976; Gregory et al. 1991; Holmden
and Muehlenbachs 1993; Muehlenbachs et al. 2003; Sengupta and Pack 2018; Zakharov and
Bindeman 2019). A common critique of these record is that they do not fully reflect seawater,
but isotopically shifted seawater that has interacted with basalts and mantle rocks upstream.
The isotopically negative oceans and modern range of temperatures is also not an easily
acceptable solution. The steady state d18O value of seawater is fundamentally a balance of
high temperature processes: seafloor hydrothermal alteration that increases d18O water (about
70–80% of all fluxes), and low-temperature processes such as weathering and low-T benthic
alteration (about 15–20% of all fluxes), which decreases d18O values. Other processes play a
minor role, and thus it is a balance and relative importance of spreading vs. weathering. Given
a large oxygen reservoir in the hydrosphere at 1.25 × 1024 g, and small fluxes measuring in
1015 g ‰/yr, ~108 yr are required to change d18O of oceans by more than 1‰ (Muehlenbachs
and Clayton 1976; Walker and Lohmann 1989).
Application of triple oxygen isotopes are called upon to solve these problems by providing
an additional equation to the oxygen isotopic system, potentially permitting resolving d18Ow
and temperature differences, along the line similar to Figure 6. However, largely because the
MWL and the fractionation and mixing equation lines are nearly parallel for any sedimentary
proxies (e.g., Fig. 6), the error of such resolution for carbonates, cherts, and shales is large.
If one assumes a modern MWL, then Archean temperatures are high, +25 to 100 °C, based on
the shale record (Fig. 18).
280 Bindeman
A B C
a b )LJJUH\VFDOHRUFRORU
0.5305
D E F
0.5305
Figure 18. Calculated environmental parameters on Earth: T (a), Δ′17O MW (b) and d18O MW (c) vs. age
based on the solution of three equations with three input parameters: d′18O shale, d′17O shale, Modern meteoric
water line (Eqn. 4), and CIA, as explained in Figure 6. Note that T and d′18Owater solve within realistic
bounds using assumptions and isotopic fractionations as in Equations (5) and (6). Furthermore, the lowest
d′18Owater values are for the 5 Ma-old clay samples from Antarctica and the 2.5–2.2 Ga synglacial Paleo-
proterozoic shales, confirming the participation of low- d′18Ow synglacial waters in diagenesis (cf. Binde-
man et al. 2016). The highest recent T and d′18Owater values are for 55.5 Ma Paleocene–Eocene thermal
maximum shales (Freiling et al. 2014). d–f represent the same solution for rocks greater than 2 Ga but by
assuming ocean of -5‰ and corresponding MWL equation: d′17O = 0.528· d′18O + 0.078 as is explained on
Figures 6 and 17. Note more realistic surface temperatures in the Archean, low T and low-d′18OMW during
2.2–2.4 Ga Paleoproterozoic glaciations in both a and c panels.
The other sedimentary archives, i.e., carbonates and cherts, suffer from syn- and post-
diagenetic shifts, along the concave up curves connecting the rock with low-d18O and high-
∆′17O waters, comparable to mixing curves in Figure 6. Using triple oxygen isotopes Liljestrand
et al. (2020) considered the diagenetic effects upon carbonates and cherts respectively; these
effects will shift ∆′17O downward and along the concave down mixing lines. Thus the inferred
high temperatures were interpreted to reflect diagenesis. In contrast, our insight into diagenetic
transformation (Fig. 9) does not return the same result for Texas shales that rather indicate that
weathering conditions are preserved.
I consider a compromised solution, globally elevated temperatures in the Archean, (perhaps
MAT = 37 °C) and slightly isotopically negative oceans, perhaps −4 to −5‰. Such values can
reconcile the balance between low- and high temperature geochemical fluxes within currently
accepted limits, with some modifications as indicated below (Fig. 18). A simple solution for
−5‰ seawater shifted away and linearly from the mantle with a slope of 0.5215 would generate
an “Archean MWL” also shifted up (Fig. 17). Sengupta and Pack (2018) proposed even steeper
shift along the slope of ~0.51 by taking into account estimated triple oxygen isotopic values of
typical end-member products (deep sea clays, altered MORB and others) and mass balances that
were based on the original Muehlenbachs (1998) work. As these balances are likely to change
(see above), we here adopted even a simple solution of hydrothermal water–rock interaction
globally following the linear path away from the mantle and toward seawater, and weathering/
diagentic processes also nearly linearly connecting crustal rocks with parental weathering waters
on meteoric water line. Linearity is explained by combination of concave down fractionation
and concave up mixing during diagenesis (Figs. 6, 9). The justification for this slope includes
consideration of lines for each of the high and low-T processes involved: 0.5212 of hydrothermal
alteration of MORB (d18O = 5.7‰, ∆′17O = −0.051‰) with seawater (0,0‰); weathering 0.522
(+14.8‰, −0.164‰ sediment) and (−7.3‰, +0.02‰) weathering water (average values in
Bindeman et al. 2019). At a combination of these slopes, any mass balance puts the line away
from the mantle or crustal rocks very close to the modern seawater as is shown on Figure 17.
This effort is by no mean final and should be refined further when balances of low and high-T
fluxes are revised (e.g. Kanzaki 2020a,b). This way we generate hypothetical “Archean meteoric
water lines” estimates with excess of D17O, the main subject of this Chapter.
The new, “Archean MWL” equation at −5‰ d18Osw becomes:
d′17O = 0.528 d′18O + 0.078 (9)
and it was used in hydrologic modeling similar to the one performed above (Fig. 6) to resolve
T and d18O and ∆′17O of meteoric waters parental to shales. Results are shown on Figure 18
and demonstrate that global temperatures stay largely below 50°±10 °C, if meteoric waters
parental to shales were initially derived from −5‰ ocean. Further tuning of Archean MWL
and global temperatures are possible with specialized and perhaps multi-proxies studies, but
this exercise here shows that using shales to reconstruct global ocean values works. Proposed
moderately high temperatures in the Archean and the Precambrian most likely also carry
preservation and sampling bias. For analogy with the modern world (Fig. 1), where most
weathering and shale production occurs in tropical environments of SE Asia, most commonly
sampled Precambrian shales likely sample the most common shales and thus record tropical
environments. Paleotemperatures of 50 ± 10 °C are no surprise, as daytime temperatures in
modern deserts routinely reach 50 °C, and ground temperature where some weathering
reactions take place can exceed 90 °C. Furthermore, temperatures during the Paleocene–
Eocene Thermal Maximum (55.6 Ma ago) were higher than the modern, semi-glacial world,
by at least 5–8 °C (Bowen et al. 2015). Therefore, the Archean temperatures perhaps need not
exceed modern by a conceivable 20 °C to explain the triple oxygen isotopic record of shales.
To justify moderately depleted −5‰ ocean, mass balance of fluxes into the hydrosphere
would need to be adjusted to balance the proportion of low and high-temperature alteration of
the oceanic crust to enable meaningful d18O changes on 107–108 yr timescales, as exchange in
mid-ocean ridge hydrothermal systems constitute the main flux of exchanging ocean. I also
think we need to be thinking outside the uniformitarian principles where modern environments
are transferred to the Archean. Walker and Lohmann (1989) suggested that low-T hydrothermal
alteration dominated the Archean, and the temporal evolution of increasing d18O is a result of
temporally increasing proportion of submarine high-T water–rock interaction. Kasting et al.
(2006) suggested that the box models of strong coupling between hydrosphere and ocean crust
(Muehlenbachs and Clayton 1972, 1976) and Holland (1984) could have varied with time, and
the relative role of low and high-T interaction at ocean floor. I point out new developments to
better understand both the magnitude of each flux, their relative ratio, and the values of the
endmembers. More importantly, it is important to think outside the box and move away from
“box models” which always have a prescribed solution of balancing fluxes, and move toward
real world dynamics of these fluxes. Kanzaki (2020a) demonstrated through 2D numerical
hydrothermal simulations that the degree of MORB alteration is non-linear with depth and in
earlier employed box models (Muehlenbachs and Clayton 1976; Muehlenbachs 1998). Kanzaki
(2020b) performed further 2D numerical simulations of water rock interaction with spreading
to argue that due to incomplete exchange, there is a strong decoupling of d18O values of rocks
282 Bindeman
and seawater. Although not explicitly stated therein, this means that oceanic crust is subducted
while largely unaltered on 107–108 yr timescales, corresponding to range of ages of oceanic
crust in the present world. Thus the overall high-T hydrothermal contribution at MOR, and the
flux of oxygen from the mantle to the hydrosphere, making it higher in d18O, is significantly
smaller than previously assumed. If so, the relative roles of continental chemical weathering
and shale production become more important in driving change in d18O of the hydrosphere
on long timescales, which are governed in part by variations in exposed landmass (Fig. 11).
The trends observed on Figure 3 call for greater and more intense weathering in the past,
perhaps due to a greater CO2 concentration, greater temperatures, and perhaps due to greater
exposure of fresh igneous and metamorphic rocks on the surface. With time the sedimentary
cover shields these fresh low-d18O rocks from weathering. Further modulations are possible due
to the advent of plant life on land. Ibarra et al. (2020) and D′Antonio et al. (2020) demonstrated
that the advent of vascular plants in the Devonian decreased overall weathering rates by a factor
of two due to increased carbon burial effects on CO2 weathering balance. Furthermore, plants
and increased sedimentary rock and soil cover may act to slow the process for d18O on a longer
term, by shielding fresh rocks from weathering (cf. Drever 1994).
I, here, further draw attention to the possibility of temporal accumulation of heavy d18O
in the crust and shales via progressive accumulation of weathering products since 4.4 Ga, and
especially temporal increase in the value of the surface d18O. This should lead to an overall
decrease in weathering d18O flux with time. For the continental weathering flux, not only
the exposed area of the crust subjected to weathering is important, but also the proportion
of preexisting sedimentary cover upon it. Sedimentary rocks not only physically blanket
crystalline igneous and metamorphic rocks available to tie heavy oxygen from the hydrosphere
via shale production as is argued above, but greater proportion of sedimentary cover may
additionally produce very small overall d18O flux to the ocean since sedimentary rocks are
already weathered and high-d18O, thus in equilibrium with hydrosphere. Periods of orogenesis
generate increased exposure of crystalline rocks and thus greater weathering flux leading to
a lower d18O hydrosphere, as compared to periods lacking mountain building which would
agree with the Phanerozoic trends on Figure 14.
Future triple O work should target well-studied Phanerozoic formations to resolve these
effects. In particular, continental breakup leads to continental flooding (young and shallow
oceans basins, warmer climate, greater ocean) and also sediment cover, diminishing the role
of fresh-rock weathering, increasing hydrosphere d18O.
This combined with an increased role of seafloor spreading during continental breakup,
leads to higher d18O value of the hydrosphere. There may be a time lag between realization of
the above torques on oxygen isotopic value of world oceans, perhaps by many tens of Myr, due
to the enormous size of reservoir. It would be interesting to compare the temporal evolution of
triple-O isotopes with the evolution of crustal residence times inferred from Nd isotopes (cf.
Michard et al. 1985). Assembly and breakup of Pangea is the most recent example of possible
realization of these torques and thus Late Proterozoic–Phanerozoic record of shales should
provide insight into this (Fig. 14).
We observe that mid Phanerozoic is accompanied by low-d18O value of shales, while more
recent and especially late Neoproterozoic, have the highest d18O shales and contemporaneous
granites. The magnitude of the change of 5–6‰ d18O could potentially be in agreement with
similar change in d18O of seawater (Wallmann 2001, 2004). Other factors that can influence
this is: colder temperatures in the mid Permian would promote greater fractionations, large
continents in mid Phanerozoic longer vapor transport and low-d18O values, also promoting
low-d18O shale values (Fig. 14).
Temporal increase in d18O of the ocean may also reflect time accumulated loss of low-
d18O interior of slabs (e.g., Bindeman et al. 2019), or progressively decreasing loss earliest
low-d18O crust, formed at high-T by interaction with hydrosphere. Added water will rehydrate
the mantle but produce minimal isotopic shift on it. Steady supply of low-d18O water to the
mantle with subducting slabs should also lead to sea level change (Wallmann 2001, 2004;
Korenaga et al. 2017). Continental emergence which happened at ~2.4 Ga (Fig. 11) may have
a component of sea level drop due to rehydration of the mantle by plate tectonic forces.
We further speculate that if the early high-∆′17O, low-d′18O water that define the earlier
hydrosphere was due to Late Veneer cometary delivery (potentially having very different
∆′17O and d′18O signatures, e.g., low-∆′17O Isua data, pending verification), this is a testable
hypothesis as such water should be reflected in sediments precipitating from it, and then
propagated into the earliest crust before dissolving in the younger record. These processes of a
loss of the originally birthmarked and lingering low-d18O crust and hydrosphere can be tested
by using ∆′17O in shales and other earliest proxies.
SUMMARY: DEFINING THE D′17O CRUSTAL ARRAY

This Chapter presented new and compiled data for D′17O values in siliciclastic sediments
and granites across the geologic history. Figure 19 presents the summary of these data plotted
against 0.528 and 0.5305 reference slopes for comparison with other materials presented in
this volume. A noticeable feature is that a slope of the trend is best reconciled with l = 0.523.
A combined fitted linear trend through shales and granites passes in the close vicinity of the
mantle. A forced fit through the mantle (assuming that crust contains many mantle-derived
rocks such as basalts and gabbros, their remelts and differentiates such as A and I type
granites), yields a line with a slope of 0.523 that passes slightly below the SMOW value
at D′17O = −0.01‰. When dataset is further split between Archean and post-Archean shales
and granites, the trends overlap within error. Modern river clays sampling the most diverse
climates (temperatures and D′18OMW values, data from Bindeman et al. 2019) provide the
glimpse into ranges of weathering conditions applicable to the past environments. Overall,
modern river clays are overlapping and conjoining with the ancient shale fields. When all 170
data points for granites, shales, and modern river clays are lumped together, these represent
silicate crust. Such “crustal D′17O–d′18O array” has the fitted line with a slope of l = 0.5235
and D′17O intercept of −0.021‰; when forced through the mantle it becomes fitted l = 0.5228
and D′17O intercept of −0.01‰, similar to fit through granites alone. Figure 19b and c provide
density plots for D′17O crustal array with utilities for other studies and Chapters in this volume.
A '18O, ‰ VSMOW B C
0 5 10 15 20 25 30
0.00
mantle
‐0.05
‐0.10
'17O 0.528 ‰
0.528
‐0.15
‐0.20
Granites (forced through the mantle):
‐0.25 Archean: ‐0.0049x‐0.0101 (=0.5231)
Post‐Archean: ‐0.0056x‐0.0059 (=0.5224)
All, without forcing through the mantle:
‐0.003x‐0.0349 (= 0.525)
‐0.30 Shales:
Archean: ‐0.077x+0.0038 (= 0.5203)
Post‐Archean: ‐0.0045x=‐0.018 (=0.5235 )
Modern River Clays:
‐0.0016x‐0.0762, R2=0.02 (=0.5264)
Figure 19. (a) Triple oxygen isotopic composition of the crustal rocks (granites, shales, and modern river
sediments) of all ages relative to the 0.528 reference slope. Various fit lines for split datasets are shown (see
inset and main text). Notice that the dataset clusters around the slope l =0.523, and intercept is slightly less
that SMOW by 0.01 to 0.008‰, broadly characterizing the silicate terrestrial crust. b–c) “crustal arrays”:
Fig. 19 color
density plots for all data in (a) relative to D′17O = 0.528 and D′17O = 0.5305 reference slopes.
284 Bindeman
ACKNOWLEDGEMENTS
This paper is benefited from sample donation by Greg Retallack, William Peck, Jade
Star Lackey, John Valley, Dave Blackwell, Axel Hofmann, Allan Keys, Shirley Dutton and
Texas drillcore facility supervisors. I thank Jim Palandri, Mike Hudak for lab assistance,
David Zakharov for help with statistical treatment, Oleg Melnik for help with equation
solving. I am also very grateful to Germain Bayon, David Zakharov and Jim Palandri for prior
collaboration and internal comments on this Chapter, external reviews by Claire Bucholz and
Sukanya Sengupta and further comments by Greg Retallack, Ann Bauer, and William Peck are
acknowledged with warm gratitude. Andreas Pack is thanked for his careful editorial handling.
Partly supported by NSF grant EAR 1833420.
REFERENCES
Anderson LJ, Morrison J (2005) Ilmenite, magnetite, and peraluminous Mesoproterozoic anorogenic granites of
Laurentia and Baltica. Lithos 80:45–60
André, L, Abraham K, Hofmann A, Monin L, Kleinhanns IC, Foley S (2019) Early continental crust generated by
reworking of basalts variably silicified by seawater. Nature Geoscience 12:769–773
Bao H, Cao X, Hayles JA (2016) Triple oxygen isotopes: fundamental relationships and applications. Annu Rev Earth
Bauer K, Vennemann T (2014) Analytical methods for the measurement of hydrogen isotope composition and water
content in clay minerals by TC/EA. Chem Geol 363:229–240
Bauer AM, Fisher CF, Vervoort JD, Bowring SA (2017) Coupled zircon Lu-Hf and U-Pb isotopic analyses of the
oldest terrestrial crust, the > 4.03 Ga Acasta Gneiss Complex. Earth Planet Sci Lett 458:37–48
Bauer AM, Reimink JR, Chacko T, Foley BJ, Shirey SB, Pearson DG (2020) Hafnium isotopes in zircons document
the gradual onset of mobile-lid tectonics Geochem Persp Lett 14:1–6
Bekker A, Holland HD (2012) Oxygen overshoot and recovery during the early Paleoproterozoic. Earth Planet Sci
Lett 317–318:295–304
Bayon G, Delvigne C, Ponzevera E, Borges AV, Darchambeau F, De Deckker P, Lambert T, Monin L, Toucanne S,
André L (2018) The silicon isotopic composition of fine-grained river sediments and its relation to climate and
lithology. Geochim Cosmochim Acta 229:147–161
Bayon G, Skonieczny C, Delvigne C, Toucanne S, Bermell S, Ponzevera E, André L (2016) Environmental Hf–Nd
isotopic decoupling in World river clays. Earth Planet Sci Lett 438:25–36
Bayon G, Freslon N, Germain Y et al. (2020) A global survey of radiogenic strontium isotopes in river sediments.
Chem Geol subm
Bindeman IN (2008) Oxygen isotopes in mantle and crustal magmas as revealed by single crystal analysis Rev
Bindeman IN, Lee JE (2018) The possibility of obtaining ultra-low-d18O signature of precipitation near equatorial
latitudes during the Snowball Earth glaciation episodes. Precambrian Res 319:211–219
Bindeman IN, Bekker A, Zakharov DO (2016) Oxygen isotope perspective on crustal evolutionon early Earth: A
record of Precambrian shales with emphasis on Paleoproterozoic glaciations and Great Oxygenation Event.
(2018) Rapid emergence of subaerial landmasses and onset of a modern hydrologic cycle 25 billion years ago.
Nature 557:545–548
Bindeman IN, Bayon G, Palandri J (2019) Triple oxygen isotope investigation of fine-grained sediments from major
world’s rivers: insights into weathering processes and global fluxes into the hydrosphere Earth Planet Sci Lett
528:115851
Bleeker W (2003) The late Archean record: a puzzle in ca 35 pieces. Lithos 71:99–134
Borisova AY, Zagrtdenov NR, Toplis MJ, Bohrson WA, Nedelec A, Safonov OG, Pokrovski GS, Ceuleneer G,
Bindeman IN, Melnik OE, Jochum KP (2020) Planetary felsic crust formation at shallow depth. arXiv:200301508
Bowen GJ (2019) The Online Isotopes in Precipitation calculator, version 31, http://waterisoutahedu/waterisotopes/
pages/data_access/oipchtml )
Bowen GJ, Maibauer BJ, Kraus MJ, Röhl U, Westerhold T, Steimke A, Gingerich PD, Wing SL, Clyde WC (2015)
Two massive, rapid releases of carbon during the onset of the Palaeocene–Eocene thermal maximum. Nat
Geosci 8:44–47
Brenner AR, Fu RR, Evans DAD, Smirnov AV, Trubko R, Rose IR (2020) Paleomagnetic evidence for modern-like
plate motion velocities at 3.2Ga. Sci Adv 6:eaaz8670
Bucholz C, Spencer CJ (2019) Strongly peraluminous granites across the Archean–Proterozoic transition. J Petrol
60:1299–1348
Bucholz CE, Jagoutz O, VanTongeren JA, Setera J, Wang Z (2017) Oxygen isotope trajectories of crystallizing melts:
Insights from modeling and the plutonic record. Geochim Cosmochim Acta 207:154–184
Bucholz CE, Stolper EM, Eiler JM, Breaks FW (2018) A comparison of oxygen fugacities of strongly peraluminous
granites across the Archean–Proterozoic boundary. J Petrol 59:2123–2156
Buick R, Rasmussen B, Krapez B (1998) Archean oil: evidence from extensive hydrocarbon generation and migration
2.5–3.5 Ga. Am Ass Petrol Geol Bull 82:50–69
Buzan JR, Huber M (2020) Moist heat stress on a hotter Earth. Annu Rev Earth Planet Sci 48:23.1–23.33
Cammack JN, Spicuzza M, Cavosie A, et al. (2018) SIMS microanalysis of the Strelley Pool Formation cherts and the
T implications for the secular-temporal oxygen-isotope trend of cherts. Precambrian Res 304:125–139
Cao X, Liu Y (2011) Equilibrium mass-dependent fractionation relationships for triple oxygen isotopes Geochim
Capuano RM (1992) The temperature dependence of hydrogen isotope fractionation between clay minerals and
water: Evidence from a geopressured system. Geochim Cosmochim Acta 56:2547–2554
Chapman AD, Saleeby JB, Eiler J (2013) Slab flattening trigger for isotopic disturbance and magmatic flare-up in the
southernmost Sierra Nevada batholith, California. Geology 41:1007–1010
Chappell BW, White AJR (1974) Two contrasting granite types. Pac Geol 8:173–4
Carlson RW, Garçon M, O’Neil J, Reimink J, Rizo H (2019) The nature of Earth’s first crust. Chem Geol 530:119321
Cavosie AJ, Valley JW, Wilde SA (2018) The oldest terrestrial mineral record: Thirty years of research on Hadean
zircon from Jack Hills, Western Australia. In: Earth’s Oldest Rocks, Elsevier
Churchman GJ, Clayton RN, Sridhar K, Jackson ML (1976) Oxygen isotopic composition of aerosol size quartz in
shales. J Geophys Res 81:381–386
Colón DP, Bindeman IN, Gerya TV (2019) Understanding the isotopic and chemical evolution of Yellowstone hot
spot magmatism using magmatic-thermomechanical modeling. J Volcanol Geotherm Res 370:13–30
Condie KC (2013) Plate Tectonics & Crustal Evolution, Pergamon Press, 3rd Edition
Criss RE (1999) Principles of Stable Isotope Distribution. Oxford U Press
D’Antonio MP, Ibarra DE ; C Kevin Boyce CK (2020) Land plant evolution decreased, rather than increased,
weathering rates. Geology 48:29–33
Delgado A, Reyes E (1996) Oxygen and hydrogen isotope compositions in clay minerals: a potential single-mineral
geothermometer. Geochim Cosmochim Acta 60:4285–4289
Dellinger M, Bouchez J, Gaillardet J, Faure L, Moureau J (2017) Tracing weathering regimes using the lithium
isotope composition of detrital sediments. Geology 45:411–414
Dhuime B, Wuestefeld B, Hawkesworth CJ (2015) Emergence of modern continental crust about 3 billion years ago.
Nat Geosci 8:552–555
Drever JI (1994) The effect of land plants on weathering rates of silicate minerals. Geochim Cosmochim Acta
58:2325–2332
Ducea MN, Chapman AD (2018) Sub-magmatic arc underplating by trench and forearc materials in shallow
subduction systems; A geologic perspective and implications. Earth Sci Rev 185:763–779
Dutton SP, Loucks RG, Ambrose WA (2015) Factors controlling permeability variation in onshore, deep Paleogene
Wilcox Sandstones in the northern Gulf of Mexico Basin: Targeting high-quality reservoirs. GCAGS J 4:1–14
Eberl DD (1993) Three zones for illite formation during burial diagenesis and metamorphism. Clays Clay Mineral 41:26–37
Eiler JM (2001) Oxygen isotope variations of basaltic lavas and upper mantle rocks. Rev Mineral Geochem 43:319–364
Eslinger, E V, Savin, S M (1976) Mineralogy and 180/160 ratios of fine-grained quartz and clay from site 323: Initial
Reports of the Deep Sea Drilling Project 35, CD Hollister, C Craddock et al. (eds), US Gov Printing Office,
Washington, DC, 489-496
Flament N, Coltice N, Rey PF (2008) A case for late-Archaean continental emergence from thermal evolution models
and hypsometry. Earth Planet Sci Lett 275:326–336
Friedman I, Lipman PW, Obradovich JD, Gleason JD, Christiansen RL (1974) Meteoric water in magmas. Science
184:1069–1072
Galloway WE, Whiteaker TL, Ganey-Curry P (2011) History of Cenozoic North American drainage basin evolution,
sediment yield, and accumulation in the Gulf of Mexico basin. Geosphere 7:938–973
Gaschnig RM, Rudnick RL, McDonough WF, Kaufman AJ, Valley J, Hu Z-C, Gao, S (2016) Compositional evolution
of the upper continental crust through time, as constrained by ancient glacial diamictites. Geochim Cosmochim
Acta 186:316–343
Gilg HA, Girard J-P, Sheppard SMF (2004) Conventional and less conventional techniques for hydrogen, oxygen
isotope analysis of clays, associated minerals and pore waters in sediments and soils. In: Handbook of Stable
Isotope Analytical techniques, Volume 1, PA de Groot, Elsevier BV, p 38–61
Goldschmidt VM (1933) Grundlagen der quantativen Geochemie. Fortschr Mineral 17:112–156
Goldstein SJ, Jacobsen SB (1988) Nd and Sr isotopic systematics of river water suspended material: implications for
crustal evolution. Earth Planet Sci Lett 87:249–265
286 Bindeman
Greber ND, Dauphas N, Bekker A, Ptáček MP, Bindeman IN, Hofmann A (2017) Titanium isotopic evidence for
felsic crust and plate tectonics 35 billion years ago Science 357:1271–1274
Gregory RT (1991) Oxygen isotope history of seawater revisited: Timescales for boundary event changes in the
oxygen isotope composition of seawater In: Stable Isotope Geochemistry: A Tribute to Samuel Epstein, HP
Taylor, JR O’Neil, IR Kaplan (eds) Geochemical Society, p 65–76
Greenwood RC, Barrat J-A, Miller MF, Anand M, Dauphas N, Franchi IA, Sillard P, Starkey NA (2018) Oxygen isotopic
evidence for accretion of Earth’s water before a high-energy Moon-forming giant impact. Sci Adv 4:eaao5928
Hacker BR, Kelemen PB, Behn MD (2011) Differentiation of the continental crust by relamination. Earth Planet Sci
Lett 307:501–516
Hamza MS, Epstein S (1980) Oxygen isotopic fractionation between oxygen of different sites in hydroxyl-bearing
silicate minerals Geochim Cosmochim Acta 44:173–182
Harris C, Faure K, Roger E, Diamond RE, Scheepers R (1997) Oxygen, hydrogen isotope geochemistry of S- and
I-type granitoids: the Cape Granite suite, South Africa. Chemical Geology 143:95–114. ( Cape Granites S Africa
551 Ma)
Hartmann J, Moosdorf N (2012) The new global lithological map database GLiM: a representation of rock properties
at the Earth surface. Geochem Geophys Geosyst 13:Q12004
Hawkesworth C.J. Cawood P.A. Dhuime B, Kemp AIS (2017) Earth’s continental lithosphere through time. Annu Rev
Earth Planet Sci 45:169–198
Hayes JM, Strauss H, Kaufman AJ (1999) The abundance of 13C in marine organic matter and isotopic fractionation
in the global biogeochemical cycle of carbon during the past 800 Ma. Chem Geol 161:103–125
Hayles J, Gao C, Cao X, Liu Y, Bao H (2018) Theoretical calibration of the triple-oxygen isotope thermometer.
Herwartz D (2021) Triple oxygen isotope variations in Earth’s crust. Rev Mineral Geochem 86:291–322
344:1146–1150
Hoffman PF (2013) The Great Oxidation and a Siderian snowball Earth: MIFS based correlation of Paleoproterozoic
glacial epochs. Chem Geol 362:143–156
Holeman JN (1968) The sediment yield of major rivers of the world. Water Resour Res 4:737–747
Holland HD (1984) The chemical evolution of the atmosphere and oceans. Princeton University Press
Holmden C, Muehlenbachs K (1993) The 18O/16O ratio of 2-billion-year-old seawater inferred from ancient oceanic
crust. Science 259:1733–1736
Hopkinson TN, Harris NB, Warren CJ, Spencer CJ, Roberts NM, Horstwood MS, Parrish RR (2017) The identification
and significance of pure sediment-derived granites. Earth Planet Sci Lett 467:57–63
Hower J, Eslinger EV, Hower ME, Perry EA (1976) Mechanism of burial metamorphism of argillaceous sediment: 1
Mineralogical and chemical evidence. GSA Bull 87:725–737
Hsieh JCC, Chadwick OA, Kelly EF, Savin SM (1998) Oxygen isotopic composition of soil water: Quantifying
evaporation and transpiration. Geoderma 82:269–293
Huber M (2012) Progress in greenhouse climate modeling. Paleontol Soc Pap 18:213–262
Ibarra DE, Rugenstein JKC, Bachan A, Baresch A, Lau KV, Thomas DL (2019) Modeling the consequences of land
plant evolution on silicate weathering Am J Sci 319:1–43
Jung S, Mezger K, Hoernes S (2001) Trace element, isotopic (Sr, Nd, Pb, O) arguments for a mid-crustal origin
of Pan-African garnet-bearing S-type granites from the Damara orogen (Namibia) Precambrian Research
110:325—355
Kanzaki Y (2020a) Interpretation of oxygen isotopes in Phanerozoic ophiolites and sedimentary rocks. Geochem
Geophys Geosystems 21:e2020GC009000
Kanzaki Y (2020b) Quantifying the buffering of oceanic oxygen isotopes at ancient midocean ridges, Solid Earth
Discuss 11:1475–1488
Kasting JF, Howard MT, Wallmann K, Veizer J, Shields G, Jaffres J (2006) Paleoclimates, ocean depth, and oxygen
isotopic composition of seawater. Earth Planet Sci Lett 252:82–93
King EM, Valley JW, Davis DW, Edwards GR (1998) Oxygen isotope ratios of Archean plutonic zircons fromgranite–
greenstone belts of the Superior Province: indicator of magmatic source. Precambrian Res 92:365–38
Knauth LP, Lowe DR (2003) High Archean climatic temperature inferred from oxygen isotope geochemistry of cherts
in the 35 Ga Swaziland Supergroup, South Africa. Bull Geol Soc Am 115:566–580
Korenaga, J, Planavsky NJ, Evans, DAD (2017) Global water cycle and the coevolution of the Earth’s interior and
surface environment. Phil Trans R Soc A 375:20150393
Kump LR (2014) Hypothesized link between Neoproterozoic greening of the land surface and the establishment of an
oxygen-rich atmosphere. PNAS 111:14062–14065
Lackey JS, Valley JW, Chen JH, Stockli DF (2008) Dynamic magma systems, crustal recycling,, alteration in the
central Sierra Nevada batholith: The oxygen isotope record. J Petrol 49:1397–1426
Lackey JS, Cecil MR, Windham CJ, Frazer RE, Bindeman IN, Gehrels GE (2012) The Fine Gold Intrusive Suite:
The roles of basement terranes, magma source development in the Early Cretaceous Sierra Nevada batholith.
Geosphere 8:292–313
Land LS (1992) Saline formation waters in sedimentary basins: Connate or diagenetic? Proceedings of the 7th Inter
national Symposium on Water–Rock Interaction, A. A. Balkema, Rotterdam, p 865–868
Land LS, Lynch FL (1996) d18O values of mudrocks: More evidence for an 18O-buffered ocean. Geochim Cosmochim
Acta 60:3347–3352
Land LS, Milliken KL, Mack LE, Lynch FL (1997) Burial metamorphism of argillaceous sediment: A re-examination.
Bull Geol Soc America 109:2–15
Lawrence JR, Taylor HP, Jr (1972) Hydrogen and oxygen isotope systematics in weathering profiles. Geochim
Lawver LA, Dalziel IWD, Norton IO, Gahagan LM (2011) The Plates 2011. Atlas of Plate Reconstructions (500 Ma
to Present Day), Plates Progress Report No 345–0811, University of Texas Technical Report No 198:189 pp
Levin NE, Raub TD, Dauphas N, Eiler JM (2014) Triple oxygen isotope variations in sedimentary rocks Geochim
Li S, Gaschnig RM, Rudnick RL (2016) Insights into chemical weathering of the upper continental crust from the
geochemistry of ancient glacial diamictites Geochim Cosmochim Acta 176:96–117
Liljestrand FL, Knoll AH, Tosca NJ, Phoebe ACohen PA, Macdonald FA, Peng Y, Johnston DT (2020) The triple
oxygen isotope composition of Precambrian chert. Earth Planet Sci Lett 537:116167
Lohmann KC, Walker JCG (1989) The d18O record of Phanerozoic abiotic marine calcite cements. Geophys Res Lett
16:319–322
Longstaffe FJ (1983) Stable isotope studies of diagenesis of clastic rocks. Geosci Canada 10:47–58
Longstaffe FJ, Schwarcz HP (1977) 18O/16O of Archean clastic metasedimentary rocks: a petrogenetic indicator for
Archean gneisses? Geochim Cosmochim Acta 41:1303–1312
Lynch FL (1997) Frio shale mineralogy and the stoichiometry of the smectite-to-illite reaction: The most important
reaction in clastic sedimentary diagenesis. Clays Clay Miner 45:618–631
Lynch FL, Mack LE, Land LS (1997) Burial diagenesis of illite/smectite in shales and the origins of authigenic quartz
and secondary porosity in sandstones. Geochim Cosmochim Acta 61:1995–2006
McLennan SM, Hemming, S, McDaniel, DK, Hanson GN (1993) Geochemical approaches to sedimentation,
provenance, tectonics In: MJ Johnsson and A Basu (Eds) Processes controlling the composition of clastic
sediments Geol Soc Am Spec Pap 284:21–40
Mackenzie FT (editor) (2005) Sediments, Diagenesis, and Sedimentary Rocks. 1st Edition. Treatise on Geochemistry,
Second Edition, Volume 7, Elsevier
Marin-Carbonne J, Chaussidon M, Robert F (2010) The silicon and oxygen isotope compositions of Precambrian
cherts: A record of oceanic paleo-temperatures? Precambrian Res 247:223–234
Mertanen S, Pesonen LJ (2012) Paleo–Mesoproterozoic assemblages of continents: paleomagnetic evidence for near
equatorial supercontinents. In: From the Earth’s Core to Outer Space. Lecture Notes in Earth System Sciences
137, I Haapala (ed) Springer-Verlag Berlin Heidelberg, p 11–35
Michard A, Gurriet P, Soudant M, Albarede F (1985). Nd isotopes in French Phanerozoic shales: external vs. internal
aspects of crustal evolution. Geochim Cosmochim Acta 49:601–610
precision oxygen triple-isotope ratio measurements of silicate rocks and minerals Chem Geol 532:119332
Milliken KL, Land LS, Loucks RG (1981) History of burial diagenesis, Frio Formation, Brazoria County, Texas Bull
Am Assoc Petrol Geol 65:1397–1413
Milliman JD, Farnsworth KL (2011) River Discharge to the Coastal Ocean: A Global Synthesis. Cambridge University
Press, Cambridge
Mix H, Chamberlain CP (2014) Stable isotope records of hydrologic change and paleotemperature from smectite in
Cenozoic western North America. Geochim Cosmochim Acta 141:532–546
Morton RA, Land LS (1987) Regional variations in formation water chemistry, Frio Formation (Oligocene), Texas
Gulf Coast. Bull Am Assoc Petrol Geol 72:191–206
Muehlenbachs K (1998) The oxygen isotopic composition of the oceans, sediments and the seafloor. Chem Geol
145:263–273
Muehlenbachs K, Clayton RN (1976) Oxygen isotopic composition of oceanic crust and its bearing on seawater. J
Geophys Res 81:4365–4369
288 Bindeman
Muehlenbachs K, Furnes H, Fonneland HC, Hellevang B (2003) Ophiolites as faithful records of the oxygen isotope
ratio of ancient seawater: The Solund-Stavfjord Ophiolite Complex as a Late Ordovician example. Geol Soc,
London, Spec Publ 218:401–414
Mulch A, Sarna-Wojcicki AM, Perkins ME, Chamberlain CP (2008) A Miocene to Pleistocene climate, elevation
record of the Sierra Nevada (California). PNAS 105:6819–6824
Nabelek PI, Russ-Nabelek C, Denison JR (1992) The generation and crystallization conditions of the Proterozoic
Harney Peak Leucogranite, Black Hills, South Dakota, USA: Petrologic and geochemical constraints. Contrib
Mineral Petrol 110:173–191
Nesbitt HW, Young GM (1982) Early Proterozoic climates and plate motions inferred from major element chemistry
of lutites. Nature 299:715–717
Newman ACD (1987) Chemistry of clays and clay minerals Mineralogical society monograph series No 6,
Mineralogical society, Longman
O’Neil JR, Chappell BW (1977) Oxygen, hydrogen isotope relations in the Berridale batholith. J Geol Soc Lond 33:
559–571
O’Neil JR, Shaw SE, Flood RH (1977) Oxygen, Hydrogen Isotope Compositions as Indicators of Granite Genesis in
the New England Batholith, Australia. Contrib Mineral Petrol 62:313–328
O’Neil J, Boyet M, Carlson RW, Paquette J-L (2013) Half a billion years of reworking of Hadean mafic crust to
produce the Nuvvuagittuq Eoarchean felsic crust. Earth Planet Sci Lett 379:13–25
Pack A (2021) Isotopic fingerprints of atmospheric molecular O2 in rocks, minerals, and melts. Rev Mineral Geochem
86:217–240
Pack A, Tanaka R, Hering M, Sengupta S, Peters S, Nakamura E (2016) The oxygen isotopecomposition of San
Parra-Avila LA, Belousova E, Fiorentini ML, Eglingerd A, Blocke S, Millera J (2018) Zircon Hf and O-isotope
constraints on the evolution of the Paleoproterozoic Baoulé-Mossi domain of the southern West African Craton.
Precambrian Res 306:174–188
Passey BH, Ji H (2019) Triple-oxygen isotope signatures of evaporation in lake waters and carbonates: A case study
implications for continental paleoclimate reconstruction. Rev Mineral Geochem 86:429-462
Payne JL, Hand M, Pearson NJ et al. (2015) Crustal thickening and clay: Controls on O isotope variation in global
magmatism and siliciclastic sedimentary rocks. Earth Planet Sci Lett 412:70–76
Peck WH (2016) Protolith carbon isotope ratios in cordierite from metamorphic and igneous rocks. American
Mineralogist 101:2279–2287
Peck WH, Valley JW, Corriveau L, Davidson A, McLelland J, Farber DA (2004) Oxygen-isotope constraints on
terrane boundaries and origin of 1.18–1.13 Ga granitoids in the southern Grenville Province. Geol Soc Am
Memoir 197:163–182
Perry EC, Jr, Ahmad SN, Swulius TM (1978) The oxygen isotope composition of 3,800 my old metamorphosed chert
and iron formation from Isukasia, West Greenland. J Geol 86:223–239
Pettijohn FJ (1957) Sedimentary Rocks, second edition, Harper, New York
Peucat JJ, Capdevila R, Drareni A, Choukroune P, Fanning CM, Bernard-Griffiths J, Fourcade S (1996) Major, trace
element geochemistry and isotope (Sr, Nd, Pb, O) systematics of an Archaean basement involved in a 2.0 Ga
very high-temperature (1000 °C) metamorphic event in Ouzzal Massif, Hoggar, Algeria. J Metamorph Geol
14:667–692
Pierrehumbert RT (1994) Thermostats, radiator fins, and the local runaway greenhouse. J Atmos Sci 52:1784–1806
Pollington AD, Kozdon R, Valley JW (2011) Evolution of quartz cementation during burial of the Cambrian Mt
Simon Sandstone, Illinois Basin: In situ microanalysis of 18O. Geology 39:1119–1122
Pope EC, Bird DK, Rosing MT (2012) Isotope composition and volume of Earths early oceans. PNAS 109:4371–4376
Ptáček MP, Dauphas N, Greber ND (2020) Chemical evolution of the continental crust from a data-driven inversion
of terrigenous sediment compositions. Earth Planet Sci Lett 539:116090
Reimink JR, Chacko T, Stern RA, Heaman LM (2016) The birth of a cratonic nucleus: Lithogeochemical evolution of
the 4.02–2.94 Ga Acasta Gneiss Complex. Precambrian Res:281:453–472
Reimink JR, Davies JH, Bauer AM, Chacko T (2020) A comparison between zircons from the Acasta Gneiss Complex
and the Jack Hills region. Earth Planet Sci Lett 531:115975
Retallack GJ (2001) Soils of the past an Introduction to Pedology, Blackwell Science, Oxford, 2nd edition
Retallack GJ (2007) Cenozoic paleoclimate on land in North America. J Geol 115:271–294
Robert F, Chaussidon M (2006) A palaeotemperature curve for the Precambrian oceans based on silicon isotopes in
cherts. Nature 443:969–72
Ronov AB, Yaroshevsky AA (1969) Chemical composition of the Earth’s crust, in the Earth’s crust and upper mantle,
ed PJ Hart, American Geophysical Union, Washington, DC
Rowley DB, Pierrehumbert R, Currie B (2001) A new approach to stable isotope-based paleoaltimetry: implications
for paleoaltimetry and paleohypsometry of the High Himalaya since the Late Miocene. Earth Planet Sci Lett
188:253–268
Rozanski R, Araguás-Araguás L, Gonfiantini R (1993) Isotopic patterns in modern global precipitation In: PK Swart,
KC Lohmann, J Mckenzie, S Savin (Eds) Climate Change in Continental Isotopic Records, Geophysical
Monograph 78, p 1–36
Rozel A, Golabek G, Jain C, Tackley PJ, Gerya TV(2017) Continental crust formation on early Earth controlled by
intrusive magmatism. Nature 545:332–335
silicate minerals: an inter-laboratory comparison of hydrothermal quartz and eclogitic garnet. Geochim
Rumble D, Bowring S, Iizuka T, Komiya T, Lepland A, Rosing MT, Ueno Y (2013) The oxygen isotope composition
of Earth’s oldest rocks and evidence of a terrestrial magma ocean. Geochem Geophys Geosystems 14:1929–1939
Rudnick RL, Gao S (2003) Composition of the continental crust. In: Treatise on Geochemistry, Volume 3. RL Rudnick
HD Holland, KK Turekian (eds) Elsevier, pp 1–64
Ryb U, Eiler JM (2018) Oxygen isotope composition of the Phanerozoic ocean and a possible solution to the dolomite
problem. PNAS 115:6602–6607
Sageman BB, Lyons TW (2004) Geochemistry of fine-grained sediments, sedimentary rocks. In: Sediments,
diagenesis, and sedimentary rocks. FT Mackenzies(Ed) Vol 7 Treatise in Geochemistry, Elsevier, Oxford, p
115–158
Savin S, Epstein S (1970a) The oxygen and hydrogen isotope geochemistry of clay minerals. Geochim Cosmochim
Acta 34:25–42
Savin S, Epstein S (1970b) The oxygen and hydrogen isotope geochemistry of ocean sediments and shales. Geochim
Savin SM, Hsieh JCC (1998) The hydrogen and oxygen isotope geochemistry of pedogenic clay minerals: principles
and theoretical background. Geoderma 82:227–253
Savin SM, Lee ML (1988) Isotopic studies of phyllosilicates. Rev Mineral Geochem 19:189–223
Schoenemann SW, Schauer AJ, Steig EJ (2013) Measurement of SLAP2 and GISP d17O and proposed VSMOW–
SLAP normalization for d17O and 17O excess. Rapid Commun Mass Spectrom 27:582–590
Schoenemann SW, Steig EJ, Ding Q, Markle BR, Schauer AJ (2014) Triple water-isotopologue record from WAIS
Divide, Antarctica: controls on glacial–interglacial changes in 17O excess of precipitation. J Geophys Res Atmos
119:8741–8763
Shackleton NJ, Kennett JP (1975) Paleotemperature history of the Cenozoic and the initiation of Antarctic glaciation;
oxygen and carbon isotope analyses in DSDP sites 277:279 and 281. Initial Rep Deep Sea Drill Proj 29:743–755
Schaltegger U, Nowak A, Ulianov A, Fisher CM, Gerdes A, Spikings R, Whitehouse MJ, Bindeman I, Hanchar JM,
Duff J, Vervoort (2019) Zircon petrochronology and 40Ar/39Ar thermochronology of the Adamello Intrusive Suite,
N Italy: Monitoring the growth and decay of an incrementally assembled magmatic system. J Petrol 60:701–722
Sharp ZD (2017) Principles of Stable Isotope Geochemistry, 2nd Edition
Sharp ZD, Gibbons JA, Maltsev O, Atudorej V, Pack A, Sengupta S, Shock EL, Knauth LP (2016) A calibration of the
triple-isotope fractionation in the SiO2–H2O system and applications to natural samples. Geochim Cosmochim
Acta 186:105–119
Shaw DB, Weather CE (1965) The mineralogical composition of shales. J Sed Petrol 35:213–222
Sheppard SMF, Gilg HA (1996) Stable isotope geochemistry of clay minerals. Clay Minerals 31:1–24
Simon L, Lécuyer C (2005) Continental recycling: The oxygen isotope point of view. Geochem Geophys Geosystems
6:Q08004
Sobolev SV, Brown M (2019) Surface erosion events controlled the evolution of plate tectonics on Earth. Nature
570:52–57
Spencer CJ, Roberts NMW Santosh M (2017) Growth, destruction, and preservation of Earth’s continental crust Earth
Sci Rev172:87–106
Spencer CJ, Partin CA, Kirkland CL, Raub TD, Liebmann J, Stern RA (2019) Paleoproterozoic increase in zircon
d18O driven by rapid emergence of continental crust. Geochim Cosmochim Acta 257:16–25
Suchecki R K, Land L S (1983) Isotopic geochemistry of burial-metamorphosed volcanogenic sediments, Great
Valley sequence, northern California. Geochim Cosmochim Acta 47:1487–1499
Surma J, Assonov S, Staubwasser M (2021) Triple oxygen isotope systematics in the hydrologic cycle. Rev Mineral
Geochem 86:401–428
Sylvester PJ (1998) Post-collisional strongly peraluminous granites. Lithos 45:29–44
Taylor SR, McLennan SM (1995) The geochemical evolution of the continental crust. Rev Geophys 33:241–265
Taylor HP, Epstein S (1962) Relationship between 18O/16O ratios in coexisting minerals of igneous and metamorphic
rocks, part 1, principles and experimental results. Geol Soc Am Bull 73:461–480
290 Bindeman
Trail D, Boehnke P, Savage PS, Liu MC, Miller ML, Bindeman I (2018) Origin, significance of Si and O isotope
heterogeneities in Phanerozoic, Archean, and Hadean zircon. PNAS 115:10287–10292
Turner S, Tonarini S, Bindeman I, Leeman WP, Schaefer BF (2007) Boron and oxygen isotope evidence for recycling
of subducted components over the past 2.5 Gyr. Nature 447:702–705
Uemura R, Barkan E, Abe O, Luz B (2010) Triple-isotope composition of oxygen in atmospheric water vapor.
Valley JW (1986) Stable isotope geochemistry of metamorphic rocks. Rev Mineral 16:445–489
Valley JW, Peck WH, King EM, Wilde SA (2002) Cool early Earth. Geology 30:351–354
Valley JW, Lackey J S, Cavosie AJ, CC Clechenko, MJ Spicuzza, MAS Basei, IN Bindeman, VP Ferreira, AN Sial, E
M King, W H Peck, A K Sinha, C S Wei (2005) 44 Billion years of crustal maturation: Oxygen isotope ratios of
magmatic zircon. Contrib Mineral Petrol 150:561–580
Veizer J, Prokoph A (2015) Temperatures and oxygen isotopic composition of Phanerozoic oceans Earth Sci Rev 146:92–104
Veizer J, Ala D, Azmy K, Bruckschen P, Buhl D, Bruhn F, Strauss, H (1999) 87Sr/86Sr, d13C and d18O evolution of
Phanerozoic seawater. Chem Geol 161:59–88
Vlaar NJ (2000) Continental emergence and growth on a cooling earth Tectonophysics 322:191–202
Wallmann K (2001) The geological water cycle and the evolution of marine d18O values. Geochim Cosmochim Acta
65:2469–2485
Wallmann K (2004) Impact of atmospheric CO2 and galactic cosmic radiation on Phanerozoic climate change and the
marine d18O record. G-cubed 5:Q06004
Walker JCG, Lohmann KC (1989) Why the oxygen isotopic composition of sea water changes with time. Geophys
Res Lett 16:323–326
Watts KE, Bindeman IN, Schmitt AK (2011) Large-volume rhyolite genesis in caldera complexes of the Snake River
Plain: Insights from the Kilgore tuff of the Heise volcanic field, Idaho, with comparison to Yellowstone and
Bruneau-Jarbidge Rhyolites. J Petrol 52:857–890
Wedepohl KH (1995) The composition of the continental crust. Geochim Cosmochim Acta 59:1217–1239
Wetmore PH, Mihai N Ducea MN (2011) Geochemical evidence of a near-surface history for source rocks of the
central Coast Mountains Batholith, British Columbia. Int Geol Rev 53:230–260
Wilde SA, Valley JW, Peck WH, Graham CM (2001) Evidence from detrital zircons for the existence of continental
crust and oceans on the Earth 4.4 Gyr ago. Nature 409:175–178
Wilkinson M, Crowley SF, Marshall JD (1992) Model for the evolution of oxygen isotope ratios in the pore fluids of
mudrocks during burial. Mar Petrol Geology 9:98–105
Wostbrock JAG, Sharp ZD (2021) Triple oxygen isotopes in silica–water and carbonate–water systems. Rev Mineral
Geochem 86:367–400
Wostbrock JAG, Cano E, Sharp ZD (2020) An internally consistent triple oxygen isotope calibration of standards for
Yeh H-W, Epstein S (1978) Hydrogen isotope exchange between clay minerals and seawater. Geochim Cosmochim
Acta 42:140–143
Yeh H-W, Eslinger E (1986) Oxygen isotopes and the extent of diagenesis of clay minerals during sedimentation and
burial in the sea. Clays Clay Miner 34:403–406
Yeh H-W, Savin SM (1976) The extent of oxygen isotope exchange between clay minerals and sea water. Geochim
Yeh H-W, Savin SM (1977) Mechanism of burial metamorphism of argillaceous sediments: 3 Oxygen isotopic
evidence. Geol Soc Am Bull 88:1321–1330
and their geochemical and cosmochemical significance Geochim Cosmochim Acta 66:1095–1104
mixing during the Moon-forming giant impact Science 351:493–496
Geology 45:667–670
Zakharov DO, Bindeman IN, Serebryakov NS, Prave AR, Azimov PYa, Babarina II (2019) Low d18O rocks in the
Belomorian belt, NW Russia, and Scourie dikes, NW Scotland: a record of ancient meteoric water captured by
the early Paleoproterozoic global mafic magmatism. Precambrian Res 333:105431
Zakharov DO, Marin-Carbonne J, Alleon J, Bindeman IN (2021) Temporal triple oxygen isotope trend recorded by
Precambrian cherts: A perspective from combined bulk and in situ secondary ion probe measurements. Rev
Zanazzi A, Judd E, Fletcher A, Bryant H, Kohn M (2015) Eocene–Oligocene latitudinal climate gradients in
North America inferred from stable isotope ratios in Perissodactyl tooth enamel. Palaeogeogr Palaeoclimatol
Palaeoecol 417:561–568
Vol. 86 pp. 291–322, 2021 9
Triple Oxygen Isotope Variations in Earth’s Crust

Daniel Herwartz
Institut für Geologie und Mineralogie
Universität zu Köln
Zülpicher Str. 49b
50674 Cologne
Germany
d.herwartz@uni-koeln.de
INTRODUCTION
Analyzing classic 18O/16O ratios in solids, liquids and gases has proven useful in almost
every branch of earth science. As reviewed in this book, additional 17O analysis generally help to
better constrain the underlying fractionation mechanisms providing a more solid basis to quantify
geologic processes. Mass independent fractionation (MIF) effects known from meteorites
and atmospheric gases (Clayton et al. 1973; Thiemens and Heidenreich 1983; Thiemens and
Lin 2021, this volume; Brinjikji and Lyons 2021, this volume) generate large 17O anomalies
(expressed as ∆17O) that can be transferred to solids (Bao 2015; Cao and Bao 2021, this volume;
Pack 2021, this volume) and to Earth′s crust (Peters et al. 2020a). However, such occurrences
are rarely observed in igneous or metamorphic rocks. Most of the literature summarized herein,
deals with much smaller 17O “anomalies” stemming from fundamental differences between
individual mass dependent fractionation mechanisms (e.g., equilibrium vs. kinetic isotope
fractionation) and mixing processes. The basic theory for these purely mass dependent triple
oxygen isotope fractionation mechanisms is long known (Matsuhisa et al. 1978; Young et al.
2002), but respective applications require high precision isotope analysis techniques that are now
available for a number of materials including silicates, sulfates, carbonates and water.
Clear differences between equilibrium and kinetic isotope fractionation effects resolved for
water (Barkan and Luz 2005, 2007) triggered novel research related to the meteoric water cycle,
where the “17Oexcess parameter” is introduced in analogy to the traditional “Dexcess parameter”
(Surma et al. 2021, this volume). Both Dexcess and 17Oexcess (= ∆′17O herein) are defined such that
they remain more or less constant for equilibrium fractionation and evolve if kinetic processes
occur. Triple oxygen isotope systematics in the meteoric water cycle are therefore similar to
dD vs. d18O systematics, but with some important differences (Landais et al. 2012; Li et al.
2015; Surma et al. 2018, Voigt et al. 2020). Chemical precipitates or alteration products capture
and preserve the triple oxygen isotopic composition of ambient water as shown for gypsum
(Herwartz et al. 2017; Gázquez et al. 2018; Evans et al. 2018), silicates (Pack and Herwartz
2014; Levin et al. 2014; Sharp et al. 2016; Wostbrock et al. 2018; Wostbrock and Sharp 2021,
this volume), weathering products (Bindeman 2021, this volume) and carbonates (Passey et al.
2014; Passey and Levin 2021, this volume;). Most studies focus on low T environments, were
fractionation factors are large and kinetic effects as well as mixing can be important generating
significant variability in ∆′17O due to purely mass dependent effects.
In igneous or metamorphic systems, temperatures are high and timescales are sufficiently
long to attain equilibrium. Hence, variability in ∆′17O should be limited. Indeed, abundant high
precision analyses of peridotites and primitive mantle melts reveal no obvious variability in ∆′17O

1943-2666/21/0086-0009$05.00 (online)
292 Herwartz
(Wiechert 2001; Spicuzza et al. 2007; Hallis et al. 2010; Tanaka and Nakamura 2013; Pack and
Herwartz 2014; Herwartz et al. 2014; Young et al. 2016; Starkey et al. 2016; Cano et al. 2020).
The exception are rocks that have interacted with water. The triple oxygen isotopic composition
of surface water is generally distinct from the rocks and minerals they interact with (Tanaka and
Nakamura 2013) and water partially transfers its isotopic composition to the alteration products.
The original water isotopic composition can be calculated from hydrothermally altered rocks
providing a tool to study paleo hydrology (Chamberlain et al. 2020) including Precambrian
seawater (Zakharov and Bindeman 2019; Peters et al. 2020b) and ancient glacial water from
snowball Earth episodes (Herwartz et al. 2015; Zakharov et al. 2017, 2019a)
Compared to the low T applications, the field of triple oxygen isotopes in high T environments
is young, partly owing to the fact that “anomalies” in ∆′17O are more difficult to resolve, when
fractionation in 18O/16O is limited. The set of recent papers summarized herein demonstrates
the potential of triple oxygen isotope systematics in hydrothermally altered, igneous and
metamorphic rocks. The next pages introduce the basics of triple oxygen isotope systematics
with a focus on purely mass dependent processes. I will then review a classic experimental
concept to acquire equilibrium fractionation factors using triple oxygen isotopes before focusing
on water–rock interaction and individual processes such as assimilation, dehydration and
decarbonation that can be traced via triple oxygen isotope systematics. Finally, I will discuss
alteration using an exotic example of lunar meteorite finds weathered in dry desert environments.
GUIDING PRINCIPALS
The theoretical framework for isotope fractionation on a quantum mechanical level was
set by Urey, Bigeleisen and Mayer demonstrating how chemical bond strength and temperature
control isotope fractionation, something widely known as “equilibrium isotope fractionation”
today (Bigeleisen and Mayer 1947; Urey 1947; Yeung and Hayles 2021, this volume; Schauble
and Young 2021, this volume). Isotopic discrimination between two substances generally
increases with decreasing temperature, providing the basis for isotope paleo-thermometers
ranging from the traditional 18O/16O carbonate paleo-thermometer (Epstein et al. 1951;
Epstein and Mayeda 1953), to a wide range of mineral–mineral thermometers used in earth
sciences (Matthews 1994). The respective isotopic temperature estimates are only accurate if
equilibrium between the isotopic substances is attained. Any deviation from equilibrium (e.g.,
due to kinetic effects, alteration or mixing) compromises temperature accuracy.
The term “kinetic isotope fractionation” is used in various ways, sometimes addressing
diffusion, or the preferential breaking of chemical bonds with light isotopes, and sometimes
simply meaning that equilibrium is not attained. Besides insufficient time to reach equilibrium,
isotopic fluxes between reservoirs can hold individual reservoirs in a steady state (e.g.,
seawater; Pack and Herwartz 2014), leading to apparent fractionation factors that are unrelated
to thermodynamic equilibrium between two phases. These various types of “kinetic isotope
fractionation” can be distinguished because the underlying physics dictate characteristic slopes
in triple oxygen isotope space, all of which differ from equilibrium. This offers an elegant way
of identifying kinetic processes and thus allows testing the equilibrium assumption.
Physical transport processes critically control effective isotopic fractionation between
individual reservoirs (Druhan et al. 2019). Simple mathematical models such as batch or
open system Rayleigh fractionation are often used to approximate isotopic fractionation
curves, but considerable complexity is expected and observed in nature (Dansgaard 1964;
Druhan et al. 2019). Even simple batch experiments designed to constrain equilibrium
fractionation factors can show artefacts of intra-experiment transport processes and kinetic
isotope effects (Matsuhisa et al. 1978).
Triple Oxygen Isotope Variations in Earth′s Crust 293
With a single isotope ratio, it is challenging to: 1) identify complex transport mechanisms;
2) verify equilibrium; or 3) to pinpoint superimposed kinetic effects or a mixing process.
The characteristic slopes and curves in triple oxygen isotope space allow to identify and
sometimes to quantify a given combination of fractionation processes. The diverse applications
summarized in this book demonstrate the versatile nature of additional 17O/16O analyses to the
traditional 18O/16O isotope system.
Definitions, reference frames and the significance of TFL′s, MWL′s and q
Quantification of triple oxygen isotope fractionation builds on traditional notations.
Isotope ratios are reported in the traditionally delta notation (Eqn. 1) relative to an international
standard. For oxygen:
i
Rsample  i Rreference
i O  i
(1)
Rreference
where “R” stands for the ratio of the rare isotope “i” (17 or 18) relative to the abundance of
16
O. Fractionation between two phases A and B is quantified via the ratio of the isotope ratios:
i
R A 1  i O A
iA-B
/16
 (2)
i
RB 1  i O B
and both fractionation factors are related by:
 

17
A-B   A-B
/16 18 /16 (3)
The key observation is that q is not constant, but somewhat dependent on the fractionation
process (mainly kinetic vs. equilibrium) and to a lesser extent on the substances involved, as
well as fractionation temperature (Fig. 1). Therefore, the d17O is expected to vary as a function
of q, and thus with the fractionation process.
To visualize the small 17O/16O variations, the D′17O (or 17Oexcess) notation is used, which
quantifies the deviation from an arbitrary reference slope (λ).
17
O ’17O    18O (4)
Here, λ = 0.528 is used as a reference slope so that D′17O = 17
Oexcess (see below). In this
logarithmic form of the delta notation, with:
i O ln 1  i O
   (5)
processes follow lines, while they would represent curves in the traditional d18O vs. d17O
space. Mixing lines, however, which are linear in d18O vs. d17O space represent curves in the
d′ reference frame.
Reservoirs A and B in Figure 2 are in equilibrium at a given temperature.
The fractionation of 18O/16O and 17O/16O are represented by the 1000 ln 18aeq and 1000 ln 17aeq,
respectively. Hence, the slope connecting the two equilibrated phases in the d′ reference frame
is ln17aA-B / ln18aA-B, which is the definition of aA-B.
If this qA-B differs numerically from the reference slope l, the D′17O of phases A and B
are not identical. As an example, phase A could be SMOW and thus part of the reference line
and B could be SiO2. The qSiO2–water = 0.5245 at 25°C (Sharp et al. 2016), hence lower than
the reference slope chosen here (0.528). Therefore, SiO2 in equilibrium with SMOW has a
negative D′17O (Fig. 2). If λ = 0.5245 was chosen as a reference slope, D′17O would be zero.
So-called ‘Terrestrial Fractionation Lines′ (TFL) that are defined by regressions through
294 Herwartz
CaCO3 high T
water Qz-water
-water min eq.
0.535 0.535
A (empirical; (empirical; B
hydroxyla on
(empirical;
upper equilibrium limit S 2016) W 2020) P&H 2014)
0.530 H2O (eq. at 11-42°C; B&L 2005) 1500 °C perido te 0.530
H₂O in gypsum - free H₂O; H 2017) 1000 °C 100 °C 1050°C
500 °C 50 °C qz-hed-mt
0.525 300 °C 0°C
0.525
proposed `MWL´ slopes

600 °C
100 °C qz-pl
low T eq. approxima on (for heavy bond partners) 0°C 620°C
0.520 diffusion of H₂O (empirical; B&L 2007)

qz-cc
CO₂-H₂O
(H 2012) qz-ep
0.520
(theory,
proposed `TFL´ slopes
λ
diffusion of H₂O in air (L 2006) 250-400 °C
θ
H 2018)
(Z 2019b)
0.515 H₂O qz-cc 0.515
(empirical,
W&S 2020)
0.510 2 0.510
high T seafloor
2
altera on
�-
0.505 0.505
�
diffusion
(Grahams law
breaking che )
dehydra on mical bonds
0.500 (C&M 2009)

0.500
0 20 40 60 80
molecular or reduced mass
Figure 1. Selected triple oxygen isotope exponents (q) for kinetic and equilibrium fractionation (Panel A)
and resulting slopes (Panel B). A) The qdiff for diffusion of molecules increases with decreasing molecular
mass, as indicated by the solid black curve labeled “diffusion” (estimated using Grahams law). The lower
limit for qdiff is 0.5 for diffusion of oxygen bound to a molecule of infinite mass. Empirical and experimen-
tal constraints for qdiff of water are 0.518 and 0.5185 (Landais et al. 2006; Barkan and Luz 2007). Breaking
chemical bonds results in low qkin especially when the oxygen bond partner is heavy, but also for dehy-
dration of serpentinite or brucite (Clayton and Mayeda 2009). The upper limit for qeq = 0.5305 at infinite
temperature (Matsuhisa et al. 1978) and equilibrium often approaches qeq = 0.5232 at low temperatures
(Dauphas and Schauble 2016; Hayles et al. 2018). Empirically determined qeq generally fall in this range
as observed for liquid–vapor (Barkan and Luz 2005); gypsum–water (Herwartz et al. 2017; Gázquez et al.
2017; Liu et al. 2019); qz–water (Sharp et al. 2018; Wostbrock et al. 2018), cc–water (Wostbrock et al.
2020b) and high T mineral equilibria (Tanaka and Nakamura 2013; Pack and Herwartz 2014; Sharp et al.
2016; Zakharov et al. 2019b). However, both empirical and theoretical estimates for qeq of qz–cc are clearly
lower than 0.5232 (Wostbrock and Sharp 2021, this volume; Hayles et al. 2018). The equilibrium qCO2–water
= 0.522 ± 0.002 (Hofmann et al. 2012) is lower, but within error identical to 0.5232. Empirical qeq estimates
for carbonate–water (Voarintsoa et al. 2020) are lower than theoretical qeq estimates for carbonate–water
(Cao and Liu 2011; Hayles et al. 2018), which are in turn lower than the empirical qcarbonate–water calibra-
tion of Wostbrock et al. (2020b). These discrepancies are either related to kinetic effects during carbonate
precipitation or to analytical artefacts. B) The effective TFL or MWL slopes (λ) vary depending on the
sample suite used to define them. Therefore, respective λ values have restricted physical meaning. Note
that effective λ can have values far outside the theoretical range as exemplified here for high T seafloor
alteration with λ ≈ 0.51 (Pack and Herwartz 2014; Sengupta and Pack 2018; Zakharov et al. 2019b) and for
hydroxylation λ ≈ 0.535 explained in the main text.
A 8 B
0.52
)= reference slope (λ) = 0.528
∆‘ O [not to scale]
δ‘¹⁷O [not to scale]
λ
e( Δ¹⁷O or
s lop ¹⁷Oexcess
nce
ere B
ref
1000ln17αeq
θ eq Δ¹⁷O or
θeq
18
α eq = ¹⁷Oexcess
/ln
ln α
17
eq
1000ln18αeq
A
δ‘¹⁸O [not to scale] δ‘¹⁸O [not to scale]
Figure 2. Isotope fractionation in triple isotope space. Materials A and B are in isotopic equilibrium with
a triple isotope exponent q lower than 0.528. Because the triple oxygen isotopic composition of A falls on
the reference line (∆′17O = 0), substance B comprises negative ∆′17O. The x and y axis in panel A are not to
scale, as otherwise the deviation from the reference slope (stippled line) would not be visible. In the d′18O
vs. ∆′17O reference frame (panel B), which is commonly used to better visualize the small offsets from a ref-
erence slope in triple oxygen isotope space, the reference slope from panel A is a horizontal line (∆′17O = 0).
unrelated terrestrial rocks often comprise values around 0.5245 (Pack et al. 2007; Rumble et
al. 2007). These TFL slopes typically include quartz to define the high d′18O region of TFL
slopes, hence the numerical similarity of some TFL slopes and the empirical qSiO2–water at low
T is not a coincidence. Apart from this, the slopes of these suites of unrelated rock samples
(TFL′s) have no meaning (Matsuhisa et al. 1978; Pack and Herwartz 2014, 2015).
Typical compositions of silicate rocks comprise negative ∆′17O and are depleted relative
to seawater ∆′17O = −5 ppm (i.e., −0.005‰) and meteoric water that cluster around the global
meteoric water line, termed MWL with positive ∆′17O (Luz and Barkan 2010; Tanaka and
Nakamura 2013). Global compilations of meteoric water actually reveal curved relationships
(Li et al. 2015; Sharp et al. 2018), hence the variation in published MWL slopes does not come
unexpected and individual regression values are of limited significance.
A common MWL slope value of 0.528 is used as a reference slope to define 17Oexcess in the
hydrological literature (Passey and Levin 2021, this volume; Surma et al. 2021, this volume).
When dealing with water–rock interaction, it is convenient if 17Oexcess = ∆′17O. Hence, it makes
sense to use the same reference slope (λ = 0.528 is always used for 17Oexcess) to define ∆′17O. Both
water and rocks should be reported on SMOW-SLAP scale (Schoenemann et al. 2013; Sharp
and Wostbrock 2021, this volume). Silicate samples are, however, measured relative to rock
standards (San Carlos Olivine, NBS28, UWG, etc.) and respective calibrations to SMOW-SLAP
scale differ by about 24 ppm on average (Pack et al. 2016; Wostbrock et al. 2020a). Here, I use a
mean estimate (Table 1) that agrees well with more comprehensive data compilations published
in this volume (Miller and Pack 2020; Sharp and Wostbrock 2020). As demonstrated below, the
choice of the scale can critically affect interpretations of water–rock interaction studies.
The intercalibration of two non-identical materials (e.g., silicates vs. water) is challenging
because non-identical methods are required for their analyses resulting in non-identical blanks
and matrix effects. Even for the same material cross contamination, pressure baseline effects
and effects tied to the design of mass spectrometers can result in scale distortion between
laboratories that can be easily corrected if two standard materials are routinely analyzed
(Schoenemann et al. 2013; Yeung et al. 2018). Miller et al. (2020) now provides two rock
standards namely SKFS (Stevns Klint Flint Standard with d18O = 33.8‰) and KRS (Khitostrov
Rock Standard with d18O = –25.02‰) that are far removed from each other and are thus
ideal for scaling (most SMOW-SLAP calibrated KRS and SKFS data is given in Sharp and
Wostbrock 2021, this volume). The same concept can be used for carbonates e.g., by using
the recent calibration of NBS18 and NBS19 (Wostbrock et al. 2020a). Data reported herein is
mostly SMOW-SLAP scaled by applying shifting and scaling factors of Pack et al. (2016) to
older data from Göttingen. Ilya Bindeman kindly provided KRS and SKFS analyses that I used
to bring all published data from their lab to the same scale (Table 1).
Mass dependent isotope fractionation in triple oxygen isotope space
Both 18aeq and 17aeq decrease with increasing temperature, while qeq increases with increasing
temperature. Therefore, equilibrated phases fall on the respective equilibrium curve in triple
oxygen isotope space (Fig. 3). Deviation from this “equilibrium curve” indicates disequilibrium.
Equilibrium fractionation factors generally range between q = 0.5232 and 0.5305
(Matsuhisa et al. 1978; Young et al. 2002; Cao and Liu 2011; Dauphas and Schauble 2016;
Hayles et al. 2016) The upper limit for equilibrium fractionation at infinite temperature is
q = 0.53052 and defined by:
1 1

m m2 (6)
 1
1 1

m1 m3
296 Herwartz
where the m1 (15.9949146), m2 (16.9991315) and m3 (17.9991604) correspond to the isotopic

masses of 16O, 17O and 18O. For heavy bond partners, the low T approximation for q often
approaches:
1 1

m1 m2
 (7)
1 1

m1 m3
and thus 0.52316 (Hayles et al. 2016; Dauphas and Schauble 2016). The grey field in Figure 1
covers most of the qeq values that are expected to occur in nature, but the lower boundary is
not strict and there are equilibrium q such as the qz–cc equilibrium (Wostbrock and Sharp
2021, this volume) that are clearly lower than predicted by Equation (7). For materials that
mainly comprise hydrogen as a bonding partner to oxygen qeq are generally high (Dauphas and
Schauble 2016). The high vibrational frequencies of O–H bonds lead to high q for water–vapor
equilibrium even at low temperatures as confirmed empirically for water qliquid–vapor = 0.529 in
the 11 to 42°C temperature range (Barkan and Luz 2005). For very small aA-B, apparent q far
outside the theoretical limits of 0.5 to 0.5305 may apply (Bao et al. 2015; Hayles et al. 2016).
A 28 B
equilibrium 0.5
∆‘ O [not to scale]
δ‘¹⁷O [not to scale]
curve 500°C 300°C

0°C
50°C
B 100°C 0.528
°C)
1000ln17αeq
(at 0
100°C
50°C
θe
α eq =
q
300°C 18
ln
500°C
17
ln α eq/
0°C
1000ln18αeq
A
δ‘¹⁸O [not to scale] δ‘¹⁸O [not to scale]
Figure 3. Equilibrium isotope fractionation between substances A and B at variable temperatures. Due to
the small temperature dependency of q, equilibrium in triple oxygen isotope space is represented by the
black curves in both panels.
Graham′s law provides a first order approximation for diffusion (Young et al. 2002). The
relative velocities of diffusing species are directly proportional to their molecular mass so that:
m 
ln  1 
m
diff   2 (8)
 m1 
ln  
 m3 
where m1, m2 and m3 correspond to the isotopologues comprising 16O, 17O and 18O respectively,
giving qdiff = 0.5142 for individual H2O molecules (Fig. 1). The details of diffusion are more
complex leading to slightly higher theoretical qdiff = 0.518 for diffusion in air, which is very
close to the empirical estimate of qdiff = 0.5185 (Barkan and Luz 2007; Landais et al. 2006).
Other kinetic processes such as the preferential breaking of bonds comprising light isotopes
scale with the reduced mass of the bond partners so that qkin for bond breaking differs from
qdiff (Fig. 1). Beyond Grahams law, the prediction of qkin can be challenging. Transition state
theory considers that chemical reactions generally proceed via transitional states (Bigeleisen
and Wolfsberg 1958). Consider a phase A with a transition state A‡ that reacts to phase B.
A  A‡  B (9)
Table 1. Reference frame: A wide range of studies calibrate rock standards to SMOW but absolute values differ
beyond analytical precision, including recent calibrations of silicate standards and air O2 on a SMOW-SLAP
scale (Pack et al. 2016; Wostbrock et al. 2020a). Here I use an intermediate scale calculated by simply adding
or subtracting 12 ppm (the average offset between SCO, NBS28 and UWG is ca. 24 ppm) to standard data of
Pack et al. (2016) and Wostbrok et al. (2020a). The mean values compare well to previous calibrations (e.g.,
SCO from Sharp et al. 2016) and considerably reduces the offset for AIR O2 between the two labs. These
average values are almost identical to the more comprehensive compilation by Sharp and Wostbrock (2020) in
this volume. I recommend using the later in future studies.
Material/Reference Shifted by: d17O SD d18O SD ∆′17O (0.528) SD
SC Olivine
Pack et al. 2016 (GZG) d′17O − 0.012 2.668 5.152 −49
Pack et al. 2016 (ISI) d′17O − 0.012 2.737 5.287 −51
Sharp et al. 2016 d′ O
17
2.745 5.310 −55
Wostbrock et al. 2020a d′17O + 0.012 2.732 5.268 −46
mean 2.72 0.04 5.25 0.07 −50 4
mean in S&W 2020 2.75 0.08 5.32 0.16 −52 14
UWG−2
Miller et al. 2019 (GZG) d′17O − 0.012 2.974 5.750 −58
Miller et al. 2019 (OU) d′ O − 0.012
17
2.974 5.750 −58
mean 2.96 0.02 5.73 0.03 −58 1
mean in S&W 2020 2.94 0.08 5.70 −64
NBS28
Miller et al. 2019 (OU) d′ O − 0.012
17
4.997 9.555 −37
Pack et al. 2017 d′ O − 0.012
17
5.074 9.712 −42
mean 5.00 0.06 9.73 0.11 −45 7
mean in S&W 2020 4.99 9.57 −49
KRS
Miller et al. 2019 (GZG) d′17O − 0.012 −13.307 −24.899 −83
Miller et al. 2019 (OU) d′17O − 0.012 −13.465 −25.200 −80
mean −13.386 0.112 −25.050 0.213 −82 2
mean in S&W 2020 −13.381 −25.023 −91
SKFS
Miller et al. 2019 (OU) d′ O − 0.012
17
17.646 33.932 −127
mean 17.528 0.167 33.705 0.322 −127 0
mean in S&W 2020 17.556 33.778 −137
AIR
Pack et al. 2017 − 12ppm d′17O − 0.012 12.253 24.150 −421
Wostbrock et al. 2020a 12ppm d′17O + 0.012 12.190 24.046 −430
mean 12.221 0.044 24.098 0.074 −425 6
mean in S&W 2020 12.204 0.060 24.077 0.070 −432 24
298 Herwartz
Because transition states (A‡) are generally not observed and cannot be easily identified such
a reaction may appear like a purely kinetic reaction from A to B. The combination of the
two individual reactions with variable a′s, however, can produce effective qeff outside the
theoretical range. Most kinetic isotope reactions will be combinations of various physical
processes resulting in variable approximations for isotope effects (Melander and Saunders
1980). Accurate predictions of qkin may serve as a tool to study the underlying physics of
kinetic effects (Cao and Bao 2017; Yeung and Hayles 2021, this volume).
The term “kinetic isotope effect” is also frequently used to say that equilibrium is not
attained, but the underlying processes remain unclear. Using carbonate formation at high
pH and low temperature (25 °C), I will exemplify how a “kinetic isotope effect” can be a
combination of mixing, diffusion and equilibration resulting in an effective slope (λ) in triple
oxygen isotope space that is outside the theoretical range for q (Fig. 4). Consider hydroxylation
(CO2 + OH− → HCO3−) as the rate limiting step for the subsequent precipitation of calcite. In
equilibrium with water, the two reactants have vastly different oxygen isotopic compositions
with d18OOH- = −38.4‰ at 25°C (Green and Taube 1963) and d18OCO2 ≈ + 40.5‰ (Beck et al.
2005). A simple mass balance with d18OHCO3- = 2/3 d18OCO2 + 1/3 d18OOH- approximates the
overall isotope effect in d18O for hydroxylation (Clark et al. 1992; Dietzel et al. 1992) resulting in
d18OHCO3- = 14.2‰. In triple oxygen isotope space, this mixing product (HCO3−) falls on a mixing
curve between CO2 and OH− resulting in a very low D′17OHCO3- of −286 ppm (Fig. 4B). Due to a
superimposed kinetic isotope effect (Devriendt et al. 2017; Sade et al. 2020), the actual kinetic
CaCO3 hydroxylation endmember has a d18OCaCO3 ≈ 10‰ (Böttcher et al. 2018), so that the
expected D′17OCaCO3 ≈ −227 ppm (using qKIE = 0.5143 for a 4‰ kinetic effect). The mixing curve
between this kinetic CaCO3 hydroxylation endmember and CaCO3 in full equilibrium defines an
apparent kinetic fractionation curve with a slope λ ≈ 0.535 (green arrow labeled “hydroxylation”
in Fig. 4), even higher than the upper theoretical limit for equilibrium fractionation.
Using a numerical approach, Guo and Zhou (2019) show how disequilibrium causes departure
from equilibrium in all directions depending on the underlying process(es) These perturbations
of equilibrium actually form tight loops. The respective kinetic qkin are not generally restricted to
low values as implied by Figure 1. For a combination of processes (e.g., equilibrium + kinetic +
mixing in the above example), the notation lkin may be more appropriate than qkin.
Disequilibrium is common in carbonates (Coplen 2007; Daëron et al. 2019; Bajnai et al.
2020). Carbonate thermometry only works because the empirical d18O – T calibrations for a
given carbonate material include “kinetic effects” (often termed “vital effects” for biogenic
carbonates) as constant offsets from true equilibrium (Coplen 2007). Most Earth surface
carbonates precipitate out of equilibrium (Daëron et al. 2019), hence disequilibrium is also
expected at elevated temperatures especially when precipitation rates or pH are high (Guo
and Zhou 2019). High precision carbonate D′17O analyses are challenging, however, and first
applications are only beginning to emerge for low T precipitates (Bergel et al. 2020; Wostbrock
and Sharp 2021, this volume; Passey and Levin 2021, this volume).
Mixing
Mixing occurs at all scales. Consider: 1) the above example of hydroxylation for
mixing on the molecular scale both for “CO2 + OH−” and “CaCO3 in eq. + kinetic CaCO3
endmember”; 2) mixing of evaporated water with unevaporated groundwater or precipitation
in desert lake systems (Herwartz et al. 2017; Voigt et al. 2020); 3) intra mineral mixing of
phytolith SiO2 precipitated in leaves from water with variably evaporitic compositions
(Alexandre et al. 2019); 4) intra mineral mixing of two quartz generations formed at opposing
temperatures and water to rock ratios (Zakharov and Bindeman 2019; Zakharov et al. 2019a);
0
typical SD A
(8 ppm)
Diff
us
-50
n io
∆‘17OVSMOW [ppm]
100°C
CaCO₃
CaCO 75°C
(eq at 25°C)
₃ eqil g
ibriu sin
m lin s
e 50°C ga
-100 H₂O-C
O₂ m 40°C ₂ de
ixing
line
CO₃2- CO
20°C
CO₂ (aq)
H2CO₃(Hydra on) HCO₃-
on
-150 line ox
yla 0°C
ing ydr
) mix H
(eq ne
C O₃ g li
- Ca ixin
on) CO
₂m
yla H-
dox
-
O ₃ (hy O
-200 CaC
15 20 25 30 35 40
δ‘18OVSMOW [‰]
50
H2O B
0
= 0.53 A
θ H2O- O H-
m
-50 OH ixi
-
45°C ng
25°C 35°C lin
e
15°C
5°C
-100 CO₃
2-
CO₂(aq)
-150 HCO₃-
ine
gl
xin
mi
-200 m
ixi
ng CaCO₃
lin (hydroxyla on)
e
-250
kin
e c
-300 HCO₃- (CO₂ + OH- mass balance)
-350
-50 -40 -30 -20 -10 0 10 20 30 40 50
δ OVSMOW [‰]
‘18
Figure 4. Illustration of predicted effective slopes in triple oxygen isotope space (i.e., d′18O vs. D′17O). In
this reference space mixing lines fall on curves, while isotope fractionation processes are linearized. Panel
A: The stippled orange line represents equilibrium fractionation of CaCO3 at the given temperatures. This
line and the equilibrium values for CO32−, HCO3−, CO2 (aq) at 25°C are estimated from published d18O vs. T
(Kim and O′Neil 1997; Beck et al. 2005) and q vs. T (Cao and Liu 2011) relationships. Kinetic effects drive
samples away from this equilibrium curve. Black and grey arrows, for example, depict expected slopes
for diffusion of CO32− or HCO3− (qdiff = 0.5052; solid gray) and CO2 (qdiff = 0.5066; black) using Grahams
law; and for diffusion of CO2 in air (qdiff = 0.5098; stippled gray) using the equations given in Landais et
al. (2006). The vector for CO2 degassing (thick black arrow) is taken form Gou and Zhou (2019). Natural
samples affected by hydration or hydroxylation are expected to fall on mixing lines between fully equili-
brated CaCO3 endmember and the respective kinetic endmembers (see Panel B and text for details). For
hydration, the mass balance is: d18OH2CO3 = 2/3 d18OCO2 + 1/3 d18OH2O. At 25°C the d18OH2CO3 of this mixture
is 27‰. The quantum mechanical estimates for the superimposed KIE of hydration (Zeebe 2014) is on the
order of 1‰ or less relative to this mass balance (Devriendt et al. 2017). Quantitatively transformation to
CaCO3 results in a hydration endmember with only 2.5 per mill lower d18O CaCO3 ≈ 26‰ than the (presumed)
equilibrium value of d18OCaCO3 = 28.5 at 25°C (Kim and O′Neil 1997). The D17OCaCO3 would be = −130 ppm
and the slope ≈ 0.533 (short blue arrow in panel A resembling mixing between “H2CO3 hydration” and
“CaCO3 eq. at 25°C”). The mass balance for hydroxylation is outlined in the main text (and illustrated in
Panel B). The estimates for triple oxygen isotope slopes critically rely on the accuracy of the oxygen isotope
fractionation factors. Using different literature values (e.g., Hofmann et al. 2012 for qCO2–water; Hayles et al.
2018 or Wostbrock et al. 2020b for qCaCO3–water; or Coplen et al. 2007 for 1000 ln18aCaCO3–water) will substan-
tially change estimated slopes e.g., for hydroxylation. This is a simplified conceptional figure.
300 Herwartz
5) mixing on the intra sample scale for metamorphic BIF′s (Levin et al. 2014); 6) intra sample
mixing of pristine MORB and the alteration product clay (Pack and Herwartz 2014); 7)
assimilation of altered rocks in magma chambers (Zakharov 2019b; Peters et al. 2020a); and
8) water–rock interaction (Herwartz et al. 2015). All of these mixing processes result in low
D′17O isotopic compositions in the mixing products. The further apart the mixing endmembers
are in d18O, the lower the resulting D′17O.
On a rock scale, minerals with contrasting isotopic compositions can isotopically mix and
re-equilibrate. In a conceptual model, Levin et al. (2014) envisioned the precipitation of quartz
and an Fe phase from ocean water at low temperature. Due to the small aeq for FeOx–water,
the respective D′17OFeOx is always close to that of the initial water (whatever the q; Fig. 5),
while the D′17Oquartz = −132 ppm for precipitation at 25°C (with qSiO2–water = 0.5243; Sharp et al.
2016; Wostbrock et al. 2018). When this bi-mineralic rock is metamorphosed (with T > 600 °C),
quartz and the iron phase re-equilibrate to high T conditions. The final D′17O compositions
fall close to the molar mean of the system, which lies on a concave mixing curve (Fig. 5).
The model elegantly explains the apparently “too low” D′17O of some (metamorphosed) banded
iron formations (Levin et al. 2014) as well as the bulk composition of a metamorphosed BIF
where the mineral separates (qz, hed, mt) re-equilibrated to a final qqz–hed–mt = 0.53 at 600 °C
(Pack and Herwartz 2014).
50
typical error
Fe (8 ppm)
phase sw
0 500
250
∆‘ O [ppm]
θeq
-50 = 0.
524
3
100
75
qz
θeq = 0.53
50
-100 hed
mt
mixin
g line 25
(bulk
)
qz
-150
0
-200
-5 0 5 10 15 20 25 30 35 40 45
δ¹⁸O
Figure 5. Model of intra-sample re-equilibration modified after Levin et al. (2014). Metamorphic BIF
data from Pack and Herwartz (2014) is recalibrated to SMOW-SLAP scale. Numbers close to tick marks
represent equilibrium temperatures for the qz–water system (Sharp et al. 2016).
Silicate rocks generally comprise positive d18O and negative D′17O, distinct from surface
water leading to resolvable mixing relationships in hydrothermal settings (Fig. 6). The
pristine water mixing endmember can be constrained even after the altered rocks have been
metamorphosed (Herwartz et al. 2015). Hydrothermally altered rocks, as well as sediments such
as carbonates, cherts, or clays comprise typical triple oxygen isotope compositions that can be
traced when involved in metamorphic or magmatic processes (Peters et al. 2020a) ultimately
leading to small heterogeneities in D′17O expected between mantle melts (Cao et al. 2019).
The guiding principles outlined above allow: 1) to identify disequilibrium; 2) to distinguish
between various types of kinetic effects; and 3) to identify mixing processes. Here, I focus on
publications that apply these principals to solve geological problems.
pris ne altered
1000lnαrock-water water (δ�water) endmember
MWL
sw 0.528
pris ne
∆‘ O final
altered rock (δ�rock)
rock (δ�rock)
water (δ�water) Earth crust
mixing lines
δ¹⁸O
Figure 6. Mixing lines between crustal rocks (gray shaded area) and low d18O meteoric water (MWL)
represent curves. The bulk triple oxygen isotopic composition of altered rock samples represents a mixture
between these two reservoirs and falls on respective mixing lines. Details of this approach are explained
in the main text.
QUANTIFYING FRACTIONATION FACTORS

As for the classic 18aA-B, temperature dependent fractionation factors for 17aA-B and thus
qA-B can be determined via modelling, by analyzing natural materials with known formation
temperatures or from laboratory experiments. Equilibration timescales in experiments are
always short compared to geological timescales. The two directional approach (Clayton et
al. 1972) and the three-isotope method (Matsuhisa et al. 1978; Matthews et al. 1983) permit
estimating equilibrium fractionation factors from unequilibrated batch type experiments,
circumventing the problem of long timescales.
The two directional approach
The substances of interest are enclosed in sealed containers and left to equilibrate for
variable periods of time, so that a time-series curve approaching equilibrium can be constructed
for batches with identical starting composition. One experimental series comprises two or
three combinations of starting compositions so that equilibrium is approached from two sides.
Ideally, the respective curves converge to a single value representing equilibrium (Fig. 7A).
20 expecta on
A 20 observa on
B
1000ln18α
1000ln18α
10 9.04 10 9.04
6.1
0 0
-10 -10
0 25 100 225 0 25 100 225
me me
Figure 7. Two directional approach. A) In the idealized case, individual experiments fall on exponential
curves that merge at the equilibrium 1000 ln a. Grey and black dots represent expected values for equili-
bration between SiO2 and water. B) In the real case example (with cristobalite as initial SiO2 phase), the
measured 1000 ln a fall on complex trajectories that are inconsistent with a gradual approach towards
equilibrium (Matsuhisa et al. 1978).
302 Herwartz
It was soon realized that there can be complications with this approach. Clayton et al. (1972)
warned that extrapolation can generate fractionation factors larger than the equilibrium value
by 1 to 2‰, while Matsuhisa et al. (1978) observed fractionation factors 3‰ too low (Fig. 7B).
Long standing inconsistencies between laboratories or between experiments and theory (Zheng
1993) reveal a certain degree of complexity within these apparently simple experiments. In
principal, triple oxygen isotope systematics can now be used to test if experiments reached near
equilibrium or not because theoretical qA-B estimates are more robust than estimates for classic
i
aA-B (Cao and Liu 2011; Hayles et al. 2018). In fact, a variety of this idea has been introduced
long ago and is known as the “three-isotope method” (Matsuhisa et al. 1978).
The three-isotope method
The method requires two starting materials with different triple oxygen isotope compositions.
Matsuhisa et al. (1978) produced water with extraordinary low ∆17O by: 1) evaporating water
to a point where the residual is extremely enriched with d18O = 6100‰; and 2) subsequent
mixing with normal distilled water to final d18O compositions of 14.4 and 27.66‰ and very
low ∆17O of −5 and −7‰. Equal molar proportions of this low ∆17O water and fine grained
SiO2 (quartz or cristobalite) are mixed and left to equilibrate at 250°C and 15 kbar in sealed
platinum capsules. At full equilibrium quartz and water must fall on the same mass dependent
fractionation line, that passes through the mass balance point of the closed system, the so called
“secondary fractionation line”. The original paper uses λ = 0.52 in d18O vs. d17O space, but for a
more exact treatment 0.527 (qqz–H2O at 250 °C; Sharp et al. 2016) should be used in d′17O vs. d′18O
coordinates (Fig. 8). Even with only one single experiment it is possible to extrapolate from the
initial isotopic compositions and the partially exchanged compositions to the final equilibrium
value (Fig. 8A). Therefore, the triple isotope method is much less labor intensive than the two
directional approach. The original result for 1000 ln 18aSiO2–water = 9.03‰ is remarkably consistent
with the latest compilation of 1000 ln 18aSiO2–water = 9.04‰ at 250 °C (Sharp et al. 2016).
Even though the approach appears to be extremely elegant, Matsuhisa et al. (1978)
demonstrate some limitations in the very same paper. In a second equilibration experiment, that
utilized cristobalite instead of quartz as the initial SiO2 phase, the measured quartz compositions
cross the secondary equilibration line at apparently lower 1000 ln 18aSiO2–waterr ≈ 6.1‰, overstep
to apparently too low ∆′17Oquartz < ∆′17Omolar mean and then slowly approach the equilibrium
ini al quartz
(t = 0) me
A -1
ini al quartz
(t = 0) m
(ex ixing
B
-1 m
(o ixin
pe lin
water quartz
-2 bs g l
cte e
d)
er in quartz in
ve e
-2 in equilibrium in equilibrium -3 d) equilibrium
(t = ∞) mean θ = 0.527 (t = ∞) me (t = ∞)
-4
∆‘ �O
∆‘ �O
eq
θeq = 0.527
-3 water in mean
∆ �O � 1000ln18α = 9.03 -5 equilibrium me
-4 -6 (t = ∞)
me -7
-5 ini al water αkin
(t = 0) -8 ini al water αeff = 6.1
18
(t = 0) first quartz prec.
5 10 15 20 15 20 25 30 35 40
δ‘¹⁸O δ‘¹⁸O
Figure 8. Three-isotope method. A) Idealized case with data from Experiment 1 of Matsuhisa et al.
(1978). The secondary mass fractionation line with θSiO2–water ≈ 0.527 (Sharp et al. 2016; Wostbrock et al.
2018) passes through the mean isotopic composition (black diamond). Water samples (circles) and quartz
(diamonds) evolve towards their fully equilibrated endmembers (black circle and diamond) that can be
approximated by linear regression (stippled lines). Curvature due to mixing is insignificant at this scale.
B) Real case example with data from experiment 2 of Matsuhisa et al. (1978) using cristobalite instead of
quartz as starting material. Symbols are equal to panel A. The white diamond represents the composition
of initial quartz if precipitated in equilibrium (with 1000ln18αqz–water = 9.03) with initial water (not sup-
ported by the data), while the light gray diamond indicates the composition of initial quartz if precipitated
with a superimposed kinetic effect from initial water (supported by the data).
composition (Fig. 8B). At equimolar proportions of SiO2 and water, only a small fraction of the
cristobalite can dissolve, equilibrate with the water and re-precipitate as quartz (Bottinga and
Javoy 1987). Therefore, the quartz formed at an early stage inherits a low ∆′17O similar to the
initial water (white diamond in Fig. 8B) resulting in the apparent overstepping of the measured
quartz samples. The expected mixing line between such an early precipitate and the initial
cristobalite crosses the secondary fractionation line at the equilibrium composition. Hence, low
D′17O quartz alone does not explain why regressions do not always pass through the equilibrium
composition. About 50% of the cristobalite is transformed to quartz within the first three minutes
(shortest experiment) and such fast reaction kinetics always hold potential for kinetic effects.
The effective fractionation factor of 1000 ln aeff = 6.1‰ observed for the early experiment (as in
the two directional approach; Fig. 7), shows that the initial SiO2 reaction product did not form
in equilibrium (Fig. 8). The initial “kinetic” quartz precipitates (light gray diamond in Fig. 8B)
drive samples away from the expected re-equilibration trend. Samples from long experimental
runs slowly approach full equilibrium due to slow re-equilibration of the initial quartz precipitates
(also see Appendix 2 of Matsuhisa et al. 1978). These results show how even a seemingly simple
batch experiment can result in inaccurate fractionation factor estimates if the underlying physics
and transport mechanisms are not accounted for (Bottinga and Javoy 1987).
The three-isotope method is applied for oxygen and a range of other three-isotope systems
(Matthews et al. 1983; Shahar et al. 2008, 2009; Beard et al. 2010; Lazar et al. 2012; Macris
et al. 2013). A rigorous mathematical investigation (Cao and Bao 2017) using kinetic theory
(Biegeleisen and Wolfsberg 1958) shows that accurate aeq can only be determined if appropriate
initial isotope compositions are chosen (Matthews et al. 1983) that minimize potential artefacts
of the method. The main point of Cao and Bao (2017), however, is to show that re-equilibration
trajectories do not necessarily follow linear slopes and that the form of these trajectories can
be used to quantify kinetic isotope effects as crudely outlined above. Hence, the triple oxygen
isotope method holds a greater potential than previously realized, providing a tool to tackle
the complexity associated to open systems (Druhan et al. 2019), multiple reservoir exchange
systems (Beard et al. 2010) and dissolution/re-precipitation (Putnis 2015).
The example of intra-rock mixing above (Fig. 5) demonstrates that the three-isotope
method can now be used in natural systems. At the present analytical precision, the natural
variability in initial ∆′17O between two materials is often sufficient to apply the concept.
Disequilibrium “frozen in” in multi component systems may eventually prove valuable to
better understand and quantify the underlying kinetics (Cao and Bao 2017), which will be
useful for the traditional concepts of “geothermometry” and “geospeedometry” (Clayton and
Epstein 1961; Garlick and Epstein 1967; Dodson 1973; Giletti 1986).
Isotope thermometry and geospeedometry
The temperature dependence of equilibrium fractionation factors provide the basis to use
coexisting minerals in magmatic and metamorphic rocks as geo-thermometers (Clayton and
Epstein 1961; Garlick and Epstein 1967). Some mineral pairs, such as quartz and magnetite
comprise relatively large 18aqz–mt fractionation factors even at high T, due to the large difference
in oxygen bond strength. The quartz–magnetite pair typically implies isotopic temperatures close
to 600 °C for rocks that cooled fast, while at slow cooling rates the oxygen isotopes of the mineral
pair can equilibrate to lower temperatures of around 550 °C (Dodson 1973). This concept of
geospeedometry (Giletti 1986) can be applied to multi mineral assemblages with variable closure
temperatures providing insight into exhumation and cooling of metamorphic rocks.
Some minerals are more susceptible to alteration than others, potentially leading to
meaningless temperature estimates or exhumation rates. Alteration can be identified by comparing
apparent equilibration (or closure) temperatures between several mineral pairs that should remain
in feasible bounds for the presumed cooling rate of a simple p–T path. Aqueous fluids, however,
304 Herwartz
are known to enhance diffusion rates (Elphick and Graham 1988; Kohn 1999) making it challenging
to decide if a given mineral pair is in equilibrium and thus if the isotopic temperatures are feasible.
Late infiltration of water could either alter one individual mineral phase (e.g., magnetite or
plagioclase) or facilitate equilibration between all phases at relatively low temperature.
By using triple oxygen isotope systematics, it can be tested if qeq falls in the expected
range (Fig. 1). Cano et al. (2020) argues that mineral pairs of lunar rocks are not in equilibrium
because the respective qapp ranging from 0.47 to 0.56 are inconsistent with equilibrium.
The respective range for terrestrial rocks (0.522 to 0.53) is generally more feasible, although
the lowermost value is suspiciously low. The temperature determined by 18aA-B dictates a
given qeq typically >0.526 for high T. If this theoretical triple oxygen isotope exponent is not
observed (within error), the mineral pair is not in equilibrium. Identifying disequilibrium for
the hydrothermal quartz–epidote mineral pair (qeq = 0.526; Zakharov et al. 2019b) is outlined
in more detail below, after introducing the basics of water–rock interaction.
WATER–ROCK INTERACTION
Hydrothermal alteration dominantly occurs in areas where hot magmatic rocks are cooled
by the convective flow of fluids (Baumgartner and Valley 2001; Shanks 2001). Seawater
penetrates the oceanic crust at the mid ocean ridges, resulting in large scale isotope exchange
between the two reservoirs (Muehlenbachs and Clayton 1976; Sengupta and Pack 2018).
Meteoric water with low d18O can imprint its isotopic composition on to rocks via fluid–rock
interaction (Taylor 1978; Criss and Taylor 1983; Pope et al. 2014; Zakharov et al. 2019b).
At high temperatures, the water to rock ratio (W/R) dictates if the isotopic composition of the
rock is driven towards that of the water (high W/R) or if the water adjusts to the composition of
the rock (low W/R). For magmatic and metamorphic waters, the rock usually governs the mass
balance of the system, while in hydrothermal systems the constant flux of water dominates
mass balance. In its simplest form, the W/R ratio is estimated via the mass balance of the mole
fractions of the oxygen reservoirs in a closed system (Taylor 1978).
X rock  irock  1  X rock   iwater X rock  frock  1  X rock  fwater (10)

where Xrock is the mole fraction of oxygen in the rock and i and f stand for the initial and final
compositions respectively. The W/R ratio is then:
W frock   irock
 i (11)
R water  ( rock  1000 ln  rock  water )
f
This equation applies to a batch system which is rarely true for hydrothermal systems.
Therefore, “open system W/R ratios” are sometimes used:
W  W  
 ln     1 (12)
R  
  R closed system 
Because the W/R ratio only accounts for the water that has actually reacted with the
rock the meaning of W/R is not straight forward (see Baumgartner and Valley 2001 for a
discussion). A full understanding of the transport and transformation rates and mechanisms
is necessary (Putnis 2015; Druhan et al. 2019) to apply multi-dimensional transport
models, which can be necessary to adequately model real world examples of water–rock
exchange (Baumgartner and Valley 2001). In addition, other fluids than water are also
relevant as addressed later in this chapter for CO2 and SO2. However, despite its simplicity,
Equation (11) represents a good first order approximation to understand water–rock interaction.
To date, all 17O studies on high temperature water–rock interaction are exclusively using this
original definition (Herwartz et al. 2015; Zakharov et al. 2017, 2019a, b, 2021, this volume;
Wostbrock et al. 2018; Sengupta and Pack 2018; Zakharov and Bindeman 2019; Wostbrock
and Sharp 2021, this volume; Peters et al. 2020a, b; Chamberlain et al. 2020).
Combined d18O and dD isotope systematics are traditionally used to study water–rock
interaction. Rocks comprise small mole fractions of hydrogen compared to oxygen, thus
small quantities of water can significantly alter the dD, while d18O only changes at elevated
W/R. Alteration trajectories in d18O vs. dD space are therefore strongly curved (Taylor 1978).
In hydrothermal systems the aim is to resolve the intertwined effects of boiling,
liquid–vapor phase separation, mixing and water–rock exchange at variable water–rock ratios
and temperature (Shanks 2001; Pope et al. 2014). Additional d17O analyses help to resolve the
individual processes and to extract information such as the initial fluid isotope composition
of ancient seawater (Sengupta and Pack 2018; Zakharov and Bindeman 2019; Liljestrand et
al. 2020; Peters er al. 2020b) or paleo-meteoric water providing insight into the respective
hydrological cycles and thus climate (Herwartz et al. 2015; Zakharov et al. 2017, 2019a;
Chamberlain et al. 2020). In the following section, I will show how the required parameters
for the mass balance model can be constrained and how individual processes can be identified.
The respective concepts could also be used to better understand contact metamorphism
(Baumgartner and Valley 2001) or the genesis of mineral deposits (Shanks 2014), or any other
field that presently uses the traditional d18O and dD isotope systematics.
Constraining input variables for the water–rock mass balance equation
For most hydrothermally altered rocks (with frock ) there are country rocks that have
escaped water–rock interaction. Analyses of these pristine rocks, or average compositions for
this rock type (e.g., MORB) closely resembles the initial rock isotopic composition (irock ) of
the altered specimen.
The isotope fractionation between the bulk rock and water 1000 ln arock–water depends on its
respective mineralogy and the temperature of water–rock interaction. For mixed mineralogy
or for ancient samples, where the hydrothermal mineralogy is not preserved, typical bulk rock
1000 ln arock–water ≈ 1–3‰ are reasonable at typical hydrothermal temperatures and the d17O
can be approximated via a high qrock–water around 0.526–0.53. For a preserved hydrothermal
mineralogy, individual mineral fractionation factors can be used. Epidote for instance, typically
forms at 250–400 °C with 1000 ln 18aepidote–H2O ≈ 0‰ at 400 °C and 1.5‰ at 250 °C (Zheng 1993;
Bird and Spieler 2004). Such small fractionation factors allow to directly trace the final water
isotopic composition (fwater) in equilibrium with the altered rock ( frock ) (Zakharov et al. 2019b).
The quartz–water fractionation at hydrothermal temperatures (1000 ln aqz–H2O) is
several permill, hence the large fractionation between quartz and epidote can be utilized as
geo-thermometer (Matthews 1994) that is accurate if the mineral pair forms in equilibrium.
While equilibrium is generally observed (Zakharov et al. 2019b), disequilibrium is also
expected to occur, simply because quartz forms over a much wider range of temperatures than
epidote. If temperatures drop below the typical epidote stability field of > 250 °C, the driving
force for water circulation (heat) decreases and effective W/R ratios are expected to decrease.
Any quartz precipitated at this late stage does not form in equilibrium with epidote and
comprises low ∆′17O (Fig. 9). Intra mineral mixing (i.e., a bulk analysis) of two generations of
quartz also decreases ∆′17O (Zakharov and Bindeman 2019; Zakharov et al. 2019b) and thus
“too low” qqz–epidote values are indicative of disequilibrium. Equilibrium qepidote–H2O = qqz–H2O =
0.5278 at 400 °C because 1000 ln 18aepidote–H2O ≈ 0‰. At 250°C the qqz–H2O = 0.527 (Sharp et al.
2016) and identical within error to an empirical qqz–epidote = 0.526 ± 0.001 estimate (Zakharov
et al. 2019b). Any lower qqz–epidote reveals disequilibrium and respective temperature or fwater
estimates using quartz are inaccurate. Therefore, epidote more reliably captures fwater during
the main hydrothermal phase (Zakharov et al. 2019b).
306 Herwartz
50 MWL typical error Borehole 504B

θeq = 0.5278 (8 ppm) epidote
quartz
sw
0 400 300
250
200
Reykjanes (Iceland)
epidote
low quartz
Tq 150 mixing with mw component
z m ixin
∆‘ O [ppm]
g li
-50 water in
MORB
ne 100
eq.with 150
MORB
50
-100 mi
xin
gl
? ine
25
-150
25
-200
qz
-5 0 5 10 15 20 25 30 35 40 45
δ¹⁸O
Figure 9. Triple oxygen isotope exchange during interaction between seawater and basalt. At high W/R
ratios and hydrothermal temperatures epidote and quartz form in equilibrium with ∆′17O close to zero.
At 400 °C, epidote is isotopically identical to the water (Matthews et al. 1994), hence the qquartz-epidote =
qquartz–water = 0.5278 (Sharp et al. 2016). Quartz precipitation during cooling and at low W/R has low ∆′17O.
Intra-mineral mixing of high and low T precipitates (stippled mixing curves) drives ∆′17O of bulk quartz
to low values (thick black arrow). The apparent q would decrease. The composition of epidote from Reyk-
janes is consistent with a meteoric water component at that site (thin black arrow); “Borehole 504B” is a
marine site. The low ∆′17O is only partially explained by late quartz precipitation at low W/R. Data is from
Zakharov et al. (2019b) and Zakharov and Bindeman (2019).
Fluids from wells that penetrate active hydrothermal systems can be used to directly
constrain fwater of these high temperature systems. Water sampled from hot springs or
geothermal power plants reveal ca. 40 ppm lower ∆′17O for fwater compared to iwater (Wostbrock
et al. 2018). In case of the Hellisheiði power plant in Iceland about 20% of the water is
isotopically altered by water–rock interaction after passing through the hydrothermal pluming
system (Wostbrock et al. 2018). Seawater derived fluids sampled from hydrothermal vents
reveal that seawater-basalt reaction has little effect on dD and shifts d18Owater from near
pristine values (≈ 0‰) at high W/R to ≈ 2‰ at low W/R ratios (Shanks 2001). Respective ∆′17O
decreases from near seawater compositions (i.e., −0.005‰; Luz and Barkan 2010) to mantle
values (i.e., ≈ −0.05‰; Fig. 9) as confirmed by recent ∆′17O analyses of high T altered oceanic
crust and alteration products (Sengupta and Pack 2018). Both fluids and altered rocks evolve
along trajectories with slopes of about 0.51 (Sengupta and Pack 2018; Zakharov et al. 2019b)
that resemble mixing lines between the pristine and altered endmembers.
From d18O systematics alone it is impossible to calculate iwater because the W/R ratio is
generally unknown. The additional d17O analyses allow to extrapolate the mixing lines towards
feasible iwater compositions. Pristine fluids entering the crust (iwater ) are either seawater with dD,
d17O and d18O all close to zero, or meteoric water with compositions close to the meteoric water
line (MWL) with dD = 8 × d18O + 10 (Craig 1961; Dansgaard 1964) and d′17O = 0.528 × d′18O
+ 0.033 (Luz and Barkan 2010). Global compilations of meteoric water actually reveal curved
relationships (Li et al. 2015; Sharp et al. 2018; Surma et al. 2021, this volume), hence for
individual studies it can be useful to define local meteoric water lines (LMWL), curves or
intervals to approximate feasible combinations of dD, d17O and d18O for the pristine iwater .
Constraining the isotopic composition of pristine (paleo-) fluids

Altered samples fall on a mixing line between the pristine endmember (irock ) and
a hypothetical rock sample altered at infinite W/R, that is offset from the pristine iwater by
1000 ln iarock–water (Fig. 6). Feasible compositions of the altered endmember fall on a line on or
sub-parallel to the meteoric water line (depending on the qrock–water). Extrapolating the mixing
line defined by irock and one or more altered samples ( frock ) to this line (identical to the MWL
herein; Fig. 6) gives the composition of the altered endmember at infinite W/R. The pristine
iwater is in equilibrium with this theoretical 100% alteration endmember, hence the calculated
composition depends on the temperature of water–rock interaction.
Herwartz et al. (2015) tested this approach for samples from Krafla, Iceland and calculated
rather low pristine d18OMW = −22 ± 4‰ forcing the authors to speculate on a potential
contribution of “ice age water” from the last glacial maximum. However, it is now clear that
the ∆′17O calibration between silicate samples relative to SMOW was inaccurate and that the
scale was also rotated compared to the SMOW-SLAP scale (Pack and Herwartz 2014; Pack
et al. 2016). Using a revised calibration of San Carlos olivine relative to SMOW from Pack
et al. (2016) gives an intercept at −14‰ in perfect agreement with local meteoric water and
consistent with directly measured well fluids and epidotes from the same site (Zakharov et
al. 2019b). Herein, I also compensate for the “scale rotation” by recalibrating all data to a
“mean SMOW-SLAP scale” (Table 1; supplementary data table), resulting in apparent d18OMW
ranging between −14 and −22‰ (Fig. 10) interpreted to represent some variation in the
meteoric water isotopic composition over the timescales of water–rock interaction. Clearly, an
accurate intercalibration between the water and silicate scale is critical in order to come to a
robust interpretation for such datasets.
Traditionally, combined d18O and dD data are used in the same way (Taylor 1978; Criss
and Taylor 1983; Pope et al. 2009). However, because the chemical behavior of oxygen differs
from hydrogen and because hydrogen is susceptible to retrograde exchange at low temperatures
(Savin and Lee 1988), combined d18O and dD datasets can be compromised. Chamberlain et
al. (2020) analyzed triple oxygen and deuterium isotopes from hydrothermally altered rocks of
the Idaho batholith. Paleo water d18O estimates derived from triple oxygen isotopes are higher
than those derived from combined d18O and dD, which is interpreted as post crystallization
exchange of hydrogen. The higher d18OMW translates into a lower Eocene paleo elevation than
originally proposed by Criss and Taylor (1983).
Especially for fossil analogue′s to Iceland that have been metamorphosed, it can be
difficult to interpret dD datasets (Bindeman and Serebryakov 2011; Zakharov et al. 2019a).
By using additional d17O analyses, however, it is also possible to “see through” metamorphism
and to constrain ancient d18OMW compositions (Herwartz et al. 2015; Zakharov et al. 2019a).
Resolving the d18O composition of snowball Earth glaciers
Ice cores represent valuable climate archives because the measured d18Oice can be
translated into temperature using global or local d18O vs. T correlations (Dansgaard 1964;
Masson-Delmotte et al. 2008). For the more distant past >1 Ma, the d18O of chemical sediments,
especially carbonates are the most common temperature proxies (Zachos 2001). On an Earth
that is entirely covered by ice, however, the atmosphere and the oceans are decoupled and
isotopic temperature proxies are almost absent for these extreme climate states (Peng et al.
2013). Triple oxygen isotopes of hydrothermally altered rocks represent a rare exception.
Hydrothermal water–rock interaction was active throughout Earth′s history, hence there
are altered rocks that interacted with low d18OMW meltwaters from ancient glaciers (Yui et al.
1995; Krylov 2007, 2008; Bindeman and Serebryakov 2011; Bindeman et al. 2014). The most
prominent example are eclogites and ultra high pressure shists from the Dabie and Sulu terranes
308 Herwartz
50 Global MWL δ18Ointercept

A
δ18Ointercept GISP
MWL
used in H 2015
0 SLAP SW
unaltered
∆‘ O [ppm]
SCO rocks
UWG
NBS28
-50
KRS
Krafla (Iceland)
altered basalt
epidote
rhyoli c glass
Sulu (China)
-100 Khitoostrov (Russia)
metamorphiosm
other locali es in Karelia (Russia)
range of pris ne water es mates dehydra on (0.505)
range of mw compiled by S 2018 bio te (dehydrated in vacuum)
-150 -199
-60 -50 -40 -30 -20 -10 0 10 20

δ¹⁸O
50
B
δ Ointercept
18
0
GISP
-50
-100
∆ O [ppm]
-150
t
metamorphism
ep
bio te
-200
rc
te
in
O
(dehydrated dehydra on in
δ 18
in vacuum) vacuum (0.505)

-250
L
-300
W
lM
01 d
H 2 use
ba
-350
Glo
L
MW
in
-400 SLAP
-60 -50 -40 -30 -20 -10 0 10 20

δ¹⁸O
Figure 10. Hydrothermally altered rocks in triple oxygen isotope space. Data is presented on a linear-
ized ∆′17O (= d′17O – λd′18O) vs. d18O scale in Panel A and on a traditional ∆17O (= d17O – λd18O) vs.
d18O scale in Panel B. Mixing lines between the unaltered and altered rock are extrapolated to the global
meteoric water line (which is a curve in Panel B). The intercepts represent the most feasible composi-
tions for a hypothetical altered endmember in equilibrium with the pristine fluid. Hence, the fluid can be
estimated from a respective 1000 ln iarock–water or 1000 ln iamineral–water. Assuming a qrock–water = 0.528 and a
1000 ln 18arock–water = 2‰ gives the stippled fields. Typical errors are 8 ppm SD for single analysis, hence the
variation within individual sites probably reflects fluctuations in the meteoric water composition. Data from
the “other localities” in Karelia including Mt. Dyadina, Kiy island, Varatskoe, Vaut Varakka, Height128
and Shueretskoe fluctuate even more representing strong climate fluctuations over the timescales of hydro-
thermal activity. Data are compiled from (Herwartz et al. 2015 (dark blue for Khitostrov); Zakharov et al.
2017, 2019a, b (light blue for Khitostrov) and normalized to SMOW-SLAP scale using standard values as
reported in Table 1. The opposing absolute values reported for rock standards relative to SMOW and SLAP
affect the calculated intercepts and thus the calculated composition of the pristine meteoric water. Using
the SMOW-SLAP scale of Pack et al. (2016) increases calculated d18OMW by ca. 1–2‰, while using the
SMOW-SLAP scale from Wostbrock et al. (2020a) decreases calculated d18OMW isotopic compositions by
ca. 1–2‰. Metamorphism seems to drive minerals susceptible to alteration (biotite) back to higher d18O
(black stippled arrow). Artificial dehydration of biotite by laser heating in vacuum drives the residual
silicate to low ∆′17O, probably along a kinetic slope around 0.505 corresponding to dehydration (yellow
stippled arrow; Clayton and Mayeda 2009).
in eastern China, which comprise d18Orock as low as −10‰ (Yui et al. 1995). In triple oxygen
isotope space, these samples fall on the “Iceland” trend implying that ancient water isotopic
compositions had been similar to those in Iceland today or in the LGM (Herwartz et al. 2015).
It is now thought that the respective water represents the Kaigas glaciation, a smaller event that
followed the global Sturtian and Marinoan glaciations in the Neoproterozoic.
The most spectacular example of fossil hydrothermal vents known to date is observed for
corundum bearing gneisses from the Belomorian belt in Karelia, northwestern Russia, which
comprise d18Orock down to ≈ −27‰ at the Khitoostrov locality (Krylov 2007, 2008; Bindeman
and Serebryakov 2011; Bindeman et al. 2014). The respective water, which is thought to be
associated to a Paleoproterozoic snowball Earth event at 2291 ± 8 Ma (Zakharov et al. 2017),
must have had a d18Owater lower than −27‰, but it is impossible to say how much lower from
the d18O systematics alone. Using triple oxygen isotope systematics, however, the d18OMW can
be approximated to ≈ −40‰ (Herwartz et al. 2015). In this case, the inaccurate calibration
of the silicate vs. water scale (see above) has little effect on the result due to the strong
curvature of the mixing line (Fig. 10). In a recent compilation (data from Herwartz et al. 2015;
Zakharov et al. 2017, 2019a) on a revised reference frame, Zakharov et al. (2019a) determine
an intercept at −38‰. After applying SMOW-SLAP normalization procedures (see above),
intercepts for individual samples from Khitoostrov range from −35 to −47.5‰ translating to
pristine glacial water isotopic compositions ranging between −35 and −50‰ for this locality
(using 1000 ln arock-water from 0 to 2.5‰).
The observed scatter of the data is far larger than the analytical uncertainty. This variability
is partly related to variable mineralogy and alteration temperatures. The main effect, however, is
probably related to isotopic fluctuations in iwater , related to climate fluctuations over the timescales
of water–rock interaction. Sharp gradients in d18OMW are expected for “snowball” or “slushball”
Earth climate situations especially at low latitudes (Bindeman and Lee 2018). Hence, the mean
d18OMW ≈ −40‰ represents a mean value over the timescale of water–rock interaction with
extremes down to −50‰ or less. Strongly variable d18OMW between individual sites with d18OMW
up to ≈ −9‰ probably represent larger fluctuations on longer timescales (Zakharov et al. 2019a).
Direct U–Pb dating of the two local generations of gabbro intrusions to 2.43–2.41 Ga (high
Mg gabbros) and 2.29 Ga (low Mg gabbros) suggest that two snowball Earth episodes are recorded
in Karelia (Zakharov et al. 2017). Direct dating holds great potential for correlating glacial events
recorded in hydrothermal rocks with diamictites from the Kola peninsula and more prominent
diamictites from South Africa and Canada. Better age control allows better constraining the paleo-
latitude of the ancient glaciers. It turns out that the Baltic Shield had been at near-equatorial to
subtropical positions at the respective dates (Mertanen et al 1999; Salminen et al 2014) providing
evidence for the extent of the ice cover (Bindeman et al. 2010; Zakharov et al. 2017).
It is tempting to translate the d18OMW into mean annual temperatures using modern
d18OMW vs. T relationships (e.g., using MAT = (d18OMW + 13.6)/0.69; Dansgaard 1964), which
implies that mean annual temperatures typical for Antarctica today persisted at tropical to
(sub)tropical latitudes. Respective calibrations for glacial times (Jouzel et al. 1997; Lee et
al. 2008) imply even lower temperatures and the isotope enabled Global Circulation Model
reveals similar to modern d18OMW vs. T relationships for very cold climate states (Bindeman
and Lee 2018). The high d18OMW ≈ −9‰ inferred for some sites in Karelia are consistent with
strong climatic gradients expected especially for a “slushball” Earth (Bindeman and Lee 2018;
Zakharov et al. 2019a) or for water–rock interaction unrelated to these extreme climate states.
Known occurrences of hydrothermal rocks altered by water derived from snowball earth
glaciers are rare. Tephra and ignimbrite deposits also capture paleo meteoric water isotopic
compositions especially when deposited directly on glacial ice (Hudak and Bindeman 2018).
A detailed model for the respective triple oxygen isotope systematics (Rempel and Bindeman
310 Herwartz
2019), essentially comes down to mixing between a pristine and an altered endmember as
outlined above. Rempel and Bindeman (2019) argue that tephra and ignimbrite deposits must
have formed frequently during snowball earth events, providing another potential archive to
study the hydrologic cycle of these enigmatic episodes in Earth history. However, interpreting
ancient hydrosphere data requires accurate knowledge of ancient seawater compositions, since
the ocean governs the mass balance of the hydrosphere.
Resolving the d18O composition of ancient seawater
Archean and Proterozoic chemical sediments have low d18O values compared to their
modern analogues. For broadly similar ocean temperatures compared to today, this shift
implies a shift in ocean water isotopic composition to values as low as −15‰ (Degens and
Epstein 1962; Perry 1967; Jaffrés et al. 2007). From a 17O perspective it now becomes clear
that seawater d18O had not been very low (Sengupta and Pack 2018; Zakharov and Bindeman
2019; Hayles et al. 2020; Liljestrand et al. 2020; Peters et al. 2020b; Sengupta et al. 2020;
Wostbrock and Sharp 2021, this volume; Zakharov et al. 2021, this volume).
The ocean water oxygen isotopic composition is argued to be at steady state with the
Earth′s crust due to water–rock interaction at various temperatures (Muehlenbachs and Clayton
1976). The most important flux of exchanged oxygen is high-T alteration at mid ocean ridges,
and the second most important fluxes are continental weathering and sea floor weathering
at low temperatures. Together with “continental growth” and “water recycling”, these fluxes
control the oxygen isotopic composition of the modern and ancient ocean (Muehlenbachs
and Clayton 1976; Gregory and Taylor 1981; Muehlenbachs 1986, 1998). Recent modelling
in triple oxygen isotope space demonstrates that a low d18O ocean would have a high ∆′17O
(Sengupta and Pack 2018) which is inconsistent with Archean chert data (Levin et al. 2014)
that fall below the present day quartz–water equilibrium curve of (Sharp et al. 2016). If the
chert had formed in equilibrium with seawater at elevated temperatures, the triple oxygen
isotopic composition would coincide with the equilibrium line, which is not the case (Sengupta
and Pack 2018). Several datasets now support a diagenetic modification of the ancient chert
(Degens and Epstein 1962), essentially representing a mixing process that drives ∆′17O to
lower values than expected both for equilibrium or a low d18O ocean (Sengupta and Pack 2018;
Hayles et al. 2020; Liljestrand et al. 2020; Sengupta et al. 2020; Wostbrock and Sharp 2021,
this volume; Zakharov et al. 2021, this volume).
An alternative approach is to study ancient oceanic crust that provides a record of high
temperature water–rock interaction with seawater (Gregory and Taylor 1981). The same triple
oxygen isotope systematics outlined above to derive the composition of ancient meteoric water
(Herwartz et al. 2015) can also be used to approximate the composition of ancient seawater (Peters
et al. 2020b). Serpentinites form via water–rock interaction of mafic rocks. Using petrological
evidence and trace element systematics, Peters et al. (2020b) demonstrates that olivine from an
ultramafic lens in the Archean Kuumiut terrane, Greenland formed via dehydration of serpentinite.
Respective d18Oolivine are as low as 1.6‰ and ultimately derived from serpentinisation of peridotite
with ancient seawater about 2.7 Ga ago. Triple oxygen isotope systematics of these samples is
most consistent with water–rock interaction at 250 to 450 °C and an ancient seawater composition
of d18O = −1‰ typically assumed for Phanerozoic seawater (Peters et al. 2020b).
Zakharov and Bindeman (2019) directly investigated 2.42 Ga old submarine basalts
from the Vetreny belt in Russia, that comprise pillow structures with preserved vein fillings
of epidote, quartz and carbonate. Respective mineral pair equilibrium d18O temperatures
range from 286 to 387°C, typical for hydrothermal temperatures, implying that the isotopic
composition of these minerals is well preserved. Epidote almost directly captures water
isotopic compositions and suggests a seawater d18O = −1.7 ± 1.1‰; ∆′17O = −0.001 ± 0.011‰
and dD = 0 ± 20‰ in the early Paleoproterozoic, consistent within error with modern ice free
world seawater (Zakharov and Bindeman 2019).
All these studies suggest only slightly negative ancient d18Oseawater and only Bindeman
(2021, this volume) still argues for moderately low d18O ≈ −5, based on shale data. Quantifying
oxygen isotopic fluctuations of seawater through time will provide insight into the relative
contributions of hydrothermal alteration and continental weathering.
RESOLVING INDIVIDUAL PROCESSES

In this section I review several processes that are potentially important when studying
hydrothermal, igneous or metamorphic rocks. Resulting vectors in triple oxygen isotope space
are summarized in Figure 11. These include: 1) boiling and phase separation; 2) assimilation; 3)
CO2 and SO2 rich fluids; 4) decarbonation (including pyrolysis); 5) dehydration and; 6) alteration.
Boiling and phase separation observed in well fluids
At hydrothermal temperatures, the fractionation between liquid and vapor is small
(1000 ln 18aliquid–vapor ≈ 1‰) and epidote data from Iceland implies that phase separation is
generally not very important for the epidote forming hydrothermal fluids (Pope et al. 2009, 2012,
2014; Zakharov et al. 2019b). Hydrothermal water–rock interaction as discussed above precedes
the boiling process, that only generates significant isotope effects at temperatures < 150 °C (Horita
and Wesolowski 1994; Shmulovich et al. 1999). The respective qliquid–vapor is around 0.529 to 0.53
(Barkan and Luz 2005; Zakharov et al. 2019b) allowing to predict the triple oxygen isotope
effects of boiling and phase separation of liquid and vapor in modern hydrothermal systems.
Well head fluids at Krafla (Iceland) are enriched in d18O and dD to a degree, which could be
interpreted as water–rock interaction at low W/R ratios of ca. 0.2. Combined triple oxygen and
hydrogen isotope systematics show, however, that a combined effect of water–rock interaction
(W/R ≈ 2) and subsequent phase separation is required to explain well fluid isotopic compositions
(Zakharov et al. 2019b). At Reykjanes (Iceland), mixing between seawater and meteoric water is an
additional complication, but well fluids appear to be “too light” in dD implying that the vapor phase
dominates the well fluids in this case (Arnórsson 1978; Pope et al. 2009; Zakharov et al. 2019b).
Assimilation of low d18O rocks
Magmatism drives hydrothermal systems and thus produces altered rocks in close
proximity to the magma chamber. In regions with low d18OMW, such syn-eruptive alteration
generates low d18Ocrust that is assimilated by the magma generating low d18O magmatic rocks
(Bindeman et al. 2012; Colón et al. 2015). For the classic example of Krafla on Iceland
assimilation of altered country rock is estimated to around 10-20% using triple oxygen and
hydrogen isotope mass balance (Elders et al. 2011; Zakharov et al. 2019b). The Scourie dikes
from the Lewisian Complex in Scotland comprise negative d18O, too low to be derived via
assimilation of high-T altered oceanic crust. The most viable mechanism is assimilation of
20 to 30% of altered crust with very low d18O ≈ −20 to −35‰ similar to those found in
Kithoostrov Karelia. Although this low d18O endmember has not been found anywhere in the
area, the triple oxygen isotope data imply that the pristine meteoric water, that imprinted its
isotopic composition onto the altered rocks, must have had d18OMW ≈ −35 ± 10‰ (Zakharov
et al. 2019a). The close relationship of the Lewisian complex with the Belomorian belt lends
support to this interpretation (Zakharov et al. 2019a).
The assimilation of country rocks into the magma chamber is traditionally studied using
radiogenic or trace elements, but respective compositions differ substantially between altered
and pristine rocks. Quantifying the amount of hydrothermally altered rock assimilated by a
magma chamber via triple oxygen isotopes thus holds potential for a better interpretation
of trace element and isotope data. Assimilation of other rock types such as carbonates or
sulfates (Peters et al. 2020a) generate unique triple oxygen isotope trends that can aid in better
quantifying specific lithologies assimilated by the magma. All vectors labeled as “mixing” in
Figure 11 represent reasonable trajectories for assimilation.
312 Herwartz
550 sulfates and nitrates

expelled H₂O (MIF)
500 (from brucite) expelled H₂O
(C & M 2009) (from serpen nite)
450 (C & M 2009)
brucite mantle lunar

50 meteoric water (S 2018) CO₂ from pyrolysis
meteorite finds of NBS19
GISP SW (Miller et al. 2002)
0 200
Ice 400
lan 100
d
-50 mgt A H₂O CO₂
car
(W bona
25
KRS & S tes
mgt C 20
-100 20
Khitos SiO )
trov NBS19 ₂a
(W 2020a) th
-150 25 igh
dehydra on S
SKFS iO W
SO₂ /R
H₂O ₂a
-200 pyrolysis tl
o
lokal w
mgt B sulfates W
mi /R
-250 xin
g mixing with (MIF)
liquid
vapour high δ18O
-300 MgO residue
mixing apparent (C & M 2009) CaO from pyrolysis
-350 low T altera on of NBS19
pyrolysis in deserts (Miller et al. 2002)
dehydra on
-400
mixing with MIF-O₂ air O₂
-450
-50 -40 -30 -20 -10 0 10 20 30 40 50
δ OVSMOW [‰]
18
Figure 11. Summary of individual processes potentially affecting triple oxygen isotope compo-
sitions of crustal rocks. These include: 1) dehydration of serpentinite and brucite (Clayton and
Mayeda 2009). For serpentinite (green triangles) the bulk composition is similar to the dehydrated
phase, while mass balance forces the residual MgO to very low ∆′17O for dehydration of brucite
(grey triangles); 2) pyrolysis of carbonate (brown diamonds) using NBS19 as an example. Re-
spective mean (and recalibrated) CaO and CO2 data from (Miller 2002) are compared to bulk
NBS19 data from Wostbrock et al. (2020a); 3) boiling and phase separation (blue arrows labeled
“vapour” and “liquid” in the inset); 4) low T alteration of mafic rocks in desert environments is
represented by lunar meteorite finds (supplementary dataset). These samples (small green dots)
are altered at low T over thousands of years with evaporating water. The effective slope (stippled
green double arrow with λ = 0.5175) of these samples does not pass through the unaltered end-
member, because these samples weathered under variable climate conditions. Individual weather-
ing trajectories (green arrows in the inset) probably had quite variable slopes. 5) precipitation of
silicates (Sharp et al. 2016) and carbonates (Wostbrock and Sharp 2021, this volume) at variable
temperatures and variable starting compositions for the water (i.e., seawater for carbonates and
water equilibrated at variable W/R ratios with MORB for silicates); and 6) several mixing pro-
cesses that can drive samples into various directions. Consider for instance mixing of magnetite A
(formed in eq. with magmatic water), magnetite B (formed from fluids derived from sulfates with
a negative MIF anomaly) and magnetite C (fromed from fluids with high XCO2 derived from car-
bonates). These mixing endmembers explain measured magnetite compositions (small dark grey
dots) within the grey triangle (Peters et al. 2020a). Mixing with light d18O altered endmembers
(e.g., Iceland or Khitostrov) may drives samples in a completely different direction. Any contribu-
tion ultimately derived from atmospheric O2 drives samples to low ∆′17O (Pack et al. 2017; Pack
2021, this volume; Sutherland et al. 2020), while material ultimately derived from ozone (e.g.,
atmospheric sulfates and nitrates) increase ∆′17O (Brinjikji and Lyons 2021, this volume; Cao and
Bao 2021, this volume; Thiemens and Lin 2021, this volume;).
Isotopic exchange with CO2 and SO2

Assimilation and decomposition of carbonates or sulfates generate large quantities of CO2
or SO2. It is well known that the isotopic composition of H2O can be modulated by equilibrium
exchange with CO2 in high XCO2 fluids (Bottinga 1968; Truesdell 1974). High d18Omagnetite
observed iron oxide deposits in Iran are probably derived from the CO2 of decomposed
Cambrian carbonates with d18Ocarbonate = 20–30‰. In this particular case, exceptionally
low ∆′17Omagnetite compositions down to −170 ppm are also observed, which are probably
inconsistent with mass dependent processes (Peters et al. 2020a). Local sulfates are known
to comprise negative MIF signatures, which are apparently passed on to the magnetite (“mgt
B” in Fig. 11). Formation temperatures of > 800 °C (Troll et al. 2019) support a magmatic
rather than a hydrothermal process. The natural MIF tracer of the sulfate rocks from the area
(comprising low ∆′17OSO4) helps to distinguish between three viable endmembers (Fig. 11),
demonstrating that SO2 can significantly alter fluid isotopic compositions just like CO2 (Peters
et al. 2020a). The study also shows that minerals can capture and preserve transient fluid
isotopic compositions, which may also be expected from large scale dehydration (Clayton and
Mayeda 2009) and decarbonation (Miller et al. 2002) reactions.
Decarbonation
The reaction products derived from the thermal decomposition of carbonate
(CaCO3 → CaO + CO2) in vacuum differ by ≈ 400 ppm in ∆′17O (Miller et al. 2002). This
apparently mass independent process (see Fig. 11) is—to my knowledge—still not understood.
Hence, it is unclear if thermal decomposition, which occurs on large scales in subduction
zones, induces ∆′17O heterogeneity. Fast re-equilibration with water and oxygen bearing
minerals at temperatures > 700 °C would rapidly diminish most of this initial ∆′17O, as also
observed in experiments where the liberated CO2 is not directly separated from solid CaO
(Miller et al. 2002). However, a residual heterogeneity may be resolvable in natural systems at
the present analytical precision.
Sun and Bao (2011b) proposed that the effect could be associated to a gas phase diffusion
mechanism, related to the magnetic properties of 17O affecting the effective collision diameter
of respective isotopologues (Sun and Bao 2011a, b). These authors observe apparent mass
independent fractionation in a thermal gradient for O2 gas but not for oxygen in melts (Sun and
Bao 2011a, b). Pyrolyses of CaCO3 in melts does not seem to generate any “MIF” signature
(Peters et al. 2020a) implying that the underlying process is indeed restricted to the gas phase.
Dehydration
The oxygen isotopic composition of hydroxyl groups and H2O within a crystal lattice
are generally distinct from other oxygen sites within a mineral (Zheng 1993; Girard and
Savin 1996). This feature has been explored as a potential intra-mineral thermometer. Rapid
dehydration under vacuum, with fast separation of the product water from the hot residual
phase, generates large kinetic isotope fractionation effects that are superimposed on the intra-
mineral fractionation. Dehydration of serpentinite, for instance, results in d18Owater that is ca.
15‰ lighter than the mineral internal d18OOH that is in turn 19‰ lighter than non-hydroxyl
oxygen (Clayton and Mayeda 2009). The large kinetic fractionation factors decrease to
near zero (i.e., equilibrium) if the reaction products are not separated or heated to higher
temperatures allowing for re-equilibration of the initial kinetic products with the anhydrous
phases (Girard and Savin 1996).
Triple oxygen isotope slopes for the dehydration of serpentinite and brucite are 0.507
and 0.503 respectively (Clayton and Mayeda 2009). Both slopes are much shallower than
expected for equilibrium q ≈ 0.527 at 300°C (Fig. 1). The initial kinetic H2O reaction products
comprise average ∆′17Orock–H2O = 570 ppm for the dehydration of serpentinite and as much as
314 Herwartz
∆′17OMgO–H2O = 810 ppm for dehydration of brucite (Fig. 11). In one analytical experiment,
I dehydrated biotite via laser heating within the vacuum chamber of the fluorination line.
The ∆′17O of the residual silica glass came out 40 ppm lower compared to biotite fluorinated
directly (Fig. 10), demonstrating that analytical protocols should be checked for such artefacts.
In this context it may be interesting to re-investigate the various techniques for the removal of
exchangeable H2O and OH from biogenic silica (Opal-A) prior to analyses (Chapligin et al. 2011;
Bandriss et al. 1998). Most biogenic and abiogenic silica samples fall on the same curve in triple
oxygen isotope space (Sharp et al. 2016; Wostbrock et al. 2018), but phytolith SiO2 seems to have
ca. 70 ppm lower ∆′17O than expected (Alexandre et al. 2019), either reflecting kinetic processes
during the SiO2 polymerization, or contrasting isotopic effects for the respective dehydroxylation
methods used (i.e., pre-fluorination vs. dehydration at 1000°C in a N2 gas stream).
The fast re-equilibration kinetics observed for d18O (Girard and Savin 1996) imply that
the preservation potential for such initial heterogeneity in ∆′17O is limited on Earth. Peters
et al. (2020b) does not observe a significant effect in olivine formed by dehydration of
serpentinites. In this case, however, only a small mole fraction of the bulk oxygen is expelled
as water. The maximum drawdown shift in ∆′17O for the residual silicate is only 21ppm
(Clayton and Mayeda 2009; Peters et al. 2020b; Fig. 11). For materials containing more
hydroxyl groups such as brucite, dehydration may well leave a measurable ∆′17O trace both
in materials that react with the expelled water and in residual anhydrous phases. Preservation
of the ∆′17O requires fast separation of the expelled water at moderate temperatures, just
high enough for the dehydration to proceed, but not too high as this would facilitate fast re-
equilibration. Anhydrous minerals with low ∆′17O would be generated, potentially explaining
the surprisingly low ∆′17O sporadically observed for quartz (Fig. 9), rhyolitic glass from Krafla
and a few peculiar gneiss samples (Fig. 10). Alternatively, these minerals document mixing
with extreme and yet unknown mixing components. Analytical artefacts related to dehydration
prior analyses (see above) are not feasible for these nominally anhydrous minerals.
Dehydration into near-vacuum and fast separation of the initial kinetic water from the
rock residue is far more feasible for extraterrestrial dehydration (or pyrolysis) reactions.
It was shown that “metamorphosed carbonaceous chondrites” (MCCs) are probably genetically
related to CM2 carbonaceous chondrite via multiple cycles of thermal processing including
dehydration and rehydration of silicate phases (Clayton and Mayeda 2009; Ivanova et al. 2013).
The proposed mechanism requires that the expelled water was at least not fully re-equilibrated,
hence some of the apparently mass independent effects observed in extraterrestrial material may
actually be derived from mass dependent processes. Heterogeneity in ∆′17O between individual
lunar lithologies and within green glass was assigned to a mass independent effect (Cano et
al. 2020) but mass dependent mechanisms should not be excluded a priori. Dehydration of a
magma fountain in vacuum would probably induce ∆′17Ogreen glass heterogeneity provided that
the magma was not completely anhydrous.
Alteration
Identifying and quantifying isotopic shifts related to alteration processes is a general
challenge for using isotopes in earth sciences. Quantitative discussions of alteration related
to cherts and carbonates are provided in recent triple oxygen isotope papers (Liljestrand et al.
2020; Zakharov et al. 2021, this volume; Wostbrock and Sharp 2021, this volume). Herein, I
summarized how altered samples can be useful to learn about alteration endmembers that inform
on paleo-environments. The assumption that apparent alteration trajectories always intersect the
pristine endmembers must not be true, however, as evident from the following exotic example.
The alteration trend of lunar meteorite finds (supplementary dataset in Herwartz et al. 2014
recalibrated herein) does not intersect with pristine compositions (∆′17O ≈ −50 to −40 ppm at d18O
≈ 5.5‰) of lunar Apollo samples (e.g., Wiechert et al. 2001; Spicuzza et al. 2007; Hallis et al. 2010;
Herwartz et al. 2014; Young et al. 2016; Greenwood et al. 2018; Cano et al. 2020). The apparent
alteration slope (stippled green double arrow in Fig. 11) plotted trough the samples (green dots in
Fig. 11) is λ = 0.5175 and passes through ∆′17Omantle = −50 ppm at d18O > 8‰. Hence, the pristine
endmember does not represent any endmember of this apparent alteration slope. Making that
assumption would indicate a pristine lunar ∆′17O composition that is > 20 ppm higher compared to
Earth′s mantle, which is not the case (Wiechert et al. 2001; Spicuzza et al. 2007; Hallis et al. 2010;
Herwartz et al. 2014; Young et al. 2016; Greenwood et al. 2018; Cano et al. 2020).
These samples are altered at low T over thousands of years with evaporating meteoric water.
Therefore, low T alteration products inherit variable ∆′17O ranging from freshwater (high ∆′17O)
to evaporitic water (low ∆′17O). The average triple oxygen isotopic composition of this water
strongly depends on the local climatic conditions (Surma et al. 2015, 2018, Herwartz et al.
2017; Gázquez et al. 2018, Voigt et al. 2020). Each individual meteorite interacted with a unique
average water composition in a unique climate at a unique average temperature. All samples
are displaced to higher d18O, because the 1000 ln aalteration product–H2O are generally high and thus
alteration endmembers generally comprise a high d18O. The ∆′17Oalteration product depends on the
average alteration temperature, ∆′17Owater and qalteration product–H2O. The alteration endmembers for
the individual samples are probably strongly variable as indicated by the solid green arrows in the
inset of Fig. 11. Hence, these samples are not expected to follow the same alteration trend and, the
concept of plotting one slope through all samples is a misconception.
In this particular case, the samples share the same starting composition and they are all
weathered in dry desert environments. Hence the overall trend with a low λ = 0.5175 could be
related to mixing of residual evaporitic water in the rock with fresh meteoric water or fog as
observed for source water in dry desert environments (Voigt et al. 2020), but the similarities in
slope could also be a coincidence.
This exotic example of alteration may represent a good analogue for low T alteration of
mafic rocks. More importantly, however, it shows how apparent alteration trajectories may deviate
from expected relationships because the samples comprise some spatial and temporal variability.
Assuming that the unaltered endmember must fall on a trend defined by a set of samples is clearly
inaccurate in the above case. Whenever a certain isotopic variability in the pristine or alteration
endmember is feasible, triple oxygen isotope compositions are expected to deviate from a uniform
slope or mixing trajectory. In the case of Karelia outlined above, the respective variability of the
samples was utilized to estimate variability in paleo water isotopic compositions.
CONCLUSIONS AND OUTLOOK

Triple oxygen isotope systematics provide a second dimension to the d18O scale. Extending
traditional models built on d18O to d17O simply requires reasonable approximations for q.
The theoretical framework developed for d18O over the past decades can be adopted. The
simplest application of additional d17O analyses is to test model assumptions such as equilibrium,
which is crucial for palaeothermometry. A second application is to constrain parameters like T,
W/R or iwater , using the additional d17O quantity (Herwartz et al. 2015). Feasible sets of input
parameters can be determined even for mathematically under constrained systems, specifying
feasible approximations for individual parameters (Gázquez et al. 2018; Liljestrand et al.
2020). Unexpected d17O values may uncover processes or reaction components that had not
been detected from the d18O systematics alone such as the role of SO2 in the formation of
magnetite (Peters et al. 2020a). Reaction kinetics dictate certain q values, relating the triple
oxygen isotopic compositions of educts and products in a unique way. Therefore, combined
d18O and d17O analyses allow to test proposed reaction mechanisms and to study reaction
rates (Cao and Bao 2017; Guo and Zhou 2019). High reaction rates and fast separation of the
316 Herwartz
reaction products is often realized in laboratories as summarized here for decarbonation and
dehydration. Respective effects in ∆′17O are quite significant and could be resolvable in natural
high T environments as well. Being able to “see through” metamorphism (Herwartz et al.
2015) or diagenesis (Wostbrock and Sharp 2021, this volume) holds great potential for more
accurate constraints on paleoenvironments if the expected variability for individual alteration
endmebers is accounted for (lunar meteorite example).
This summary exhibits the potential of triple oxygen isotopes in more quantitatively
constraining geological processes. The field of purely mass dependent 17O systematics
is rapidly expanding, but is still in a “tool development phase”, especially with respect to
igneous, metamorphic and hydrothermal rocks. Any field presently using d18O to study
water–rock interaction will benefit from additional d17O analyses.
One of the greatest potentials is to reconstruct paleowater isotopic compositions. This is
interesting for paleoclimate studies throughout Earth′s history, but especially for resolving
long standing debates related to Precambrian hydrology. Accurate reconstructions for seawater
and meteoric water provide a tool to investigate the hydrological cycle in deep time offering
insights into paleoclimate and geodynamic regimes.
In this contribution I focused on silicates but other oxygen bearing materials such as
carbonates, sulfates, phosphates, metal oxides etc. all comprise unique oxygen isotope
systematics and thus characteristic d18O vs. ∆′17O compositions that are unique to their
respective formation mechanisms and conditions (Schauble and Young 2021, this volume).
ACKNOWLEDGEMENTS
I thank Andreas Pack and Ilya Bindeman for inviting me to write this chapter and for their
editorial handling. Zach Sharp and David Zakharov provided excellent reviews that substantially
improved the quality of this manuscript. I also thank Claudia Voigt and David Bajnai for their
suggestions to improve the manuscript and Addi Bischoff for kindly providing the lunar meteorite
samples. Financial support was provided by the DFG via Project HE 6357/2-1.
REFERENCES
Alexandre A, Webb E, Landais A, Piel C, Devidal S, Sonzogni C, Coupal M, Mazur JC, Pierre M, Prié F, Valle-
Coulomb C, Outrequin C, Roy J (2019) Effects of leaf length and development stage on the triple oxygen
isotope signature of grass leaf water and phytoliths: insights for a proxy of continental atmospheric humidity.
Biogeosciences 16:4613–4625
Arnórsson S (1978) Major element chemistry of the geothermal sea-water at Reykjanes and Svartsengi, Iceland.
Mineral Mag 42:209–220
Bajnai D, Guo W, Spötl C, Coplen TB, Methner K, Krsnik E, Gieschler E, Hansen M, Henkel D, Price GD,
Raddatz J, Scholz D, Fiebig J (2020) Dual clumped isotope thermometry resolves kinetic biases in
carbonate formation temperatures. Nature communications (accepted)
Bao H (2015) Sulfate: A time capsule for Earth′s O2, O3, and H2O. Chem Geol 395:108–118
Spectrom 19:3737–3742
Baumgartner LP, Valley JW (2001) Stable isotope transport and contact metamorphic fluid flow. Rev Mineral
Geochem 43:415–467
Beard BL, Handler RM, Scherer MM, Wu L, Czaja AD, Heimann A, Johnson CM (2010) Iron isotope fractionation
between aqueous ferrous iron and goethite. Earth Planet Sci Lett 295:241–250
Beck WC, Grossman EL, Morse JW (2005) Experimental studies of oxygen isotope fractionation in the carbonic acid
system at 15°, 25°, and 40°C. Geochim Cosmochim Acta 69:3493–3503
Bergel SJ, Barkan E, Stein M, Affek HP (2020) Carbonate 17O excess as a paleo-hydrology proxy: Triple oxygen
isotope fractionation between H2O and biogenic aragonite, derived from freshwater mollusks. Geochim
Bigeleisen J, Mayer MG (1947) Calculation of equilibrium constants for isotopic exchange reactions. J Chem Phys
15:261–267
Bigeleisen J, Wolfsberg M (1958) Theoretical and experimental aspects of isotope effects in chemical kinetics. In:
Prigogine I, Debye P (eds) Advances in Chemical Physics. John Wiley & Sons, Inc., Hoboken, NJ, USA, pp 15–76
Bindeman IN, Serebryakov NS (2011) Geology, petrology and O and H isotope geochemistry of remarkably 18O
depleted Paleoproterozoic rocks of the Belomorian Belt, Karelia, Russia, attributed to global glaciation 2.4Ga.
Bindeman IN, Lee J-E (2018) The possibility of obtaining ultra-low-d18O signature of precipitation near equatorial
latitudes during the Snowball Earth glaciation episodes. Precambrian Res 319:211–219
Bindeman IN, Schmitt AK, Evans DAD (2010) Limits of hydrosphere-lithosphere interaction: Origin of the lowest-
known d18O silicate rock on Earth in the Paleoproterozoic Karelian rift. Geology 38:631–634
Bindeman I, Gurenko A, Carley T, Miller C, Martin E, Sigmarsson O (2012) Silicic magma petrogenesis in Iceland
by remelting of hydrothermally altered crust based on oxygen isotope diversity and disequilibria between zircon
and magma with implications for MORB: Silicic magma petrogenesis in Iceland. Terra Nova 24:227–232
Bindeman IN, Serebryakov NS, Schmitt AK, Vazquez JA, GuanY, Azimov PY, Astafiev BY, Palandri J, Dobrzhinetskaya
L (2014) Field and microanalytical isotopic investigation of ultradepleted in 18O Paleoproterozoic “Slushball
Earth” rocks from Karelia, Russia. Geosphere 10:308–339
Bird DK, Spieler AR (2004) Epidote in geothermal systems. Rev Mineral Geochem 56:235–300
Böttcher ME, Neubert N, Escher P, Von Allmen K, Samankassou E, Naegler TF (2018) Multi-isotope (Ba, C, O)
partitioning during experimental carbonatization of a hyper-alkaline solution. Geochemistry 78:241–247
carbon dioxide–water. J Phys Chem 72:800–808
Bottinga Y, Javoy M (1987) Comments on stable isotope geothermometry: the system quartz–water. Earth Planet Sci
Lett 84:406–414
Brandriss ME, O′Neil JR, Edlund MB, Stoermer EF (1998) Oxygen isotope fractionation between diatomaceous
silica and water. Geochim Cosmochim Acta 62:1119–1125
Brinjikji M, Lyons JR (2021) Mass-independent fractionation of oxygen isotopes in the atmosphere. Rev
Cao X, Bao H (2021) Small triple oxygen isotope variations in sulfate: Mechanisms and applications. Rev
Cao X, Bao H (2017) Redefining the utility of the three-isotope method. Geochim Cosmochim Acta 212:16–32
ocean island basalts. Acta Geochim 38:327–334
Chamberlain CP, Ibarra DE, Lloyd MK, Kukla T, Sjostrom D, Gao Y, Sharp ZD (2020) Triple oxygen isotope
systematics of meteoric hydrothermal systems—implications for paleoaltimetry. Geochem Persp Lett 15:2026
Chapligin B, Leng ML, Webb E, Alexandre A, Dodd JP, Ijiri A, Lücke A, Shemesh A, Abelmann A, Herzschuh U,
Longstaffe FJ, Meyer H, Moschen R, Okazaki Y, Rees NH, Sharp ZD, Sloane HJ, Sonzogni C, Swann GEA,
Sylvestre F, Tyler JJ, Yam R (2011) Inter-laboratory comparison of oxygen isotope compositions from biogenic
silica. Geochim Cosmochim Acta 75:7242–7256
Clark ID, Fontes JC, Fritz P (1992) Stable isotope disequilibria in travertine from high pH waters: Laboratory
investigations and field observations from Oman. Geochim Cosmochim Acta 56:2041–2050
Clayton RN, Epstein S (1961) The use of oxygen isotopes in high-temperature geological thermometry. J Geol 69:447–452
Clayton RN, O′Neil JR, Mayeda TK (1972) Oxygen isotope exchange between quartz and water. J Geophys Res
77:3057–3067
Colón DP, Bindeman IN, Stern RA, Fisher CM (2015) Isotopically diverse rhyolites coeval with the Columbia River
Flood Basalts: evidence for mantle plume interaction with the continental crust. Terra Nova 27:270–276
Coplen TB (2007) Calibration of the calcite–water oxygen-isotope geothermometer at Devils Hole, Nevada, a natural
laboratory. Geochim Cosmochim Acta 71:3948–3957
Craig H (1961) Isotopic variations in meteoric waters. Science 133:1702–1703
318 Herwartz
Criss RE, Taylor HP Jr (1983) An 18O/16O and D/H study of Tertiary hydrothermal systems in the southern half of the
Idaho batholith. GSA Bull 94:640–663
Daëron M, Drysdale RN, Peral M, Huyghe D, Blamart D, Coplen TB, Lartaud F, Zanchetta G (2019) Most Earth-
surface calcites precipitate out of isotopic equilibrium. Nat Commun 10:429
Degens ET, Epstein S (1962) Relationship between O18/O16 ratios in coexisting carbonates, cherts, and diatomites:
geological notes. AAPG Bull 46:534–452
Devriendt LS, Watkins JM, McGregor H V. (2017) Oxygen isotope fractionation in the CaCO3–DIC–H2O system.
Dietzel M, Usdowski E, Hoefs J (1992) Chemical and 13C/12C-and 18O/16O-isotope evolution of alkaline drainage waters
and the precipitation of calcite. Appl Geochem 7:177–184
Dodson MH (1973) Closure temperature in cooling geochronological and petrological systems. Contrib Mineral
Petrol 40:259–274
Druhan JL, Winnick MJ, Thullner M (2019) Stable isotope fractionation by transport and transformation. Rev Mineral
Geochem 85:239–264
Elders WA, Friðleifsson GÓ, Zierenberg RA, Pope EC, Mortensen AK, Guðmundsson Á, Lowenstern JB, Marks NE,
Owens L, Bird DK, Reed M (2011) Origin of a rhyolite that intruded a geothermal well while drilling at the
Krafla volcano, Iceland. Geology 39:231–234
Elphick SC, Graham CM (1988) The effect of hydrogen on oxygen diffusion in quartz: evidence for fast proton
transients? Nature 335:243–245
Epstein S, Buchsbaum R, Lowenstam H, Urey HC (1951) Carbonate–water isotopic temperature sclae. J Geol 62:417–426
Epstein S, Mayeda TK (1953) Variation in 18O content of waters from natural sources. Geochim Cosmochim Acta 4:213–224
Evans NP, Bauska TK, Gázquez-Sánchez F, Brenner M, Curtis JH, Hodell DA (2018) Quantification of drought
during the collapse of the classic Maya civilization. Science 361:498–501
Garlick GD, Epstein S (1967) Oxygen isotope ratios in coexisting minerals of regionally metamorphosed rocks.
Gázquez F, Evans NP, Hodell DA (2017) Precise and accurate isotope fractionation factors (a17O, a18O, aD) for water
and CaSO4 × 2H2O (gypsum). Geochim Cosmochim Acta 198:259–270
Gázquez F, Morellón M, Bauska T, Herwartz D, Surma J, Moreno A, Staubwasser M, Valero-Garcés B, Delgado-
Huertas A, Hodell DA (2018) Triple oxygen and hydrogen isotopes of gypsum hydration water for quantitative
paleo-humidity reconstruction. Earth Planet Sci Lett 481:177–188
Giletti BJ (1986) Diffusion effects on oxygen isotope temperatures of slowly cooled igneous and metamorphic rocks.
Girard J-P, Savin SM (1996) Intracrystalline fractionation of oxygen isotopes between hydroxyl and non-hydroxyl sites
in kaolinite measured by thermal dehydroxylation and partial fluorination. Geochim Cosmochim Acta 60:469–487
Green M, Taube H(1963) Isotopic fractionation in the OH−–H2O exchange reaction. J Phys Chem 67:1565–1566
Greenwood RC, Barat JA, Miller MF, Anand M, Dauphas N, Franchi IA, Sillard P, Starkey NA (2018) Oxygen isotopic
evidence for accretion of Earth′s water before a high-energy Moon forming giant impact. Sci Adv 4:eaao5928
Gregory RT, Taylor HP (1981) An oxygen isotope profile in a section of cretaceous oceanic crust, Samail Ophiolite,
Oman: evidence for d18O buffering of the oceans by deep (>5 km seawater hydrothermal circulation at mid-
ocean ridges. J Geophys Res 84 (B4):2737–2755
Hallis LJ, Anand M, Greenwood RC, Miller MF, Franchi IA, Russell SS (2010) The oxygen isotope composition,
petrology and geochemistry of mare basalts: Evidence for large-scale compositional variation in the lunar
mantle. Geochim Cosmochim Acta 74:6885–6899
Hayles J, Yeung LY, Homann M, Banerjee A, Jiang H, Shen B, Lee CT (2020) Three billion year secular
evolution of the triple oxygen isotope composition of marine chert. Submitted to Earth Planet Sci Lett;
Non peer reviewed version available at EarthArXiv 10.31223/osf.io/n2p5q
344:1146–1150
snowball Earth from D17O systematics of hydrothermal rocks. PNAS 112:5337–5341
Hofmann MEG, Horváth B, Pack A (2012) Triple oxygen isotope equilibrium fractionation between carbon dioxide
and water. Earth Planet Sci Lett 319–320:159–164
Hudak MR, Bindeman IN (2018) Conditions of pinnacle formation and glass hydration in cooling ignimbrite sheets from
H and O isotope systematics at Crater Lake and the Valley of Ten Thousand Smokes. Earth Planet Sci Lett 500:56–66
Ivanova MA, Lorenz CA, Franchi IA, Bychkov AY, Post JE (2013) Experimental simulation of oxygen isotopic
exchange in olivine and implication for the formation of metamorphosed carbonaceous chondrites. Meteorit
Planet Sci 48:2059–2070
Jaffrés JBD, Shields GA, Wallmann K (2007) The oxygen isotope evolution of seawater: A critical review of a long-standing
controversy and an improved geological water cycle model for the past 3.4 billion years. Earth-Sci Rev 83:83–122
Jouzel J, Alley RB, Cuffey KM, Dansgaard W, Grootes P, Hoffmann G, Johnsen SJ, Koster RD, Peel D, Shuman CA,
Stievenard M (1997) Validity of the temperature reconstruction from water isotopes in ice cores. JGR Oceans
102(C12):26471–26487
Kim ST, O′Neil JR (1997) Equilibrium and nonequilibrium oxygen isotope effects in synthetic carbonates. Geochim
Kohn MJ (1999) Why most “dry” rocks should cool “wet.” Am Mineral 84:570–580
Krylov DP (2007) O18 depletion in corundum bearing rocks from North Karelija (the Baltic Shield). Water–rock
Interact 7:87–89
Krylov DP (2008) Anomalous 18O/16O ratios in the corundum-bearing rocks of Khitostrov, northern Karelia. Dokl
Earth Sci 419A:453–456
Landais A, Barkan E, Yakir D, Luz B (2006) The triple isotopic composition of oxygen in leaf water. Geochim
Landais A, Steen-Larsen HC, Guillevic M, Masson-Delmotte V, Vinther B, Winkler R (2012) Triple isotopic composition
of oxygen in surface snow and water vapor at NEEM (Greenland). Geochim Cosmochim Acta 77:304–316
Lazar C, Young ED, Manning CE (2012) Experimental determination of equilibrium nickel isotope fractionation
between metal and silicate from 500°C to 950°C. Geochim Cosmochim Acta 86:276–295
Lee J-E, Fung I, DePaolo DJ, Otto-Bliesner B (2008) Water isotopes during the Last Glacial Maximum: New general
circulation model calculations. J Geophys Res 113:D19109
implications for continental paleoclimate reconstruction. Rev Mineral Geochem 86:429-462
Liu T, Artacho E, Gázquez F, Walters G, Hodell D (2019) Prediction of equilibrium isotopic fractionation of the
gypsum/bassanite/water system using first principles calculations. Geochim Cosmochim Acta 244:1–11
Macris CA, Young ED, Manning CE (2013) Experimental determination of equilibrium magnesium isotope
fractionation between spinel, forsterite, and magnesite from 600 to 800 °C. Geochim Cosmochim Acta 118:18–32
Masson-Delmotte V, Hou S, Ekaykin A, Jouzel J, Aristarain A, Bernardo RT, Bromwich D, Cattani O, Delmotte
M, Falourd S, Frezzotti M (2008) A review of Antarctic surface snow isotopic composition: observations,
atmospheric circulation, and isotopic modeling. J Clim 21:3359–3387
Matthews A, Goldsmith JR, Clayton RN (1983) On the mechanisms and kinetics of oxygen isotope exchange in
quartz and feldspars at elevated temperatures and pressures. Geol Soc Am Bull 94:396–412
Matthews A (1994) Oxygen isotope geothermometers for metamorphic rocks. J Metamorph Geol 12:211–219
Melander L, Saunders WH (1987) Reaction Rates of Isotopic Molecules. Robert E. (Ed.) Krieger Publishing Company:
Malabar, FL
Mertanen S, Halls HC, Vuollo JI, Personen LJ, Stepanov VS (1999) Paleomagmatism of 2.44 Ga mafic dykes in Russia
Karelia, eastern Fenoscandian Shield—implications for continental reconstructions. Prec Res 98:197–221
Miller MF, Franchi IA, Thiemens MH, Jackson TL, Brack A, Kurat G, Pillinger CT (2002) Mass-independent
fractionation of oxygen isotopes during thermal decomposition of carbonates. PNAS 99:10988–10993
Miller MF, Pack A, Bindeman IN, Greenwood RC (2020) Standardizing the reporting of Dʹ17O data from high
Muehlenbachs K (1986) Alteration of the oceanic crust and the 18O history of seawater. Rev Mineral Geochem
16:425–444
320 Herwartz
145:263–273
Muehlenbachs K, Clayton RN (1976) Oxygen isotope composition of the oceanic crust and its bearing on seawater.
Pack A (2021) Isotopic fingerprints of atmospheric molecular O2 in rocks, minerals and melts. Rev Mineral Geochem
86:217–240
Pack A, Herwartz D (2014) The triple oxygen isotope composition of the Earth mantle and understanding D17O
Pack A, Herwartz D (2015) Observation and interpretation of D17O variations in terrestrial rocks - Response to the
comment by Miller et al. on the paper by Pack and Herwartz (2014). Earth Planet Sci Lett 418:184–186
Carlos olivine on the VSMOW2-SLAP2 scale: San Carlos olivine on the VSMOW2-SLAP2 scale. Rapid
Pack A, Höweling A, Hezel DC, Stefanak MT, Beck AK, Peters STM, Sengupta S, Herwartz D, Falco L (2017) Tracing
the oxygen isotopic composition of the upper Earth′s atmosphere using cosmic spherules. Nat Comm 8:15702
Peng Y, Bao H, Zhou C, Yuan X, Luo T (2013) Oxygen isotope composition of meltwater from a Neoproterozoic
glaciation in South China. Geology 41:367–370
Perry EC (1967) The oxygen isotope chemistry of ancient cherts. Earth Planet Sci Lett 3:62–66
Peters ST, Alibabaie N, Pack A, McKibbin SJ, Raeisi D, Nayebi N, Torab F, Ireland T, Lehmann B (2020a) Triple
oxygen isotope variations in magnetite from iron-oxidedeposits, central Iran, record magmatic fluid interaction
Peters STM, Szilas K, Sengupta S, Kirkland CL, Garbe-Schönberg D, Pack A (2020b) >2.7 Ga metamorphic peridotites from
southeast Greenland record the oxygen isotope composition of Archean seawater. Earth Planet Sci Lett 544:116331
Pope EC, Bird DK, Arnórsson S, Fridriksson T, Elders WA, Fridleifsson GÓ (2009) Isotopic constraints on ice age
fluids in active geothermal systems: Reykjanes, Iceland. Geochim Cosmochim Acta 73:4468–4488
Pope EC, Bird DK, Rosing MT (2012) Isotope composition and volume of Earth′s early oceans. PNAS 109:4371–4376
Pope EC, Bird DK, Arnórsson S (2014) Stable isotopes of hydrothermal minerals as tracers for geothermal fluids in
Iceland. Geothermics 49:99–110
Putnis A (2015) Transient porosity resulting from fluid–mineral interaction and its consequences. Rev Mineral
Geochem 80:1–23
Rempel AW, Bindeman IN (2019) A model for the development of stable isotopic water signatures of tephra deposited
on ice following subglacial caldera collapse. J Volcanol Geotherm Res 377:131–145
Sade Z, Yam R, Shemesh A, Halvey I (2020) Kinetic fractionation of carbon and oxygen isotopes during BaCO3
precipitation. Geochim Cosmochim Acta 280:395–422
Salminen J, Halls HC, Mertanen LJ, Personen J, Vuollo J, Söderlund U (2014) Paleomagnetic and geochronological
studies on Paleoproterozoic diabase dykes of Karelia, East Finland - Key for testing the Superia supercraton.
Savin SM, Lee M (1988) Isotopic studies of phyllosilicates. Rev Mineral Geochem 19:189–223
Schauble EA, Young ED (2021) Mass dependence of equilibrium oxygen isotope fractionation in carbonate,
nitrate, oxide, perchlorate, phosphate, silicate, and sulfate minerals. Rev Mineral Geochem 86:137–178
Schoenemann SW, Schauer AJ, Steig EJ (2013) Measurement of SLAP2 and GISP d17O and proposed VSMOW-
SLAP normalization for d17O and 17Oexcess: Measurement of SLAP2 and GISP d17O values. Rapid Commun
Sengupta S, Pack A (2018) Triple oxygen isotope mass balance for the Earth′s oceans with application to Archean
Sengupta S, Peters STM, Reitner J, Duda JP, Pack A (2020) Triple oxygen isotopes of cherts through time.
Chem Geol 554:119789
Shahar A, Young ED, Manning CE (2008) Equilibrium high-temperature Fe isotope fractionation between fayalite
and magnetite: An experimental calibration. Earth Planet Sci Lett 268:330–338
Shahar A, Ziegler K, Young ED, Ricolleau A, Schauble EA, Fei Y (2009) Experimentally determined Si isotope fractionation
between silicate and Fe metal and implications for Earth′s core formation. Earth Planet Sci Lett 288:228–234
Shanks WC (2001) Stable isotopes in seafloor hydrothermal systems: vent fluids, hydrothermal deposits, hydrothermal
alteration, and microbial processes. Rev Mineral Geochem 43:469–525
Shanks WC (2014) Stable isotope geochemistry of mineral deposits. In: Treatise on Geochemistry. Elsevier, p 59–85
Sharp ZD, Wostbrock JAG (2021) Standardization of the triple oxygen isotope system: waters, silicates,
carbonates, air and sulfate minerals. Rev Mineral Geochem 86:137–178
Shmulovich KI, Landwehr D, Simon K, Heinrich W (1999) Stable isotope fractionation between liquid and vapour in
water–salt systems up to 600°C. Chem Geol 157:343–354
Sutherland KM, Wostbrock JAG, Hansel CM, Sharp ZD, Hein JR, Wankel SD (2020) Ferromanganese crusts as
Spicuzza MJ, Day JMD, Taylor LA, Valley JW (2007) Oxygen isotope constraints on the origin and differentiation of
the Moon. Earth Planet Sci Lett 253:254–265
Starkey NA, Jackson CR, Greenwood RC, Parman S, Franchi IA, Jackson M, Fitton JG, Stuart FM, Kurz M, Larsen LM
Sun T, Bao H (2011a) Non-mass-dependent 17O anomalies generated by a superimposed thermal gradient on a
rarefied O2 gas in a closed system. Rapid Commun Mass Spectrom 25:20–24
Sun T, Bao H (2011b) Thermal-gradient-induced non-mass-dependent isotope fractionation. Rapid Commun Mass
Surma J, Assonov S, Herwartz D, Voigt C, Staubwasser M (2018) The evolution of 17O-excess in surface water of the
arid environment during recharge and evaporation. Sci Rep 8:4972
Surma J, Assonov S, Staubwasser M (2021) Triple oxygen isotope systematics in the hydrologic cycle. Rev
Tanaka R, Nakamura E (2013) Determination of 17O-excess of terrestrial silicate/oxide minerals with respect to Vienna
Standard Mean Ocean Water (VSMOW): 17O-excess of terrestrial silicate/oxide minerals. Rapid Commun Mass
Taylor (1978) Oxygen and hydrogen isotope studies of plutonic granitic rocks. Earth Planet Sci Lett 38:177–210
Thiemens MH, Lin M (2021) Discoveries of mass independent isotope effects in the Solar System: past, present and
future. Rev Mineral Geochem 86:35–95
Thiemens MH, Heidenreich JE (1983) The mass-independent fractionation of oxygen: a novel isotope effect and its
Troll VR, Weis FA, Jonsson E, Andersson UB, Majidi SA, Högdahl K, Harris C, Millet MA, Chinnasamy SS,
Kooijman E, Nilsson KP (2019) Global Fe–O isotope correlation reveals magmatic origin of Kiruna-type
apatite-iron-oxide ores. Nat Commun 10:1712
Truesdell AH (1974) Oxygen isotope activities and concentrations in aqueous salt solutions at elevated temperatures:
Consequences for isotope geochemistry. Earth Planet Sci Lett 23:387–396
Urey HC (1947) The thermodynamic properties of isotopic substances. J Chem Soc 562–581
Voarintsoa NR, Barkan E, Bergel S, Vieten R, Affek HP (2020) Triple oxygen isotope fractionation between CaCO3
Voigt C, Herwartz D, Dorador C, Staubwasser M (2020) Triple oxygen isotope systematics of evaporation and mixing
processes in a dynamic desert lake system. Hydrol Earth Syst Sci Discuss https://doi.org/10.5194/hess-2020-255
Wiechert U (2001) Oxygen isotopes and the Moon-forming giant impact. Science 294:345–348
Wostbrock JA, Sharp ZD, Sanchez-Yanez C, Reich M, van den Heuvel DB, Benning LG (2018) Calibration and
Wostbrock JAG, Brand U, Coplen TB, Swart PK, Carlson SJ, Sharp ZD (2020b) Calibration of carbonate–water triple oxygen
isotope fractionation: seeing through diagenesis in ancient carbonates. Geochim Cosmochim Acta 288:369–388
Geochem 86:367–400
Yeung L, Hayles JA (2021) Climbing the top of Mount Fuji: Uniting theory and observations of oxygen triple isotope
systematics. Rev Mineral Geochem 86:97–135
Yui T-Fu, Rumble DIII, Lo C-H (1995) Unusually low d18O ultra-high-pressure metamorphic rocks from the Sulu
Terrain, eastern China. Geochim Cosmochim Acta 59:2859–2864
322 Herwartz
Zachos J (2001) Trends, rhythms, and aberrations in global climate 65 Ma to present. Science 292:686–693
Geology 45:667–670
Zakharov DO, Bindeman IN, Serebryakov NS, Prave AR, Azimov PY, Babarina II (2019a) Low d18O rocks in the
Zakharov DO, Bindeman IN, Tanaka R, Friðleifsson GÓ, Reed MH, Hampton RL (2019b) Triple oxygen isotope systematics
as a tracer of fluids in the crust: A study from modern geothermal systems of Iceland. Chem Geol 530:119312
Zakharov DO, Marin-Carbonne J, Alleon J, Bindeman IN (2021) Triple oxygen isotope trend recorded by
Precambrian cherts: A perspective from combined bulk and in situ secondary ion probe measurements.
Rev Mineral Geochem 86:323–365
Zeebe RE (2014) Kinetic fractionation of carbon and oxygen isotopes during hydration of carbon dioxide. Geochim
Acta 57:1079–1091
Vol. 86 pp. 323–365, 2021 10
Triple Oxygen Isotope Trend Recorded by

Precambrian Cherts: A Perspective from Combined
Bulk and in situ Secondary Ion Probe Measurements
D.O. Zakharov, J. Marin-Carbonne, J. Alleon
Institute of Earth Sciences
University of Lausanne
Lausanne
1015 Switzerland
david.zakharov@unil.ch, johanna.marincarbonne@unil.ch, julien.alleon@unil.ch
I.N. Bindeman
Department of Earth Sciences
University of Oregon
Eugene, OR 97403
USA
bindeman@uoregon.edu
INTRODUCTION
The surface temperature of a planet is one of the key parameters that define its potential
habitablity. On Earth, the temperature regime sustained by the seawater column provided one
of the necessary conditions for the evolution and prosperity of life. Ancient marine chemical
sediments such as cherts and carbonates offer an opportunity to reconstruct the seawater
temperature throughout geologic history, due to temperature-dependent fractionation of
oxygen isotope ratios between the aqueous and mineral species. Pioneered by early work on
the fractionation of 18O/16O ratio between calcite and seawater (McCrea 1950; Epstein et al.
1951; Urey et al. 1951), isotope investigations of marine carbonates paved our knowledge of the
paleoclimate and advanced our understanding of the feedbacks that exist in nature, especially
those that are driven by the rise in atmospheric pCO2 (Raymo and Ruddiman 1992; Veizer et
al. 1999; Berner and Kothavala 2001; Royer et al. 2004; Miller et al. 2005). The isotope ratio
of oxygen measured in marine foraminifera and carbonate sediments is a trusted proxy for
ocean temperatures during the past ~50 Myr (Imbrie et al. 1984; Berner and Kothavala 2001;
Lisiecki and Raymo 2005). Older sediments, including the early Cenozoic ones (Sexton et al.
2006; Raymo et al. 2018), and especially those exposed on land, require a great deal of detailed
analyses and careful treatment of uncertainties that originate from secondary processes such
as late diagenesis and metamorphism that together obscure the primary signal generated in the
seawater column. Siliceous sediments such as cherts, diatoms, and porcelanites are also used
as powerful temperature-sensitive seawater and lacustrine records (Brandriss et al. 1998; Leng
and Marshall 2004; Swann and Leng 2009). However, these samples, especially cherts, cannot
compare to the precision offered by carbonates because the rock is generated during diagenesis
and crystallization of microquartz from siliceous ooze sediments. Due to the conditions imposed
by this diagenetic recrystallization within the sediment, which might take place several million
years after the deposition, the microquartz oxygen isotope ratios do not record the seawater
column temperature, where the initial silica precipitation occurred biotically or abiotically.

1943-2666/21/0086-0010$05.00 (online)
324 Zakharov et al.
Nonetheless, regarding the ancient pre-Cenozoic marine record Kolodny and Epstein (1976)
rightfully noted that “in spite of these limitations, it is of interest to study δ18O values of the
cherts as a possible imperfect indication of temperature variation of the ocean floor during the
Earth’s history beyond the time for which there are well-preserved carbonate skeletal remains.”
The key observation was made based on studies since the early 1970’s: the average
18
O/16O ratio of pre-Cenozoic and older carbonates and cherts decreases with age at the rate of
about 2‰ per every 500 Ma (Fig. 1; Shields and Veizer 2002; Prokoph et al. 2008; Veizer and
Prokoph 2015 and references therein). The 18O/16O ratio of Archean cherts and carbonates is
10−12‰ lower than the average of Phanerozoic counterparts, which, expressed in δ18O relative
to Vienna Standard Mean Oceanic Water (VSMOW), has a value of +28‰ (see Fig. 1). The
conventional delta notations used here are defined as 103∙[Rsample/RVSMOW – 1], where R stands
for isotope ratios of oxygen and hydrogen, i.e., 18O/16O, 17O/16O and D/H. This secular trend
has been interpreted in three different ways: i) precipitation in hotter oceans (see Fig. 1; up to
55–85 °C in Knauth and Lowe 2003; Robert and Chaussidon 2006); ii) much lower 18O/16O
ratio of seawater while ocean temperature remains similar to modern (e.g., Kasting et al. 2006);
iii) diagenetic obliteration of the primary signal (Degens and Epstein 1964; Ryb and Eiler
2018). A combination of these three hypotheses is also a possibility (for review refer to Jaffrés
et al. (2007) and Discussion in Bindeman 2021, this volume). The first two interpretations
have very different and important implications for the secular evolution of Earth’s systems
during the vast majority of geologic history until about 400 Ma (Fig. 1). Significantly higher
temperatures in Precambrian oceans require high concentrations of atmospheric greenhouse
gases testing the extent of the ‘faint young Sun paradox’ (Kasting 1993; Pavlov et al. 2000;
Berner and Kothavala 2001). Such an explanation would be at odds with the multimillion year-
long snowball Earth glacial episodes of the Paleo- and Neoproterozoic (Hoffman et al. 1998;
carbonates
cherts
30
1000lnαsilica-water at 25 °C
d18O, ‰ VSMOW
20
1000lnαsilica-water at 70 °C
10
0 ice-free world seawater

0 1000 2000 3000
Age, Ma
Figure 1. The oxygen isotope trend defined by carbonates (blue) and cherts (green) from the Archean
(to the right) to present day (at origin). The values are compiled from Shields and Veizer (2002), Knauth
(2005), and Prokoph et al. (2008). The moving averages (solid bands) are plotted using the local polyno-
mial regression LOESS method. The dotted lines approximate the composition of least-altered samples,
drawn based on the formula: mean + upper limit of the interquartile range multiplied by 1.5, i.e., without
the outliers. The temperature-dependent fractionation of δ18O between silica and water in equilibrium is
shown on each side of the plot assuming that seawater oxygen isotope composition stayed around −1‰.
Triple Oxygen Isotope Trend Recorded by Precambrian Cherts 325
Kirschvink et al. 2000). On the other hand, a significantly lower 18O/16O ratio of seawater is
difficult to maintain on geologically long time scales, as spreading at mid-ocean ridges and
seawater–basalt reactions produce fluids that are buffered by the oxygen isotope composition
of the oceanic crust (Muehlenbachs and Clayton 1972; Holland 1984). This would imply a
different isotope budget of the terrestrial hydrosphere in which dominant fluxes of exchanged
oxygen are not controlled by high-temperature reactions with mantle-derived rocks (i.e., oceanic
crust). However, previous investigations of ophiolites and Precambrian altered seafloor rocks
do not provide any evidence for such conditions on the geologic timescale (Muehlenbachs and
Clayton 1972; Holland 1984; Holmden and Muehlenbachs 1993; Muehlenbachs 1998; Furnes
2004; Turchyn et al. 2013; Sengupta and Pack 2018; Zakharov and Bindeman 2019).
Besides the explanations entailing significant differences in the ocean temperature or in
its oxygen isotope composition, this trend might also be attributed to the secular evolution of
diagenetic conditions. Recent advances in clumped carbonate measurements (Eiler 2011;
Cummins et al. 2014; Watkins and Hunt 2015; Bergmann et al. 2018; Ryb and Eiler 2018)
enable researchers to estimate the temperature of solid-state reordering in carbonate structure
independently of the isotope ratio of equilibrium fluid, which places constraints on the preservation
of original temperature signals making the effect of late diagenetic recrystallization resolvable.
However, there is no method like clumped isotopes that can help to resolve the extent of diagenetic
obliteration for siliceous sediments, yet low δ18O meteoric waters and/or metamorphic overprints
(Perry and Lefticariu 2003; Liljestrand et al. 2020) are also frequently invoked to explain the
low 18O/16O values of Precambrian chert. The effects of diagenetic transformations recorded
in ancient siliceous sediments have been previously demonstrated with silicon and oxygen
isotope in situ measurements by secondary ion probe mass spectrometry (SIMS). Such studies
(Robert and Chaussidon 2006; Marin et al. 2010; Marin-Carbonne et al. 2012; Stefurak et al.
2015; Cammack et al. 2018) highlighted the complexity of deposition and preservation of cherts
through time. While diagenetic effects are undoubtedly important for biogenic and chemical
precipitates, there is a potential for at least partial preservation of original seawater signals. These
previous in situ-based studies suggest that the original marine signals could be recognized by a
combination of methods, e.g., high-resolution measurements of isotope ratios, trace elements,
and non-traditional approaches such as triple oxygen and clumped isotopes.
Nonetheless, it is important to stress that the temporal oxygen isotope trend (Fig. 1) is
defined not only by carbonates and cherts, but also by shales (Bindeman et al. 2016, 2018),
phosphates (Karhu and Epstein 1986), and iron oxides (Galili et al. 2019), all of which show
a similar first-order transition from low to high 18O/16O ratios across the geologic timescale,
potentially reflecting a secular evolution of the hydrosphere. The oxygen isotope temporal
trend remains significant, yet its interpretation is one of the long-lasting debates in isotope
geochemistry that has not been fully reconciled.
With the advancing ability to measure triple oxygen isotopes with the precision of 0.01‰ or
better, the fractionation of two isotope ratios, 17O/16O and 18O/16O, gives rise to the opportunity
to distinguish between processes that accompany crystallization, deposition and alteration of
marine sediments (Matsuhisa et al. 1978; Rumble et al. 2007; Cao and Liu 2011; Pack and
Herwartz 2014; Sharp et al. 2016). The recent studies (Levin et al. 2014; Sengupta and Pack
2018; Hayles et al. 2019; Liljestrand et al. 2020) presented high-precision triple oxygen
isotope measurements of cherts spanning in age from Archean to Phanerozoic. These studies
demonstrate the simultaneous change in 17O/16O and 18O/16O values of cherts across geologic
history. In theory, it could be possible to simultaneously constrain the temperature and the fluid
isotope composition using temperature-dependent calibrations and seawater evolution equations.
Conclusions about insignificantly different seawater oxygen-isotope composition across the
geologic time scale arise in the community, mostly calling for alteration of primary signals
recorded by cherts (see Sengupta and Pack 2018; Liljestrand et al. 2020).
326 Zakharov et al.
In this chapter, we focus on understanding the triple oxygen isotope signals recorded
in ancient siliceous sediments deposited during the first two thirds of geologic history,
particularly during the Archean (3.8–2.5 Ga) and the Paleoproterozoic (2.5–1.6 Ga), as it
should be possible to reconstruct original isotope signatures of marine signals provided careful
investigation of diagenetic conditions and provenance. In general, the Precambrian (4–0.5 Ga)
encompasses the period of Earth’s evolution when conditions critical for sustaining complex
life were established, permitting the biogeochemical cycles and biological diversification that
gave our planet its modern appearance (Broecker and Broecker 1985). Below we present an
overview of some emblematic Precambrian chert formations. We also provide an overview
of how triple oxygen isotopes can be used to gain additional insights into the origin of
silica. Further, new oxygen isotope data is presented. In this study we utilized two distinct
methods that operate on different scales (macro and micro) with the goal to develop a better
understanding of what signals are preserved in ancient cherts and how to interpret them. The
availability of published triple oxygen isotope calibrations (Sharp et al. 2016; Wostbrock et
al. 2018), combined with the large fractionation between silica and water at low temperatures,
offers a good opportunity to take a close look at the Precambrian chert record and attempt to
understand the overall increase in 18O/16O values across the geologic timescale. Furthermore,
silica is a simple matrix suitable for high-precision SIMS analysis of d18O with several μm
resolution (Valley and Graham 1996; Kelly et al. 2007; Marin et al. 2010; Pollington et al.
2011; Cammack et al. 2018). The large radius SIMS is currently incapable of generating d17O
with the required ±0.01‰ precision, but a combination of bulk triple oxygen isotopes based
on ~1 mg of material with high spatial resolution SIMS analyses for d18O is currently the
best available avenue to investigate the role of original depositional vs. secondary diagenetic
signals recorded in the chert samples. We pursue this in this Chapter.
PART I: PRECAMBRIAN CHERTS AS AN ARCHIVE OF

ANCIENT SILICA CYCLE
Chert is a common sedimentary rock that is composed mostly of silica (> 90 wt. % SiO2),
present either as opal, amorphous silica, or quartz. Despite this simple chemical formula
and relatively simple silica oxygen isotope fractionation, chert could be formed as a direct
precipitate from water, or by silicification, i.e., replacement of pre-existing rock. Precambrian
cherts are particularly interesting because they represent an ancient archive of marine silica
cycle prior to the advent of silica-secreting marine organisms such as radiolarians, sponges and
diatoms (Fig. 2). Modern seawater has a very low silica content, around 0.1 mmol/kg of H4SiO4,
that is biotically converted to opaline silica, opal-A, and then deposited on the seafloor. After
the deposition, silica undergoes a series of transitions to an aggregate of cristobalite and opal
(opal-CT) and finally to microcrystalline quartz (Hesse 1988; Maliva and Siever 1988; Maliva
et al. 1989). Voluminous chert deposits in the Precambrian document abiotic precipitation of
silica from silica-saturated seawater and/or silicification reactions that occurred close to the
seafloor interface (Fig. 2). The upper limit of dissolved marine silica throughout geologic
history can be perhaps be taken as the point of saturation of seawater with amorphous silica
that would occur at the concentration of around 3 mmol/kg, 22 °C and pH of 7.9 (Okamoto
et al. 1957; Siever 1992). Multiple prior studies (Knauth and Lowe 2003; Van den Boorn et
al. 2007) also have highlighted the complex origin of chert, especially during the Archean.
It is common to find Archean cherts as partially formed through silicification of a precursor
lithology (Van den Boorn et al. 2007). The precursors could be recognized by the relicts of
carbonates, evaporitic minerals, and altered volcanic fragments embedded into cherts, partially
or completely silicified. In other cases, bedded cherts appear to be purely chemical precipitates
from a seawater solution; even within the same stratigraphic context these differences are
apparent (Knauth and Lowe 1978). Furthermore, unlike the Phanerozoic chert deposits,
A B
riverine riverine
bio.silica hydrothermal amorphous

silica?
auth.silica low-T auth.silica

input low-T
exchange
C 5-35 Precipitation Silica dissolved Ice-free seawater

(H4SiO4)
δ18O = -1 ‰
2
Δ17O ≈ 0 ‰
Opal A
Temperature (°C)
Sediment porewater
Diagenesis
Opal CT Amorphous
δ O from -4 to +1 ‰
18
40 Dissolution- precursor
Precipitation Δ O > 0 ‰ (?)
17
60 Fibrous
quartz
80
Hydrothermal
Crystallization Microquartz fluids
100-350 MORB-buffered:
δ O = +3 ‰
18
Δ17O0.528 = -0.05 ‰
Re-crystallization
Megaquartz Quartz vein
Metamorphism
Figure 2. Schematic and simplified drawing of Phanerozoic (A) and Precambrian (B) silica cycles. The
key distinction is the presence of a biologically-mediated silica cycle in the Phanerozoic and its absence
in the Precambrian. In addition, early Precambrian continents likely lacked vast subaerial exposures of
continental crust available for subaerial weathering. The important fluxes of dissolved silica are riverine
and hydrothermal contributions. In addition, low-temperature off-axis interaction between seawater, pore-
waters, siliceous sediments (oozes), and mid-ocean ridge basalts also contribute to the marine silica budget.
In the Precambrian, when silica-secreting critters were absent, the marine H4SiO4 concentration was likely
higher, causing abiotic precipitation of amorphous silica. Panel C depicts deposition of opaline silica,
which is accompanied by submarine diagenetic reactions between siliceous sediments and pore fluids with
phase transitions to opal-A, opal-CT, and crystallization of quartz. The panel shows the ways to achieve
precipitation and early diagenesis of modern (solid triangle) and Precambrian silica (dotted). The figure
is adapted and modified from (Knauth 1994; Marin-Carbonne et al. 2013). The effects of late diagenesis,
metamorphism, and late hydrothermal cycling of fluids are shown with precipitation of megaquartz and
quartz veins, which frequently occur in samples of Precambrian cherts. The oxygen isotope values of pos-
sible silica-saturated fluids in modern oceans are shown.
many Archean cherts form fracture-filling, dike-like structures often interpreted as a result of
hydrothermal delivery of Si induced by chemical interaction between seawater and seafloor
(Knauth and Lowe 1978; Paris et al. 1985; Nijman and Valkering 1998; Hofmann and Bolhar
2007; Hofmann and Harris 2008). Vast occurrences of silicification and replacement textures
(e.g., komatiite with quartz-formed spinifex texture) inform of silica influx from submarine pore
or vent fluids and by adsorption onto pre-exiting mineral surfaces (Siever 1992). Production
of silica-rich fluids is commonly attributed to hydrothermal, exhalative activity caused by
seawater–rock interaction at greenstone belts (Paris et al. 1985; Konhauser et al. 2001). This
presents evidence that Archean cherts formed via multiple pathways of dissolved silica from
seawater to seafloor and sub-seafloor. Therefore, reconstructing oceanic paleotemperature
from a silicified volcanic precursor or a hydro-thermal silica deposit might not produce an
accurate temperature proxy of the ocean water above.
328 Zakharov et al.
The origin of Precambrian chert is further complicated by microtextural diversity and

micrometer-scale isotope heterogeneities. Petrographic and in situ studies have revealed that
even at a micrometer scale cherts can be highly heterogeneous in oxygen and silicon isotopes,
and thus have recorded multiple precipitation events (Knauth 1994; Marin-Carbonne et al.
2012; Stefurak et al. 2015; Cammack et al. 2018). Cherts frequently visibly combine multiple
morphologies: microcrystalline quartz, chalcedonic, megaquartz, and quartz veins (Knauth
1994). The most abundant silica phase is typically microquartz, which consists of α-quartz
crystals with size ranging between 2 to 20 μm, irregular boundaries, and sweeping extinction
(Folk and Weaver 1952; Hattori et al. 1996; Maliva et al. 2005). Microquartz and chalcedonic
quartz are considered as the first silica phase formed during chert precipitation and thus
the most pristine (Knauth and Lowe 1978; Knauth 1994; Marin et al. 2010). Megaquartz
is composed of relatively large crystals with size greater than 20 µm across with clear
granoblastic textures. It forms by secondary, replacement process, as void-filling cements,
and also appears in heavily metamorphosed species. As chert might have experienced further
post-depositional fluid circulation, quartz veins are often abundant and consist of quartz grains
with sizes between 10 and 200 µm. Fluid inclusions trapped in such quartz indicate that veins
formed as a result of fluid circulation at high temperatures, often between 150−300 °C (Harris
et al. 2009; Marin-Carbonne et al. 2011; Farber et al. 2015). These observations document the
prolonged history of fluid cycling and quartz recrystallization that occurred after the seafloor
deposition of marine silica (Fig. 2). Thus, it is important to document the texturally different
generations of silica phases and their isotope composition by SIMS.
The collection of early Precambrian cherts
As outlined above, it is critical to understand the petrographic and geologic context of
chert samples that are used to study the isotope composition of the terrestrial hydrosphere
on the time scales ranging from Archean to modern days. Extending to as early as ~3.8 Ga
and especially common in Archean greenstone belts, ancient cherts have been the subject of
numerous studies (Perry and Tan 1972; Knauth and Epstein 1976; Knauth and Lowe 2003;
Perry and Lefticariu 2003; Knauth 2005; Robert and Chaussidon 2006) that focused on the
temporal trend defined by their oxygen isotope values and the implications for the global
evolution of the planet. Being spatially and temporally close to banded iron formations (BIFs),
Archean and Proterozoic cherts are still actively studied with the goal of understanding the
geochemical cycles intrinsic to the first half of the Earth’s history, including the evolution of
atmospheric oxygen content and its consequences on marine species (Holland 1984; Siever
1992; Canfield 1998; André et al. 2006; Poulton and Canfield 2011; Bekker et al. 2014). Further,
cherts are well known to finely preserve morphological details of fossilized microorganisms in
the geological record dating back to the Paleoproterozoic eon (Javaux and Lepot 2018). The
emblematic Precambrian chert formations (see Fig. 3) that have been extensively studied came
from four different Archean greenstone belts and Proterozoic basins:
1. The 3.8 Ga Isua Greenstone Belt (Ueno et al. 2002; van Zuilen et al. 2003; André et
al. 2006; Whitehouse and Fedo 2007; Heck et al. 2011; Brengman et al. 2016)
2. The 3.5–3.4 Ga Warawoona Group from the Pilbara Craton (Knauth and Lowe 1978;
Robert and Chaussidon 2006; Van den Boorn et al. 2007; Marin-Carbonne et al.
2012; Cammack et al. 2018).
3. The 3.4–3.3 Ga Onverwacht Group from the Barberton Greenstone Belt (Knauth and
Lowe 1978; Robert and Chaussidon 2006; Abraham et al. 2011; Marin-Carbonne et
al. 2011; Geilert et al. 2014; Stefurak et al. 2015).
4. The 1.9 Ga Paleoproterozoic Gunflint Formation of the Animikie group (Winter and
Knauth 1992; Robert and Chaussidon 2006; Marin et al. 2010).
Dresser Chert
(3. 48- 3.46 Gyr-Old)
Warawoona Group
Pilbara Craton
Australia
A B C
Buck Reef Chert

(3.41 Gyr-Old)
Onverwacht Group,
Barberton Greenstone
Belt
South Africa
D E F
Mendon Chert
(3.28 Gyr-Old)
Onverwacht Group,
Barberton Greenstone
Belt
South Africa
G H I
Gunflint Chert
(1.88 Gyr-Old)
Animike group
Canada
J K L
Figure 3. Emblematic Archean and Proterozoic chert formations. A–Landscape view of the Dresser For-
mation at the North Pole locality and the outcrops of (B) stromatolitic chert, and (C) stromatolitic and
bedded cherts. D–Landscape view of the Buck Reef Chert with (E) white and black bedded chert, and
(F) evaporitic chert. G–Outcrops of the Mendon chert with laminated black and white cherts (H and I).
J–Kakabeka Falls, Ontario in Canada, a locality for the Gunflint Formation. (K) Schreiber Beach locality
with outcrops of stromatolitic chert. (L) A detailed view of the Schreiber Beach stromatolitic chert. Photo-
graphs are courtesy of N. Olivier, P. Sans-Jofre, C. Thomazo and R. Shapiro.
Here we will review the main characteristics of these emblematic chert formations, but
we will not discuss the Isua chert as these rocks have experienced a higher metamorphic grade
(amphibolite facies) and have a more complex history of post-depositional modifications
(Dauphas et al. 2004; André et al. 2006). In this study we do not consider the chert bands
associated with banded iron formations (BIFs). They are also very common in the Precambrian
and host large amounts of silica and iron oxides, reflecting high amounts of H4SiO4 and Fe2+
dissolved in ancient seawater. Although their deposition occurred in deep and calm water of
the Precambrian oceans, the origin of BIF-associated chert deposits is often considered to be
different from the bedded cherts that were deposited in shallower environments. Moreover,
the BIF silica has very low δ30Si (down to −4‰; Chakrabarti et al. 2012) and very low δ18O
values (down to 7‰; Heck et al. 2011) that potentially suggest a different pathway of silica
precipitation, involving hydrothermal fluids, diagenesis at high temperature, precipitation
induced by temperature variation in seawater column, and/or adsorption of silica on iron
oxides (Konhauser et al. 2017; Schad et al. 2019).
330 Zakharov et al.
The Warawoona Group (3.52–3.43 Ga, Pilbara Craton, Australia). The Warawoona
Group is located in the Pilbara Craton (Western Australia) and consists of alternating volcanic
units with sedimentary successions and it is well-known for the abundance of well-preserved
stromatolites (Fig. 3; Allwood et al. 2006). The Strelley Pool Formation unit consists of a
sedimentary succession that was initially deposited between 3.43 and 3.35 Ga in a shallow
marine environment (Hickman 2008). There, seven distinct stromatolite facies (see Fig. 3) that
vary within paleoenvironments of a shallow-water facies carbonate platform were described
(Allwood et al. 2007). Detailed multi-scale morphological and geochemical studies of the
stromatolites and organic matter have led to a broad consensus on the presence of early life
in this Formation (Wacey 2010; Bontognali et al. 2012; Flannery et al. 2018). The underlying
Dresser Formation (3.5 Ga) consists of an alternating succession of chert, evaporite, and pillow
basalt (Van Kranendonk 2006). The base unit consists of silicified carbonate. Most of the units
in the Pilbara craton have experienced hydrothermal fluid circulation and metamorphism (Lowe
1983; Buick and Dunlop 1990; Hickman 2008; Wacey 2010; Sugitani et al. 2013), thus high-
temperature vent fluids are suggested to be the source of the silicification of the chert. The Dresser
Formation has experienced extensive hydrothermal alteration in the range of 250−300 °C that has
led to the crystallization of barite (Van Kranendonk et al. 2008; Harris et al. 2009). One sample
from the Dresser Formation, PPRG 006, has been analyzed in this study. This sample has been
previously extensively studied for organic matter characterization (e.g., Binet et al. 2008) and for
O and Si isotope composition (Robert and Chaussidon 2006; Marin-Carbonne et al. 2011, 2012).
Onverwacht Group (3.55–3.26 Ga, Kaapvaal Craton, South Africa). The Onverwacht Group
(Fig. 3) is the oldest unit of the Barberton Greenstone Belt in the Kaapvaal craton and consists of
a sequence of ultramafic and volcanic rocks alternating with sedimentary chert layers (Lowe and
Byerly 1999). The Onverwacht includes several chert layers: the Sandspruit, Theespruit, Komati,
Hooggenoeg, Kromberg, and Mendon Formations. Cherts from the Onverwacht Group have
various proposed origins, from silicified volcanic rocks, like the Middle Marker chert (Lanier
and Lowe 1982), hydrothermal exhalative products (de Wit et al. 1982; Hofmann and Harris
2008), marine chemical sediments (Knauth and Lowe 1978; Tice and Lowe 2006). Bulk O and
Si isotope studies on 3 different volcanic chert successions from the Onverwacht Group have
also suggested that cherts were late stage silicified rocks formed from the alteration of basalt
(Abraham et al. 2011). Further O and Si isotope studies on the Buck Reef chert of the Kromberg
Formation have reported stratigraphic changes and support a marine origin for this chert (Hren
et al. 2009; Geilert et al. 2014; Stefurak et al. 2015). The Buck Reef chert (3.416 Ga; Fig. 3)
consists of the largest accumulation of pure and carbonaceous silica deposits of the Archean rock
record (Kruener et al. 1991; Lowe and Byerly 1999; de Vries et al. 2010). As some authors have
found no evidence for hydrothermal activity (Tice and Lowe 2004), others have proposed that
part of the Buck Reef chert was silicified during late hydrothermal circulation (Hofmann and
Bolhar 2007; Hofmann and Harris 2008). One of the samples (PPRG 193) in current study came
from a black band in the Buck Reef chert of the Kromberg Formation.
The Mendon Formation (3.34−3.26 Ga) is also a part of the Onverwacht group (Fig. 3)
and is characterized by 50−75 m of black, black and white, and ferruginous chert (Lowe and
Byerly 1999; Trower and Lowe 2016), and was deposited by slow sedimentation in quiet
water settings during a period of local volcanic quiescence (Stefurak et al. 2014). Previous
studies have highlighted the influence of post depositional fluid circulation, especially on
S and Fe isotope compositions (Busigny et al. 2017; Marin-Carbonne et al. 2020), and the
metamorphic overprint likely occurred at 300 °C (Lowe and Byerly 1999). Extensive study of
O and Si isotope ratios in some Mendon chert samples has revealed Si isotope variations on the
micrometer scale (Stefurak et al. 2015) and systematic texture-specific δ30Si differences, while
there is no obvious correlation between the chert texture and δ18O values measured in the
same area. This study proposes that the bulk Si isotope ratios are likely preserved but cannot
be used as a robust quantitative paleothermometer due to the occurrence of inter-sediment
fractionations distributed across multiple diagenetic silica phases. A previous study that
combined fluid inclusion thermometry and O–Si isotope measurements by ion probe (Marin-
Carbonne et al. 2011) also found heterogenous compositions. Further, a detailed petrographic
observation of the Mendon chert from the drillcore BARB4 have led the authors to propose
a new mode of silica deposition as granules (Stefurak et al. 2014). Three samples (PPRG
198, PPRG 200 and Onverwacht 2 of 16.09.65) have been previously studied for oxygen
and silicon isotope composition (Robert and Chaussidon 2006, Marin-Carbonne et al. 2011,
2012) and are used in this study for triple oxygen isotope analysis. Despite the preservation of
sedimentological features in some of the Barberton Belt cherts, it is important to keep in mind
that all of the units of the Onverwacht Group underwent metamorphism under greenschist
facies conditions at temperatures around 300−350 °C. Numerous intrusions were emplaced
during and after the deposition of the Onverwacht Group of the Barberton Belt (de Ronde and
de Wit 1994; Lowe 1994) and promoted an elevated geothermal gradient, episodes of contact
metamorphism, and fluid circulation (e.g., de Ronde et al. 1994), all of which affected to some
extent the preservation of original marine signals.
Gunflint cherts (1.88 Ga, Canada). The Gunflint Iron Formation (Fig. 3) is the middle
unit of the Animikie group of Northwestern Ontario and is conformably overlain by the Rove
Formation (Floran and Papike 1978). The stratigraphy of the Gunflint Iron Formation was
previously described by several authors (e.g., Goodwin 1956; Simonson 1987) who defined
several members and lithofacies. Winter and Knauth (1992) defined facies ranging from deep
to shallow water environments. According to these studies, the Gunflint chert was deposited
in seawater with no evidence for a mixing between seawater and local precipitation (Carrigan
and Cameron 1991; Winter and Knauth 1992). The Gunflint Formation has experienced late
hydrothermal circulation and metamorphism due to the Penokean Orogeney (40 Myr after
the deposition of the Gunflint) and from the Duluth intrusion (1.1 Ga). These events induced
maturation of organic matter and isotope re-equilibration between constituent minerals of the
nearby Biwabik Iron Formation of the same Animikie group (Hyslop et al. 2008; Marin et al.
2010), yet the Gunflint Formation remained relatively well-preserved with a temperature of
metamorphism under 200 °C (Alleon et al. 2016). The 5 chert samples that have experienced
different diagenetic temperatures have been selected from the Precambrian Paleobiology
Research Group (PPRG 1,289 Schreiber Beach and PPRG 1,284 Frustration Bay) and
from the collection of Stanley Awramick (3 of 06/30/84, 5 of 06/28/84 and 1c of 06/29/84).
The samples studied here have already been previously investigated in great detail by Marin et
al. (2010), Alleon et al. (2016), and Marin-Carbonne et al. (2012).
The chert of the Gunflint Formation occupies a special place in Precambrian paleontology.
The discovery in the 1950s and 1960s of fossils of microorganisms in the Gunflint cherts (Tyler
and Barghoorn 1954; Barghoorn and Tyler 1965), coupled with geochronological analysis of
ash-hosted zircons (1878 ± 2 Ma; Fralick et al. 2002), changed the way scientists view the earliest
records of life on Earth (Knoll 1992). Morphologically distinct filamentous and spheroidal
microfossils were described in the Gunflint cherts, and were interpreted to be the remains of
oxygenic photosynthetic bacteria (cyanobacteria), sulfate reducing bacteria, iron-mineralizing
microorganisms, and heterotrophs (Barghoorn and Tyler 1965; Cloud et al. 1965; Awramik and
Barghoorn 1977; Strother and Tobin 1987; Lepot et al. 2017). Carbon isotopic compositions of
fossils identified as Huroniospora and Gunflintia minuta species were reported to be consistent
with oxygenic photosynthesis (House et al. 2000; Williford et al. 2013), although anoxygenic
photosynthesis, chemoautotrophy, or heterotrophy were also considered as plausible metabolisms
(Lepot et al. 2017). The excellent state of preservation and the diversity of fossilized forms in
the Gunflint cherts have led to a universal agreement on their authenticity, and they were thus
frequently used as a reference for evaluating the origin of organic microstructures in Archean
rocks (Schopf et al. 2002; Wacey et al. 2012; Brasier et al. 2015; Alleon et al. 2018).
332 Zakharov et al.
Silica precipitation and diagenesis, and the effect of oxygen isotopes

Our understanding of sedimentary chert genesis has little evolved since early work in
the 1970s (Knauth and Epstein 1975; Murata et al. 1977; Hesse 1988; Maliva and Siever
1988; Maliva et al. 1989, 2005; Matheney and Knauth 1993; Knauth 1994). Transfer of
marine oxygen isotope signatures starts with growth of skeletal silica, frustule in diatoms,
at seawater-column temperatures. After diatoms die, the frustules are deposited on the
seafloor to form siliceous ooze. This then undergoes a series of transitions at bottom-water
and subseafloor temperatures: from opaline frusta, to opal-A and opal-CT, to cristobalite and
finally to microquartz. These transitions are accompanied by dissolution–precipitation and
by the interaction between the siliceous sediment and pore fluids at the temperatures from
20 to 80 °C. The timing of the transitions is then governed by the burial rates and the local
geothermal gradient (Kolodny and Epstein 1976; Hesse 1988; Behl et al. 1994; Geilert et al.
2014). Even in marine cores recovered from Tertiary and Cretaceous sediments that have never
undergone high-temperature metamorphism and were never exposed to the low δ18O meteoric
waters intrinsic to subaerial environments, the phase transitions from amorphous precursor to
microquartz are accompanied by several ‰ shifts in δ18O (Knauth and Epstein 1975; Tatzel et
al. 2015; Yanchilina et al. 2020). Shifts of around 15‰ were documented by Yanchilina et al.
(2020). The well-studied Miocene chert from the Monterey Shale Formation is a key locality
showing the evolution of oxygen isotope ratios also varying 10‰ across these phase transitions
(Murata et al. 1977). Moreover, an about 8‰ change in δ18O values is documented in diatom
frustules from antemortem to post-mortem states (Dodd et al. 2012). These variations are best
explained by step-wise maturation of the sediment that starts at sediment–water interface and
continues deep within the sediment at the local diagenetic conditions.
As cherts contain 1−2 wt. % water in form of hydroxyl groups bonded with atoms of silicon,
their δD values have been used as an additional constraint for temperature and fluid sources
(Knauth and Epstein 1975). Interaction of cherts with diagenetic and meteoric fluids then can
be documented from coupled oxygen and hydrogen isotope analysis. The δ18O–δD relationship
measured in Phanerozoic formations exposed on land presents an illustrative example of meteoric
water influencing the composition of chert (Knauth and Epstein 1976). Moreover, combined
δ18O–δD spatial profiling of a Jurassic chert nodule reveled antiphase multiple-‰ variability
across a distance of 10−20 mm, emphasizing the role of temperature during diagenesis (Sharp et
al. 2002). Thus, the crystallized microquartz does not reflect equilibrium with pristine seawater
in a straightforward way. A logical question arises: how can one still use cherts across geological
history if diagenetic processes produce 10‰ shifts in δ18O?
As described above, modern chert formation is initiated by precipitation of amorphous
silica, and its early transformation to opal-A and opal-CT (Hesse 1988, 1989). In the
Precambrian, so far there is no evidence for precipitation as amorphous silica or as opal-CT,
thus the isotope effect of these transitions is not well-established for ancient cherts. The likely
precursor of Archean chert is amorphous silica gel, either precipitated directly from seawater or
within the seafloor lithologies (Fig. 3). Following the pattern of modern amorphous silica, the
Precambrian silica gels were likely then totally dissolved and recrystallized during diagenesis
into microquartz (Winter and Knauth 1992; Knauth 1994). Logically, it is likely that the
dissolution–precipitation of amorphous silica gel during diagenesis also induced modification
of the oxygen isotope composition of quartz, similar to Phanerozoic silica deposits (Kolodny
and Epstein 1976; Murata et al. 1977; Jones and Knauth 1979; Knauth 1979).
A potential for isotope integrity of marine signals in Precambrian cherts is highlighted
by several microscale studies that collected oxygen isotope compositions measured from 10
to 20 μm pits by SIMS (Fig. 4; Marin et al. 2010; Marin-Carbonne et al. 2011, 2012, 2013;
secondary laser
ion probe fluorination
30
A B
d'18O, ‰ VSMOW
25 microquartz
20
megaquartz
15
10
5 50
1E-8 1E-6 spot size (μm)
1E+3
material used (μg)
3f
1270
1280HR
Figure 4. A–the effect of sample size on the δ′18O values determined by secondary ion probe (SIMS; left) and
by bulk laser fluorination GS-IRMS (right side). Depending on the spot size and the scale of heterogeneity,
the SIMS analysis is capable of documenting isotopically diverse generations of quartz present in chert sam-
ples. The spot sizes of 5 and 50 μm (normally 10–20 μm) of the primary beam are used to illustrate the differ-
ence; the size of the beam is more important relative to the scale of heterogeneity. The curves are probability
density functions representing random sampling of a population with dominantly high δ18O values (24−25‰)
and a smaller population with values of 17−18‰ (shown with grey bands). The bulk value determined by tra-
ditional laser fluorination GS-IRMS would present a very tight distribution centered around the mean value.
B and C–Microscale determinations of oxygen isotope ratios by SIMS in a sample from the Gunflint Forma-
tion (Schreiber Beach) across multiple petrographic varieties. The δ18O values are marked in ‰ VSMOW
next to the analyzed pits (red dots; scaled with size about 20 μm). The composition with δ18O = +24‰ is con-
sidered as the closest to the value of the primary microquartz. D–reflected light image of the exact same area
depicted in C. This sample was previously analyzed by Cameca ion probes 3f, 1270 at CRPG-CNRS (Nancy,
France), and, the most recently, by 1280HR at the SwissSIMS facility (University of Lausanne). The arrows
point to the pits left by the primary ion beam during the analysis by the respective ion probes.
334 Zakharov et al.
Cunningham et al. 2012; Stefurak et al. 2015; Cammack et al. 2018). These studies recognized
multiple generations of quartz spanning a range of δ18O values, up to 14‰ within some samples.
These samples are mounted within epoxy as pieces of chert several cm across, along with
smaller grains of standards placed in the middle of the mount. The heterogeneous compositions
are present on the scale of 10−100 μm and are sometimes accompanied by variations in trace
element concentrations and cathodoluminescence textures but are often not visible otherwise.
These observations somewhat undermine the utility of the measurements done by conventional
gas-source isotope ratio mass spectrometers (GS-IRMS) as such analyses involve bulk liberation
of O2 from 1−2 mg samples using the laser fluroination technique. A combination of texturally
controlled triple oxygen with SIMS measurements holds some promise that is explored below.
Triple oxygen isotopes in silica
The previously proposed interpretations of the 18O/16O trend (Fig. 1), be it higher temperature
of oceans, lower δ18O of seawater, or diagenetic overprint, are not mutually exclusive due to
the inability to uniquely reconstruct both fluid sources and temperatures from a single isotope
ratio. Adding a second isotope ratio, 17O/16O, provides new opportunities. Here we focus on our
new and the recently published triple oxygen isotope datasets of Precambrian cherts (Fig. 5)
in the context of their microscale heterogeneities recorded by SIMS. Other isotope systems,
such as hydrogen or silicon, could be used as an additional control on the nature of oxygen
isotope fractionation (e.g., Robert and Chaussidon 2006; Hren et al. 2009). However, these
systems might be decoupled from each other, given that silicon isotopes appear to display large
fractionations controlled by precipitation mechanisms and rates (Geilert et al. 2014, 2015) and
that hydrogen isotopes are easily reset at low temperatures (Wenner and Taylor 1973; Knauth
and Epstein 1976; Kolodny and Epstein 1976; Kyser and Kerrich 1991). In more recent years,
the triple oxygen isotope ratios 17O/16O and 18O/16O have been used to gain new insights into
geologic processes, including temperature-dependent fractionation between fluids and minerals
as well fluid source fingerprinting (Levin et al. 2014; Pack and Herwartz 2014; Passey et al.
2014; Herwartz et al. 2015; Sharp et al. 2016, 2018; Zakharov et al. 2017, 2019a,b; Bindeman et
al. 2018; Wostbrock et al. 2018; Zakharov and Bindeman 2019; Peters et al. 2020a,b).
The slope of silica–water fractionation in the triple oxygen isotope coordinates is now
calibrated and varies with temperature. For example, at temperatures from 25 to 400 °C, the
slope of silica–water equilibrium fractionation varies from 0.5244 to 0.5278 (Fig. 5; Sharp
et al. 2016; Wostbrock et al. 2018). This slope is expressed in δ′17O–δ′18O coordinates,
where each axis is a linearized notation of conventionally defined delta values, i.e.,
δ′17/18O‰ = 103 ∙ ln(1 + δ17/18O/103). The δ′17O-excess, expressed as Δ′17O = δ′17O − 0.528 ∙ δ′18O,
is a convenient way to demonstrate the offset between a point and a reference line with the slope
of 0.528 in the triple oxygen isotope coordinates. That means that for a given fluid composition,
the value of Δ′17Oquartz–Δ′17Owater would be small at high temperatures, approaching the slope
of 0.528. At infinitely high temperature, equilibrium fractionation in the triple oxygen isotope
system approaches the slope of 0.5305 (Matsuhisa et al. 1978). Since the analytical uncertainty
on the Δ′17O values is about 5‒10 ppm (0.005‒0.010‰), equilibrium quartz and fluids would
have indistinguishable values at temperatures above about 400 °C, while δ18Oquartz–δ18Owater is
+4.5‰. At temperatures below 40 °C, the δ′18O and Δ′17O of quartz are at least 32.5‰ higher and
0.1‰ lower than that of equilibrium fluid, respectively, as shown with the isotope composition
of diatoms and sponge spiculas (Pack and Herwartz 2014; Sharp et al. 2016).
In theory, having two isotope ratios, 17O/16O and 18O/16O, it is possible to solve for two
unknowns, i.e., equilibrium fluid composition and temperature. In such case, the equations
could represent evolution of marine pore fluids, or meteoric water array, and silica–water
equilibrium fractionation. While the latter has been calibrated (Sharp et al. 2016) and the meteoric
40
modern seafloor silica A
d'18O, ‰
VSMOW
30
20
10
B
−0.05
D'17O0.528, ‰
VSMOW
−0.15
modern seafloor silica
−0.25
0 1000 2000 3000
Age, Ma
C
meteoric
water
seawater
± ice 300
0.0
D'17O0.528, ‰ VSMOW
|
200
|
|
|
150 3000
| |
Age, Ma
|
100 qu 2000
| | art
MORB z-w
|
75
| ate 1000
|
re
qu
50 il. (
25 °C 0
−0.1 | | | )
90%
80%
| |
25
|
5 °C
Levin et al 2014 mixing | |
5 5 °C
Sharp et al 2016
−0.2
Hayles et al 2019
modern
Liljestrand et al 2020 seafloor silica
This study
0 10 20 30 40
d'18O, ‰ VSMOW
Figure 5. Triple oxygen isotope data on cherts compiled from previous studies (Levin et al. 2014; Hayles
et al. 2019; Liljestrand et al. 2020) and from this study recalibrated to a common silicate standard UWG−2
(Sharp and Wostbrock 2021, this volume). The data from (Levin et al. 2014) was recalibrated using the
two-point scheme using the UWG−2 and NBS28 standards. Panel A presents the same general trend in the
triple oxygen isotope dataset as in the previous conventional 18O/16O analyses (see Figure 1). The accom-
panied Δ′17O values are shown in (B). Panel C shows the δ′18O–Δ′17O values of cherts color coded by their
age including modern oceanic diatomaceous silica and one chert sample from (Sharp et al. 2016). The solid
curved line shows the isotope composition of quartz in equilibrium with modern seawater. The dotted line
shows silica–water fractionation at the same temperature range but in equilibrium with ice-free seawater.
The dashed line to the left shows equilibrium fractionation with meteoric water (δ′18O = −18‰). Calibra-
tion after Sharp et al. (2016). An arbitrary selection of equilibrium temperatures (5 through 300) is shown
with numbers in °C. As an example of mixed compositions, we show mixing between cherts in equilibrium
with seawater and meteoric water with a blue dash dotted curve and the percent of meteorically derived
component is indicated. The region of meteoric waters is shown in the left upper corner after analyses
presented in (Luz and Barkan 2010; Li et al. 2015; Surma et al. 2018; Passey and Ji 2019; Tian et al. 2019).
The composition of mid-ocean ridge basalt (MORB) is shown with an open square.
waters was characterized in detail (Landais et al. 2008; Luz and Barkan 2010; Passey and Ji 2019;
Tian et al. 2019). The exact effects of diagenesis are not known and there is no published analyses
of marine pore fluids that we are aware of. The early measurements of cherts were presented
by Levin et al. (2014), including some Archean and Cenozoic samples. In more recent years,
silica–water equilibrium fractionation has been constrained by measurements of modern marine
diatoms and geothermal silica (Pack and Herwartz 2014; Sharp et al. 2016; Wostbrock et al. 2018)
336 Zakharov et al.
as well as by theoretical determination of fractionation factors (Cao and Liu 2011; Hayles et
al. 2018). These studies provide grounds to explore the triple oxygen isotope imprint of the
terrestrial hydrosphere and diagenetic conditions recorded by cherts. One can expect that
Precambrian cherts would reflect, at least partially, the original marine signals as well as the late
diagenetic and metamorphic signals, resulting in a mixture of equilibrium compositions. The
recent studies (Hayles et al. 2019; Liljestrand et al. 2020) presented high-precision triple oxygen
isotope measurements of cherts spanning in age from the Archean to the Cenozoic (see Fig. 5).
These papers demonstrated that the Archean cherts are 15−20‰ lower in δ′18O and about 0.1‰
higher in Δ′17O than those of Phanerozoic age (Fig. 5). The effect of higher temperature of the
Archean seawater or during diagenesis cannot be easily ruled out because the data do not tightly
follow the silica–water fractionation curve. Instead, cherts plot as a ‘cloud’ under the curve (see
Fig. 5). Such compositions could be reconciled by mixing arrays between primary and secondary
silica. These arrays are curved and concave upward in δ′18O–Δ′17O coordinates. Combined with a
probabilistic treatment of diagenetic effects, the study by Liljestrand et al. (2020) showed that the
triple oxygen isotope values of cherts can be explained by recrystallization, emphasizing that the
δ′18O value of seawater did not have to change significantly to replicate the entire dataset.
In essence, mixing between primary and secondary signals generated at different
temperatures results in masking the original equilibrium signal. Meteoric water as the source of
fluids responsible for low δ′18O values of the ancient silica deposits has been suggested numerous
times previously (Degens and Epstein 1964; Knauth and Epstein 1976; Robert and Chaussidon
2006; Tartèse et al. 2018). Triple oxygen isotope fractionation during the hydrologic cycle (Luz
and Barkan 2010; Li et al. 2015) is now sufficiently characterized resulting in a quasilinear
relationship for δ′18O–Δ′17O values, similar to the meteoric line in conventional δ18O–δD
coordinates (Craig 1961). Moreover, multiple stages of diagenesis and metamorphism in theory
would not permit using this treatment of ancient samples to reconstruct the composition of a
single reservoir, such as ancient seawater-derived fluids. In this paper we take the triple oxygen
isotope study of cherts one step further by attempting to resolve the influence of primary signals
and diagenetic overprints on a case-by-case basis. First, we describe the in situ techniques such
as secondary ion probe and Raman spectroscopy that we combine to get a detailed insight into
the complexity of isotope signatures recorded by the same set of samples.
Secondary ion mass spectrometry
Secondary ion mass spectrometry (SIMS) has been extensively used to determine oxygen
and silicon isotope compositions at high spatial resolutions (McKeegan 1987; Hervig et al.
1992; Valley and Graham 1996; Bindeman and Valley 2001; Peck et al. 2001; Kelly et al.
2007; Bindeman et al. 2008; Eiler 2011). Since the development of large ion microprobes
(series 1270−1300 from Cameca for example), it has been possible to measure oxygen and
silicon isotopes with high precision : 0.2‰ and 0.4‰ (2σ) for oxygen and silicon isotopes
respectively (Kita et al. 2009; Whitehouse and Nemchin 2009), and therefore SIMS have been
applied to various research questions: igneous processes (e.g., Eiler et al. 1997; Gurenko and
Chaussidon 2002; Appleby et al. 2008), paleotemperature reconstruction from foraminifera
and coral (for example, Rollion-Bard et al. 2003; Kozdon et al. 2018), fluid–rock interactions
during metamorphism (Valley and Graham 1996; for recent examples see, Bonamici et al. 2014;
Bégué et al. 2019), and biogeochemical cycles (Alexandre et al. 2015). The pioneer study of
Robert and Chaussidon (2006) have measured both O and Si isotopes by SIMS on a set of chert
samples spanning from the Phanerozoic to the Archean and highlighted a secular correlation
between O and Si isotope compositions. That study used a large number of cherts but performed
only a few measurements per sample (3 to 5 depending on the samples) and used the average
value for the interpretation of the secular trend. Marin et al. (2010) and Heck et al. (2011)
followed the analytical procedure described by Robert and Chaussidon (2006) with more
than 400 analyses in various chert samples from the Archean to the Proterozoic and observed
extreme isotopic variability at microscale for oxygen and silicon (see Fig. 4).
While it is possible to measure triple oxygen isotope ratios by SIMS (Tartèse et al. 2018;
Bellucci et al. 2020), the applications remain within the realm of meteoritics, where the
Δ′17O signals are large, spanning over 1‰ (e.g., Krot et al. 2010). Due to the relatively large
uncertainties (± 0.2‰) in 17O/16O measured by SIMS, meaningful measurements of terrestrial
materials are limited. The required precision of ±0.01‰ is achieved by the traditional gas-
source isotope ratio mass spectrometers like the MAT 253 used here, yet the spatial resolution
is compromised by the relatively large sample size of 1−2 mg (several mm across), compared
to several nanograms analyzed by SIMS (see Fig. 5).
Raman spectroscopy
Further insight into cherts’ post-depositional histories can be provided by Raman
spectroscopy. Raman microspectroscopy is a nondestructive spatially resolved technique that
yields combined chemical and structural information on both inorganic and organic phases at
the micrometer scale, and is therefore widely used in Earth sciences (Dubessy et al. 2012).
As Precambrian cherts are almost exclusively made of quartz, they generally lack mineral
assemblages that can be used for mineral thermometry. Yet, these rocks usually contain
carbonaceous matter that has been often characterized using Raman spectroscopy in order to
inform on its thermal history (Tice et al. 2004; Schopf and Kudryavtsev 2005; Wacey et al.
2012; Sugitani et al. 2013; Alleon et al. 2016, 2019).
Raman spectral features of carbonaceous matter were indeed shown to evolve systematically
with metamorphic grade (Wopenka and Pasteris 1993; Yui et al. 1996). Over geologic timescales,
the structural evolution of carbonaceous matter is essentially governed by the maximum
temperature reached during metamorphism (Beyssac et al. 2002, 2004). By studying carbonaceous
matter in rocks with well-constrained P–T histories, Beyssac et al. (2002) first quantified a
phenomenological relation between Raman spectral characteristics and peak metamorphic
temperatures from 330–650 °C. The authors decomposed the Raman signal into four different G
and D bands using a pseudo-Voigt function. In lower temperature contexts (200‒300 °C), Raman
spectra are more complex and additional bands are used to describe the signal and estimate the
peak temperature conditions that carbonaceous matter experienced (Rahl et al. 2005; Lahfid et
al. 2010). Figure 6 illustrates the impact of increasing metamorphic temperature on the Raman
spectra of carbonaceous matter preserved in Precambrian cherts studied here.
In the present study, we used Raman spectroscopy for the formations studied here by SIMS
and by laser fluorination GS-IRMS for triple oxygen isotopes. To estimate the upper limit for
the temperature of secondary dissolution–reprecipitation of quartz we utilized samples from the
Gunflint Formation (Alleon et al. 2016), Dresser Formation (Delarue et al. 2016), and the newly
collected chert samples of the Mendon Formation (Fig. 6). Raman data indicate that Gunflint
organic matter experienced maximum burial temperature conditions between 160 and 220 °C
without clear petrographic evidence for recrystallization of microquartz at this temperature (see
Alleon et al. 2016a for details). Consequently, Gunflint microquartz likely recrystallized from a
silica precursor at temperature below, at temperatures ranging from 130 to 170 °C as proposed
previously (Marin et al. 2010). On the other hand, Archean organic matter from Dresser and
Mendon Formations experienced higher peak temperatures during metamorphism (Fig. 6).
Recrystallization of quartz in Dresser and Mendon cherts could be bracketed by the range
300‒360 °C, according to Raman spectral characteristics of carbonaceous matter. The analytical
details of Raman spectroscopy used here are described in the Methods section.
338 Zakharov et al.
D1
G + D2
D3 +
3.3 Ga 360 °C
Mendon
increasing peak temperature

D1 G + D2
increasing structural order

D3
3.4 Ga 300 °C
Strelley Pool
G + D2
D1
D4 D3
1.9 Ga
Gunflint 160 °C
Schreiber Beach -
800 1000 1200 1400 1600 1800
Raman shift (cm-1)
Figure 6. Evolution of Raman spectra of Precambrian organic matter with increasing metamorphic grade. Ra-
man data and estimated peak temperatures for Gunflint and Strelley Pool organics are from Alleon et al. (2016)
and Alleon et al. (2018), respectively. Raman data for Mendon organics were acquired for the present study.
Estimation of the peak temperature was done following the methodology described by Kouketsu et al. (2014).
PART II: COMBINED SIMS AND

TRIPLE OXYGEN ISOTOPE ANALYSIS OF PRECAMBRIAN CHERTS
We use a unique collection of Precambrian cherts from the Dresser, Mendon and Gunflint
Formations, 3.5, 3.2 and 1.9 Ga, respectively, that were previously characterized in great
detail by in situ SIMS measurements of O and Si isotopes and trace elemental (Al, Ti, Fe, K)
concentrations (Marin et al. 2010; Marin-Carbonne et al. 2011, 2012, 2013). With the detailed
spatial data set (see Fig. 7), we attempt to ‘see-through’ the effects of provenance, secondary
processes such as early and late diagenetic and metamorphic transformations that together
define the triple oxygen isotope composition. At first, the procedure involves identification of
the low-Al and low-Ti compositions and their δ18O and δ30Si values from the SIMS data set.
This subset of data is interpreted as the earliest low-temperature generation of silica, textually
identifiable as microquartz of early diagenetic origin. We then estimate the corresponding
δ′18O–Δ′17O compositions of this generation. Further, we resolve the triple oxygen isotope
composition of secondary quartz generations that resulted from post-depositional dissolution
and reprecipitation induced by burial and regional metamorphism.
In addition, to test for the hydrothermal origin of Archean silica (Perry and Lefticariu
2003; Van den Boorn et al. 2007), we provide triple oxygen isotope measurements of modern
amorphous silica that precipitated in a geothermal pipeline at the Reykjanes system of
Iceland at temperatures of 180−190 °C (Hardardottir 2011). A similar work was also done by
Wostbrock et al. (2018) in a different Icelandic geothermal system with the goal to calibrate
the silica–water fractionation. In Reykjanes, seawater-dominated fluid reacts with basalt at
high temperature (350 °C). Upon depressurization and boiling of the fluid during geothermal
exploration, abundant amorphous silica precipitates as scales on the walls of the pipes at a
lower temperature of 188 °C (Hardardottir 2011). The silica scale was sampled from well
RN−9. This temperature regime and shallow water conditions are commonly envisioned for
the Archean silica-generating environment (De Ronde et al. 1997; Perry and Lefticariu 2003;
Van den Boorn et al. 2007). Consequently, we compare the modern-day amorphous silica scale
from Reykjanes to the reconstructed composition of primary silica represented by the Dresser
and Mendon Formation samples, where input of hydrothermal silica might have been high.
Mendon Dresser Gunflint

25 A 3
C
d'18O, ‰ VSMOW (SIMS)
d30Si, ‰ NBS28 (SIMS)

2
2500
2000
Al2O3 ppm
1
20
1500
0
1000
−1 500
15 Formation:
3.5 Ga Dresser −2
3.3 Ga Mendon
3.4 Ga Kromberg
15 20 15 20 15 20
10 1.9 Ga Gunflint
Mendon Dresser Gunflint

3
D
d30Si, ‰ VSMOW (SIMS)
3
B
d30Si, ‰ NBS28 (SIMS)

2
2
1 1 300
0
TiO2 ppm
−1 0 200
−2
−1 100
Schreiber Beach
Frustration Bay
−2
1c of 06.29.84
Onverwacht 2
5 of 06.28.84
3 of 06.30.84
PPRG 200
PPRG 198
PPRG 193
PPRG 006
15 20 15 20 15 20
d'18O, ‰ VSMOW (SIMS)
Figure 7. The inventory of oxygen (A) and silicon (B) isotope data determined by secondary ion spectros-
copy (SIMS) in the samples used for triple oxygen isotope analyses. The trace elemental concentrations
were measured from the same pits also by SIMS. Aluminum (as Al2O3) and titanium (TiO2) concentrations
are shown as color-coded points plotted against their O and Si isotope ratios (C and D) and were incorpo-
rated here from the previous study (Marin-Carbonne et al. 2012).
Finally, we provide new hydrogen isotope and water content data for the same set of
samples. We explore the relationship between the reconstructed triple oxygen and hydrogen
isotope compositions as OH− groups contained in microcrystalline quartz can be reflective
of the secondary equilibrium fluids originating during different stages of diagenetic
transformation of silica. In modern microquartz collected from the seafloor, hydrogen isotope
compositions reflect temperature-dependent fractionation with interstitial porewaters being
the likely equilibrium fluid (Knauth and Epstein 1975). However, in older counterparts of
Cenozoic age, the hydrogen isotope compositions of microquartz exhibit interaction with
meteoric water, forming an array of values parallel to the meteoric water line (Craig 1961).
Given that a similar meteoric water line exists for triple oxygen isotopes (e.g., Luz and Barkan
2010), its correlation with δD values would help to disentangle the effect of post-depositional
alteration from the original marine signal.
Methods
Triple oxygen isotope measurements. The triple oxygen and hydrogen isotope analyses
were carried out at the University of Oregon Stable Isotope Lab. The methodology was
previously described in Zakharov et al. (2019b). Briefly, small chips of chert (1.5−2 mg)
excluding visible quartz veins were placed in a stainless steel holder inside an oven at 130 °C
held under vacuum by a roughing pump for several hours. Later, they were placed in a stainless
steel vacuum chamber and pre-treated with BrF5 overnight to remove the absorbed moisture
and reactive compounds. After laser-assisted fluorination of samples, the extracted oxygen
was purified from fluorination byproducts by a series of cryogenic traps and reaction with
Hg-vapor in a mercury diffusion pump. Then oxygen gas was trapped on a 5 Å molecular sieve
by cooling to liquid nitrogen temperature. The released O2 was carried though a GC-column
by He flow at 30mL/minute and room temperature. After about 3 minutes of elution time, the
O2 was trapped on another 5 Å molecular sieve at liquid nitrogen temperature. The gas was
further trapped on another, smaller volume 5 Å molecular sieve immersed in liquid nitrogen and
introduced into a MAT 253 gas-source isotope ratio mass spectrometer (GS-IRMS) at 50−60 °C.
Each measurement consisted of 40 cycles of sample-reference gas comparisons in dual-inlet
mode with intermittent equilibration of pressure in the bellows of the mass spectrometer.
340 Zakharov et al.
In the study we used the previously characterized silica standards (Miller et al. 2020;
Wostbrock et al. 2020) to monitor the accuracy of triple oxygen isotope measurements.
Four UWG−2 garnet and two San Carlos (SCO) olivine measurements were made during
the analytical session between 24.09.2020 and 27.09.2020. The mean value of the
measured standards yielded (mean ± 1 standard deviation): δ17OUWG−2 = 3.137 ± 0.073,
δ18OUWG−2 = 6.109 ± 0.159‰, δ17OSCO = 2.911 ± 0.032, and δ18OSCO = 5.684 ± 0.052‰. In
addition, we ran a chert standard SKFS (Stevns Klint Flint Standard; Miller et al. 2020) that
returned δ17OSKFS = 17.300 ± 0.012‰, δ18OSKFS = 33.333 ± 0.003‰ reported as mean ± standard
error for a single measurement. We used the UWG−2 and SKFS values presented in this volume
to calibrate the measurements to the VSMOW-SLAP scale (δ17OUWG−2 = 2.94, δ18OUWG−2 = 5.7
and δ17OSKFS = 17.57, δ18OSKFS = 33.81; see Sharp and Wostbrock 2021, this volume). The two-
point calibration is used to address the instrument-specific non-linearity of the scale. However,
for the point of comparison with other datasets, where SKFS was not included, most figures
in this chapter present values relative to UWG−2. So far no significant scale compression/
stretching was found by analyses of SCO, UWG−2 and SKFS in other laboratories (Miller et
al. 2020; Sharp and Wostbrock 2021, this volume).
Hydrogen isotopes. Hydrogen isotopes were analyzed at the University of Oregon using
a high temperature thermal conversion elemental analyzer (TC/EA) connected to the MAT
253 in a continuous flow mode, where gases from samples and standards are transported in
He carrier gas (Sharp et al. 2001; see recent application, Hudak and Bindeman 2020). Each
solid sample and standard were wrapped in a silver foil capsule and dried in the vacuum oven
overnight, then transported to the auto sampler where they were purged with He carrier gas.
Two datasets were produced for hydrogen isotopes using different amounts of cherts analyte.
The first session included between 3−4 mg of chert in each capsule and second dataset included
7−9 mg of chert. To test for homogeneity, we ran the same cherts but different pieces. In the
TC/EA’s furnace lined with a glassy carbon column, samples underwent pyrolysis at 1450 °C,
and all of the H2O in the minerals was pyrolyzed to H2 and CO gas. Extracted gas was carried
by He into a gas chromatograph and further introduced into the mass spectrometer via the
CONFLOIII device. Mica standards, USGS57 and USGS58 with δD = −91 and −28‰ (Qi et
al. 2017), respectively, were included in each analytical session to monitor the accuracy of
analysis and to normalize data to VSMOW scale.
SIMS measurements of O and Si isotopes. The oxygen isotope compositions of two chert
samples from the Gunflint Formation (Schreiber Beach and Frustration Bay) and one Archean
sample (PPRG 193) from the Kromberg Formation were determined using the Cameca IMS 1270
ion microprobe at UCLA (Los Angeles, USA) and the Cameca 1280 HR (SwissSIMS, Lausanne,
Switzerland) during 3 different sessions, using procedures previously described (Marin-
Carbonne et al. 2012). The Schreiber Beach sample was also analyzed for silicon isotopes. A
meticulous microscopic observation has been carried out in all the samples in order to select the
purest microquartz area, free of any visible oxides and carbonates. Oxygen and silicon isotopic
compositions are reported here as per mil deviations from the international standards: VSMOW
(Vienna Standard Mean Ocean Water) for oxygen and NBS28 for silicon using the delta notation.
In each mount, in-house quartz standards QZCRWU (δ18O = +24.5‰; Marin 2009), and
UNIL-Q1 (δ18O = +9.8‰; Seitz et al. 2017) were embedded in epoxy together with the fragments
of cherts. The fragments were sputtered with a Cs+ primary beam of 2–25 nA intensity, 10 kV
acceleration voltage, and of size ranging from 10 to 30 µm for oxygen and silicon isotopic
measurements, respectively. The electron gun was used for charge compensation. Negative
secondary ions were accelerated at 10 kV. The mass resolving power was set at ~ 4000. The
ions 16O−, 18O−, 28Si− and 30Si− were detected in multicollection mode using faraday cups; L’2
trolley for 16O− and 28Si− (using a 1010 Ω resistor) and H1 trolley for 18O− and 30Si− (using a 1011 Ω
resistor). The centering of the secondary beam in the field diaphragm and of the magnetic field
were done automatically during the analyses. A pre-sputtering time of one minute was added for
the silicon isotope measurements. With such conditions, a counting statistic ≤ ±0.1‰ is obtained
after few minutes counting (40 cycles of 5 s acquisition time and 60 s of pre-sputtering). The
external reproducibility for oxygen and silicon isotope measurements obtained on the QZCRWU
and UNIL-Q1 quartz during the three sessions of analyses was ±0.2 and ±0.4‰ at one sigma.
Raman spectroscopy. Raman data were obtained using a Horiba Jobin Yvon LabRAM 800
HR spectrometer (University of Lausanne, Switzerland) in a confocal configuration equipped
with an Ar+ laser (532 nm) excitation source and a CCD detector. Raman microspectroscopy
measurements were performed at constant room temperature, directly on freshly exposed surfaces
of chert samples, to characterize the degree of structural organization of carbonaceous matter. The
laser beam was focused on the sample with a 300 µm confocal hole using a long working distance
×50 objective (NA = 0.70). This configuration provided a ≈2 µm spot size for a laser power
delivered at the sample surface below 1 mW, thereby preventing irreversible laser-induced thermal
damage. The collected Raman spectra were used to estimate the peak metamorphic temperature
experienced by organic matter, following the methodology proposed by Kouketsu et al. (2014).
Results
Oxygen isotope values. In Figure 8, we compare the bulk isotope values plotted against
the average values of δ′18O determined by SIMS as well as the δ30Si. In Figure 9, the triple
oxygen isotope values are shown in δ′18O–Δ′17O coordinates with a reference slope of 0.528,
along with the samples of amorphous silica scale from Reykjanes. The Archean Mendon and
Dresser Formation cherts have δ′18O values between 14 and 16‰ consistent with the previous
values reported by SIMS in the exact same samples. Their Δ′17O values cluster at −0.05 ± 0.02‰
(Fig. 9). The early Paleoproterozoic cherts of the Gunflint Formation have distinctly higher δ′18O
3
B
d30Si mean, ‰ NBS28 (SIMS)
2
30
A 1
y = 4(2) + 0.83 (0.11 ) × x
0
d'18O mean, ‰ VSMOW (SIMS)
25 −1
−2
15.0 17.5 20.0 22.5

20 d'18O, ‰ VSMOW (bulk & SIMS)
3
C
d30Si mean, ‰ NBS28 (SIMS)
2
Formation:
15 3.5 Ga Dresser 1
3.4 Ga Kromberg
3.3 Ga Mendon 0
1.9 Ga Gunflint
1
−1
1:
10
10 15 20 25 30 −2
d'18O, ‰ VSMOW (bulk)

−0.100 −0.075 −0.050 −0.025
D'17O0.528, ‰ VSMOW (bulk)
Figure 8. Comparison of the data measured in bulk by laser fluorination GS-IRMS in this study versus the
average values of the same samples measured by SIMS. The vertical error bar in each plot represents the
total range of values measured by SIMS (e.g., length = δ′18Omax − δ′18Omin). Panel A shows oxygen isotope
compositions of cherts measured by two different methods with the least-square regression line and 95%
confidence intervals shown with dashed lines. The equation of the line is displayed in the upper left cor-
ner with standard errors shown in parentheses. The black solid line has slope of 1 and xy-intercept of 0.
The oxygen isotope values versus average δ30Si is shown in Panel B. The vertical bars show the range of
δ30Si determined by SIMS. The symbols with black rims are plotted as δ′18O values determined in bulk
analyses. The symbols without the black rim were plotted using the average δ′18O values determined by
SIMS. The Δ′17O values plotted against δ30Si are shown in C. The horizontal error bars in C are analytical.
342 Zakharov et al.
0.1
UWG-SKFS normalization:
0.0
UWG-2
−0.1 n=5
0.1 mean ± 2SD
SKFS
−0.2
meteoric w
ater line
−0.3
0.0
D'17O0.528, ‰ VSMOW
ice-free world
−10 0 10 20 30 40
seawater 175 80
vent MORB
fluids 175
eq. 40
quart
−0.1 with MORB
80 equil z-water
. (T
°C)
SFKS 40
Modern silica:
Reykjanes
−0.2 pipe scale
Precambrian cherts:
3.5 Ga Dresser SIMS spread:
3.4 Ga Kromberg
Dresser, Kromberg,
−0.3 3.3 Ga Mendon
1.9 Ga Gunflint Mendon Gunflint
Previous study:
2.5−3.5 Ga
1.8−2.5 Ga
−10 0 10 20 30 40
d'18O, ‰ VSMOW
Figure 9. The triple oxygen isotope values of the samples measured in this (large circles) and previous
studies. Small circles are cherts from Liljestrand et al. (2020), small squares are values from Hayles et
al. (2019). The silica–water equilibria are shown with curved lines projecting from modern ice-free sea-
water (solid; δ′18Osw = −1‰), modern seawater (dashed) and seawater-derived hydrothermal fluids (blue
dotted). The array of hydrothermal vent fluids that approach the composition of fluids in equilibrium with
mid-ocean ridge basalts (MORB) is shown with the thick blue dotted line. The spread of SIMS values
in samples measured here is shown at the bottom of the plot with density distributions. The tick marks
near each other on different curves all correspond to the same temperatures of silica–water equilibrium at
40, 80 and 175 °C. All of the data shown is calibrated to the UWG−2 standard. Only samples containing
> 80 wt.% quartz are shown for cherts from Liljestrand et al. (2020). The inset in the upper right corner
shows our data normalized using a two-point calibration using UWG−2 and SKFS standards (Miller et al.
2020; updated values in Sharp and Wosbrock 2021, this volume), demonstrating the negligible effect of
scale compression issues. The red points depict ‘stretched’ values, while black points are normalized us-
ing the single-point normalization (same as in the large plot).
values ranging between 19 and 25‰. Their Δ′17O values are varying within −0.08 ± 0.01‰,
significantly lower than that of the Mendon and Dresser Formations. The two samples of the
Reykjanes geothermal scale have δ′18O of +14.2 and +14.6‰, accompanied by a Δ′17O of
−0.03‰. The results of triple oxygen isotope analyses are reported in Supplementary Table 1.
The new SIMS values are reported for 2 samples from Schreiber Beach and Frustration
Bay (Gunflint Formation) and one sample PPRG193 (from the Kromberg Formation) in
Supplementary Table 2. The results of previous investigations by Marin-Carbonne et al. (2012)
are compiled in Supplementary Table 2 as well. PPRG193 returned the range of δ′18O values
between +14.2 and +15.2‰, with an average of +14.8 ± 0.3‰ (n = 23). The δ′18O values range
between +18 and +26‰, with an average value of +23.7 ± 1.8‰ (mean ± σ; n = 90) for the two
Gunflint samples, Schreiber Beach and Frustration Bay. We found that in the Schreiber Beach
sample the δ′18O values vary on a scale of ~100 μm by 8‰ (Fig. 4). The megaquartz segregations
are particularly low in δ′18O, as low as 18‰, while microquartz and chalcedony reach values
of +26‰. These are shown in Figure 4B. The δ30Si in Schreiber Beach varies between +0.8
and +2.5‰ with an average value of +1.9 ± 0.4‰ relative to NBS28. The linear regression
(Fig. 8) constructed based on the δ′18O bulk vs. mean δ′18O SIMS values yields a slope of
0.8 ± 0.1 and a y-intercept of 4 ± 2‰ (mean ± 1se). It is worth noting that the previous SIMS
measurements were specifically designed to preferentially target microcrystalline δ′18O quartz

resulting in somewhat higher (1−2‰) mean δ18O values compared to the measurements by
bulk laser fluorination (Fig. 8). This suggests that bulk fluorination analyses inevitably contain
fragments of altered and recrystallized cherts with δ′18O values lower than that of microquartz.
The overall δ′18O range of the samples is consistent with the values determined in previous
studies (Hayles et al. 2019; Liljestrand et al. 2020). However, the Δ′17O are mostly slightly
elevated compared to previous analyses of triple oxygen isotope values in Archean and early
Paleoproterozoic cherts produced by Liljestrand et al. (2020), but agree better with the values
determined by Hayles et al. (2019). To further improve the comparison, we only plotted samples
from Liljestrand et al. (2020) that contain > 80 % quartz. The previously measured cherts from
Hayles et al. (2019) shown in Figure 9 are from the Gunflint Formation and Onverwacht group
that makes comparison even more accurate. Further, we attempted to re-normalize the previously
published analyses to the single scale defined by UWG−2 (see Figs. 5 and 9). Despite these
procedures, we cannot explain these differences in the datasets as purely related to calibration.
We thus suggest that these differences in Δ′17O measured here and in previous studies are at
least partly a result of natural variations between samples. This is suggested by the excellent
agreement between the samples that come from the same well-preserved Gunflint Formation
from this study and from the study of Hayles et al. (2019). The samples from the Kromberg
Formation reported here and in Liljestrand et al. (2020) are more difficult to compare because
the formation includes multiple chert layers with variable δ18O values (Knauth and Lowe 2003).
Further, we observe a positive correlation between the δ′18O values determined in bulk
samples and the average of the δ30Si values determined by SIMS (Fig. 8). Consequently, the
Δ′17O values correlate negatively with δ30Si. This relationship between the triple oxygen
isotope and silicon isotope values (Fig. 8) has not been identified in the previous studies. We
suggest that the δ′18O–Δ′17O–δ30Si correlation documents preservation of distinct differences
between samples that, in turn, can be used to decipher the processes recorded in isotope signals.
The δD values and water content. The hydrogen isotope values range between −110 and
−60‰ with water contents between 0.1 and 1.2 wt. % (Fig. 10). The Mendon and Dresser
cherts have the lowest water contents, below 0.2 wt. %, and δD ranging from −100 to −60‰.
The early Paleoproterozoic Gunflint cherts contain 0.3−1.2 wt. % water and δD values between
−110 and −90‰. Within each group of samples there is no significant relationship between
hydrogen and oxygen isotope ratios or the water contents (Fig. 10). We observe no systematic
difference between the duplicated analysis of the same chert samples using 3−4 mg or 7−9 mg
of analyte. The δD values are variable within ~10‰ between the replicates with the exception
of two samples where the difference is close to 20‰, likely reflecting natural variability as the
analytical precision of this method is ±1‒3‰ (Fig. 10). The water content of the duplicated
measurements varies within < 0.05 wt. % with exception of two samples that gave a larger
difference in δD. In these samples the water content exceeds 0.6 wt.%.
Discussion: various origins of Precambrian cherts and record of seawater
We observe that the cherts from each age group are defined by ranges of values that
have distinct differences: Archean samples are generally lower in δ′18O and higher in Δ′17O
compared to the early Paleoproterozoic Gunflint cherts, consistent with the values determined
in the previous studies of Precambrian chert (Knauth 2005; Hayles et al. 2019; Liljestrand et
al. 2020). We interpret the scatter of δ′18O–Δ′17O values registered within each formation as
the result of precipitation of silica from different sources of fluids during: I) seafloor deposition
and seafloor-diagenesis, and II) later interaction with fluids and recrystallization during burial
metamorphism. We consider marine pore fluids or hydrothermal vent fluids as the possible
primary sources of silica in category I, while fluids of category II would be those derived from
meteoric sources or crustal/basinal fluids.
344 Zakharov et al.
0 A 1.0
r
te crustal °C “Line A”
wa 5
dD, ‰ VSMOW
fluids 17
ric 0.1
t eo line °C
−50 e 80
m ic
0.1| zo
1.0 leo
W/R |
| Pa ic
= 2.3 ss
−100 ra
°C t-Ju
5 os
17 °C
p
°C
αD/H (Gilg ans Sheppard, 1996) 80 20
−150 α18,17O/16O (Sharp et al., 2016)
−10 0 10 20 30
d'18O, ‰ VSMOW
0 B
1.0 2.3
dD, ‰ VSMOW
0.1
−50 0.1
2.3 |
mete line
| 1.0
|
−100
oric
wate
175 °
175 °
80 °C
20 °C
−150
r
C
C
−0.10 −0.05 0.00 0.05

D'17O0.528, ‰ VSMOW
−50 C 25.0 D
d'18O, ‰ VSMOW
dD, ‰ VSMOW
22.5
−70
20.0
−90
17.5
−110 15.0
12.5
0.0 0.3 0.6 0.9 1.2 0.0 0.3 0.6 0.9 1.2
H2O, wt. % H2O, wt. %
−0.02 E Dresser
D'17O0.528, ‰ VSMOW
Kromberg
−0.04 Mendon
Gunflint
−0.06
−0.08
−0.10
0.0 0.3 0.6 0.9 1.2
H2O, wt. %
Figure 10. The hydrogen isotope data plotted against the triple oxygen isotope values. The δD–δ′18O diagram
(A) shows the composition of cherts from Dresser, Kromberg, Mendon and Gunflint Formations plotting in the
restricted range of values. The field of Paleozoic cherts is from (Knauth and Epstein 1976). The meteoric water
line is shown (Craig 1961) with parallel lines depicting the isotope composition of equilibrium silica. Frac-
tionation factors for oxygen (Sharp et al. 2016) and hydrogen (Sheppard and Gilg 1996) are combined to define
the isotope value of chert in equilibrium with meteoric waters. The thin dotted line shows the composition of
silica in equilibrium with meteoric water (δ′18O = −10‰) at different temperatures. Line ‘A’ from (Knauth and
Epstein 1976) depicts cherts in equilibrium with the modern ocean. In addition, we show the field of crustal
fluids. Such fluids are generated as a result of reaction with crustal rocks. As an example, we show two different
meteoric waters reacting with crustal rocks at different water–rock ratios (marked with numbers 0.1 through
2.3). Panel (B) shows the relationship across the hydrogen and triple oxygen isotope ratios. The water content
plotted against the isotope parameters (C–E) display the difference between the Archean (Mendon, Kromberg,
Dresser) and the Paleoproterozoic (Gunflint) cherts. The low water content of the Archean cherts and their
generally high δD values are clearly different from the more water-rich samples of the Gunflint Formation.
There is no correlation observed between water content and the isotope ratios when individual formations are
considered. The duplicated measurements of the same sample are shown as connected points (see Methods).
Archean cherts: Dresser, Kromberg and Mendon Formations. The SIMS determinations
defined a range of heterogeneous δ18O values that span between +14 and +20‰ in our collection
of Archean cherts. Among these samples, the δ30Si values span between around −2‰ and
+3‰ (Fig. 7), mostly encompassing the values of igneous rocks (−1‰), modern seawater
(between +1 and +3‰), hydrothermal silica deposits (between −3 and 0‰) and marine abiotic
sediments (between 0 and +2‰; see recent compilation in Kleine et al. 2018). In the context
of the Precambrian, the silicon isotope values below 0‰ have been interpreted to reflect the
hydrothermal origin of the silica initially precipitated from vent fluids (Van den Boorn et al.
2007, 2010; Marin-Carbonne et al. 2012). Even lower δ30Si compositions are measured in
BIFs (between −4 and 0‰; Konhauser et al. 2017). Further, given that the Al and Ti content
of measured microquartz samples is quite low (see Fig. 7), we suggest that the involvement
of silicified material with igneous origin δ30Si = −1‰ in our samples is limited. Meanwhile,
microquartz with δ30Si above 0‰ is thought to represent a seawater component (Van den Boorn et
al. 2010; Stefurak et al. 2015). Although δ30Si within each sample tends to be extremely variable
and texture-specific (Stefurak et al. 2015), the average values correlate with the average δ′18O
values of the same samples (see Figs. 7 and 8). Among the samples from our collection, only one
sample (PPRG 200) from the Mendon Formation has an average δ30Si of above 0‰, which could
be used as an indication of the highest input of the marine signal. The δ′18O values determined
by SIMS in this sample are also the highest (up to +20‰) compared to the other samples. We
thus suggest that the triple oxygen isotope values of the samples with the average negative δ30Si
values and δ18O of around +14‰ can be used to trace the isotope values of seawater-derived vent
fluids, while the value of PPRG 200 should be closest to marine microquartz.
We find that in triple oxygen isotope space (Fig. 11) most samples, even PPRG200,
plot below the equilibrium curve that projects from ice-free world seawater. Based on the
possible fluid sources resulting from seawater–basalt reactions and the Raman-based record of
metamorphism, we construct a field of possible equilibrium and mixed compositions (Fig. 11).
Assuming that the original equilibrium fluids were derived from seawater with a δ′18O ± 2‰
of modern seawater, encompassing hydrothermal vent fluid compositions or ice-free world
seawater (Fig. 9), the primary Archean microquartz with low δ30Si values is best explained
by high-temperature equilibrium fractionation at 150−170 °C. The sample PPRG 200 with a
high δ30Si value seems to be also affected by hydrothermal input as its Δ′17O value is quite
low, shifted towards the compositions in equilibrium with vent fluids. Using the Raman-based
estimate of the metamorphic overprint of 300−360 °C, the secondary low-Δ′17O quartz formed
in equilibrium with fluids with δ′18O of crustal values (+3 to +5‰) and with Δ′17O close to
−0.05‰. Although quartz did not necessarily recrystallize at the temperature recorded by Raman
spectroscopy, we suggest that it formed at relatively high temperature due to the presence of
fluid inclusions hosted in cherts from Onverwacht group that homogenize at temperatures up to
300 °C (Marin-Carbonne et al. 2011). In addition, we show a low temperature quartz with d18O
of +22‰ (Fig. 11) that could have precipitated from minimally modified seawater as suggested
by previous detailed investigations of Barberton Belt cherts (Knauth and Lowe 2003)
We suggest that initial precipitation of microquartz might have occurred in equilibrium
with vent fluids, or from a mixture of seawater and vent fluid sources. The Δ′17O of primary
silica deposits would be dependent on temperature and the water–rock ratios during seawater–
basalt reaction. Exchange reactions at low water–rock ratios induce oxygen isotope shifts
experienced by vent fluids towards the composition of rocks. In our dataset, a sample from the
Kromberg Formation for example plots on the modern seawater equilibrium curve at 150 °C
(Fig. 11) , which could be explained by the minimal interaction between seawater and basalts,
i.e., high water–rock ratios (mass ratio of 2 and above), resulting in negligible oxygen isotope
shifts of vent fluids (e.g., Shanks and Seyfried 1987). Further interaction with crustal and
basinal fluids (i.e., those that reacted with crustal rocks during burial and metamorphism)
346 Zakharov et al.
and precipitation of equilibrium quartz would also contribute to lowering the bulk Δ′17O value.
Although there is no unique way to reconstruct the late generation quartz compositions, a
combination of mixing arrays such as shown in Fig. 11 might help explaining the low Δ′17O
signature of many chert samples of the Archean that underwent metamorphism.
meteoric waters submarine vent

fluids A
0.0 l
l
2.3 l 175 | 150
2.3 l | 125 100
1.0 l
1.0 80
|
0.3 |
0.3
l
l
?|
0.1 |
40
−0.1 crustal 300 |
crustal rocks PPRG200

|
25
|
fluids 80 |
D'17O0.528, ‰ VSMOW
10 °C
−0.2 30
positive δ Si only Mendon
all samples Kromberg
−0.3 Dresser
meteoric waters porewater

fluids B
0.0 l
l 175
2.3 l l |
1.0 l
175 100 80
0.3
l
40? 60
|
0.1
|
|
−0.1 MORB 170

|
60
25
|
| |
25 10|
80 |
| 10 °C
40
−0.2
microquartz
megaquartz
Gunflint
−0.3
−10 0 10 20 30 40
d'18O, ‰ VSMOW
Figure 11. Triple oxygen isotope diagram with Archean (A) and Paleoproterozoic (B) samples divided into
separate panels. Based on isotope ratios measured by SIMS, average δ30Si values and peak metamorphic
temperatures, we attempt to resolve provenance and distinguish between the early and late generations
of quartz (see Discussion). The quartz–water fractionation is shown with the concave-down curves. The
solid curve depicts equilibrium with ice-free world seawater. The dotted curves show silica in equilibrium
with fluids that underwent reaction either with oceanic crust (blue; submarine vent fluids) or continental
crustal rocks (grey; crustal fluids). In the Archean (A), where hydrothermal cycling of seawater might have
been important, the silica might have inherited some of the low Δ′17O values (blue dotted lines) and low
δ30Si from vent fluids (shown with the blue field). Those match the δ′18O values determined by SIMS in
low-δ30Si samples. Further, precipitation of quartz from crustal fluids at 300 °C during metamorphism is
shown with a curve projecting from the respective field. This is the lowest range of metamorphic condi-
tions registered by the Raman spectroscopy and fluid inclusion data (Marin-Carbonne et al. 2011) in our
collection of Archean cherts. Mixing with quartz that precipitated at 80 °C is shown with the concave-up
dashed-dotted line, as some cherts in the Barberton Belt reach +22‰ in δ′18O (Knauth and Lowe 2003).
The red curved arrow and question mark show a possible trajectory taken by cherts that formed as a mixture
of seawater and hydrothermally sourced silica. The question mark shows seawater-derived cherts that
originally precipitated at ~80 °C. In case of the Gunflint Formation (B), several samples plot near the equi-
librium fractionation line with seawater at ~70 °C. The δ′18O of such samples match the δ′18O measured by
SIMS in the early microquartz. The marine porewater fluids are shown with the pink field and equilibrium
silica is depicted with the pink dotted line. The composition of secondary quartz is constructed based on
the δ′18O value of +18‰ measured by SIMS in megaquartz segregations. The equilibrium temperatures
of 130 and 170 °C are picked as the possible temperature range of quartz recrystallization in presence of
crustal fluids. Alternatively, equilibrium with pristine meteoric water would yield a similar composition
of secondary quartz, however at much lower temperatures, around 40 °C (shown with a question mark).
These trends are demonstrative of possible origins of low Δ′17O values in the studies’ samples and intend
to describe the variability determined by SIMS in this study and by previous determinations within cherts
of the same age (Hayles et al. 2019; Liljestrand et al. 2020).
The early Paleoproterozoic Gunflint chert. Similarly to the procedure outlined above,
we use the isotopic values determined by SIMS to distinguish the triple oxygen isotope
values of the early from the late generations of silica. In the case of the Gunflint Formation,
the Al2O3 and TiO2 contents are uniformly low, rarely exceeding 500 and 50 ppm (Fig. 7),
respectively, generally reflecting the well-preserved early Paleoproterozoic silica deposits and
the absence of silicified lithologies in our collection. The cherts of the Gunflint Formation
used here were collected from Schreiber Beach and Frustration Bat, localities far-removed
from the Duluth intrusion and thus minimally affected by hydrothermal activity and passing
all microanalytical criteria presented previously (Marin-Carbonne et al. 2011, 2012). In this
case, the earliest generation is best explained by early transformation of the marine silica
precursor into microquartz acquiring the high δ30Si values between 0 and +3‰ (Fig. 7). The
new SIMS measurements presented here show that the microquartz has values between 23 and
26‰, while some (but not all) megaquartz segregations with granoblastic textures have δ′18O
vales as low as 18‰ (Fig. 4). The textural and modeling evidence allowed us to distinguish the
signal defining the original early diagenetic silica corresponding to the highest δ′18O values,
that should be close to the mean of the range (Marin-Carbonne et al. 2012).
Using the criteria outlined in Marin-Carbonne et al. (2012) and our new measurements,
we assign the δ′18O value of +24‰ as the isotope signature of the early diagenetic silica,
i.e., the most primary microquartz. The corresponding Δ′17O of this microquartz is −0.08‰
as evident from bulk laser fluorination measurements (Fig. 11). The lower range of δ′18O
values previously measured by SIMS is interpreted as the late stage cycling of fluids during
metamorphism (Marin et al. 2010). The triple oxygen isotope values corresponding to this
generation of quartz are δ′18O = +18‰ and were measured directly in this study (see Fig. 4B).
One sample (5 of 06.28.84) measured for triple oxygen isotope composition yields a δ′18O of
+19.7‰, close to the granoblastic quartz measured by SIMS, and Δ′17O = −0.092‰. Such
a composition could be used to derive the general trend induced by alteration during the
metamorphic event at the temperature recorded by Raman spectroscopy. Then the Gunflint
chert can be explained by primary high-δ′18O microquartz that precipitated in equilibrium
with fluids with δ′18O of −4 to −1‰, similar to the values of modern sediment-hosted pore
fluids (e.g., Aagaard et al. 1989). The corresponding equilibrium temperature recorded by this
generation is between 60 and 80 °C, depending on the exact choice of the equilibrium fluid
value. The secondary generation of quartz formed via dissolution–reprecipitation at elevated
temperatures between 130 and 170 °C during metamorphism, as constrained by Raman
spectroscopy of organic matter (Alleon et al. 2016). The reconstructed equilibrium fluid yields
a δ′18O between +2 and +5, and a Δ′17O of around −0.08‰ (Fig. 11).
Triple oxygen isotope signature of the primary fluids. In each chert-bearing formation
studied here, the depositional environment is undoubtedly marine, thus we assume that the main
reservoirs responsible for the precipitation of primary silica and early diagenesis are: I) seawater,
II) submarine pore fluids, and III) hydrothermal vent fluids (see Fig. 11). The distinction in
slopes of these processes in δ′18O–Δ′17O space is generated via different temperature and
different water–rock balances of exchange reactions between seawater and seafloor lithologies.
The isotope signatures of the fluid reservoirs are then imprinted in the primary microquartz as
shown with their triple oxygen isotopes of modern oceanic quartz, geothermal silica, and chert
(Fig. 12). As an example of transfer of isotope signature from the seawater-derived vent fluids, we
use the Reykjanes scale silica measured in this study and the previously presented compositions
of oceanic quartz and geothermal fluids (Zakharov and Bindeman 2019; Zakharov et al. 2019b).
These samples record the triple oxygen isotope composition of seawater that underwent isotope
exchange with basalt and might serve as a point of comparison for mechanisms of Precambrian
silica precipitation that involved hydrothermal cycling (Fig. 13).
348 Zakharov et al.
Hydrothermal fluids, Reykjanes silica scale: a comparison with Archean silica. In the
case of the studied Archean formations, our interpretation of the primary hydrothermal origin
of cherts conforms to the previous results based on the field relationships, isotope studies
and trace elemental concentrations (de Ronde et al. 1994; Van den Boorn et al. 2007, 2010;
Hofmann and Harris 2008; Cammack et al. 2018; Alleon et al. 2019; Lowe et al. 2019).
These suggest that many Archean chert deposits and silicified basalts were produced during
reactive circulation of seawater rather than precipitation of silica in the deep ocean from cold
seawater. However, we should note that the high-δ′18O cherts (up to +22‰) of the Barberton
Belt presented by Knauth and Lowe (2003) carry a potential to provide insights into a lower
temperature environment, where a marine-dominated signal can be better distinguished.
From reaction with basalt at high temperatures, modern submarine vent fluids are 1−2‰
higher in δ′18O than ambient seawater (Shanks 2001) and 0.01−0.03‰ lower in Δ′17O (Zakharov
and Bindeman 2019; Zakharov et al. 2019b). Consequently, we use the geothermal silica scales
to demonstrate the possible triple oxygen isotope signature carried by silica-saturated fluids at
~180−190 °C. Our measurements are in close agreement with this temperature. They yield an
equilibrium temperature of 175 °C and equilibrium seawater-derived fluids reacted at high water–
rock ratio (Fig. 12). These fluids initially resulted from circulation in hot basaltic wall rocks
at temperature of around 320−350 °C and subsequently cooled to about 188 °C (Hardardottir
2011). The cooling is accompanied by boiling, steam loss, and precipitation of amorphous silica
along with precipitation of minor carbonate and sulfides on the walls of the geothermal pipes
downstream. For more information on these samples refer to Hardardottir (2011). Some meteoric
water component has been inferred in the Reykjanes system, although the salinity of the fluid is
that of seawater (Ólafsson and Riley 1978; Pope et al. 2009). Despite this potential involvement
of local precipitation, we do not see a significant enrichment of Δ′17O values in our Reykjanes
scales (Fig. 9). Some of the Precambrian cherts from the Barberton Belt and the Pilbara craton
form dikes, fracture-filling aggregates, and inter-bedded layers in silicified komatiitic basalts
(e.g., Hofmann and Bolhar 2007; Hofmann and Harris 2008; Lowe et al. 2019) and they might
document a similar style of silica precipitation as these modern silica scales.
The well fluid compositions of the Reyakjanes geothermal field were measured previously
with δ′18O between −0.2 and +1.1‰ and Δ′17O between −0.025 and −0.015‰ (Zakharov et al.
2019b). Such values are in agreement with the seawater–basalt reaction that took place at water–
rock ratios of about 1 (Fig. 12). These might be viewed as the parental fluids for the silica scales as
some of the samples were extracted from the same well (RN−9). The reconstructed compositions
of the earliest generation silica in the Archean Mendon, Kromberg and Dresser Formations are
accordingly consistent with precipitation from such a vent fluid (Fig. 12). Lowering the water–
rock ratio of seawater–basalt reaction would increase the δ′18O and lower the Δ′17O of chert-
forming fluids. At much higher temperatures (300−350 °C), the effect of low water–rock ratio
is depicted with the values of oceanic quartz crystals extracted from the altered oceanic crust
drilled at the Oceanic Drilling Project site 504B (Alt et al. 1996) that were previously analyzed
in Zakharov and Bindeman (2019). With these measurements, the seawater evolution line upon
reaction with basalt is clearly constructed and has a slope of around 0.51 in the δ′17O–δ′18O
coordinates, which gives its steep negative slope in δ′18O–Δ′17O coordinates (Fig. 12A).
In the case of the Archean, the reconstructed δ′18O values of vent fluids at the Barberton
Belt range between +0.9 to +1.6‰, as constrained from the fluid inclusion microthermometry
and oxygen isotope measurements of hydrothermal quartz hosted in altered komatiitic rocks
spatially associated with the cherts (De Ronde et al. 1997; Farber et al. 2015). The Δ′17O
value of such a fluid would be dependent on the initial value of seawater and the extent
of water–rock reaction.If the seawater value was 0 or −1‰, then the Δ′17O of such fluids
would plot along the same trajectory as the modern fluids, without a significant difference.
0.1
A
Archean
seawater? Arche
equil an
ibrium
silica
345 ?
0.0 175
Reykjanes
scale silica 85
Reykjanes
Oceanic quartz 55
Eocene
−0.1 fluids
Z2019
ODP504B marine chert
Z&B2019 3.5-3.3 Ga S2016
1.9 Ga
D'17O0.528, ‰ VSMOW
cherts Cretaceous
cherts 5 °C
chert (SKFS)
−0.2 25
pore water fluids

−0.3 hydrothermal fluids
0 10 20 30 40
0.1 meteoric
water continent-free
seawater?
B
200
0.0 cr
to
w
us ard 175
ta s l
l in ow
te hydroth.
55
ra W/R
cti input
−0.1 on -> 25
175 5 °C
3000
crustal
−0.2
Age, Ma
2000 rocks 55
1000 25
5 °C
−0.3 0
−20 0 20 40
d'18O, ‰ VSMOW
Figure 12. Panel A depicts the systematics of the modern oceanic silica phases. Precipitation of original
silica from seawater or seawater-derived fluids (pore and hydrothermal fluids) is shown with pink and
blue curves encompassing the temperature range between 5 °C and infinity (intermediate temperatures are
shown), with the δ′18O fluid values between +3 and −4‰ of modern seawater. As an example of hydrother-
mal vent fluids we show the composition of seawater-dominated well fluids (shown as Z2019; Zakharov et
al. 2019) and amorphous silica scale from the pipelines of the Reykjanes system (this study). Further, the
oceanic quartz extracted from high-temperature altered oceanic crust (ZB2019; Zakharov and Bindeman
2019) depicts equilibrium with evolved seawater-derived fluids at 345 °C. The seawater evolution path
induced by hydrothermal circulation is shown with blue dotted lines using modern and ice-free seawater
values (collectively shown as an oval). Modern marine chert (S2016; Sharp et al. 2016) and Cretaceous
chert (SKFS standard measured here) are shown with squares. The samples studied here are shown with
shaded circles. The Archean seawater path is shown with the dotted arrow and the equilibrium quartz
composition is shown with a dotted curve (model after Sengupta and Pack, 2018; see main text below).
In Panel B we show the compilation of chert values (Levin et al. 2014; Hayles et al. 2019; Liljestrand et
al. 2020; this study) and the modern oceanic silica samples from panel A (black unfilled shapes). The
overarching feature of these datasets is their Δ′17O lower than silica in equilibrium with seawater-derived
sources (as depicted in A). Secondary quartz in equilibrium with pristine meteoric waters of different com-
position is shown with an array of grey curves. We also show quartz crystallized from fluids that underwent
isotope exchange with crustal rocks at variable water–rock ratios. Precipitation of quartz from pristine or
modified meteoric water would require temperatures mostly below 55 and 175 °C, respectively, to explain
mixing between primary and secondary quartz. In addition, we show the seawater evolution path in absence
of continental weathering and with enhanced oceanic weathering (red arrow; see main text). The compo-
sition of such seawater becomes similar to that of meteoric water; however, it does not extend below the
value of ~ −4‰ (see text). The contribution from hydrothermal cycling of seawater is shown with the blue
arrow and equilibrium silica composition is outlined with the blue dotted curve.
350 Zakharov et al.
Thus, the positive δ′18O values of the fluids would be accompanied by negative Δ′17O values
toward the composition of basalt (δ′18O = +5.6 and Δ′17O = −0.06‰; Eiler et al. 2001; Cano et
al. 2020). The compositions of analyzed Archean cherts in fact are more consistent with this
origin of the fluids resulting from exchange with basaltic rocks. This trend is consistent with
the hypothesis of a vent-fluid origin of at least some of the Archean cherts, supported by their
low δ30Si values. Further, this suggests that the seawater value must have not been significantly
different in the Archean from the range between −2 and 0‰ (Holmden and Muehlenbachs
1993; Furnes et al. 2007; Pope et al. 2012; Zakharov and Bindeman 2019). However, it is not
possible to reconstruct the ambient seawater temperature directly from these measurements as
the vent fluid temperature is not reflective of ambient conditions.
Interstitial pore fluids. Contrary to the Archean counterparts, the Proterozoic cherts of the
Gunflint Formation are found within shallow to deep marine sequences indicating precipitation
from the seawater column and with minimal contribution of vent fluid solutions. The high δ30Si
values measured in post-Archean cherts are also in agreement with limited contribution from
basalt-derived hydrothermal silica and primary contribution by diagenetic solutions initially
sourced from seawater dissolved silica (Tatzel et al. 2015). In modern siliceous sediments the
crystallization of microquartz is accompanied by interaction with marine pore water fluids.
Consequently, we propose that the triple oxygen isotope composition of the Gunflint chert
sheds light on the composition of sediment-hosted pore water fluids at the temperature of
early marine diagenesis. Pore fluids in oceanic crust are routinely measured (e.g., Aagaard
et al. 1989; Hesse et al. 2000), yielding δ′18O values several‰ lower, between −4 to −1‰,
compared to that of the ambient seawater. Using the corresponding Δ′17O shift measured in
authigenic minerals (Sengupta and Pack 2018; Bindeman et al. 2019; Bindeman 2021, this
volume), we propose that the value of pore fluids must be positive in Δ′17O compared to the
parental seawater value. Since such fluids are likely representative of the equilibrium fluids
for the early diagenetic microquartz that formed within seafloor sediment, the corresponding
equilibrium temperature is 60−80 °C (Figs. 11 and 12). Analogously to the modern pore water
fluids retrieved from the siliceous sediments, the signature of the Gunflint Formation chert is
best explained by a fluid with δ′18O of around −2‰ (Fig. 12).
Secondary fluids: burial metamorphism and post-depositional recrystallization. The textural
complexity of silica phases and the several per mil range in δ′18O and δ30Si values within each
sample reflect several stages of interaction with diagenetic and secondary fluids, and dissolution–
reprecipitation at temperatures different from the initial precipitation. Depending on the history of
individual formation, such fluids could originate from local meteorically charged ground waters, as
well as water expelled from hydrous minerals of surrounding lithological units at higher temperature
during phase transitions (e.g., illite–smectite or smectite–chlorite transitions; see Bindeman 2021,
this volume). Upon further burial and metamorphism, crustal fluids might become important. Their
isotope compositions are dominated by rock-buffered reactions at high temperature with crustal
rocks that on average have a δ′18O of around +11‰ (Bindeman et al. 2019).
It is evident that the Archean cherts studied here were modified by a metamorphic event
which affected all lithologies of the hosting greenstone belts (e.g Xie et al. 1997; Knauth and
Lowe 2003). Assuming that quartz recrystallized in metamorphic conditions (at least ~300 °C
for Mendon and Dresser Formations, as documented by Raman spectroscopy and fluid inclusion
studies; Marin-Carbonne et al. 2011), the reconstructed fluids would have to have δ′18O values
of around +3‰ to explain the equilibrium secondary silica with a low Δ′17O value (Fig. 11).
These fluids would be similar to those formed at low water–rock ratios and could be calculated
in the triple oxygen isotope space given high-temperature fractionation with crustal rocks
(water–rock ratios below ~0.3; Fig. 11). The proposed fluids with a δ′18O around +3−5‰
would have a Δ′17O of −0.1‰, which are best explained by rock-buffered reactions that
occur in the crust, i.e., crustal fluids (Wilkinson et al. 1995; Baumgartner and Valley 2001;
Bindeman 2021, this volume). Similarly, in the case of the Gunflint Formation, the late
generation silica could have formed at temperatures as high as 170 °C, as indicated by Raman
spectroscopy data (Alleon et al. 2016), which yields the corresponding equilibrium fluids with
a δ′18O of +2 to +5‰ and a Δ′17O of −0.10‰. Such a fluid signature could be a result of basinal
brines (also known as formation waters) contributing to the late diagenetic transformations
(e.g., Clayton et al. 1966; Hitchon and Friedman 1969).
In fact, many previous studies explained the low δ′18O values of Precambrian cherts
by diagenetic interaction with meteoric waters (see Fig. 12). With the triple oxygen isotope
approach, it seems difficult to explain all of the low Δ′17O values by interaction with pristine
meteoric water because in such cases the precipitation and dissolution of microquartz had
to occur at relatively low temperatures (40−70 °C; Fig. 12) , in equilibrium with fluids with
δ′18O between −25 and 0‰ (Liljestrand et al. 2020), and presumably at high water–rock ratios
above 1. The modified crustal fluids would be able to induce the same effect by precipitation of
quartz at temperatures between 55 and 175 °C (see Fig. 12). The sign of change in δ′18O from
original microquartz to altered chert would depend on its original value and the temperature of
interaction with crustal fluids, yet the Δ′17O values would change in one direction, negatively
(see Fig. 12). Further evidence for interaction between such fluids and sedimentary silica has
been documented using in situ SIMS analysis of quartz overgrowth formed in sandstones that
experienced several kilometers of burial depth (Williams et al. 1997; Pollington et al. 2011).
Further, in other geological situations, where metamorphism involves other lithologies, it is
common to observe increased δ18O values due to interaction with crustal fluids. For example,
the early Paleoproterozoic Scourie dikes of NW Scotland with initially low δ18O magmatic
values spanning between −2 to +2‰ experienced recrystallization in presence of crustal
fluids near shear zones and in high-grade metamorphosed parts of the terranes. The attendant
increase in the δ18O bulk value is 4−5‰ (Cartwright and Valley 1992; Zakharov et al. 2019a).
Unlike cherts, these rocks are composed of multiple minerals, thus making identification of
preserved and metamorphosed precursors more manageable. While direct measurements of
basinal and crustal fluids for triple oxygen isotopes have not been reported yet, their Δ′17O
must be low due to exchange with rocks that are isotopically distinct from the surface sources
of fluids (Fig. 12). Due to high-temperature fractionation at several kilometers burial depth, the
crustal fluid Δ′17O would be governed by that of rocks, between −0.06 and −0.1‰, spanning
between mantle-derived and metamorphic rocks (see Bindeman 2021, this volume).
Water content and cherts’ hydrogen isotope composition. Measured hydrogen isotope
compositions of cherts give us an opportunity to cross-validate the model of the origin of
secondary quartz, since the bulk δD values of chert are more sensitive to the involvement
of meteoric water (Fig. 10). This is confirmed by homogeneous δD values measured from
multiple chunks of the same samples that only vary about ±10‰ on average. The generally
low δD values between −110 and −60‰ likely point towards involvement of meteorically
derived fluid (Fig. 10). Even if we assume that water is held by phases with large H2O–OH
fractionations defined by low-temperature equilibrium (Sheppard and Gilg 1996), we still
expect that the low δD values of cherts trace involvement of meteoric waters. Alteration of δD
values of chert could also readily occur under ambient temperature due to the low hydrogen
content and small grain size of microquartz. Even Jurassic and Cretaceous cherts exhibit very
low and variable δD compositions owing to low-temperature exchange with local precipitation
(Fig. 10; Knauth and Epstein 1976). Moreover, the presence of organic material with low δD
values in the cherts probably influenced the measured values. For example, the reported total
organic carbon in the Dresser Formation varies on the order of 0.1 % (Ueno et al. 2004), and
would be higher for the well-preserved Gunflint Formation. The weakly bonded hydrogen
would have been a subject to modification due to thermal maturation of organic molecules and
devolatilization, favoring heavy (high δD) H2O loss.
352 Zakharov et al.
Triple oxygen isotope evolution of seawater

Both interstitial pore fluids and vent fluids are considered here as possible sources of
equilibrium fluids during the transformation of amorphous silica into microquartz. Such fluids
originate from interaction between pristine seawater and oceanic crust, however, their isotope
compositions are directly related to the primary seawater signature. We can first consider the
secular evolution of seawater composition in terms of dominant processes that control the
isotope ratios of the terrestrial hydrosphere (Fig. 13). The contributions from high-temperature
alteration of oceanic crust and from low-temperature weathering reactions provide the two
largest volumes of exchanged oxygen delivered into the oceans. The effect produced by each
process has been considered in numerous previous publications (Muehlenbachs and Clayton
1972; Holland 1984; Muehlenbachs 1998; Kasting et al. 2006; Kanzaki 2020), including those
focused on the triple oxygen isotope ratios (Sengupta and Pack 2018; Bindeman 2021, this
volume; Wostbrock and Sharp 2021, this volume). The dominant amount of exchanged oxygen
is provided by high-temperature reactions between seawater and oceanic crust at mid-ocean
ridges that exert positive shifts in δ′18O and negative in Δ′17O, close to +1.5‰ and −0.03‰,
respectively (Fig. 13). These isotope shifts are clearly detected by the previously measured
δ18O values of altered oceanic crust (Alt et al. 1986, 1996; Alt and Teagle 2000; Alt and Bach
2006) and hydrothermal fluids (Shanks and Seyfried 1987; Bach and Humphris 1999; Shanks
2001). These measurements show that the oceanic crust is lower in 18O/16O than pristine
unaltered MORB, while vent fluids are higher compared to ambient seawater. They evolve
along similar trajectories in δ′18O–Δ′17O space but with opposite sign.
Next are the weathering reactions that provide less voluminous but significant amounts of
exchanged oxygen. Traced by isotope measurements of weathering products such as submarine
clays and terrigenous sediments (Bindeman et al. 2018, 2019; Sengupta and Pack 2018),
the collective input of weathering is characterized by values −3‰ in δ′18O and +0.040‰ in
Δ′17O compared to the values of contemporaneous hydrosphere. These are average values for
the combined submarine and terrestrial weathering processes, with the difference shown in
Figure 13. Most importantly, these contributions deliver fluids that are shifted in the opposite
direction compared to the fluids derived from alteration of oceanic crust. Today these processes
balance seawater at its modern value of around 0‰ (Fig. 13). The relative contributions of
these fluxes through time is still poorly understood, especially regarding the off-axis alteration
of oceanic crust, and may undergo revisions in the near future. We currently adopt the flux
figures (g/yr) of Muehlenbachs (1998).
The isotope contributions of other processes such as cycling of water through the
mantle wedge at subduction zones, growth of continental crust, and increasing proportion
of exposed sedimentary blanket complete the isotope budget of terrestrial hydrosphere,
however the volume of exchanged oxygen produced by these processes is small, less than
10 %, compared to the effect of seawater–basalt reaction and weathering (Muehlenbachs
1998). Moreover, the isotope effects produced by these process are difficult to quantify by
direct measurements. Given that about 3 km2 of oceanic floor is being produced each year that
subsequently undergoes high-temperature alteration at depths between 1 and 3 km (Alt and
Teagle 2000), the seawater–basalt reaction provides the dominant flux of exchanged oxygen
into the terrestrial hydrosphere, accounting for 60 % when considered against the contribution
from weathering. That is 1.8∙1016 g/year of exchanged oxygen produced by the hydrothermal
alteration of oceanic crust (Muehlenbachs 1998). The combined estimated flux of weathering
in submarine and terrestrial settings equates to 1.2∙1016 g/year, as estimated from the thickness
of low-temperature altered crust and the suspended load of clay fraction carried by rivers
(Holland 1984; Muehlenbachs 1998). Even in the modern world, these figures are subject to
uncertainties due to our inability to directly estimate how much material actually undergoes
isotope exchange. It is further complicated by the presence of diverse lithologies of the exposed
0.
51
6±
0.
00
0.04 Fcw = 1∙1016 g/yr
D'17O0.528 , ‰ VSMOW
1
c
we ont
at ine
he nt
rin al
0.02 Fow = oce
g
0.2∙10 g/yr wea anic
16
the
ring steady ±continental glacial ice
state
0.00 hy h
no 0.5 d ig
c 1
we ontin 9 al roth h-T
te e
ath en ra rm
eri tal
ng
tio a
n l
−0.02 Fha = 1.8 ∙1016 g/yr
−4 −2 0 2 4
d'18O, ‰ VSMOW
Figure 13. The effects and fluxes (g/yr) of the main contributions to the isotope budget of the terrestrial
hydrosphere visualized in the triple oxygen isotope coordinates. The high temperature alteration of oceanic
crust (Fha) represents the largest flux of exchanged oxygen into the oceans (flux figures from Muehlen-
bachs 1998). The slope of the isotope exchange produced by hydrothermal input is constrained based on
previous studies of hydrothermal fluids and epidotes (Zakharov et al. 2019b). The values of exchanged
water produced by oceanic and continental weathering (Fow and Fcw, respectively) are based on previous
measurements of weathered products (Sengupta and Pack 2018; Bindeman et al. 2019). The relative pro-
portion of the two weathering fluxes in the modern mass-balance model is shown with the cross on the
dotted line. The effect of converting 10% of the hydrosphere into continental glacial ice is shown with
the double-sided arrow yielding the seawater δ′18O value of about +0.5‰. A potential explanation for
the moderately low 18O/16O ratio of Archean seawater (e.g., Bindeman 2021, this volume) is possible by
enhanced weathering fluxes and/or lower hydrothermal activity. The outputs of the Monte-Carlo simulation
of the steady state solution with variable fluxes are shown with the concentrically colored-coded density
distribution. The warmer colors represent higher relative density distribution. The continental weathering
and hydrothermal inputs were varied between 50 and 150% of their modern-day values, while oceanic
weathering was varied between 50 and 300%. The outputs of the simulation form an array with a slope of
0.516 ± 0.001 in δ′17O–δ′18O coordinates. In addition, if all continental weathering is suppressed (0 g/yr),
the values of seawater will vary along a slope of 0.519 (shown with dashed arrow).
continental crust that contain already-weathered material that is in isotope equilibrium with
the weathering fluids. As for the Precambrian world, the spreading rates are not known, while
contributions from weathering might have been enhanced by CO2 rich atmospheres. However,
the enhanced weathering would be counter-balanced by limited subaerial exposure of Archean
continental land (Flament et al. 2013; Bindeman et al. 2018). Consequently, it is difficult
to imagine that the weathering flux of exchanged oxygen was overwhelming relative to the
hydrothermal input in the early Precambrian, causing the dramatically lower 18O/16O ratios of
seawater. A potential solution is offered by a hypothesis in which Archean mid-ocean ridges
were exposed above seawater, providing a significantly higher proportion of low-temperature
interaction between seawater and the ocean floor (Walker and Lohmann 1989).
Nonetheless, we can operate with these simple estimates to reproduce the array of
possible compositions of ancient seawater, in a manner similar to the study of Sengupta
and Pack (2018). In this simplistic approach, we consider that the ratio of fluxes produced
by hydrothermal and weathering processes is what exerts the main control on the isotope
composition of seawater. A budget that is dominated by the hydrothermal contribution
would produce the seawater that is positive in δ′18O and negative Δ′17O, i.e., shifted towards
composition of fluids in equilibrium with mid-ocean ridge basalts. In the other scenario, where
weathering contributions exceed their modern proportion, the isotope ratio of seawater would
evolve along a steep slope in δ′18O–Δ′17O coordinates (Fig. 13; note that the corresponding
354 Zakharov et al.
slope is gentle in δ′17O–δ′18O coordinates). The change in the isotope composition of seawater
governed by these contributions is described as the sum of flux figures (Fi) multiplied by the
isotope effects (Δδ18/17Oi) exerted on contemporaneous seawater by the respective processes:
d( 18 /17 O hyd )   Fi   18 /17 Oi 

 
dt  mhyd
 
where δ18/17Ohyd and mhyd are the isotope value and mass of oxygen contained in the hydrosphere,
which is essentially seawater and continental glacial ice. The steady state is reached after
about 150 Myr with the values of seawater being δ′18O = −0.2‰ and Δ′17O = +0.001‰. In
addition, we performed a Monte-Carlo simulation of this model, where the contributing fluxes
were randomly changed to anywhere from 50 to 150 % of their modern values using uniform
distributions. To make this simulation applicable to the Archean, when the Earth was mostly
covered by oceanic crust, we extended the upper variability limit of the oceanic weathering
flux to 300 % of its modern value. The locus of steady state solutions is shown in Figure 13.
The direction of changing compositions varies along the slope of 0.516 ± 0.001 when expressed
in δ′17O–δ′18O coordinates, similar to what was found by Sengupta and Pack (2018). The δ′18O
in the output of the simulations varies within −2 to +1‰, and is accompanied by variation in
Δ′17O between −0.01 and +0.02‰ (Fig. 13).
Further, if we consider the Archean world, where continental exposure might have been
minimal (Flament et al. 2013; Bindeman et al. 2018), the fluxes of exchanged oxygen would
be merely controlled by equilibrium fractionation between seawater and minerals of basaltic
composition at low (off-axis) and high (on-axis) temperatures. In such a case, seawater would
possibly evolve along a simple path between values around −4 and +1.5‰ with the slope
of 0.519 in the triple oxygen isotope system as defined by equilibrium fractionation factors
(Fig. 13). This limited variability might offer reconciliation for some low 18O/16O ratios
recorded by marine sediments, however not many low Δ′17O compositions in the present
datasets can be exclusively explained that way (Fig. 12).
This is a characteristic array of seawater values that can be used as a theoretical constraint
when considering the triple oxygen isotope evolution of the sedimentary record. That array
is shown with the measurements of cherts in Fig. 12. If the change is small, for example the
Archean seawater was −4‰, the accompanying shift in δ′18O–Δ′17O would be similar to
that exhibited by the difference between seawater and near-coastal meteoric water (Fig. 12).
Our reconstructed primary compositions agree with precipitation from seawater-derived
fluids, either vent fluids or marine pore fluids (Fig. 12), and a range of seawater δ′18O values
between −4 and +2‰ is permitted. Such a range of seawater values can offer only a limited
range of solutions explaining the carbonate and chert records without invoking higher ocean
temperatures (> 60 °C; see Fig. 1). However, given that carbonates are even more soluble than
quartz, the complexity of diagenetic interactions with fluids must bear a significant impact
on the isotope signature of ancient sediments. Moreover, recent clumped isotope studies of
early Phanerozoic carbonates show that many samples experienced diagenetic alteration
(Cummins et al. 2014; Bergmann et al. 2018; Ryb and Eiler 2018), and that the preserved
samples do not bear evidence for exceptionally high oceanic temperatures or for drastically
low seawater δ18O. Similarly to microanalytical investigations of cherts, ancient (post-
Cenozoic) carbonates require detailed imaging and trace elemental measurements to define
the best preserved domains (Grossman 2012). Bergmann et al. (2018) provided estimates
for Ordovician and Cambrian seawater that range between −2.5 and + 3‰ in δ18O and with
temperatures between 26 and 38 °C. However, the recent iron oxide sedimentary record
(Galili et al. 2019) presents yet another challenging evidence for the change in oxygen isotope
values of seawater through time, where Cambrian and Precambrian time was characterized
by seawater with δ18O at least 5‰ lower than today, inconsistent with the clumped isotope
studies of the early Phanerozoic and with the model presented in Figure 13. Though the iron
oxide record has not been explored in detailed by in situ methods, it presents an interesting
low-solubility archive of both the redox and isotope evolution of terrestrial oceans.
In our suite of samples, the oxygen isotope variations do not require a significantly
different triple oxygen isotope composition of seawater outside the realm of variable fluxes
(Fig. 13). Recrystallization during late diagenesis or metamorphism is likely to disturb the
primary signal shifting the triple oxygen isotope composition downward in terms of Δ′17O,
towards the composition of fluids that reacted with crustal rocks.
FUTURE DIRECTION AND CONCLUSIONS

Fractionation of triple oxygen isotopes between fluids and minerals is a powerful tool
recently added to the arsenal of geochemists studying ancient marine record. Theoretically,
the δ′18O–Δ′17O values measured in a pristine microquartz that formed during the last silica
transformation within the subseafloor setting should reflect the temperature of the transition and
the isotope signature of marine pore water fluids. In the realm of Precambrian chert deposits
that underwent multiple episodes of recrystallization induced by diagenesis and metamorphism,
the measured δ′18O–Δ′17O values likely represent a mixture of equilibrium compositions
generated asynchronously. This is evident from the SIMS analyses that reveal several-‰
spatial heterogeneity in δ′18O values within each sample. Sometimes, but not always, the
heterogeneities are accompanied by petrographic distinction, e.g., low δ′18O veins in high δ′18O
microquartz. The distance between several-‰ heterogeneities varies on the scale of 10−100 μm.
Consequently, the currently available triple oxygen isotope values of cherts produced by bulk
laser fluorination GS-IRMS measurements of ~1 mm3 samples are unlikely to give completely
accurate reflections of the original marine pore fluids and equilibrium temperatures. The
question then is, which composition measured by SIMS corresponds to the seafloor-deposited
silica that underwent just the early stage of marine diagenesis? Is it possible to reconstruct
the δ′18O–Δ′17O values of this phase? We suggest that future triple oxygen isotope endeavors
should include careful investigation of chert samples with identification of multiple generations
of silica. Then, selection of the best-preserved domains could be assessed by trace element
measurements, CL-imaging, and δ′18O and δ30Si values measured in situ. Following that, the
triple oxygen isotope composition should be measured on these distinct generations of quartz,
perhaps using micro-drilling of the samples or partial fluorination. Additional triple oxygen
isotope measurements of modern silica sediments carried out across the phase transitions, aided
by measurements of local geothermal gradient and pore water fluids (see example in Yanchilina
et al. 2020) would greatly assist our understanding of the ancient silica cycle.
In this study we examined the triple oxygen isotope compositions of Precambrian cherts
accompanied by SIMS measurements. We provided an attempt to disentangle triple oxygen
isotope signals of early and late quartz generations in the Precambrian cherts of the well-
studied Dresser, Kromberg, Mendon and Gunflint Formations using SIMS-determined δ′18O
and δ30Si values that span several ‰ within the samples. Aided by measurements of Al and Ti
concentrations, δ30Si values, and Raman spectroscopy, we provide estimates for the triple oxygen
isotope signature of the early microquartz and late-stage recrystallized quartz on a case-by-case
basis. Using the Raman spectroscopy of organic matter hosted in these samples, we derived the
maximum possible temperature at which recrystallization might have occurred. Further, by using
δD values measured in the same samples we tested for the involvement of meteoric water during
the recrystallization of the cherts. We found that the data set is best explained by the following:
356 Zakharov et al.
• The Archean 3.5 Ga Dresser, 3.4 Ga Kromberg and 3.2 Ga Mendon cherts contain
low-Al microquartz that was likely the earliest generation of silica precipitated in an
environment with mixed hydrothermal and seawater input. The triple oxygen isotope
signature of samples with negative δ30Si values is consistent with precipitation of
microquartz at 150−170 °C and some contributions from vent fluids. The extrapolation
of vent fluids towards pristine seawater does not require a hydrosphere significantly
different from the modern-day and ice-free world values. This interpretation applies
to the samples studied here, which does not mean that all Archean cherts formed from
mixed hydrothermal solutions. However, each chert deposit needs to be evaluated for the
influence of marine vs hydrothermal contributions to its isotope composition.
• The triple oxygen isotope signature of modern amorphous silica scale precipitated in
pipelines of the seawater-dominated hydrothermal system at Reykjanes, Iceland is used
here as a possible analogue of Archean chert deposits that resulted from reaction between
seawater and basalt. The Reykjanes silica formed at ~188 °C due to high-temperature
reaction between basalt and seawater. The δ′18O−Δ′17O values of such silica are is in
good agreement with this temperature and previous measurements of the Reykjanes well
fluids. Based on the triple oxygen isotopes, we suggest that at least some of the Archean
chert deposits might reflect a similar regime of silica precipitation. Such cherts would
have δ′18O−Δ′17O values defined by high-temperature exchange reactions with basaltic
rocks at different water–rock ratios. At decreasing water–rock ratios, the silica acquires
Δ′17O values lower than that of silica in equilibrium with pristine seawater.
• The Paleoproterozoic Gunflint chert contains a high-δ′18O generation of microquartz
that is thought to originally crystallized from marine siliceous sediment based on trace
element concentrations and silicon isotope measurements. We interpret the triple oxygen
isotope signature of such microquartz as a record of marine pore fluids. The δ′18O−Δ′17O
of such quartz is consistent with an equilibrium temperature of 60−80 °C, given that the
pore fluid values had δ′18O of around −2‰.
• Triple oxygen isotope values of secondary quartz resulting from dissolution–
reprecipitation of original material indicates involvement of crustal fluids, akin to
modern basinal brines. The temperature of the imposed signature is estimated from
Raman-spectroscopy at 330 ± 30 °C and 160 ± 30 °C for the Archean and Paleoproterozoic
formations, respectively. These estimates help to assess the preservation state of the
original microquartz. They do not necessarily indicate that oxygen isotope exchange
occurred at these temperatures, however in the case of Onverwacht group cherts, these
temperatures are close to the homogenization temperatures measured in chert-hosted fluid
inclusions (Marin-Carbonne et al. 2011). These temperatures also allow us to suggest that
the fluids altering primary signals might have ranged between +2 and +5‰, and that their
Δ′17O values are low, close to −0.1‰, characteristic of crustal origin. Such fluids would
cause cherts to become low in Δ′17O, between −0.06 and −0.1‰, offering a generalized
explanation for some of the low-Δ′17O cherts present in the global dataset (Fig. 12).
Triple oxygen isotope investigation of Precambrian cherts, even in combination with high-
resolution methods, demonstrates their polygenic nature and limited utility to directly read
isotopic values of the seawater and the Earth’s surface temperatures. Despite the complexity of
chert deposition and its cycle through diagenetic and metamorphic transformations, the recorded
secular isotope trend is still best explained by the temperature control, but not in a straightforward
way as a reflection of ocean temperatures. The high-temperature early diagenetic and
hydrothermal processes fueled by circulation of seawater within seafloor sediments can account
for the signature of some Archean cherts without the need to invoke a significantly different
oxygen isotope value of the terrestrial hydrosphere. Future triple oxygen isotope investigations
of cherts should involve samples that underwent careful examination in the context of geological
setting, trace elemental and mineralogical composition, as well as microscale isotope variations.
ACKNOWLEDGEMENTS
We are thankful to Stanley Awramcik, Bill Schopf, François Robert for donating their
samples to this study and previous studies, and Mike Hudak for measuring the δD values in
cherts. We also thank Vigdís Hardardóttir for granting the samples of Reykjanes silica scales
for triple oxygen isotope analyses and Mark Reed for facilitating the access to the samples.
JMC and JA thank the European Research Council (ERC) under the European Union’s Horizon
program (STROMATA, grant agreement 759289). INB and DOZ thank the NSF grant EAR
#1833420. We are grateful to the reviewers Daniel Herwartz and Zach Sharp for their input,
comments and suggestions that helped to improve the manuscript. We also greatly appreciate
Dylan Colón’s help proofreading the manuscript.
REFERENCES
Aagaard P, Egeberg PK, Smalley PC (1989) Diagenetic reactions in Leg 104 sediments inferred from isotope and
major element chemistry of interstitial waters. Proc Ocean Drill Program Sci Results 104:273–280
Abraham K, Hofmann A, Foley SF, Cardinal D, Harris C, Barth MG, André L (2011) Coupled silicon–oxygen isotope
fractionation traces Archaean silicification. Earth Planet Sci Lett 301:222–230
Alexandre A, Basile-Doelsch I, Delhaye T, Borshneck D, Mazur J-C, Reyerson P, Santos GM (2015) New highlights of
phytolith structure and occluded carbon location: 3-DX-ray microscopy and NanoSIMS results. Biogeosciences
12:863–873
Alleon J, Bernard S, Le Guillou C, Marin-Carbonne J, Pont S, Beyssac O, McKeegan KD, Robert F (2016) Molecular
preservation of 1.88 Ga Gunflint organic microfossils as a function of temperature and mineralogy. Nat Commun 7:1–11
Alleon J, Bernard S, Le Guillou C, Beyssac O, Sugitani K, Robert F (2018) Chemical nature of the 3.4 Ga Strelley
Pool microfossils. Geochem Perspect Lett 37–42
Alleon J, Flannery DT, Ferralis N, Williford KH, Zhang Y, Schuessler JA, Summons RE (2019) Organo-mineral
associations in chert of the 3.5 Ga Mount Ada Basalt raise questions about the origin of organic matter in
Paleoarchean hydrothermally influenced sediments. Sci Rep 9:16712
Allwood AC, Walter MR, Kamber BS, Marshall CP, Burch IW (2006) Stromatolite reef from the Early Archaean era
of Australia. Nature 441:714–718
Allwood AC, Walter MR, Burch IW, Kamber BS (2007) 3.43 billion-year-old stromatolite reef from the Pilbara
Craton of Western Australia: ecosystem-scale insights to early life on Earth. Precambrian Res 158:198–227
Alt JC, Bach W (2006) Oxygen isotope composition of a section of lower oceanic crust, ODPHole 735B. Geochem
Geophys Geosyst 7:Q12008
Alt JC, Teagle DAH (2000) Hydrothermal alteration and fluid fluxes in ophiolites and oceanic crust. In Ophiolites and
oceanic crust: new insights from field studies and the Ocean Drilling Program. Geological Society of America
Alt JC, Muehlenbachs K, Honnorez J (1986) An oxygen isotopic profile through the upper kilometer of the oceanic
crust, DSDPHole 504B. Earth Planet Sci Lett 80:217–229
Alt JC, Kinoshita H, Stokking LB, Michael PJ (Eds.) (1996) Proc Ocean Drill Program Sci Results Vol 148. Ocean
Drilling Program
André L, Cardinal D, Alleman LY, Moorbath S (2006) Silicon isotopes in 3.8 Ga West Greenland rocks as clues to the
Eoarchaean supracrustal Si cycle. Earth Planet Sci Lett 245:162–173
Appleby SK, Graham CM, Gillespie MR, Hinton RW, Oliver GJH, Eimf (2008) A cryptic record of magma mixing in
diorites revealed by high-precision SIMS oxygen isotope analysis of zircons. Earth Planet Sci Lett 269:105–117
Awramik SM, Barghoorn ES (1977) The Gunflint microbiota. Precambrian Res 5:121–142
Bach W, Humphris SE (1999) Relationship between the Sr and O isotope compositions of hydrothermal fluids and the
spreading and magma-supply rates at oceanic spreading centers. Geology 27:1067–1070
Barghoorn ES, Tyler SA (1965) Microorganisms from the Gunflint Chert. Science 147:563–577
Baumgartner LP, Valley JW (2001) Stable isotope transport and contact metamorphic fluid flow. Rev Mineral
Geochem 43:415–467
Bégué F, Baumgartner LP, Bouvier A-S, Robyr M (2019) Reactive fluid infiltration along fractures: Textural
observations coupled to in-situ isotopic analyses. Earth Planet Sci Lett 519:264–273
Behl R, Garrison RE, Iijima I, Abed A, Garrison, R (1994) The origin of chert in the Monterey Formation of California
(USA). In: Siliceous, Phosphatic and Glauconitic Sediments of the Tertiary and Mesozoic: Proceedings of the
29th International Geological Congress, Part C, p 101–132
Bekker A, Planavsky N, Rasmussen B, Krapez B, Hofmann A, Slack J, Rouxel O, Konhauser K (2014) Iron
formations: Their origins and implications for ancient seawater chemistry. In: Treatise on Geochemistry, Vol.
12, Elsevier, p 561–628
358 Zakharov et al.
Bellucci JJ, Whitehouse MJ, Nemchin AA, Snape JF, Kenny GG, Merle RE, Bland PA, Benedix GK (2020) Tracing
martian surface interactions with the triple O isotope compositions of meteoritic phosphates. Earth Planet Sci
Lett 531:115977
Bergmann KD, Finnegan S, Creel R, Eiler JM, Hughes NC, Popov LE, Fischer WW (2018) A paired apatite and
calcite clumped isotope thermometry approach to estimating Cambro–Ordovician seawater temperatures and
isotopic composition. Geochim Cosmochim Acta 224:18–41
Berner RA, Kothavala Z (2001) Geocarb III: A revised model of atmospheric CO2 over Phanerozoic Time. Am J Sci
301:182–204
Beyssac O, Goffé B, Chopin C, Rouzaud JN (2002) Raman spectra of carbonaceous material in metasediments: a new
geothermometer. J Metamorph Geol 20:859–871
Beyssac O, Bollinger L, Avouac J-P, Goffé B (2004) Thermal metamorphism in the lesser Himalaya of Nepal
determined from Raman spectroscopy of carbonaceous material. Earth Planet Sci Lett 225:233–241
Bindeman IN, Valley JW (2001) Low-d18O rhyolites from Yellowstone: Magmatic evolution based on analyses of
zircons and individual phenocrysts. J Petrol 42:1491-517
Bindeman IN, Fu B, Kita NT, Valley JW (2008) Origin and evolution of silicic magmatism at yellowstone based on
ion microprobe analysis of isotopically zoned zircons. J Petrol 49:163–193
Bindeman IN, Bekker A, Zakharov DO (2016) Oxygen isotope perspective on crustal evolution on early Earth: A
record of Precambrian shales with emphasis on Paleoproterozoic glaciations and Great Oxygenation Event.
Nature 557:545–548
world’s rivers: Insights into weathering processes and global fluxes into the hydrosphere. Earth Planet Sci Lett
528:115851
Binet L, Gourier D, Derenne S (2008) Potential of EPR imaging to detect traces of primitive life in sedimentary rocks.
Bonamici CE, Kozdon R, Ushikubo T, Valley JW (2014) Intragrain oxygen isotope zoning in titanite by SIMS:
Cooling rates and fluid infiltration along the Carthage–Colton Mylonite Zone, Adirondack Mountains NY, USA.
J Metamorp Geol 32:71–92
Bontognali TR, Sessions AL, Allwood AC, Fischer WW, Grotzinger JP, Summons RE, Eiler JM (2012) Sulfur
isotopes of organic matter preserved in 3.45-billion-year-old stromatolites reveal microbial metabolism. PNAS
109:15146–15151
Brandriss ME, O’Neil JR, Edlund MB, Stoermer EF (1998) Oxygen isotope fractionation between diatomaceous
Brasier MD, Antcliffe J, Saunders M, Wacey D (2015) Changing the picture of Earth’s earliest fossils (3.5–1.9 Ga)
with new approaches and new discoveries. PNAS 112:4859–4864
Brengman LA, Fedo CM, Whitehouse MJ (2016) Micro-scale silicon isotope heterogeneity observed in hydrothermal
quartz precipitates from the >3.7 Ga Isua Greenstone Belt, SWGreenland. Terra Nova 28:70–75
Broecker WS, Broecker WS (1985) How to Build a Habitable Planet. Eldigio Press Palisades, New York
Buick R, Dunlop JSR (1990) Evaporitic sediments of early Archaean age from the Warrawoona Group, North Pole,
Western Australia. Sedimentology 37:247–277
Busigny V, Marin-Carbonne J, Muller E, Cartigny P, Rollion-Bard C, Assayag N, Philippot P (2017) Iron, sulfur
isotope constraints on redox conditions associated with the 3.2 Ga barite deposits of the Mapepe Formation
(Barberton Greenstone Belt, South Africa). Geochim Cosmochim Acta 210:247–266
Cammack JN, Spicuzza M, Cavosie A, Van Kranendonk M, Hickman A, Kozdon R, Orland I, Kitajima K, Valley J
(2018) SIMS microanalysis of the Strelley Pool Formation cherts and the implications for the secular-temporal
oxygen-isotope trend of cherts. Precambrian Res 304:125–139
Canfield DE (1998) A new model for Proterozoic ocean chemistry. Nature 396:450–453
13:270–274
Carrigan WJ, Cameron EM (1991) Petrological and stable isotope studies of carbonate and sulfide minerals from
the Gunflint Formation, Ontario: evidence for the origin of early Proterozoic iron-formation. Precambrian Res
52:347–380
Cartwright I, Valley JW (1992) Oxygen-isotope geochemistry of the Scourian complex, northwest Scotland. J Geol
Soc 149:115–125
Chakrabarti R, Knoll AH, Jacobsen SB, Fischer WW (2012) Si isotope variability in Proterozoic cherts. Geochim
Clayton R, Friedman I, Graf D, Mayeda T, Meents W, Shimp N (1966) The origin of saline formation waters: 1.
Isotopic composition. J Geophys Res 71:3869–3882
Cloud PE, Gruner JW, Hagen H (1965) Carbonaceous rocks of the Soudan Iron Formation (Early Precambrian).
Science 148:1713–1716
Cummins RC, Finnegan S, Fike DA, Eiler JM, Fischer WW (2014) Carbonate clumped isotope constraints on Silurian
ocean temperature and seawater δ18O. Geochim Cosmochim Acta 140:241–258
Cunningham LC, Page FZ, Simonson BM, Kozdon R, Valley JW (2012) Ion microprobe analyses of δ18O in early
quartz cements from 1.9 Ga granular iron formations (GIFs): A pilot study. Precambrian Res 214:258–268
Dauphas N, Van Zuilen M, Wadhwa M, Davis AM, Marty B, Janney PE (2004) Clues from Fe isotope variations on
the origin of early Archean BIFs from Greenland. Science 306:2077–2080
de Ronde CE, de Wit MJ (1994) Tectonic history of the Barberton greenstone belt, South Africa: 490 million years of
Archean crustal evolution. Tectonics 13:983–1005
de Ronde CE, de Wit MJ, Spooner ET (1994) Early Archean (> 3.2 Ga) Fe-oxide-rich, hydrothermal discharge vents
in the Barberton greenstone belt, South Africa. Geol Soc Am Bull 106:86–104
De Ronde CEJ, Channer DMD, Faure K, Bray CJ, Spooner ETC (1997) Fluid chemistry of Archean seafloor
hydrothermal vents: Implications for the composition of circa 3.2 Ga seawater. Geochim Cosmochim Acta
61:4025–4042
de Vries ST, Nijman W, de Boer PL (2010) Sediment Geol of the Palaeoarchaean Buck Ridge (South Africa), Kittys
Gap (Western Australia) volcano-sedimentary complexes. Precambrian Res 183:749–769
de Wit MJ, Hart R, Martin A, Abbott P (1982) Archean abiogenic and probable biogenic structures associated with
mineralized hydrothermal vent systems and regional metasomatism, with implications for greenstone belt
studies. Econ Geol 77:1783–1802
Degens ET, Epstein S (1964) Oxygen and carbon isotope ratios in coexisting calcites and dolomites from recent and
ancient sediments. Geochim Cosmochim Acta 28:23–44
Delarue F, Rouzaud J-N, Derenne S, Bourbin M, Westall F, Kremer B, Sugitani K, Deldicque D, Robert F (2016) The
Raman-derived carbonization continuum: A tool to select the best preserved molecular structures in Archean
kerogens. Astrobiology 16:407–417
Dodd JP, Sharp ZD, Fawcett PJ, Brearley AJ, McCubbin FM (2012) Rapid post-mortem maturation of diatom silica
oxygen isotope values. Geochem Geophys Geosyst 13:Q09014
Dubessy J, Caumon M-C, Rull F (2012) Raman Spectroscopy Applied to Earth Sciences and Cultural Heritage. The
Mineralogical Society of Great Britain and Ireland
Eiler JM (2001) Oxygen isotope variations of basaltic lavas and upper mantle rocks. Rev Mineral Geochem 43:319–364
Eiler JM (2011) Paleoclimate reconstruction using carbonate clumped isotope thermometry. Quat Sci Rev 30:3575–3588
Eiler JM, Farley KA, Valley JW, Hauri E, Craig H, Hart SR, Stolper EM (1997) Oxygen isotope variations in ocean
island basalt phenocrysts. Geochim Cosmochim Acta 61:2281–2293
Epstein S, Buchsbaum R, Lowenstam H, Urey HC (1951) Carbonate–water isotopic temperature scale. Geol Soc Am
Bull 62:417–426
Farber K, Dziggel A, Meyer FM, Prochaska W, Hofmann A, Harris C (2015) Fluid inclusion analysis of silicified
Palaeoarchaean oceanic crust–A record of Archaean seawater? Precambrian Res 266:150–164
Flament N, Coltice N, Rey PF (2013) The evolution of the 87Sr/86Sr of marine carbonates does not constrain continental
growth. Precambrian Res 229:177–188
Flannery DT, Allwood AC, Summons RE, Williford KH, Abbey W, Matys ED, Ferralis N (2018) Spatially-resolved
isotopic study of carbon trapped in 3.43 Ga Strelley Pool Formation stromatolites. Geochim Cosmochim Acta
223:21–35
Floran RJ, Papike JJ (1978) Mineralogy and petrology of the Gunflint Iron Formation, Minnesota–Ontario: Correlation
of compositional and assemblage variations at low to moderate grade. J Petrol 19:215–288
Folk RL, Weaver CE (1952) A study of the texture and composition of chert. Am J Sci 250:498–510
Fralick P, Davis DW, Kissin SA (2002) The age of the Gunflint Formation, Ontario, Canada: single zircon U–Pb age
determinations from reworked volcanic ash. Can J Earth Sci 39:1085–1091
Furnes H (2004) Early Life Recorded in Archean Pillow Lavas. Science 304:578–581
Furnes H, de Wit M, Staudigel H, Rosing M, Muehlenbachs K (2007) A vestige of earth’s oldest ophiolite. Science
315:1704–1707
Galili N, Shemesh A, Yam R, Brailovsky I, Sela-Adler M, Schuster EM, Collom C, Bekker A, Planavsky N,
Macdonald FA, Préat A (2019) The geologic history of seawater oxygen isotopes from marine iron oxides.
Science 365:469–473
Geilert S, Vroon PZ, Roerdink DL, Van Cappellen P, van Bergen MJ (2014) Silicon isotope fractionation during abiotic
silica precipitation at low temperatures: Inferences from flow-through experiments. Geochim Cosmochim Acta
142:95–114
360 Zakharov et al.
Geilert S, Vroon PZ, Keller NS, Gudbrandsson S, Stefánsson A, van Bergen MJ (2015) Silicon isotope fractionation
during silica precipitation from hot-spring waters: evidence from the Geysir geothermal field, Iceland. Geochim
Goodwin AM (1956) Facies relations in the Gunflint iron formation [Ontario]. Econ Geol 51:565–595
Grossman EL (2012) Applying oxygen isotope paleothermometry in deep time. Paleontolog Soc Pap 18:39–68
Gurenko AA, Chaussidon M (2002) Oxygen isotope variations in primitive tholeiites of Iceland: evidence from a
SIMS study of glass inclusions, olivine phenocrysts and pillow rim glasses. Earth Planet Sci Lett 205:63–79
Hardardottir V (2011) Metal-rich Scales in the Reykjanes Geothermal System, SW Iceland: sulfide minerals in a
seawater-dominated hydrothermal environment. PhD thesis, University of Ottawa, Canada
Harris AC, White NC, McPhie J, Bull SW, Line MA, Skrzeczynski R, Mernagh TP, Tosdal RM (2009) Early
Archean hot springs above epithermal veins, North Pole, Western Australia: New insights from fluid inclusion
microanalysis. Econ Geol 104:793–814
Hattori I, Umeda M, Nakagawea T, Yamamoto H (1996) From chalcedonic chert to quartz chert; diagenesis of chert
hosted in a Miocene volcanic-sedimentary succession, central Japan. J Sediment Res 66:163–174
Hayles JA, Yeung LY, Homann M, Banerjee A, Jiang H, Shen B, Lee C-TA (2019) Three billion year secular
evolution of the triple oxygen isotope composition of marine chert. preprint, https://eartharxiv.org/n2p5q/
download?format=pdf
Heck PR, Huberty JM, Kita NT, Ushikubo T, Kozdon R, Valley JW (2011) SIMS analyses of silicon and oxygen
isotope ratios for quartz from Archean and Paleoproterozoic banded iron formations. Geochim Cosmochim
Acta 75:5879–5891
Hervig RL, Williams P, Thomas RM, Schauer SN, Steele IM (1992) Microanalysis of oxygen isotopes in insulators
by secondary ion mass spectrometry. Int J Mass Spectrom Ion Processes 120:45–63
snowball Earth from Δ 17 O systematics of hydrothermal rocks. PNAS 112:5337–5341
Hesse R (1988) Diagenesis# 13. Origin of chert: diagenesis of biogenic siliceous sediments. Geosci Canada 15 171–192
Hesse R (1989) Silica diagenesis: origin of inorganic and replacement cherts. Earth Sci Rev 26:253–284
Hesse R, Frape SK, Egeberg PK, Matsumoto R (2000) 12. Stable isotope studies (Cl, O, and H) of interstitial waters
from site 997, Blake Ridge gas hydrate field, West Atlantic 1. Proc Ocean Drill Program Sci Results 164:129–137
Hickman AH (2008) Regional review of the 3426–3350 Ma Strelley Pool Formation, Pilbara Craton, Western
Australia. West Australia Geol Surv Rec 2008:15
Hitchon B, Friedman I (1969) Geochemistry and origin of formation waters in the western Canada sedimentary
basin—I. Stable isotopes of hydrogen and oxygen. Geochim Cosmochim Acta 33:1321–1349
Hoffman PF, Kaufman AJ, Halverson GP, Schrag DP (1998) A Neoproterozoic snowball Earth. Science 281:1342–1346
Hofmann A, Bolhar R (2007) Carbonaceous cherts in the Barberton Greenstone Belt and their significance for the
study of early life in the Archean record. Astrobiology 7:355–388
Hofmann A, Harris C (2008) Silica alteration zones in the Barberton greenstone belt: A window into subseafloor
processes 3.5–3.3 Ga ago. Chem Geol 257:221–239
Holland HD (1984) The Chemical Evolution of the Atmosphere and Oceans. Princeton University Press
House CH, Schopf JW, McKeegan KD, Coath CD, Harrison TM, Stetter KO (2000) Carbon isotopic composition of
individual Precambrian microfossils. Geology 28:707–710
Hren M, Tice M, Chamberlain C (2009) Oxygen and hydrogen isotope evidence for a temperate climate 3.42 billion
years ago. Nature 462:205–208
Hudak MR, Bindeman IN (2020) Solubility, diffusivity, and O isotope systematics of H2O in rhyolitic glass in
hydrothermal temperature experiments. Geochim Cosmochim Acta 283:222–242
Hyslop EV, Valley JW, Johnson CM, Beard BL (2008) The effects of metamorphism on O and Fe isotope compositions
in the Biwabik Iron Formation, northern Minnesota. Contrib Mineral Petrol 155:313–328
Imbrie J, Hays JD, Martinson DG, McIntyre A, Mix AC, Morley JJ, Pisias NG, Prell WL, Shackleton NJ (1984) The
orbital theory of Pleistocene climate: support from a revised chronology of the marine d18O record
controversy and an improved geological water cycle model for the past 3.4 billion years. Earth Sci Rev 83:83–122
Javaux EJ, Lepot K (2018) The Paleoproterozoic fossil record: implications for the evolution of the biosphere during
Earth’s middle-age. Earth Sci Rev 176:68–86
Jones DL, Knauth PL (1979) Oxygen isotopic and petrographic evidence relevant to the origin of the Arkansas
Novaculite. J Sediment Res 49:581–597
Kanzaki Y (2020) Interpretation of oxygen isotopes in Phanerozoic ophiolites and sedimentary rocks. Geochem
Geophys Geosyst 21:e2020GC009000
Karhu J, Epstein S (1986) The implication of the oxygen isotope records in coexisting cherts and phosphates. Geochim
Kasting JF (1993) Earth’s early atmosphere. Science 259:920–926
Kasting JF, Howard MT, Wallmann K, Veizer J, Shields G, Jaffrés J (2006) Paleoclimates, ocean depth, and the
oxygen isotopic composition of seawater. Earth Planet Sci Lett 252:82–93
Kelly JL, Fu B, Kita NT, Valley JW (2007) Optically continuous silcrete quartz cements of the St. Peter Sandstone:
High precision oxygen isotope analysis by ion microprobe. Geochim Cosmochim Acta 71:3812–3832
Kirschvink JL, Gaidos EJ, Bertani LE, Beukes NJ, Gutzmer J, Maepa LN, Steinberger RE (2000) Paleoproterozoic
snowball Earth: Extreme climatic and geochemical global change and its biological consequences. PNAS
97:1400–1405
Kita NT, Ushikubo T, Fu B, Valley JW (2009) High precision SIMS oxygen isotope analysis and the effect of sample
topography. Chem Geol 264:43–57
Kleine BI, Stefánsson A, Halldórsson SA, Whitehouse MJ, Jónasson K (2018) Silicon and oxygen isotopes unravel
quartz formation processes in the Icelandic crust. Geochem Perspect Lett 7:5–11
Knauth P (2005) Temperature and salinity history of the Precambrian ocean: implications for the course of microbial
evolution. In: Geobiology: Objectives, Concepts, Perspectives, N Noffke (ed), Elsevier, p 53–69
Knauth PL (1979) A model for the origin of chert in limestone. Geology 7:274–277
Knauth PL (1994) Petrogenesis of chert. Rev Mineral 29:233–256
Knauth PL (2005) Temperature and salinity history of the Precambrian ocean: implications for the course of microbial
evolution. Palaeogeogr Palaeoclimatol Palaeoecol 219:53–69
Knauth PL, Epstein S (1975) Hydrogen and oxygen isotope ratios in silica from the JOIDES Deep Sea Drilling
Project. Earth Planet Sci Lett 25:1–10
Knauth PL, Epstein S (1976) Hydrogen and oxygen isotope ratios in nodular and bedded cherts. Geochim Cosmochim
Acta 40:1095–1108
Knauth PL, Lowe DR (1978) Oxygen isotope geochemistry of cherts from the Onverwacht Group (3.4 billion years),
Transvaal, South Africa, with implications for secular variations in the isotopic composition of cherts. Earth
Planet Sci Lett 41:209–222
Knauth PL, Lowe DR (2003) High Archean climatic temperature inferred from oxygen isotope geochemistry of
cherts in the 3.5 Ga Swaziland Supergroup, South Africa. Geol Soc Am Bull 115:566–580
Knoll AH (1992) The early evolution of eukaryotes: a geological perspective. Science 256:622–627
Kolodny Y, Epstein S (1976) Stable isotope geochemistry of deep sea cherts. Geochim Cosmochim Acta 40:1195–1209
Konhauser KO, Phoenix VR, Bottrell SH, Adams DG, Head IM (2001) Microbial–silica interactions in Icelandic
hot spring sinter: possible analogues for some Precambrian siliceous stromatolites. Sedimentology 48:415–433
Konhauser KO, Planavsky NJ, Hardisty DS, Robbins LJ, Warchola TJ, Haugaard R, Lalonde SV, Partin CA, Oonk
PB, Tsikos H, Lyons TW (2017) Iron formations: A global record of Neoarchaean to Palaeoproterozoic
environmental history. Earth Sci Rev 172:140–177
Kouketsu Y, Mizukami T, Mori H, Endo S, Aoya M, Hara H, Nakamura D, Wallis S (2014) A new approach to develop the
Raman carbonaceous material geothermometer for low-grade metamorphism using peak width. Island Arc 23:33–50
Kozdon R, Kelly DC, Valley JW (2018) Diagenetic attenuation of carbon isotope excursion recorded by planktic
foraminifers during the Paleocene–Eocene thermal maximum. Paleoceanogr Paleoclimatol 33:367–380
Krot AN, Nagashima K, Ciesla FJ, Meyer BS, Hutcheon ID, Davis AM, Huss GR, Scott ER (2010) Oxygen isotopic
composition of the Sun and mean oxygen isotopic composition of the protosolar silicate dust: Evidence from
refractory inclusions. Astrophys J 713:1159
Kruener A, Byerly GR, Lowe DR (1991) Chronology of early Archaean granite–greenstone evolution in the
Barberton Mountain Land, South Africa, based on precise dating by single zircon evaporation. Earth Planet Sci
Lett 103:41–54
Kyser K, Kerrich R (1991) Retrograde exchange of hydrogen isotopes between hydrous minerals and water at low
temperatures. In: Stable Isotope Geochemistry: a Tribute to Samuel Epstein, HP Taylor, JR O’Neil; IR Kaplan
(eds), Geochem Soc Spec Pub 3, p 409–422
Lahfid A, Beyssac O, Deville E, Negro F, Chopin C, Goffé B (2010) Evolution of the Raman spectrum of carbonaceous
material in low-grade metasediments of the Glarus Alps (Switzerland). Terra Nova 22:354–360
Lanier WP, Lowe DR (1982) Sedimentology of the Middle Marker (3.4 Ga), Onverwacht Group, Transvaal, South
Africa. Precambrian Res 18:237–260
Leng MJ, Marshall JD (2004) Palaeoclimate interpretation of stable isotope data from lake sediment archives. Quat
Sci Rev 23:811–831
Lepot K, Addad A, Knoll AH, Wang J, Troadec D, Béché A, Javaux EJ (2017) Iron minerals within specific microfossil
morphospecies of the 1.88 Ga Gunflint Formation. Nat Commun 8:14890
362 Zakharov et al.
Lisiecki LE, Raymo ME (2005) A Pliocene–Pleistocene stack of 57 globally distributed benthic δ18O records.
Paleoceanogr 20:PA1003
Lowe DR (1983) Restricted shallow-water sedimentation of Early Archean stromatolitic and evaporitic strata of the
Strelley Pool Chert, Pilbara Block, Western Australia. Precambrian Res 19:239–283
Lowe DR (1994) Accretionary history of the Archean Barberton greenstone belt (3.55−3.22 Ga), southern Africa.
Geology 22:1099–1102
Lowe DR, Byerly GR (1999) Stratigraphy of the west-central part of the Barberton Greenstone Belt, South Africa.
Geol Soc Am Spec Pap 329:1–36
Lowe DR, Drabon N, Byerly GR (2019) Crustal fracturing, unconformities, and barite deposition, 3.26–3.23 Ga,
Barberton Greenstone Belt, South Africa. Precambrian Res 327:34–46
Maliva RG, Siever R (1988) Diagenetic replacement controlled by force of crystallization. Geology 16:688–691
Maliva RG, Knoll AH, Siever R (1989) Secular change in chert distribution: a reflection of evolving biological
participation in the silica cycle. Palaios 4:519–532
Maliva RG, Knoll AH, Simonson BM (2005) Secular change in the Precambrian silica cycle: insights from chert
petrology. Geol Soc Am Bull 117:835–845
Marin J (2009) Composition isotopique de l’oxygène et du silicium des cherts Précambriens: implications Paléo-
environnementales. PhD Thesis, Institut National Polytechnique de Lorraine
Marin J, Chaussidon M, Robert F (2010) Microscale oxygen isotope variations in 1.9 Ga Gunflint cherts: assessments
of diagenesis effects and implications for oceanic paleotemperature reconstructions. Geochim Cosmochim Acta
74:116–130
Marin-Carbonne J, Chaussidon M, Boiron M-C, Robert F (2011) A combined in situ oxygen, silicon isotopic, fluid
inclusion study of a chert sample from Onverwacht Group (3.35 Ga, South Africa): new constraints on fluid
circulation. Chem Geol 286:59–71
Marin-Carbonne J, Chaussidon M, Robert F (2012) Micrometer-scale chemical, isotopic criteria (O and Si) on
Marin-Carbonne J, Faure F, Chaussidon M, Jacob D, Robert F (2013) A petrographic and isotopic criterion of the
state of preservation of Precambrian cherts based on the characterization of the quartz veins. Precambrian Res
231:290–300
Marin-Carbonne J, Busigny V, Miot J, Rollion-Bard C, Muller E, Drabon N, Jacob D, Pont S, Robyr M, Bontognali
TR (2020) In Situ Fe, S isotope analyses in pyrite from the 3.2 Ga Mendon Formation (Barberton Greenstone
Belt, South Africa): Evidence for early microbial iron reduction. Geobiology 18:306–325
Matheney RK, Knauth PL (1993) New isotopic temperature estimates for early silica diagenesis in bedded cherts.
Geology 21:519–522
McKeegan KD (1987) Oxygen isotopes in refractory stratospheric dust particles: proof of extraterrestrial origin.
Science 237:1468–1471
Miller KG, Kominz MA, Browning JV, Wright JD, Mountain GS, Katz ME, Sugarman PJ, Cramer BS, Christie-Blick
N, Pekar SF (2005) The Phanerozoic record of global sea-level change. Science 310:1293–1298
145:263–273
Muehlenbachs K, Clayton RN (1972) Oxygen isotope studies of fresh and weathered submarine basalts. Can J Earth
Sci 9:172–184
Murata KJ, Friedman I, Gleason JD (1977) Oxygen isotope relations between diagenetic silica minerals in Monterey
Shale, Temblor Range, California. Am J Sci 277:259–272
Nijman W, Valkering ME (1998) Growth fault control of Early Archaean cherts, barite mounds and chert–barite veins,
North Pole Dome, Eastern Pilbara, Western Australia. Precambrian Res 88:25–52
Okamoto G, Okura T, Goto K (1957) Properties of silica in water. Geochim Cosmochim Acta 12:123–132
Ólafsson J, Riley JP (1978) Geochemical studies on the thermal brine from Reykjanes (Iceland). Chem Geol 21:219–237
Pack A, Herwartz D (2014) The triple oxygen isotope composition of the Earth mantle and understanding ΔO17
Paris I, Stanistreet IG, Hughes MJ (1985) Cherts of the Barberton greenstone belt interpreted as products of submarine
exhalative activity. J Geol 93:111–129
Pavlov AA, Kasting JF, Brown LL, Rages KA, Freedman R (2000) Greenhouse warming by CH4 in the atmosphere
of early Earth. J Geophys Res: Planets 105:11981–11990
Peck WH, Valley JW, Wilde SA, Graham CM (2001) Oxygen isotope ratios and rare earth elements in 3.3 to 4.4 Ga
zircons: Ion microprobe evidence for high δ18O continental crust and oceans in the Early Archean. Geochim
Perry EC, Jr, Lefticariu L (2003) Formation and geochemistry of Precambrian cherts. Treatise Geochem 7:407
Perry EC, Jr, Tan F (1972) Significance of oxygen and carbon isotope variations in early Precambrian cherts and
carbonate rocks of southern Africa. Geol Soc Am Bull 83:647–664
Peters ST, Szilas K, Sengupta S, Kirkland CL, Garbe-Schönberg D, Pack A (2020a) > 2.7 Ga metamorphic peridotites
from southeast Greenland record the oxygen isotope composition of Archean seawater. Earth Planet Sci Lett
544:116331
Peters ST, Alibabaie N, Pack A, McKibbin SJ, Raeisi D, Nayebi N, Torab F, Ireland T, Lehmann B (2020b) Triple
oxygen isotope variations in magnetite from iron-oxide deposits, central Iran, record magmatic fluid interaction
Pollington AD, Kozdon R, Valley JW (2011) Evolution of quartz cementation during burial of the Cambrian Mount
Simon Sandstone, Illinois Basin: In situ microanalysis of δ18O. Geology 39:1119–1122
Pope EC, Bird DK, Arnórsson S, Fridriksson T, Elders WA, Fridleifsson GÓ (2009) Isotopic constraints on ice age
fluids in active geothermal systems: Reykjanes, Iceland. Geochim Cosmochim Acta 73:4468–4488
Pope EC, Bird DK, Rosing MT (2012) Isotope composition and volume of Earth’s early oceans. PNAS 109:4371–4376
Poulton SW, Canfield DE (2011) Ferruginous conditions: a dominant feature of the ocean through Earth’s history.
Elements 7:107–112
Prokoph A, Shields GA, Veizer J (2008) Compilation and time-series analysis of a marine carbonate δ18O, δ13C,
87
Sr/86Sr and δ34S database through Earth history. Earth Sci Rev 87:113–133
Qi H, Coplen TB, Gehre M, Vennemann TW, Brand WA, Geilmann H, Olack G, Bindeman IN, Palandri J, Huang L,
Longstaffe FJ (2017) New biotite and muscovite isotopic reference materials, USGS57 and USGS58, for δ2H
measurements–A replacement for NBS 30. Chem Geol 467:89–99
Rahl JM, Anderson KM, Brandon MT, Fassoulas C (2005) Raman spectroscopic carbonaceous material thermometry
of low-grade metamorphic rocks: Calibration and application to tectonic exhumation in Crete, Greece. Earth
Raymo ME, Ruddiman WF (1992) Tectonic forcing of late Cenozoic climate. Nature 359:117–122
Raymo ME, Kozdon R, Evans D, Lisiecki L, Ford HL (2018) The accuracy of mid-Pliocene δ18O-based ice volume
and sea level reconstructions. Earth Sci Rev 177:291–302
Robert F, Chaussidon M (2006) A palaeotemperature curve for the Precambrian oceans based on silicon isotopes in
cherts. Nature 443:969–972
Rollion-Bard C, Blamart D, Cuif J-P, Juillet-Leclerc A (2003) Microanalysis of C and O isotopes of azooxanthellate
and zooxanthellate corals by ion microprobe. Coral Reefs 22:405–415
Royer DL, Berner RA, Montañez IP, Tabor NJ, Beerling DJ (2004) CO2 as a primary driver of phanerozoic climate.
GSA Today 14:4–10
Schad M, Halama M, Bishop B, Konhauser KO, Kappler A (2019) Temperature fluctuations in the Archean ocean
as trigger for varve-like deposition of iron and silica minerals in banded iron formations. Geochim Cosmochim
Acta 265:386–412
Schopf JW, Kudryavtsev AB (2005) Three-dimensional Raman imagery of precambrian microscopic organisms.
Geobiology 3:1–12
Schopf JW, Kudryavtsev AB, Agresti DG, Wdowiak TJ, Czaja AD (2002) Laser–Raman imagery of Earth’s earliest
fossils. Nature 416:73–76
Seitz S, Baumgartner LP, Bouvier A-S, Putlitz B, Vennemann T (2017) Quartz reference materials for oxygen isotope
analysis by SIMS. Geostand Geoanal Res 41:69–75
Sexton PF, Wilson PA, Pearson PN (2006) Microstructural and geochemical perspectives on planktic foraminiferal
preservation:“Glassy” versus “Frosty.” Geochem Geophys Geosyst 7:Q12P19
Shanks WC (2001) Stable isotopes in seafloor hydrothermal systems: vent fluids, hydrothermal deposits, hydrothermal
alteration, and microbial processes. Rev Mineral Geochem 43:469–525
Shanks WC, Seyfried WE (1987) Stable isotope studies of vent fluids and chimney minerals, southern Juan de Fuca
Ridge: Sodium metasomatism and seawater sulfate reduction. J Geophys Res: Solid Earth 92:11387–11399
Sharp ZD, Wostbrock JAG (2021) Standardization for the triple oxygen isotope system: waters, silicates, carbonates,
air, and sulfates Rev Mineral Geochem 86:137–178
364 Zakharov et al.
Sharp ZD, Atudorei V, Durakiewicz T (2001) A rapid method for determination of hydrogen and oxygen isotope ratios
from water and hydrous minerals. Chem Geol 178:197–210
Sharp ZD, Durakiewicz T, Migaszewski ZM, Atudorei VN (2002) Antiphase hydrogen and oxygen isotope periodicity
in chert nodules. Geochim Cosmochim Acta 66:2865–2873
Sharp ZD, Gibbons JA, Maltsev O, Atudorei V, Pack A, Sengupta S, Shock EL, Knauth PL (2016) A calibration of
Sheppard SMF, Gilg HA (1996) Stable isotope geochemistry of clay minerals: “The story of sloppy, sticky, lumpy,
tough” Cairns-Smith (1971). Clay Minerals 31:1–24
Shields G, Veizer J (2002) Precambrian marine carbonate isotope database: Version 1.1: Geochem Geophys Geosyst
3:1, doi 10.1029/2001GC000266
Siever R (1992) The silica cycle in the Precambrian. Geochim Cosmochim Acta 56:3265–3272
Simonson BM (1987) Early silica cementation and subsequent diagenesis in arenites from four early Proterozoic iron
formations of North America. J Sediment Res 57:494–511
Stefurak EJ, Lowe DR, Zentner D, Fischer WW (2014) Primary silica granules—A new mode of Paleoarchean
sedimentation. Geology 42:283–286
Stefurak EJ, Fischer WW, Lowe DR (2015) Texture-specific Si isotope variations in Barberton Greenstone Belt cherts
record low temperature fractionations in early Archean seawater. Geochim Cosmochim Acta 150:26–52
Strother PK, Tobin K (1987) Observations on the genus huroniospora barghoorn: Implications for paleoecology of the
gunflint microbiota. Precambrian Res 36:323–333
Sugitani K, Mimura K, Nagaoka T, Lepot K, Takeuchi M (2013) Microfossil assemblage from the 3400 Ma Strelley
Pool Formation in the Pilbara Craton, Western Australia: results form a new locality. Precambrian Res 226:59–74
Swann GE, Leng MJ (2009) A review of diatom δ18O in palaeoceanography. Quat Sci Rev 28:384–398
Tartèse R, Chaussidon M, Gurenko A, Delarue F, Robert F (2018) Insights into the origin of carbonaceous chondrite
organics from their triple oxygen isotope composition. PNAS 115:8535–8540
Tatzel M, von Blanckenburg F, Oelze M, Schuessler JA, Bohrmann G (2015) The silicon isotope record of early silica
diagenesis. Earth Planet Sci Lett 428:293–303
Tian C, Wang L, Tian F, Zhao S, Jiao W (2019) Spatial and temporal variations of tap water 17O-excess in China.
Tice MM, Lowe DR (2006) The origin of carbonaceous matter in pre−3.0 Ga greenstone terrains: A review and new
evidence from the 3.42 Ga Buck Reef Chert. Earth Sci Rev 76:259–300
Tice MM, Bostick BC, Lowe DR (2004) Thermal history of the 3.5–3.2 Ga Onverwacht and Fig Tree Groups, Barberton
greenstone belt, South Africa, inferred by Raman microspectroscopy of carbonaceous material. Geology 32:37–40
Trower EJ, Lowe DR (2016) Sedimentology of the 3.3 Ga upper Mendon Formation, Barberton Greenstone Belt,
South Africa. Precambrian Res 281:473–494
Turchyn AV, Alt JC, Brown ST, DePaolo DJ, Coggon RM, Chi G, Bédard JH, Skulski T (2013) Reconstructing the
oxygen isotope composition of late Cambrian and Cretaceous hydrothermal vent fluid. Geochim Cosmochim
Acta 123:440–458
Tyler SA, Barghoorn ES (1954) Occurrence of Structurally Preserved Plants in Pre-Cambrian Rocks of the Canadian
Shield. Science 119:606–608
Ueno Y, Yurimoto H, Yoshioka H, Komiya T, Maruyama S (2002) Ion microprobe analysis of graphite from ca. 3.8
Ga metasediments, Isua supracrustal belt, West Greenland: relationship between metamorphism and carbon
isotopic composition. Geochim Cosmochim Acta 66:1257–1268
Ueno Y, Yoshioka H, Maruyama S, Isozaki Y (2004) Carbon isotopes and petrography of kerogens in ∼3.5-Ga
hydrothermal silica dikes in the North Pole area, Western Australia1 1Associate editor: GCody. Geochim
Urey HC, Lowenstam HA, Epstein S, McKinney CR (1951) Measurement of paleotemperatures and temperatures of
the Upper Cretaceous of England, Denmark, and the southeastern United States. Geol Soc Am Bull 62:399–416
Valley JW, Graham CM (1996) Ion microprobe analysis of oxygen isotope ratios in quartz from Skye granite: healed
micro-cracks, fluid flow, and hydrothermal exchange. Contrib Mineral Petrol 124:225–234
Van den Boorn S, van Bergen MJ, Nijman W, Vroon PZ (2007) Dual role of seawater and hydrothermal fluids in Early
Archean chert formation: evidence from silicon isotopes. Geology 35:939–942
Van den Boorn S, Van Bergen M, Vroon P, De Vries S, Nijman W (2010) Silicon isotope and trace element constraints
on the origin of 3.5 Ga cherts: implications for Early Archaean marine environments. Geochim Cosmochim
Acta 74:1077–1103
Van Kranendonk MJ (2006) Volcanic degassing, hydrothermal circulation and the flourishing of early life on Earth:
A review of the evidence from c. 3490−3240 Ma rocks of the Pilbara Supergroup, Pilbara Craton, Western
Australia. Earth Sci Rev 74:197–240
Van Kranendonk MJ, Philippot P, Lepot K, Bodorkos S, Pirajno F (2008) Geological setting of Earth’s oldest fossils in
the ca. 3.5 Ga Dresser Formation, Pilbara Craton, Western Australia. Precambrian Res 167:93–124
van Zuilen MA, Lepland A, Teranes J, Finarelli J, Wahlen M, Arrhenius G (2003) Graphite and carbonates in the 3.8
Ga old Isua supracrustal belt, southern West Greenland. Precambrian Res 126:331–348
Veizer J, Prokoph A (2015) Temperatures and oxygen isotopic composition of Phanerozoic oceans. Earth Sci Rev 146:92–104
Veizer J, Ala D, Azmy K, Bruckschen P, Buhl D, Bruhn F, Carden GA, Diener A, Ebneth S, Godderis Y, Jasper T
(1999) 87Sr/86Sr, δ13C and δ18O evolution of Phanerozoic seawater. Chem Geol 161:59–88
Wacey D (2010) Stromatolites in the 3400 Ma Strelley Pool Formation, Western Australia: examining biogenicity
from the macro-to the nano-scale. Astrobiology 10:381–395
Wacey D, Menon S, Green L, Gerstmann D, Kong C, Mcloughlin N, Saunders M, Brasier M (2012) Taphonomy of
very ancient microfossils from the 3400 Ma Strelley Pool Formation and 1900 Ma Gunflint Formation: New
insights using a focused ion beam. Precambrian Res 220:234–250
Walker JC, Lohmann KC (1989) Why the oxygen isotopic composition of sea water changes with time. Geophys Res
Lett 16:323–326
Watkins JM, Hunt JD (2015) A process-based model for non-equilibrium clumped isotope effects in carbonates. Earth
Wenner DB, Taylor HP (1973) Oxygen and hydrogen isotope studies of the serpentinization of ultramafic rocks in
oceanic environments and continental ophiolite complexes. Am J Sci 273:207–239
Whitehouse MJ, Fedo CM (2007) Microscale heterogeneity of Fe isotopes in > 3.71 Ga banded iron formation from
the Isua Greenstone Belt, southwest Greenland. Geology 35:719–722
Whitehouse MJ, Nemchin AA (2009) High precision, high accuracy measurement of oxygen isotopes in a large lunar
zircon by SIMSChem Geol 261:32–42
Wilkinson JJ, Jenkin GRT, Fallick AE, Foster RP (1995) Oxygen and hydrogen isotopic evolution of Variscan crustal
fluids, south Cornwall, UK. Chem Geol 123:239–254
Williams LB, Hervig RL, Bjørlykke K (1997) New evidence for the origin of quartz cements in hydrocarbon reservoirs
revealed by oxygen isotope microanalyses. Geochim Cosmochim Acta 61:2529–2538
Williford KH, Ushikubo T, Schopf JW, Lepot K, Kitajima K, Valley JW (2013) Preservation and detection of
microstructural and taxonomic correlations in the carbon isotopic compositions of individual Precambrian
microfossils. Geochim Cosmochim Acta 104:165–182
Winter BL, Knauth PL (1992) Stable isotope geochemistry of cherts and carbonates from the 2.0 Ga Gunflint
Iron Formation: implications for the depositional setting, and the effects of diagenesis and metamorphism.
Wopenka B, Pasteris JD (1993) Structural characterization of kerogens to granulite-facies graphite: applicability of
Raman microprobe spectroscopy. Am Mineral 78:533–557
Geochem 86:367–400
Xie X, Byerly GR, Ferrell Jr RE (1997) IIb trioctahedral chlorite from the Barberton greenstone belt: crystal structure
and rock composition constraints with implications to geothermometry. Contrib Mineral Petrol 126:275–291
Yanchilina AG, Yam R, Kolodny Y, Shemesh A (2020) From diatom opal-A δ18O to chert δ18O in deep sea sediments.
Yui T-F, Huang E, Xu J (1996) Raman spectrum of carbonaceous material: A possible metamorphic grade indicator
for low-grade metamorphic rocks. J Metamorph Geol 14:115–124
Geology 45:667–670
Zakharov DO, Bindeman IN, Serebryakov NS, Prave AR, Azimov PY, Babarina II (2019a) Low δ18O rocks in the
Zakharov DO, Bindeman IN, Tanaka R, Friðleifsson GÓ, Reed MH, Hampton RL (2019b) Triple oxygen isotope
systematics as a tracer of fluids in the crust: A study from modern geothermal systems of Iceland. Chemical
Geology 530:119312
Acta 57:1079–1091
Vol. 86 pp. 367–400, 2021 11
Triple Oxygen Isotopes in Silica–Water and

Carbonate–Water Systems
Jordan A.G. Wostbrock* and Zachary D. Sharp
Department of Earth and Planetary Sciences
University of New Mexico
Albuquerque, NM 87131
USA
jogibbons7@unm.edu
zsharp@unm.edu
INTRODUCTION
The field of stable isotope geochemistry began with the recognition that the oxygen isotope
composition of ancient carbonates could be used as a paleothermometer (Urey 1947; Urey et
al. 1951). As stated by Urey (1947), “Accurate determinations of the Ol8 content of carbonate
rocks could be used to determine the temperature at which they were formed”. This concept
was based on the temperature dependence for the oxygen isotope fractionation between calcite
and water. Urey realized that if a mass spectrometer with sufficient precision could be built,
a method of reproducibly extracting oxygen from solid carbonate could be developed, and the
isotope fractionations between calcite and water could be quantified, then the oxygen isotope
composition of ancient carbonates could be used to determine the temperature of their formation.
This idea led to the carbonate–water temperature scale (McCrea 1950; Epstein et al.
1951, 1953; Urey et al. 1951). The oxygen isotope compositions of marine carbonates are
sensitive to temperature (increase with decreasing formation temperature), so that ancient
seawater temperatures could be determined by measuring the oxygen isotope composition
of the carbonate. Urey et al. (1951) recognized a number of potential problems with the
carbonate thermometer, including: 1) the possibility that organisms form out of oxygen
isotope equilibrium, a term he called the ‘vital effect’; 2) the possibility that shells might
not preserve their oxygen isotope composition over geological time and undergo some sort
of post-mortem diagenesis; and 3) the uncertainty in the estimate of the oxygen isotope
composition of the formation water. To this last point, the oxygen isotope composition of the
ancient seawater is estimated and not measured directly. What is the possibility that the ocean
has changed its oxygen isotope composition through time? Urey et al. (1951) concluded
that any significant changes would have occurred before the Cambrian, although he states
that ‘it is a conclusion based on little more than prejudice’. In spite of these caveats, the
carbonate paleothermometer has been immensely successful and applied in many thousands
of studies. It also led to the development of other mineral–water thermometers such as
silica–water thermometry. However, the concerns raised by Urey remain, despite numerous
efforts to ascertain diagenesis and quantify any changes in the ocean oxygen isotope
composition through time. Recently, the rare 17O/16O ratio has begun to be measured and
included in geologic studies. The addition of this new variable has allowed researchers to
discern equilibrium vs. disequilibrium processes governing precipitation of a mineral and
determine whether alteration occurred in a sample after its initial precipitation.
1943-2666/21/0086-0011$05.00 (online)
368 Wostbrock & Sharp
While triple oxygen isotope measurements (measuring both the 17O/16O and 18O/16O ratios)
have been performed on a variety of materials, this chapter reviews the current literature on the
triple oxygen isotope fractionation between silica–water and carbonate–water. An overview of both
equilibrium and disequilibrium fractionation is given through theory, empirical calibrations, and
laboratory experiments. We then apply these different fractionation regimes to geologic samples.
We also give a brief overview of how the silica–water and calcite–water triple oxygen isotope
fractionation curves can be used to reconstruct past seawater oxygen isotope values. We conclude
the chapter by calculating and discussing potential uses of a quartz–calcite fractionation curve.
MINERAL–WATER OXYGEN ISOTOPE THERMOMETERS

18
O/16O fractionation
Oxygen isotope paleothermometers are based on the temperature dependent fractionation
between authigenic minerals and their formation water. For marine sediments, we generally
assume that the formation water is seawater and that the oxygen isotope composition of the
seawater has not varied through time. Obviously, this is a critical assumption that has been
questioned and evaluated extensively and is discussed later in this chapter. Oxygen isotope
paleothermometers have been developed for carbonates, silicates, phosphates and sulfates.
In this communication, we address only the silica–water and carbonate–water paleotemperature
equations. We begin with an overview of the generic 18O/16O thermometer.
One of the first calibrations of the calcite–water oxygen isotope thermometer took the
form (Epstein et al. 1951, 1953)
t (°C) = 16.5 – 4.3 d + 0.14 d2 (1)
In this equation δ is the isotopic composition of the CO2 evolved from the carbonate relative
to the δ value of CO2 evolved from a reference carbonate (Peedee belemnite or PDB).
The δ18O value is defined, in per mil (‰) notation, as
 Rsa 
d18O =   1   1000 (2)
R
 std 
where R is the 18O/16O ratio of the sample (sa) and standard (std). A similar equation can be
written for the d17O value, in which case R is the 17O/16O value.
Equation (1) is simply a quadratic best fit to empirical data and is not based on any
theoretical considerations. Bigeleisen and Goeppert Mayer (1947) showed that the log of
the equilibrium fractionation between diatomic molecules should be proportional to T −1
and T −2 (in Kelvin) at low and high temperatures, respectively. The system becomes more
complicated with more complex phases, especially when water is considered (Criss 1991).
Empirical data are often fit to 1/T, 1/T 2 or a combination of these two. In mineral–water
fractionation equations, a constant is often included as a consequence of the high vibrational
frequencies of water (Bottinga and Javoy 1973). An appropriate fit to the temperature
dependence of fractionation between phase i and water is given by:
a  106 (3)
1000 ln i H2O  c
T2
R 1000  i
Here i H2O  i and in delta notation i H2O  . An alternative form for
RH2O 1000  H2O
mineral–water fractionation is to add a 1/T term to the equation (e.g., Zheng 1991):
Silica–Water and Carbonate–Water Systems 369
a  106 b  103 (4)

1000 ln i H2O   c
T2 T
The constant in Equations (3) and (4) lead to equations that are not valid at high temperatures
because 1000 ln α values approach zero, and not c, at infinite temperatures. An alternative
expression is to fit the data to a polynomial of 1/T, such as
a  106 b  103 (5)

1000 ln i H2O  
T2 T
The justification of using the form of Equation (3) is apparent when we consider actual
data. Figure 1 shows experimental and empirical fractionation data for quartz–water and
carbonate–water as a function of 1/T2. The data are approximated by a linear fit and a y intercept
(the c term in Eqn. 3) of 3.4‰ (quartz) and 2.95‰ (calcite) (Fig. 1A). Obviously, this relationship
fails at very high temperatures, where the 1000 lnα must approach zero. A better fit is expected
if we use a polynomial of 1/T and 1/T 2 as predicted from theory (Clayton and Kieffer 1991).
A polynomial fit for quartz–water and carbonate–water is shown in Figure 1B. The constant is
zero in these fits so that 1000 ln α goes to infinity at infinite temperatures. Both curves have the
same general sigmoidal form, indicating that the constant in Equation (3), which is likely valid at
lower temperatures (e.g., Bottinga and Javoy 1973) must decrease with increasing temperature.
The quartz–water system has been experimentally and empirically studied over a uniquely
wide temperature range. Quartz–water exchange experiments have been made between 250 °C
and 800 °C (see Sharp et al. 2016 for a compilation). Low temperature empirical fractionation
of diatoms and abiogenic silica have been measured over the temperature range of 0 to almost
100 °C (Kita et al. 1985; Leclerc and Labeyrie 1987; Shemesh et al. 1995; Brandriss et al.
1998; Schmidt et al. 2001; Dodd and Sharp 2010). All low- and high-temperature quartz–water
fractionation data can be fit to Equation (5), where a = 4.28 ± 0.07 and b = −3.5 ± 0.2
(R2 = 0.9978) (Sharp et al. 2016). Eliminating the c constant allows for the fractionation
equation to be extrapolated to infinite temperature. In contrast, carbonate–water experimental
and empirical data are better fit to the form of Equation (3) up to ~1000 K (O′Neil et al.
1969; Wostbrock et al. 2020b). Equation (5) does not adequately fit both the high temperature
experimental results and low temperature synthesis and empirical data. In this paper we use
the general form of Equation (4). Based only on fitting published fractionation data, we set the
c term to zero for quartz–water, and the b term to zero for carbonate–water.
)LJXUH
Figure 1. Quartz–water (green diamond) and calcite–water (blue circle) oxygen isotope fractionation.
A) Best fit using Equation (3) (a × 106/T2 + c) with a negative y-intercept. B) Best fit using Equation
(5) (a × 106/T2 + b × 103/T) and a y-intercept of zero. The quartz–water data are better approximated by
Equation (5) (Fig. 1B), whereas the carbonate–water data is better fit by Equation (3). Higher order
polynomial fits are not warranted, as the errors in the coefficients become very large.
)LJXUH
Triple oxygen isotope fractionation

The equilibrium fractionation of 17O/16O relative to 18O/16O between two phases A and B
is given by the equation
 

17O A  B  18O A  B (6)
In a linearized format, where δ′ = 1000 ln (δ/1000 + 1) (Hulston and Thode 1965; Miller 2002),
δ′A − δ′B = 1000 ln αA–B and Equation (6) can be recast as
d′17OA − d′17OB = θ(d′18OA − d′18OB) (7)
The θ value varies with temperature ranging from 0.5305 at T = ∞ to 0.526, 0.524, and
0.518 − 0.52 at T = 298 K for calcite–water (Wostbrock et al. 2020b), quartz–water (Cao and
Liu 2011; Sharp et al. 2016; Hayles et al. 2018), and quartz–calcite (Hayles et al. 2018 and this
chapter), respectively. To a first approximation, the temperature dependence of θ for phases A
and B is given by (Sharp et al. 2016)

θΑ−Β = 0.5305  (8)
T
where ε is a constant, although higher order polynomials are indicated from theory (Cao and
Liu 2011; Hayles et al. 2018; Guo and Zhou 2019). Combining Equations (4), (7) and (8) gives
the equilibrium d17O fractionation equation between phases A and B:
17 16  a  106 b  103  
1000 ln  A O/B O
 17O A  17O B   2
  c   0.5305   (9)
 T T   T 
For all terrestrial samples not affected by mass independent fractionation processes
(e.g., Thiemens 2006), δ′17O varies with δ′18O by the empirical equation known as the
Terrestrial Fractionation Line or TFL
d′17O = ld′18O + g (10)
In Equation (10), λ is an empirical best fit to natural data and not tied to any thermodynamic
relationship and γ is the y intercept that is generally assumed to be zero. Similar to other
chapters in this volume, throughout this work we use a λ value of 0.528.
Departures from the linear relationship given in Equation (10) are subtle and expressed
using the ∆′17O value, given by the equation
D′17O = d′17O – λ d′18O (11)
The equations governing the mineral–water isotope fractionation for the three oxygen isotopes
are the following:
a  106 b  103 (12)

18Omineral  18O water   c
T2 T
 a  106 b  103  
17Omineral  17O water   2
  c   0.5305   (13)
 T T   T 
and
 a  106 b  103   
17Omineral  17O water    18O water   2
  c   0.5305     (14)
 T T  T 
For minerals in equilibrium with seawater similar to Standard Mean Ocean Water (SMOW,
δ17O = δ18O = 0), Equation (14) reduces to
 a  106 b  103   
17Omineral   2
  c   0.5305     (15)
 T T  T 
Combining Equations (12) and (13) allows us to calculate the relationship between the Δ′17O
and δ′18O values of a mineral (subscript r) in equilibrium with water (subscript w). The solution
is the following:

17O r  17O w  18O r  18O w  
 
 
c  18O r  18O w   
    18O w  
(16)
 0.5305 
)LJXUH 
  
500 b  b 2  4 a c  18O r  18O w  

Equilibrium triple oxygen isotope fractionation—the Δ′17O–δ′18O plot
The benefit of the triple oxygen isotope system is apparent when we plot the Δ′17O value
of a mineral against the δ′18O value using Equation (16) (Fig. 2). The triple oxygen isotope plot
(Δ′17O vs. δ′18O) allows us to clearly distinguish equilibrium or disequilibrium precipitation for a
given sample. Figure 2 is constructed for quartz in equilibrium with ocean water, but an equally
relevant figure could be generated for calcite or any other low temperature phase for which
triple oxygen isotope fractionation data are available. The solid black curve represents the locus
of δ′18O and Δ′17O values of silica in equilibrium with seawater equal to SMOW (yellow star).
All silica samples in oxygen isotope equilibrium with SMOW must plot on the solid black curve.
Note, that seawater with a δ18O value of 0‰ and Δ′17O value of −0.005‰ can also be used to
represent seawater based on Luz and Barkan (2010). In this chapter, we clearly state whether
we are using seawater with a Δ′17O value of 0 or −0.005‰. Equilibrium temperatures (in °C)
decrease along the curve with increasing δ′18O values and decreasing Δ′17O values.
)LJXUH
Figure 2. Δ′17O–δ′18O plot for silica–water. The solid black curve shows the combined Δ′17O–δ′18O values
of silica in equilibrium with SMOW (Standard Mean Ocean Water, δ18O = Δ′17O = 0). The position along the
line is determined by the temperature of equilibration. The grey line is for a seawater with a δ′18O value of
–1‰ (ice free conditions). The dashed curves are the same as the equilibrium fractionation curve (black
solid curve) only inverted. The curve′s origin is placed at the silica triple oxygen isotope value and the dashed
lines represent the water in equilibrium with the silica. The red and green circles represent two hypothetical
samples. The dashed line emanating from the green circle passes through the seawater composition with
a 20 °C temperature. The dashed line from the red circle does not intersect seawater. Therefore, the sample
represented by the red circle cannot be in equilibrium with seawater at any temperature. It could be in equilib-
rium with a different water composition, such as the pink square.
The triple oxygen isotope values of two hypothetical samples are shown in Figure 2 by
the red and green circles. Both have δ′18O values of 37.9‰. With δ′18O values alone, the two
samples are indistinguishable and correspond to a formation temperature of 20 °C. The addition
of the Δ′17O value clearly provides more information. The green circle lies on the equilibrium
curve, suggesting that the sample is indeed in equilibrium with seawater at 20 °C. The red
circle does not plot on the seawater equilibrium curve. Therefore, this sample cannot be in
equilibrium with seawater (δ18O and Δ′17O = 0) at any temperature. Adding the 17O isotope
unambiguously demonstrates disequilibrium precipitation with modern seawater and/or
equilibrium precipitation with a different water oxygen isotope composition. Lowering the
δ18O value of the ocean by ~1‰ during interglacial periods is shown by the grey equilibrium
curve (Fig. 2). The Δ′17O value of an ice-free ocean should be close to 0, so the curve moves
to −1‰ and does not explain the position of the red circle. As described later in this chapter,
the region where the red circle plots in Figure 2 is diagnostic of diagenesis.
The quartz-seawater equilibrium curve (solid grey and black curves; Fig. 2) can be inverted
such that the origin of the curve is placed on the measured δ′18O–Δ′17O value of the sample
(dashed lines; Fig. 2). In this case, the dashed line defines the water composition in equilibrium
with that sample. For the green circle, the intersection of the dashed line with SMOW occurs
at 20 °C, the apparent equilibrium temperature. The dashed line emanating from the red circle
does not pass through SMOW, again indicating that there is no temperature at which this
sample can be in equilibrium with seawater. The red circle could be in equilibrium with a water
having a δ18O value of –5‰ and δ′17O value of 0‰ (pink square; Fig. 2) or any water with a
triple oxygen isotope value the plots on the dashed line emanating from the red circle.
KINETIC EFFECTS RESULTING IN OXYGEN ISOTOPE DISEQUILIBRIUM

Rapid precipitation of a mineral at low temperature may result in kinetic effects that lead
to disequilibrium oxygen isotope fractionations. The effect was recognized in the first synthesis
study of calcite for the purpose of determining the a18OCaCO3–H2O fractionation (McCrea 1950).
He found that the δ18O value of rapidly precipitated calcite approached that of the carbonate
ion in solution. Disequilibrium has been seen in many natural samples as well. Cave calcite
(speleothems) and travertines can precipitate out of oxygen isotope equilibrium due to rapid
CO2 degassing (Gonfiantini et al. 1968; Mickler et al. 2006; Carlson et al. 2020). Biogenic
calcite can precipitate out of oxygen isotope equilibrium due to the rapid precipitation of
calcite resulting in incomplete equilibration between the carbonate ion CO32 and water
(McConnaughey 1989), or by changing the relative concentrations of the dissolved carbonate
species by changing pH (Spero et al. 1997; Zeebe 1999). On the basis of clumped carbonate
isotope measurements (∆47), it has been suggested that most natural calcites are out of oxygen
isotope equilibrium (Daëron et al. 2019).
The problem appears to be minimal for the silica–water system. Silica–water fractionations
are mostly insensitive to biogenic vs. abiogenic conditions (Shemesh et al. 1992; Wostbrock et
al. 2018), although there is clear evidence that diatoms precipitate out of equilibrium with water
(Schmidt et al. 1997, 2001; Dodd et al. 2012). However, diatoms appear to re-equilibrate with
water within a year post-mortem (Dodd et al. 2012; Tyler et al. 2017). The general insensitivity
of silica precipitation to kinetic isotope effects is probably related to the simple chemistry of
precipitation. Silica forms from H4SiO4 with an intermediate SiOOH+ ion (Crundwell 2017).
Solubilities of quartz are extremely low at pH below 9 with H4SiO4 being the principle dissolved
species (Krauskopf 1956). Precipitation is generally slow, so that there is minimal disequilibrium
between the dissolved species and crystallizing silica.
Disequilibrium effects with dissolved inorganic carbon

Carbonates, on the other hand, have been shown on numerous occasions to precipitate out
of thermodynamic equilibrium (Gonfiantini et al. 1968; Shackleton et al. 1973; McConnaughey
1989; Kim and O′Neil 1997; Spero et al. 1997; Zeebe 1999, 2014; Mickler et al. 2004, 2006;
Beck et al. 2005; Dietzel et al. 2009; Tremaine et al. 2011; Watkins et al. 2013; Devriendt et al.
2017; Bajnai et al. 2018; Daëron et al. 2019; Carlson et al. 2020). The primary complication
that exists for carbonates is that the oxygen isotope fractionation is not simply controlled by
the water oxygen isotope composition, but rather by the fractionation between calcite and all
dissolved inorganic carbon (DIC) species: CO2(aq), H2CO3, HCO3 , and CO32 . The relevant
proportions of these DIC species are strongly controlled by pH, and there is a large oxygen
isotope fractionation between them. The 1000 ln a18Ο values (at 25 °C) range from 31.1‰ for
HCO3 –H2O to 24.2‰ for CO32–H2O (Beck et al. 2005). Combining these results with the
acalcite–water demonstrates that HCO3 has a higher δ18O value than coexisting calcite, whereas
CO32 has a much lower δ18O value than calcite. Dissolved inorganic carbon is mainly present
as HCO3 at intermediate pH, but CO32 at high pH.
When all species are in oxygen isotope equilibrium, pH has no effect on fractionation between
calcite and water. Calcite precipitation is governed by the reaction (Garrels and Christ 1965)
Ca 2+ + CO32 = CaCO3 (17)

2
and, as long as the CO ion is in equilibrium with water, and calcite precipitates in equilibrium
3
with all DIC species, then the calcite is also in equilibrium with water. The equilibrium
δ18O value of all dissolved carbonate species is determined by the δ18O value of water, which
is overwhelmingly the predominant oxygen reservoir.
Several phenomena lead to disequilibrium oxygen isotope values during calcite
precipitation. First, a transient disequilibrium between the dissolved carbonate species and
water will occur if the pH of the system is rapidly changed. This happens for example, when
CO2 is added to or lost from the system. When subterranean CO2-saturated waters reach the
surface, they rapidly degas CO2, pH increases and calcite (travertine) precipitates. Second,
kinetic effects at the surface of a growing calcite crystal may lead to non-equilibrium if the
rates of incorporation and detachment of the HCO3 and CO32 ions into the carbonate lattice
are different (Watkins et al. 2013). Because the relative proportions of DIC species are strongly
controlled by pH, the δ18O values of some shell-secreting organisms are affected by the pH of
the solution (Spero et al. 1997; Zeebe 1999; Adkins et al. 2003).
For the reaction
HCO3 ↔ H+ + CO32 (18)
2
isotopic equilibrium is reached almost instantaneously because the protonation of CO and 3
deprotonation of HCO3 are extremely rapid. Equilibration between the bicarbonate ion and
water and dissolved CO2 (or H2CO3) occurs far more slowly. Equilibrium between H2O and
DIC will occur via CO2 hydration and hydroxylation by the reactions
CO2 + H2O ↔ H2CO3 ↔ H+ + HCO3 (19)
and

CO2 + OH− ↔ HCO3 (20)
Equation (19) controls equilibration rates at moderate pH, Equation (20) becomes important
at high pH with higher OH− concentrations. Oxygen isotope equilibration between H2O
and DIC requires hours to days, depending on temperature and pH (Mills and Urey 1940;
McConnaughey 1989; Beck et al. 2005; Uchikawa and Zeebe 2012; Watkins et al. 2013)1.
1
Mills and Urey (1940) found very similar rates for DIC–water equilibration as later work.
The δ18O value of the sum of the hydrated DIC species is given by
 O( DIC )
18

18
    
OH2CO3 H 2 CO3   18OHCO3 HCO3   18OCO32 CO32   (21)
H 2 CO3   HCO3   CO32  
where the square brackets are the concentration of the dissolved species (Zeebe 1999).
Note that dissolved CO2 is not considered in our computation of the δ18OΣ(DIC) value. If all dissolved
species are in equilibrium with water, then the δ18OΣ(DIC) value will change with pH (Fig. 3).
At low pH (below about 6.5), the δ18OΣ(DIC) value will be equal to that of H2CO3 (δ18O = 39.5‰),
which makes up the vast majority of the total DIC. At very high pH (above about 9), the δ18OΣ(DIC)
value will be equal to the CO32 ion (δ18O = 18.4‰), which is the predominant DIC species.
If the relative incorporation of the DIC species into the carbonate lattice is kinetically controlled,
then the oxygen isotope composition of the precipitating calcite will be affected by pH. If all
DIC were to be incorporated into a fast-growing calcite, then the δ18O value of the calcite would
simply equal the δ18OΣ(DIC). In Figure 3, the δ18O value of the calcite would track along the
δ18OΣ(DIC) curve (thick black line). Spero et al. (1997) and Bijma et al. (1999) found a clear linear
relationship between pH and the δ18O values of foraminifera, independent of symbiont activity
and temperature. There is a well-established kinetic-based pH–δ18O relationship for carbonates
that is controlled by non-equilibrium fractionation during hydration and hydroxylation of CO2
or by rate-dependent incorporation of different DIC components.
Oxygen isotope disequilibrium can occur due to rapid CO2 degassing or ingassing, such as
production or consumption of CO2 within the cell membranes by respiration (McConnaughey
1989). Some researchers suggest that slowly precipitated calcites in caves closer represent
thermodynamic calcite–water equilibrium, represented by the higher 1000 ln acalcite–water values
than that of most biogenic calcite and laboratory precipitation experiments (Coplen 2007;
Tremaine et al. 2011; Daëron et al. 2019).
)LJXUH
Figure 3. Effect of pH on the relative abundance of DIC species and on the δ18O value of total dissolved
inorganic carbon. At low pH the δ18OΣ(DIC) is nearly identical to that of the H2CO3 species (δ18O = 39.5‰),
whereas at very high pH, it is equal to that of theCO32 ion (δ18O = 18.4‰). The above diagram is calcu-
)LJXUH
lated assuming oxygen isotope equilibrium and the total DIC concentration Σ[DIC] = [H2CO3] + [HCO3 ]
+ [ CO32]. For this example, δ18Owater = 0‰, T = 25 °C, isotope fractionation data from Zeebe (1999).
Application to the triple oxygen isotope system

Guo and Zhou (2019) estimated both equilibrium and kinetic triple oxygen isotope
fractionations in the H2O–DIC–CO2 system using ln a18Oeq values between dissolved carbonate
species from earlier theoretical estimates (Hill et al. 2014). The triple oxygen isotope exponent, κ,
is similar for all dissolved carbonate species and calcite, but different for CO2 (κ = ln17β/ln18β,
and is equivalent to the θ value between the phase and monoatomic oxygen; Cao and Liu 2011).
They considered three processes that would disturb the equilibrium conditions of a system
and modelled how the triple oxygen isotope system evolved as it returned to equilibrium. The
three processes are: 1) a sudden change in DIC oxygen isotope composition due to dissolution
of a water-soluble carbonate, 2) a rapid loss of CO2 by degassing, and 3) an increase in CO2
concentration (AKA ingassing or absorption) as might occur inside a cell during CO2 respiration.
When the system is perturbed, the δ18O and δ17O values of all DIC species initially change
and, over time, return to oxygen isotope equilibrium with the water. However, the rates of exchange
are different for 17O and 18O, so that there are subtle variations in the θ and Δ′17O values of the
dissolved carbonate species during the perturbation (Fig. 4). The pathways shown in Figure 4
illustrate the transient changes in the δ18O and Δ′17O values of HCO3 . We can consider the HCO3
ion to control the δ18O value of the carbonate triple oxygen isotope composition during one of
)LJXUH at moderate pH (6.5–9) and temperature (20–30 °C). Rapid loss of
these perturbation episodes
CO2 increases the θ, Δ′17O, and δ18O values (Fig. 4), which then return to equilibrium over a
period of several hours. Absorption of CO2 has the opposite effect, lowering the θ, Δ′17O and δ18O
values (Fig. 4) which again return to equilibrium in several hours. Changing the DIC concentration
and oxygen isotope composition (by addition of sodium bicarbonate in the example) results in
a curved trajectory in Δ′17O–δ18O space (Fig. 4) where the θ and Δ′17O values initially increase,
and then decrease while the δ18O values continuously decreases and whole system approaches
equilibrium over a period of several hours to days (depending on pH and temperature).
As an example, rapid surface degassing of a subterranean fluid with high p(CO2) will raise
the pH of the fluid and cause rapid deposition of carbonates (travertines). The red curve in
Figure 4 shows that under such a degassing process, we might expect the δ18O and Δ′17O values
to increase. Preliminary results of the triple oxygen isotope composition of travertines from
actively precipitating 10–20 °C springs in Tierra Amarilla, NM have higher Δ′17O and δ′18O
values, 0.05‰ and 2–5‰, respectively, than expected from equilibrium precipitation from its
formation fluid (unpublished results from a senior thesis at the University of New Mexico).
)LJXUH

Figure 4. Changes in the δ O and Δ′ O values of dissolved HCO3 during degassing of CO2 (red curve),
18 17 
absorption of CO2 (yellow

curve) and introduction of a disequilibrium source of DIC (such as dissolution
of Na2CO3–blue dashed curves). The starting point of the equilibration path following introduction of DIC
(blue curves) is controlled by the δ18O and Δ′17O of the starting DIC, as illustrated by three different starting
points. Figure is reformatted after modelling data in Guo and Zhou (2019). Temperature, pH and isotopic
composition of the water also have an effect.
Fosu et al (2020) measured the triple oxygen isotope composition of a large variety of
carbonates and found a surprisingly wide range of Δ′17O values. For example, they measured
two cold seep carbonates with Δ′17O values that differed by 0.13‰ between each other. This
is more than double the Δ′17O range of carbonate samples thought to be in equilibrium with
seawater (see Fig. 13 later in this chapter). Cold seep carbonates have been shown to be out of
clumped isotope equilibrium (Loyd et al. 2016). One cold seep sample has a Δ′17O value that
is about 0.05‰ higher than the Δ′17O equilibrium value expected with seawater while the other
sample has a Δ′17O value that is 0.10‰ lower than expected for equilibrium formation with
seawater (δ17O = δ18O = Δ′17O = 0). Perhaps the triple oxygen isotope compositions are also
showing disequilibrium values. Research addressing how disequilibrium processes affect the
triple oxygen isotope values of the precipitating mineral is clearly an area of future research.
SILICA–WATER FRACTIONATION
Calibration of the triple oxygen isotope silica–water thermometer
The 18O/16O fractionation between silica and water and its temperature dependence is
extremely well studied. There are at least 7 experimental fractionation studies of quartz–water at
high temperatures (see Sharp et al. 2016 for a compilation) and a number of empirical studies of
naturally-formed abiogenic and biogenic amorphous silica at low to moderate temperatures (Kita
et al. 1985; Leclerc and Labeyrie 1987; Shemesh et al. 1995; Brandriss et al. 1998; Schmidt et
al. 2001; Dodd and Sharp 2010). Low temperature diatom silica initially form out of equilibrium
with their host water (Brandriss et al. 1998; Dodd and Sharp 2010), but then undergo an early
maturation phase, such that the silica reaches oxygen isotope equilibrium within the first year or
two post-mortem (Schmidt et al. 2001; Dodd et al. 2012; Tyler et al. 2017).
Wostbrock et al. (2018) compared the triple oxygen isotope values of abiogenic and
biogenic silica as well as amorphous and microcrystalline silica and found no measurable
differences in the δ18O fractionation. Therefore, we include both the high temperature
equilibration experiments and the low temperature empirical fractionation data from biogenic
and abiogenic silica to derive an SiO2–H2O oxygen isotope fractionation equation valid at
temperatures above 273 K, given by
4.28  106 3500 (22)

1000 ln 18OSiO2 H2 O  
T2 T
The low temperature anchor on Equation (22) is primarily controlled by Antarctic diatom
data. Shemesh et al. (1992) recommended only using Antarctic diatoms for the empirical
silica–water fractionation to avoid complications of upwelling effects seen in other regions.
Equation (22) fits the published low temperature diatom data well (Fig. 5), suggesting that the
equation is valid down to 0 °C.
The temperature dependent 17O/16O fractionation for quartz–water has been determined
empirically by measuring the triple oxygen isotope compositions of quartz and abiogenic
and biogenic amorphous silica (coexisting water was also measured when possible) over a
temperature range of 0 to over 100 °C (Sharp et al. 2016; Wostbrock et al. 2018). The θ–T
relationship can be calculated using Equation (8) with the best fit given by
1.85 (23)

qz – wt 0.5305 
T
The θ–T results are shown in Figure 6 along with two theoretical calibrations (Cao and Liu
2011; Hayles et al. 2018). There is generally good agreement between the empirical and
theoretical estimates, with the latter having slightly higher θ values at low temperatures.
)LJXUHSilica–Water and Carbonate–Water Systems 377
)LJXUH
Figure 5. Comparison of the quartz–water fractionation equation (solid curve) with low temperature dia-
tom data (Leclerc and Labeyrie 1987; Sharp et al. 2016). All analyses fit the equation to within 4 °C.
)LJXUH
)LJXUH
Figure 6. Measured θ values for natural samples and best fit for a θ vs. 1/T relationship. Data from Sharp
et al. (2016) and Wostbrock et al. (2018). Also shown are the theoretical curves of Cao and Liu (2011) and
Hayles et al. (2018).
The ΔΔ′17O fractionations (Δ′17Osilica − Δ′17Owater) as a function of temperature are governed by

Equation (14). For silica–water, we have
 4.28  106 3.5  103   1.85 

17Osilica  17O water    18O water   2
   0.5305   (24)
 T T  T 
where λ is assigned a value of 0.528 in this chapter. Equation (24) allows us to construct the
Δ′17O–δ′18O equilibrium curves for silica–water (Fig. 2).
Silica in the terrestrial environment
Figure 7 shows the quartz–water triple oxygen isotope plot with the equilibrium curves
inverted (dashed lines) so that the origin is placed at the silica oxygen isotope composition
for three silica samples. This allows us to estimate the oxygen isotope compositions of the
formation waters in a terrestrial (non-marine) setting in equilibrium with a given silica sample.
The green diamond is a modern sinter (amorphous silica) from Yellowstone National Park,
Wyoming. Both the neoform silica and coexisting water (green circle) were sampled. A water
temperature of 47 °C was measured at the time of collection. The silica and water appear to
be in triple oxygen isotope equilibrium. They both lie on the same equilibrium curve with a
)LJXUH
Figure 7. Triple isotope compositions of low-T terrestrial silica samples. Data from Sharp et al. (2016). The
Yellowstone sinter and Keokuk geode have oxygen isotope data for the coexisting waters and correspond to
formation temperatures of 40–50 °C. The Moroccan geode has a similar oxygen isotope composition to the
)LJXUH
other two samples suggesting similar conditions of formation. Meteoric water field from Sharp et al. (2018).
corresponding temperature of 42 °C. The measured water is ~10‰ higher than local meteoric
water, suggesting the sampled water has undergone some degree of evaporation and/or
hydrothermal exchange with the host rock. The blue diamond is microcrystalline quartz from a
Keokuk geode that retained fossil water in its core (blue circle). We do not know if the core water
represents the formation water. However, the triple oxygen isotope values for the quartz and the
water do in fact suggest equilibrium at a temperature of ~45 °C, suggesting the core water could
be representative of the initial formation water. The purple diamond is microcrystalline quartz
found in a stalactitic geode from Morocco (Sharp et al. 2016). We do not have coexisting water
for this sample, but recognizing the Δ′17O value of the sample is slightly lower than the other
two samples, we estimate a formation temperature of ~40 °C. Data from all the samples suggest
that crystalline quartz can form at temperatures as low as 40 to 50 °C.
CARBONATE–WATER FRACTIONATION
Analytical method
The standard method for determining the δ18O value of carbonates was derived in 1950
(McCrea 1950). Carbonates are reacted with 100% phosphoric acid at a constant temperature,
producing CO2 gas from the breakdown of the carbonate. Only 2/3 of the oxygen in the
carbonate is liberated as CO2 gas. However, if the decarbonation is performed at a constant
temperature, then the fractionation (δ18OCO2 (ACID)–δ18Ocarbonate or αCO2 (ACID)–carbonate) should be
constant, although there are subtle second order effects that should be considered (Sharp 2017).
In order to calculate the actual δ18O value of the carbonate on the VSMOW-SLAP scale, it is
necessary to know the αCO2 (ACID)–carbonate fractionation factor (Sharp and Wostbrock 2021, this
volume). This value is determined by liberating 100% of the oxygen from the carbonate either
through complete fluorination of the carbonate (Sharma and Clayton 1965) or decarbonation
of the carbonate as CO2 followed by fluorination of the remaining CaO (Kim and O′Neil 1997;
Kim et al. 2007). In these carbonate studies, CO2 and/or a combination of CO2 and O2 gas were
produced. The O2 was converted to CO2 by reaction with a heated graphite rod and all CO2
was recombined for the measurement of the δ18O value of the total carbonate. The difference
between the δ18O values of the CO2 produced by reaction with phosphoric acid and of the total
carbonate gives us the a18OCO2 (ACID)–carbonate value.
These procedures cannot be used to determine the δ17O value of carbonates. There is
an isobaric interference at mass 45 (12C16O17O) by 13C (13C16O16O). The d17O value must
be measured on O2 gas unless an ultra-high precision mass spectrometer is being used that
can measure the O+ fragment of CO2 (Adnew et al. 2019). The problem with extracting O2
gas from carbonates by fluorination is that intermediate compounds, such as CO and COF2,
are produced and they are very resistant to fluorination. Wostbrock et al. (2020a) modified
the Sharma and Clayton (1965) procedure of carbonate fluorination to quantitatively extract
O2. The method involves conventional fluorination in Ni reaction tubes heated to 750 °C for
four days. Wostbrock et al. (2020a) measured the triple oxygen isotope values of international
standards as well as the CO2 liberated by phosphoric acid digestion. These data provide us
with the a17OCO2 (ACID)–carbonate value of 1.00535 when using the a18OCO2 (ACID)–carbonate value of
1.01025 (Wostbrock et al. 2020a) . With these data, other laboratories that indirectly measure
the δ17O value of CO2 generated by phosphoric acid digestion can calculate the δ17O value of
the total carbonate. Alternative methods that start with the CO2 gas released by phosphoric
acid digestion include the following: converting CO2 to H2O followed by fluorination (Passey
et al. 2014), equilibrating CO2 gas with O2 using a hot Pt catalyst (Mahata et al. 2013; Barkan
et al. 2015; Fosu et al. 2020), and analyzing the O+ fragment of CO2 generated in the ion
source using a high resolution mass spectrometer (Adnew et al. 2019). Laser spectroscopy that
measures the δ17O value of CO2 is a promising new technology that has yet to be adapted to
wide-scale use (Sakai et al. 2017). We refer readers to Passey and Levin (2021, this volume)
and Sharp and Wostbrock (2021, this volume) for more information about procedures used to
measure the triple oxygen isotope values of carbonates.
Triple oxygen isotope fractionation in the calcite (aragonite)–water system
Debate is ongoing regarding which type of carbonate (synthetic, biogenic, or abiogenic)
best represents equilibrium oxygen isotope fractionation (Coplen 2007; Bajnai et al. 2018;
Brand et al. 2019) or if naturally forming carbonates even reach equilibrium with the waters
in which they form (Daëron et al. 2019). Wostbrock et al. (2020b) examined each category of
carbonates (biogenic marine carbonate, biogenic and abiogenic marine aragonite, abiogenic
Devils Hole calcite, synthesized calcite with and without carbonic anhydrase) to determine
differences in triple oxygen isotope fractionation using the carbonate fluorination method.
The synthesis experiments consisted of slowly precipitating calcite at a constant temperature
following the method of Kim and O′Neil (1997) either with or without carbonic anhydrase
(CA) as a catalyst (Silverman 1973; Uchikawa and Zeebe 2012; Watkins et al. 2013). The
natural samples included marine biogenic calcite and marine abiogenic and biogenic aragonite
that had known water δ18O values and growth temperatures. They found no statistically
significant difference in the 1000 ln α18Occ-wt between synthetic, biogenic, and abiogenic calcite
or aragonite but the 1000 ln α18Occ-wt values were at the higher end of that in reported literature
(1000 ln α18Ocarb-wt = 29.0 at 25 °C). Wostbrock et al. (2020b) used all the samples from each
category, combined with high temperature experimental data from O′Neil et al. (1969) to derive
the 1000 ln α18Ocarb-wt portion of the triple oxygen isotope fractionation (Eqn. 14) given by
18 16 2.84  106 (25)

1000 ln  carbO / water
O
  2.96
T2
A difference was seen in the 1000 ln α17O values between calcite–water and aragonite–water
(Wostbrock et al. 2020b). Calcite–water has higher θ values and smaller ΔΔ′17Ocarb–wt
(Δ′17Ocarbonate − Δ′17Owater) values than aragonite–water (Fig. 8). However, these conclusions are
based on aragonite samples from a very narrow range of temperatures (25–26 °C) and should
be considered preliminary until a wider temperature range is measured. The calcite–water
triple oxygen isotope fractionation equation from Wostbrock et al. (2020b) is:
 2.84  106  –1.39 

17Ocarb – 17O wt   2
– 2.96   0.5305 –   (26)
 T  T 
It is unclear whether any of the carbonates presented in Wostbrock et al. (2020b) represents
true thermodynamic equilibrium. Isotopic equilibrium will not be reached in a natural setting
if there is disequilibrium between the dissolved carbonate species (see earlier section “Effects
of dissolved inorganic carbon”). If one of the dissolved species is out of equilibrium with
the others and with the water, then the precipitated carbonate may reflect that disequilibrium
(Guo and Zhou 2019). Regardless of whether Equation (26) represents ‘true’ thermodynamic
equilibrium or not, it does show a correlation between ΔΔ′17Ocarb–wt and temperature that
should be applicable any naturally formed carbonates.
The triple oxygen isotope fractionation equations for calcite have also been calculated
from theory (Cao and Liu 2011; Hayles et al. 2018; Guo and Zhou 2019). There is a systematic
increase in the ΔΔ′17Ocarb–wt value as a function of decreasing temperature, similar to what is seen
for quartz and what is predicted from theory (Fig. 8; Hayles et al. 2018; Guo and Zhou 2019).
As noted earlier, there was no statistical difference in the ΔΔ′17Ocarb–wt values between the
different calcite groups presented in Wostbrock et al. (2020b). There was a statistical difference
between the calcite samples (blue circles and squares, Fig. 8) and the aragonite samples (brown
circles, Fig. 8) where the best fit of the aragonite samples (brown line, Fig. 8) is lower at all
temperatures, compared to the best fit of the calcite samples (blue line, Fig. 8). The calcite results
agree well with theoretical estimates (black line, Fig. 8; Hayles et al. 2018; Guo and Zhou 2019).
However, theoretical calculations predict there to be little to no difference between the Δ′17O
values of aragonite and calcite (Guo and Zhou 2019; Schauble and Young 2021, this volume).
Two other studies (Bergel et al. 2020; Voarintsoa et al. 2020) also derived θ values using
freshwater aragonitic mollusks (green diamonds, Fig. 8) and synthetic calcite/aragonite
(orange crosses, Fig. 8), respectively, using the Pt catalyzed CO2–O2 equilibration method.
In Figure 8, data from both studies, measured on O2 derived from CO2, were converted to
the total carbonate using the published a 17OCO2 (ACID)–carbonate of 1.00535 (Wostbrock et al.
2020a). The ΔΔ′17Ocarb–wt values from both studies are significantly larger and both studies
do not report a significant variation with temperature (Fig. 8). Although the reason for the
differences between the studies is unknown and warrants further investigation, one possible
)LJXUH
explanation could be a kinetic fractionation effect during the CO2–O2 equilibration method
(Fosu et al. 2020; Passey and Levin 2021, this volume) used in both the Bergel et al. (2020)
)LJXUH
Figure 8. The ΔΔ′17Ocarb-wt as a function of temperature from published studies. Symbols are the average
of multiple measurements of a particular sample and error bars are ±1 s.d of average of the reported data in
the study. Theoretical ΔΔ′17Ocarb-wt–T relationship (thick black line) from two studies overlap.
and Voarintsoa et al. (2020) studies. Also, the synthesized calcite and calcite/aragonite mixture
studies from Wostbrock et al. (2020b) and Voarintsoa et al. (2020), respectively, noted different
times before the first crystals appeared during the experiment (3 days vs. 15–30 min (at 35 °C)
to 1–2 days (at 10 °C), respectively). The rapid precipitation at 35 °C may have resulted in some
kinetic effect where the precipitating carbonate incorporated the oxygen isotope values of the
DIC, instead of precipitating in equilibrium with the water. The DIC species were most likely
in equilibrium with the water since they waited 24 hours before adding the sodium bicarbonate
and calcium chloride solution (Voarintsoa et al. 2020). Nevertheless, this explanation does not
explain the differing results between Bergel et al. (2020) which used freshwater aragonitic
mollusks and Wostbrock et al. (2020b) who use measured marine calcite shells and aragonite
coral, shells, ooids and sediment. Perhaps freshwater aragonitic mollusks tend to form further
out of equilibrium with its formation water than marine calcite and aragonite. Again, more
research is necessary to better address these differences.
Applications to natural marine samples
Triple oxygen isotope values have also been measured in carbonates found in soil (Passey
et al. 2014), tooth enamel and egg shells (Passey et al. 2014), lakes (Passey and Ji 2019) and
metamorphic settings (Fosu et al. 2020) and more information can be found in various chapters
in this book (e.g., Passey and Levin 2021, this volume). In this chapter, we will focus on calcite
and aragonite that formed in a marine setting.
Figure 9 shows the triple oxygen isotope compositions of various marine carbonates from
published data (Passey et al. 2014; Fosu et al. 2020; Wostbrock et al. 2020b). Note, Passey et
al. (2014) measures the triple oxygen isotope composition of CO2 evolved from phosphoric
acid digestion at 90 °C. Only an a17OCO2 (ACID)–carbonate at 25 °C is known (Wostbrock et al.
2020a). Therefore, we use the published a17OCO2 (ACID)–carbonate of 1.00535 to correct the CO2
triple oxygen isotope values to total carbonate, but this correction may not be completely
accurate. In this communication we add two additional triple oxygen isotope analyses of Early
Triassic ammonites from the Western Interior Seaway of North America (University of New
Mexico, Earth and Planetary Sciences collections, donation by Jim Jenks). The Smithian
ammonoid (Anasibirites sp.) comes from the Crittenden Springs (NE Nevada) locality (Jenks
and Brayard 2018). The shell preservation of the ammonoids from this locality is believed
to be exceptional, as many specimens have original color patterns preserved. The Spathian
ammonoids were collected from the Hot Springs locality from SE Idaho (Guex et al. 2010) and
are assigned to the Columbites biozone. These ammonoids are commonly found in carbonate
concretions. The triple oxygen isotope values of the Smithian ammonite (blue rhombohedron,
Fig. 9) are 12.624, 24.227, and −0.094‰ for d17O, d18O, and Δ′17O, respectively. The triple
oxygen isotope values of the Spathian ammonite (blue pentagon, Fig. 9) are 13.490, 25.867,
and −0.084‰ for d17O, d18O, and Δ′17O, respectively.
Modern carbonates, six aragonite coral samples (red circle, yellow diamond, and four
green squares), four modern brachiopod shells comprised of calcite (green diamonds) and
one calcitic oyster (red diamond) are shown in Figure 9. The calcitic oyster appears to be
in equilibrium with modern seawater, within reported error, and suggests a formation
temperature of ~20 °C, similar to the measured water temperature of 24 °C (Passey et al. 2014).
The calcitic brachiopod shells (green diamonds) also plot in equilibrium with the
seawater–calcite fractionation curve (Note, these brachiopod samples were among the
samples used to derive the fractionation curve in Wostbrock et al. 2020b). Two aragonitic coral
samples (red circle and yellow diamond) have triple oxygen isotope values that plot below
the calcite–seawater fractionation curve, suggesting a larger aragonite–water fractionation
for Δ′17O than that of calcite–water. This is similar to the conclusions between the Δ′17O
fractionation of calcite–water and aragonite–water presented in Wostbrock et al. (2020b).
Figure 9. The Δ′17O values vs. the δ′18O values of marine carbonates in relation to the equilibrium triple
oxygen isotope fractionation curve for calcite (black curve) in equilibrium with modern seawater (yellow
star, δ18O = 0 and Δ′17O = −0.005‰; Luz and Barkan, 2010). Equilibration temperatures are marked along
the black curve. The seawater curve is bracketed by the modern range (–2 to +2‰) of seawater oxygen
isotope compositions for equatorial and mid-latitude oceans (grey lines; Schmidt et al. 1999). The range of
modern brachiopod δ18O values (26.5 to 35.5‰) is represented by the light grey box (Brand et al. 2019).
Data from Wostbrock et al. (2020b) are green symbols: modern marine brachiopods (diamonds), modern
marine aragonite samples (squares) and blue symbols: Cretaceous belemnite (diamond); brachiopods:
Mid Ordovician, (X), Silurian (square), Mid Devonian (circle), Late Pennsylvanian (triangle), and Mid
Maastrichtian (Late Cretaceous; cross); ammonites: Smithian (rhombohedron) and Spathian (pentagon).
Data from Passey et al. (2014) are red symbols: Permian brachiopod (square), modern estuarine oyster
normalized to seawater (diamond), modern coral (Porites porites, circle). Data from Fosu et al. (2020) are
yellow symbols: modern aragonitic deep sea coral (diamond), late Cretaceous chalk comprised of coc-
coliths (circle), modern cold seep high Mg calcite (square). An additional cold seep calcite plots off the
chart and has a Δ′17O value of −0.18‰.
Of the fossil carbonate samples shown, only two, a Silurian brachiopod (blue square) and a
Cretaceous belemnite from the Peedee formation (PDB, blue diamond), plot close to the seawater
fractionation curve (Fig. 9). These samples are consistent with having precipitated in equilibrium
with seawater with a δ18O value of 0 and Δ′17O value of −0.005‰ (Luz and Barkan 2010).
The PDB sample corresponds with a formation temperature of 12 °C while the Silurian brachiopod
corresponds to 28 °C. All other samples plot to the left of the seawater–calcite fractionation curve
and, therefore, cannot be in equilibrium with 0‰ seawater. Assuming that the Phanerozoic
seawater value was similar to today (−1 to 0‰), as suggested by the triple oxygen isotope values of
the Silurian brachiopod sample and the Cretaceous belemnite, then all of the triple oxygen isotope
values of these samples are best explained by diagenesis. Discussion of what these altered samples
tell us about the ancient ocean is presented in “Application to carbonate and silica sediments”.
CONSIDERATIONS NECESSARY TO INTERPRET ANCIENT SEDIMENTS

Traditional oxygen isotope studies have used the oxygen isotope composition of ancient
carbonates and silicates to reconstruct the temperature of the seawater over time, as well as
other proxies such as phosphates and shales (see Sharp 2017; Bindeman 2021, this volume, for
more information on additional proxies). The accuracy of oxygen isotope paleothermometry is
critically dependent on the assumed δ18O value of the ancient ocean. While it is generally agreed
upon that the ocean is buffered to its present value by a complex interplay of plate tectonics
and erosion (Muehlenbachs and Clayton 1976; Muehlenbachs 1986), the δ18O value of past
seawater is still debated. There is a secular trend seen in the carbonate and silica (chert) record
with decreasing δ18O values with increasing age (Lowenstam 1961; Degens and Epstein 1962;
Keith and Weber 1964; Perry 1967; Knauth and Epstein 1976; Veizer and Hoefs 1976; Knauth
and Lowe 1978, 2003; Perry et al. 1978; Popp et al. 1986; Veizer et al. 1986, 1989, 1997,
1999; Lohmann and Walker 1989; Brand 2004; Knauth 2005; Prokoph et al. 2008; Veizer
and Prokoph 2015; Ryb and Eiler 2018). See Figure 1 in Zakharov et al. (2021, this volume).
For example, the δ18O value of Archean cherts are generally 15‰ lower than modern (Knauth
and Epstein 1976; Knauth and Lowe 1978, 2003; Perry et al. 1978; Winter and Knauth 1992).
The explanation for the secular trend falls under three main hypotheses (Degens and Epstein
1962) and a fourth, less talked about hypothesis:
1. The ocean was warmer in the Archean than it is today (Knauth and Epstein 1976,
1986; Knauth and Lowe 1978, 2003; Winter and Knauth 1992; Perry and Lefticariu 2007).
This hypothesis implies the d18O value of seawater remains constant (~0‰) over time
and the temperature of Archean seawater was 50–80 °C and cooled over time. Assuming
a linear cooling, the warm ocean theory would imply that the Cambrian seawater was
about 40–60 °C warmer than modern seawater during the explosion of life, a direct
contradiction to reasonable temperature estimates of survivability of most prokaryotes
(Brock 1985). Additionally, when applied to the Archean, this hypothesis complicates the
faint sun paradox (Sagan and Mullen 1972), although the faint sun paradox itself has been
recently called into question (Rosing et al. 2010). Seawater modelling studies based on
the oxygen isotope values of altered oceanic crust support this idea that seawater has been
buffered to 0‰ for most of Earth′s history (Muehlenbachs and Clayton 1976; Gregory
and Taylor 1981; Muehlenbachs 1986; Gregory 1991; Holmden and Muehlenbachs 1993,
1998; Zakharov and Bindeman 2019). Various studies using carbonate clumped isotope
(Δ47) values also suggest the ocean was buffered to 0‰, albeit with varying temperature
estimates (20–70 °C), since at least the Cambrian (Eiler 2011; Finnegan et al. 2011;
Cummins et al. 2014; Ryb and Eiler 2018; Price et al. 2020).
2. The d18O value of seawater was lower in the past (Perry 1967; Perry and Tan 1972;
Veizer and Hoefs 1976; Veizer et al. 1997, 1999; Hren et al. 2009; Veizer and Prokoph
2015). This hypothesis implies the d18O value of Archean seawater was as low as −15‰.
Models to explain the low d18O value of the ancient ocean have been developed by
assuming very different ratios of high to low temperature alteration flux rates compared
to the modern (Wallmann 2001; Kasting et al. 2006; Jaffrés et al. 2007; Kanzaki 2020).
Ocean temperatures would be similar to modern over the geologic rock record. The idea
of an ocean with lower d18O values than modern seawater is contradicted by oxygen
isotope studies of ancient ophiolites, where the oxygen isotope values of the altered rocks
is governed by the oxygen isotope value of the hydrothermal water (initially seawater) at
an assumed constant temperature over time (Muehlenbachs and Clayton 1976; Gregory
and Taylor 1981; Muehlenbachs 1986, 1998; Gregory 1991). Recent seawater oxygen
isotope compositional modelling by Sengupta and Pack (2018) suggest that a 100-fold
increase of the continental weathering flux rate would be required to lower the ocean to
−8‰. The authors could not reconstruct a seawater with the a d18O value of −15‰ using
any feasible changing scenario of known or estimated flux rates.
3. Ancient samples have been overprinted by diagenesis (Gao and Land 1991; Wenzel et al.
2000; Brand 2004; Marin-Carbonne et al. 2012, 2014; Sengupta and Pack 2018; Liljestrand
et al. 2020). Diagenetic alteration generally occurs at elevated temperatures and/or in the
presence of waters with a lower d18O value of seawater (Popp et al. 1986; Lohmann and
Walker 1989; Banner and Hanson 1990; Gao and Land 1991; Marshall 1992; Wadleigh
and Veizer 1992; Kah 2000; Swart 2015; Ahm et al. 2018). However, there are rare
instances where the d18O value of the diagenetic fluid may be higher than modern seawater
(Grossman et al. 1991), particularly in evaporative basinal porewaters (Swart 2015).
In general, the d18O values of older sediments have lower d18O values. The argument that
diagenesis explains the low d18O value of ancient sediments suggests that the diagenesis is a
continual process, such that older materials have been more strongly affected than younger
ones or that alteration conditions themselves have changed through time.
4. Hydrothermal systems formed much of the Archean chert record (De Ronde et al.
1997; Sengupta et al. 2020; Zakharov et al. 2021, this volume). The triple oxygen isotope
data are explained by this idea, but some geologic and petrographic evidence does not
support high temperature, hydrothermal deposition. For example, some stratigraphic
evidence of laterally continuous chert beds and low temperature facies preservation such
as rip-up clasts suggest that at least some of the ancient chert was deposited in shallow,
near surface settings during the Archean (Knauth 1979; Lowe 1983; Maliva et al. 1989,
2005; Tice et al. 2004; Tice and Lowe 2006; Perry and Lefticariu 2007).
To complicate things further, there are some studies where the δ18O of seawater is proposed
to have been up to 3‰ higher in the past (Longinelli et al. 2002, 2003; Johnson and Wing
2020). In summary, the δ18O value of seawater over the geologic record is still an open question.
Triple oxygen isotope measurements help constrain the fluid in which the silicate or carbonate
formed or can help determine if the sample is preserving primary depositional information
because samples must fall on a unique curve in Δ′17O–δ′18O space for a given composition of
water (Fig. 2). Departures from this line are explained by changes in the triple oxygen isotope
composition of the ocean and/or diagenesis, explained in further detail below. Samples that lie
to the left of the seawater equilibrium curve have been altered by a diagenetic fluid of meteoric
origin. Samples that lie above the modern seawater equilibration curve have either formed in an
ocean with a lower δ′18O–higher Δ′17O value or, rarely, undergone alteration with a fluid with
a higher δ18O value than seawater. The triple oxygen isotope composition of ancient materials
should therefore shed light on which processes might explain their low δ18O values.
Modelling changing ocean composition in the past
Sengupta and Pack (2018) modeled how the Δ′17O and δ′18O values of seawater
would change in response to changing rates high- to low-temperature alteration. Similar to
Muehlenbachs (1998), they determined that continental weathering and high temperature
hydrothermal alteration during seafloor spreading were the two principal fluxes that drive the
seawater oxygen isotope change. They obtained a λ value of 0.51 for changing the continental
weathering and high temperature seafloor alteration fluxes. Liljestrand et al. (2020) suggested
that at least 90% of the total alteration must come from low temperature continental weathering
in order to have Archean seawater with a δ18O value of −15‰. They propose a λ of 0.524 for
low temperature seafloor alteration (Liljestrand et al. 2020), slightly higher than the combined
high- and low-temperature alteration λ proposed in Sengupta and Pack (2018). Bindeman
(2021, this volume) measured the triple oxygen isotope composition of shales and proposes an
intermediate λ value of 0.521. For simplicity, we use the λ value of 0.51 in Figure 10 to show
how the seawater equilibrium fractionation curve would change over time if low temperature
alteration governed Archean seawater oxygen isotope composition. Sediments that formed in
equilibrium with a seawater composition plotting on the λ = 0.51 line would then plot above
the modern equilibrium curve (shaded grey region, Fig. 10).
Hypothetical triple oxygen isotope values of samples are shown as circles in Figure 10.
The green circle can be interpreted as being in equilibrium with modern seawater. The blue
circle could be in equilibrium with seawater with a δ′18O value of −4‰ and a Δ′17O value of
0.075‰. The red circle in Figure 5 could only be in equilibrium with the ocean if the δ′18O
value was significantly higher than today as suggested by Johnson and Wing (2020). It is not
unreasonable that continental erosion would have been lower in the Archean due to a low
abundance of continental crust (Bindeman et al. 2018), in which case the ocean would move
)LJXUH
Figure 10. Δ′17O–δ′18O plot of changing seawater composition. The line labeled λ = 0.51 is the trajectory
of seawater composition for changing high- to low-temperature alteration ratios (Sengupta and Pack 2018).
All circles represent hypothetical data. The green circle is in equilibrium with modern water at 70°C.
The blue circle is the Δ′17O–δ′18O value of silica in equilibrium with a modelled lower δ′18Oseawater value
of −4‰ at 70 °C.)LJXUH
The red circle is the Δ′17O–δ′18O value of silica in equilibrium with a modelled higher
δ′18Oseawater value of +2‰ at 70 °C (Johnson and Wing 2020). For similar temperatures of formation, triple
oxygen isotope values of silica should have higher Δ′17O values and lower δ18O values than modern silica
if the δ18O value of seawater was lower in the past and lower Δ′17O values and higher δ18O values than
modern silica if the δ18O value of seawater was higher in the past.
along the λ = 0.51 line towards higher δ18O values (Johnson and Wing 2020) and lower Δ′17O
values As discussed below, a shift to higher δ18O values for the ancient ocean then leads to
higher temperature estimates based on the mineral–water fractionations. It is therefore argued
that ancient cherts may have formed in high temperature hydrothermal system (De Ronde et
al. 1997; Sengupta et al. 2020; Zakharov et al. 2021, this volume). Alternatively, the red circle
in Figure 5 may be the result of diagenesis, as explained in the following section.
Modelling triple oxygen isotope trends during diagenesis
Low temperature diagenesis of marine sediments generally drives the d18O values lower
and ∆′17O values higher (Sharp et al. 2018). Several variables control the oxygen isotope
composition during diagenetic alteration including, temperature, initial rock and fluid oxygen
isotope compositions, and the fluid/rock (F/R) ratio. Nevertheless, the δ′18O–∆′17O field of
most diagenetically altered rock is surprisingly limited (Fig. 11; Sharp et al. 2018).
The effect of alteration can be determined using a simple mass-balance mixing model.
A fraction of water is allowed to equilibrate with a rock at a given temperature following
Taylor (1978). The bulk composition is given by the equation
dxObulk = Xwater (dxOwater initial) + (1−Xwater) (dxOrock initial) (27a)
and
dxObulk = Xwater (dxOwater final) + (1−Xwater) (dxOrock final) (27b)
dxO can be either δ18O or d17O. The final dxO value of the rock is determined by the additional
equation
1000   x Ofinal rock (28)

x 
1000   x Ofinal water
)LJXUH

)LJXUH
Figure 11. Alteration trajectories in Δ′17O–δ′18O space. The rock starts at δrinit (black circle) and moves
along the solid curve with increasing fluid/rock ratio. The infiltrating fluid has a composition at δMW
(white star), equilibrates with rock and has a composition δwinit (white circle). The re-equilibrated water
follows the dotted curved path from δwinit towards lower δ′18O and higher D′17O values as the F/R ratio
increases. The final rock and water compositions in this example correspond to an Xwater value of 0.6,
resulting in the δwfinal (white diamond) and δrfinal (black diamond) The equilibrium oxygen isotope frac-
tionations at 100 °C are shown by the doubled arrow lines.
Combining Equations (27a), (27b), and (28) leads to the relationship between the initial rock
and water oxygen isotope compositions, the fluid/rock ratio and the final isotopic composition
of the rock:
 x O rock final 

1000 X   X   x O rock initial  X   x O water initial   x O rock initial  1000X  (29)
X    X
A typical alteration pathway at 100 °C is shown in Figure 11. The fractionation between
the water and rock is shown by the arrows δ′r–δ′w and Δ′r–∆′w. As long as temperature is
held constant, these fractionations will not change. The initial rock is silica in equilibrium
with seawater water at 20 °C (black circle and δrinit). The meteoric water that infiltrates the
rock has initial δ18O and Δ′17O values of –6 and +0.03‰ (white star and δMW). The infiltrating
meteoric water always has a constant composition (δMW) and equilibrates with the rock as
it enters the system. When the alteration process first begins, an infinitely small amount of
water enters the system and equilibrates with the rock. The fraction of oxygen from the water
relative to the total water–rock oxygen reservoir is near zero. In the model, this first ‘aliquot’
of water will equilibrate with the overwhelmingly large oxygen reservoir of rock and have a
composition given by δwinit (white circle, Fig. 11). This is the oxygen isotope composition of
the equilibrated water when the fluid–rock interaction process first begins to take place.
As more fluid enters the system, the F/R ratio increases and the oxygen isotope
composition of the fluid and rock track up the solid (rock) and dotted (water) curves. As the
F/R ratio approaches an infinite value, the oxygen isotope composition of the fluid is equal
to that of infiltrating meteoric water (δMW). The infiltrating water ‘overwhelms’ the buffering
capacity of the rock. The δ′18O and Δ′17O values the rock will be in equilibrium at 100 °C with
the δMW (white star, Fig. 11). The final oxygen isotope composition of the rock at the end of
the alteration process is a function of the F/R ratio. In Figure 11, the white and black diamonds
illustrate an intermediate condition corresponding to an Xwater = 0.6 or a fluid–rock ratio of
1.5 (Xwater/Xrock = 0.6/0.4). The variables that control the alteration pathways are the temperature
of alteration, the δ18O and Δ′17O values of the infiltrating water, and the initial oxygen isotope
composition of the rock (which is a function of its formation temperature). Figure 12 shows
various alteration trajectories for different water compositions and temperatures.
In general, we expect to see altered marine rocks lying to the left of the seawater equilibrium
curve, shown by the lavender region in Figure 13. This field is generated by considering a
wide range of reasonable infiltrating fluid triple oxygen isotope compositions, alteration
temperatures, and initial rock triple oxygen isotope (black curves, Fig. 13). For case studies see
)LJXUH
Herwartz et al. (2015), Zakharov and Bindeman (2019), Chamberlain et al. (2020), and Herwartz
(2021, this volume). The alteration field is mostly distinct from the ‘changing seawater’ field
generated by lowering the δ18O value of ancient seawater by changing high- to low-temperature
alteration ratios (shaded grey region, Fig. 13). These two fields—alteration in lavender and
)LJXUH
Figure 12. Alteration trajectories for a rock starting with a δ′18O–Δ′17O of δrinit. Dashed and solid lines are
for interaction with an infiltrating meteoric water (δMW) with δ′18O values of −10‰ and −6‰, respectively
)LJXUH
and Δ′17O values of 0.03‰. The rock trajectories are thick curves, fluids are thin curves.
)LJXUH
Figure 13. Generalized fields for sedimentary silica. The lavender field represents the general region
expected for alteration based on numerous trajectories for different water compositions and temperatures
(black curved lines). The grey region is calculated for lower oxygen isotope compositions of seawater
predicted if high- to low-temperature alteration ratios changed dramatically in the past.
lowering ocean values in grey—represent the generalized regions where we can expect most
sediment phases to plot. Overall, samples that lie below and to the left of the equilibrium line
are explained by diagenesis. There are alteration scenarios that would result in a positive shift
for both the δ18O and Δ′17O values. This may be seen in hot buoyant fluids that infiltrate a cooler
overlying sedimentary pile (see final section of this chapter).
Application to carbonate and silica sediments
Phanerozoic calcite. Wostbrock et al. (2020b) reported the triple oxygen isotope values of
five ancient brachiopods and one belemnite from the Phanerozoic. Only two samples, a Silurian
brachiopod (formation temperature ~28 °C) and a Cretaceous belemnite (formation temperature
~12 °C), have triple oxygen isotope values that can be in equilibrium with seawater similar to a
modern ocean oxygen isotope composition (Fig. 9). The remaining brachiopods from that study
and the two ammonites presented in this manuscript all have δ18O and Δ′17O values that plot to
the left of the modern seawater triple oxygen isotope fractionation curve, suggesting alteration
by a meteoric water. It is only with the addition of the Δ′17O value that we are able to conclude
that these ancient brachiopods do not preserve triple oxygen isotope compositions and could
not have formed in seawater similar to modern oceans (grey box, Fig. 9).
We can use the fluid/rock alteration model to ‘see through’ the alteration and calculate the
original triple oxygen isotope composition of the samples, prior to alteration (Fig. 14). For the
Ordovician brachiopod, Wostbrock et al. (2020b) were able to fit the measured triple oxygen
isotope data using a meteoric fluid as the alteration fluid with a δ18O value between −15 and
−10‰, a Δ′17O value of +0.03‰, and an alteration temperature between 35 and 100 °C. Under
these conditions, the back-calculated initial δ18O value of the brachiopod was between 30 and
32.5‰, corresponding to a formation temperature of 10 to 20 °C, assuming the seawater had a
similar triple oxygen isotope composition as modern oceans. Using the same fluid–rock mixing
model, we calculated the primary composition of the other altered brachiopod and ammonite
samples (Fig. 14). All the brachiopod shells formed in water ranging between 10 and 20 °C,
similar to the 0–30 °C water temperature range of modern brachiopods (Brand et al. 2019). For
the Smithian and Spathian (Early Triassic) ammonites, we find that both ammonites are altered in
spite of the high visual preservation quality of the fossil shell. Nevertheless, we can back-calculate
formation temperatures of 10 and 20 °C for the Spathian and Smithian age samples, respectively.
These agree well with the 10 °C difference suggested using the δ18O values of conodonts (Sun
et al. 2012). Overall, we do not see evidence of lower δ18O values of the ocean, nor higher
temperatures in the Phanerozoic, although the number of samples measured to date is still small.
Ancient silica. Sedimentary bulk silica is found throughout the rock record from 3.5 Ga
through to the present day (Knauth 2005). Chert is inherently the product of a diagenetic
process where dissolved silica is reprecipitated and evolves from Opal-A to Opal-CT
to microcrystalline quartz. Generally, the term “chert” refers to microcrystalline quartz.
The amount of time it takes for Opal-A to transform to microcrystalline quartz is still an active
area of research (Yanchilina et al. 2020). The ocean has been undersaturated with respect
to silica since the appearance of silica-secreting radiolarians and siliceous sponges in the
Phanerozoic and later diatoms in the Jurassic. Most Phanerozoic silica deposits originated as
biogenic Opal-A, which later transformed to microcrystalline quartz generally in an offshore
environment (Maliva et al. 1989; Perry and Lefticariu 2007). Prior to the Phanerozoic, the
ocean may have been close to or at silica saturation, and marine silica is thought to have mostly
precipitated through abiogenic (or inorganic) processes. Preservation of sedimentary features
in Precambrian and earlier outcrops gives credence to the idea that some transformations
can occur in the near subsurface not long after deposition (Knauth 1973; Knauth and
Lowe 1978; Lowe 1983; Maliva et al. 1989, 2005; Tice et al. 2004; Tice and Lowe 2006;
Perry and Lefticariu 2007). The ubiquitous presence of chert throughout geologic time and
the assumed lower susceptibility to diagenesis than carbonate has made the oxygen isotope
composition of chert an attractive sample to use as a paleo-indicator of ancient seawater.

Figure 14. Back-calculated initial triple oxygen isotope compositions of altered brachiopod and ammonite
samples. Alteration fluid triple oxygen isotope value are written on each graph and pathways were mod-
elled for alteration temperatures of 35 (blue lines), 50 (orange lines), and 100 (red lines) °C. A) The Mid-
Ordovician brachiopod formed between 10 and 20 °C. To demonstrate the fact that the fluid–rock interac-
tion model is under constrained, Wostbrock et al. (2020b) used two different alteration fluids to calculate
two different potential primary calcite triple oxygen isotope values and show the range of results the model
can produce. For B–E, we only show one potential solution for the fluid–rock mixing model for the sake of
)LJXUH
simplicity. B) The Mid Devonian brachiopod calculated primary calcite triple oxygen isotope composition
suggests an initial precipitation temperature of ~18 °C. C) The Late Pennsylvanian brachiopod calculated
primary calcite triple oxygen isotope composition suggests an initial precipitation temperature of ~15 °C.
D) The Mid Maastrichtian (Late Cretaceous) brachiopod calculated primary calcite triple oxygen isotope
composition suggests an initial precipitation temperature of ~10 °C. E) The Smithian ammonite (rhombo-
hedron) calculated primary calcite triple oxygen isotope composition suggests an initial precipitation tem-
perature of ~20 °C. The Spathian ammonite (pentagon) calculated primary calcite triple oxygen isotope
composition suggests an initial precipitation temperature of ~10 °C.
Four studies have published values on the triple oxygen isotope composition of cherts from
the Archean through the Phanerozoic (Levin et al. 2014; Liljestrand et al. 2020; Sengupta et al.
2020; Zakharov et al. 2021, this volume) and the results are presented in Figure 15. First, the vast
majority of all chert samples that have been measured for triple oxygen isotope analysis plot to
the left of the modern silica-seawater equilibrium fractionation line. Perhaps most importantly,
all chert samples from the Archean (dark blue icons in Fig. 15) plot at far lower δ18O values than
other samples and to the left of the modern silica–seawater equilibrium fractionation line. The
fact that they do not plot above the equilibrium line indicates that their unique δ18O–Δ′17O values
are not due to lower δ18O seawater compositions related to changing high- to low-temperature
alteration ratios (Figs. 12 and 13). Therefore, the secular trend seen in the δ18O values of chert is
not compatible with an Archean seawater δ18O value of −8 to −15‰ unless the alteration trend of
λ = 0.51 (Sengupta and Pack 2018) is not correct. Other studies suggest higher λ values of 0.521
(Bindeman 2021, this volume) and 0.524 (Liljestrand et al. 2020). Using a λ value of 0.524 and
a δ18O value of −8‰ for the ocean can explain only a few of the Proterozoic chert data (solid
red line, Fig. 16) with a formation water temperature of 30 to 40 °C. However, even this high λ
value cannot reconcile any Archean chert or the majority of the published chert dataset (Fig. 16).
Almost none of the chert data can be explained by equilibration with a modern seawater
oxygen isotope composition. Chert samples from the Phanerozoic presented in Sengupta et
al. (2020), represented by the yellow diamonds in Figure 15, are explained by dissolution
and equilibrium re-precipitation with a meteoric–marine fluid mixture with a δ18O value
of −3‰ and a Δ′17O value of +0.010‰ (dashed purple line with purple star, Fig. 15;
Sengupta et al. 2020). This implies the samples are preserving a precipitation temperature
between 20 and 30 °C, suggesting near-surface dissolution and re-precipitation. Similarly,
Zakharov et al. (2021, this volume) suggest the first generation of microcrystalline quartz
in a Proterozoic chert sample (light blue crosses in Fig. 14) formed in equilibrium with
Figure 15. Published Δ′17O and δ′18O values of ancient chert from the Archean through Phanerozoic from
Levin)LJXUH
et al. (2014, squares) Liljestrand et al. (2020, circles), Sengupta et al. (2020, diamonds), Zakharov
et al. (2021, this volume, crosses). Yellow colors are younger cherts while blue colors represent older
cherts. One interpretation for the chert data suggest alteration by a fluid with a lower triple oxygen isotope
value than the modern ocean (SMOW, yellow star). Using a fluid–rock mixing model, the precipitation
temperature of the primary chert was calculated as 30 °C (Liljestrand et al. 2020; orange line and triangle)
and 10 °C (Sengupta and Pack 2018; pink lines and hexagon). Sengupta et al. (2020) proposed that Pha-
nerozoic chert (yellow icons) generally form by dissolution and equilibrium re-precipitation after burial in
a meteoric-marine water mixture (purple star and dashed lines). The initial formation temperatures are
then 30 and 40 °C. Zakharov et al. (2021, this volume) suggests a similar precipitation scenario with pore
fluid during the Proterozoic (light blue icons) and formation temperatures greater than 60 °C.
)LJXUH
Figure 16. Various seawater triple oxygen isotope compositions can describe published Δ′17O and δ′18O
values of ancient chert from the Archean through Phanerozoic from Levin et al. (2014, squares) Liljestrand
et al. (2020, circles), Sengupta et al. (2020, diamonds), Zakharov et al. (2021, this volume, crosses).
Yellow colors are younger cherts while blue colors represent older cherts. The triple oxygen isotope val-
ues of chert are not consistent with seawater having had a lower δ18O value in the past (red stars) if cor-

responding change in δ18O and Δ′17O following either the trend with the lowest λ value of 0.51 modelled

by Sengupta and Pack (2018; dashed red line) or the trend with the highest λ value of 0.524 suggested by
)LJXUH
Liljestrand et al. (2020; solid red line). A possible explanation is that the δ18O value of seawater was higher
in the Archean compared to modern or that Precambrian chert tended to form in a hydrothermal system
with a higher δ18O value than modern seawater. The fluid could have a δ18O value of +1‰ (purple star
and dashed purple line) as proposed in Sengupta et al. (2020) and Zakharov et al. (2021, this volume) or
have a δ18O value as high as +3‰ as proposed in Johnson and Wing (2020; green star and dashed line).
Either seawater δ18O scenario suggests high formation temperatures between 150 and 300 °C for Archean
chert (orange circle) and 70 to 150 °C for Proterozoic chert (purple circle).
interstitial pore fluid that is similar to slightly lower than the modern seawater oxygen isotope
composition at a temperature between 50 and 80 °C, representative of the temperature of
early diagenesis. Alteration of Phanerozoic cherts has long been known to commonly occur
and are actually probably the norm (Knauth and Epstein 1976). It is not surprising, therefore,
to expect alteration by meteoric waters to have occurred in older samples.
All current triple oxygen isotope studies of chert suggest alteration by secondary fluids.
Sengupta and Pack (2018) and Liljestrand et al. (2020) suggest the initial precipitation of the
silica occurred in seawater with a δ18O value similar to modern seawater and at temperatures
below 30 °C. Both studies used fluid–rock interaction models (Figs. 11 and 12) to calculate
primary chert compositions. Liljestrand et al. (2020) constrained the initial silica formation
to have occurred at 30 °C, with a subsequent diagenetic alteration at 200 °C with a fluid
of −16.5‰. Sengupta and Pack (2018) (using renormalized data from Levin et al. 2014)
concluded that alteration by a fluid with a δ18O value of −6‰ at 50 °C generally explains the
Phanerozoic chert data. In contrast, an alteration fluid of 10‰ at 180 °C explains the Archean
cherts and a chert precipitation temperature of −10 °C.
Sengupta et al. (2020) used a Bayesian model to calculate two different alteration scenarios
for their Archean and Proterozoic chert data. For the Archean samples they separated samples
into two groups, one with δ18O values between 13 and 16‰ and one group with δ18O values
between 18.7 and 19.3‰. For the first Archean group, they concluded that the silica precipitated
in seawater with an oxygen isotope composition similar to modern seawater, and was
subsequently altered by a fluid with a δ18O value of −16.5‰ at an alteration temperature of
50 °C and W/R ratio of 1. Their Bayesian model was inconclusive on initial silica precipitation
temperature. For the second group of Archean chert, the Bayesian model again predicted initial
precipitation in seawater with an oxygen isotope composition similar to modern at 80 °C and was
subsequently altered with a fluid with a δ18O value of –14‰ at either 50 or 150 °C. The Bayesian
model was ambiguous as to the W/R ratio. Proterozoic cherts fit within both Archean diagenetic
models. Sensitivity in the Bayesian model allows for the initial silica from either the Archean
or Proterozoic samples to have formed in temperatures as low as 10 °C (Sengupta et al. 2020).
Although diagenesis can explain the above data, Sengupta et al. (2020) also consider the
hypothesis that the Archean and Proterozoic samples are retaining primary oxygen isotope values
in which the chert equilibrated with a fluid that had a higher δ18O value than modern seawater
(Fig. 16). This idea is also favored by Zakharov et al. (2021, this volume). Sengupta et al. (2020)
acknowledge that they cannot distinguish between whether the samples formed in seawater with
higher δ18O and lower Δ′17O values than modern seawater (Johnson and Wing 2020) due to
higher hydrothermal water flux in the Archean (Isley 1995; Lowell and Keller 2003) or whether
the samples formed in high temperature hydrothermal vent fluids with ambient seawater similar
to modern (Bindeman et al. 2018; Zakharov and Bindeman 2019; Peters et al. 2020).
Zakharov et al. (2021, this volume) use a modern hydrothermal vent in Iceland as an
analog to an Archean vent-like setting with fluids having a δ18O value of 0.9 to 1.6‰ (De
Ronde et al. 1997; Farber et al. 2015) and temperatures between 150 and 300 °C. Sengupta et
al. (2020) suggest a similar fluid (either vent or ambient seawater) oxygen isotope composition
with a δ18O value of +1‰ and Δ′17O value of −0.040‰, corresponding to a temperature of
formation for the Archean cherts of 150 to 220 °C and 75–150 °C for Proterozoic cherts
(purple star and dashed line, Fig. 16). They explain the lower Proterozoic fluid temperatures
result from a lowering of the geothermal gradient. Johnson and Wing (2020) argue that, due to
the absence of continents, the Archean ocean could have a δ18O value as high as +3‰ (green
star and dashed line, Fig. 16). Following a λ ≈ 0.51 for the changing seawater trend presented
in Sengupta and Pack (2018), an Archean seawater with a δ18O value of +3‰ would then have
a Δ′17O value of about −0.05‰. Seawater with a δ18O value of +3‰ also fits the Archean
and Proterozoic chert samples (blue symbols in Fig. 16) at similar temperatures proposed by
Sengupta et al. (2020) and Zakharov et al. (2021, this volume). Interestingly, seawater with a
δ18O value of +3‰ does not fit the majority of the Phanerozoic chert samples. If the hypothesis
that most Precambrian chert formed in seawater with a higher δ18O value than modern, perhaps
chert formation processes in the Phanerozoic were different, due to the presence of marine
silica-secreting organisms (Maliva et al. 1989, 2005; Perry and Lefticariu 2007).
Overall, most Archean and Proterozoic chert that has currently been measured for triple
oxygen isotope values could have formed or been altered by high-temperature vent fluids with
δ18O values higher than modern seawater. Alternatively, the data can be explained by having
initial silica precipitation at less than 30 °C in marine water with a modern oxygen isotope
composition, and subsequent diagenetic alteration by meteoric waters (Sengupta and Pack 2018;
Liljestrand et al. 2020). This latter interpretation agrees with the classic δ18O–δD study of Knauth
and Epstein (1976) where altered samples have a δD/δ18O slope similar to the meteoric water
line. Examining Archean and Proterozoic chert that preserve sedimentary facies, such as draping
and rip up clasts, indicating formation in a near-shore marine setting at or near the seafloor would
provide additional information. Based on the data published thus far, however, we conclude that
the data do not suggest Archean or Proterozoic seawater with a lower δ18O value than modern.
TRIPLE OXYGEN ISOTOPES OF COEXISTING QUARTZ AND CALCITE

We can combine the ΔΔ′17O fractionation equations (Eqns. 24 and 26) for quartz–water and
calcite–water to derive the equilibrium oxygen isotope fractionation between quartz and calcite
(Fig. 17). The qqz–cc values are less than those of quartz–water or calcite–water. At 0°C, our
calculated qqz–cc of 0.518 compares well with the theoretical estimate of 0.520 (Hayles et al. 2018).
Figure 17. Δ′17O–δ′18O plot for quartz and calcite and the equilibrium qqz–cc values from 0 to 70 °C.
)LJXUH
We consider three carbonate lithologies that contain authigenic quartz: the well-known
Herkimer Diamond quartz crystals hosted in carbonate (New York, USA), doubly terminated
quartz crystals in early–middle transition Triassic marine pelagic limestone (North Dobrogea,
Romania), and chert nodules from a Cretaceous limestone (Étretat, France). The Herkimer
Diamond and marine pelagic limestone from North Dobrogea, Romania both contain glassy,
optically clear quartz crystals. The Romanian samples do not contain sponge spicules, although
they are found elsewhere in the section, suggesting that the quartz crystals are the product
of dissolution and reprecipitation of the sponge spicules. The Étretat, France sample is not
glassy quartz, but rather fine-grained chert (data from Pack and Herwartz 2014). The other two
samples are data from Sharp et al. (2016) and all data are normalized to the same San Carlos
olivine scale (see Sharp and Wostbrock 2021, this volume).
The two authigenic quartz samples can be explained by dissolution of amorphous silica
and recrystallization as quartz during a period of meteoric water infiltration (Fig. 18). If a
small amount of meteoric water entered the system, it could dissolve the reactive, metastable
amorphous silica and cause it to reprecipitate without appreciably interacting with the host
calcite. Using a simple mass balance model following Equation (29), we can calculate the initial
Δ′17O and δ′18O values of the amorphous silica for both the Romanian and New York samples
with differing F/R ratios and reaction temperatures. For the Romanian samples, the δ18O value
of the calcite is ~33.3‰ (not shown in Fig. 18), corresponding to a formation temperature of
~10 °C. Two reactions paths are illustrated at diagenesis temperatures of 45 and 70 °C using a
starting quartz value in equilibrium with seawater at 10 °C. For the Herkimer samples, alteration
temperatures approaching 100 °C are expected (not shown in Fig. 18). Calculated F/R ratios for
both samples are high, but it is important to recognize that this is the effective F/R ratio. If the
calcite had only minimal interaction with the infiltrating fluid, then the ratio of oxygen in the
fluid to quartz could be high even if there was only a small amount of fluid.
)LJXUH
Figure 18. Isotopic compositions of authigenic euhedral quartz crystals hosted in carbonates. The isotope
data suggest some degree of alteration by meteoric fluids. Pathways for interaction between a sample
formed at ~10 °C (as evidenced by carbonate δ18O values and a meteoric fluid (δ18O = −6‰, Δ′17O = 0.03‰.
Alteration paths shown for temperatures of 45 °C and 70 °C.
The French chert samples plot above the equilibrium curve for quartz and seawater
(Fig. 19). It is not possible to explain this sample in terms of equilibrium with ocean water
or diagenesis with meteoric water. Instead, we suggest a scenario of fluids migrating from
deeper hotter sections into overlying sediments. The overall process shown in Figure 19 is the
following: First, a carbonate is precipitated at 35 °C in equilibrium with seawater at composition
 illustrated by the red diamond. The sample is buried and heated to 100 °C. Interstitial fluid
in equilibrium with the heated carbonate re-equilibrates to an oxygen isotope composition at
 (red pentagon). The curve between  and  is the oxygen isotope equilibrium curve for
carbonates with the origin at the carbonate value (position ; blue solid line). This fluid then
buoyantly infiltrates the overlying sediment at a lower temperature and precipitates quartz at
 (red hexagon). The black curve between  and  is the oxygen isotope equilibrium curve
for quartz with a fluid at . The geological scenario is reasonable and provides an explanation
for how samples can plot above the mineral–seawater equilibrium curve.
)LJXUH
Figure 19. Possible reaction sequence to explain the high Δ′17O value (red circle) of carbonate-hosted
chert from Étretat, France (data from Pack and Herwartz 2014). 1) Carbonates form with an isotopic
composition at . 2) The sample is heated and the pore fluid buffered by the carbonate is shown at . 3)
The fluid infiltrates the cooler overlying sediment and precipitates quartz at . The grey curve shows the
equilibrium quartz–water fractionation with seawater.
CONCLUSION
Recent studies have clearly demonstrated the usefulness in making triple oxygen isotope
measurements compared to measuring the δ18O value alone. Triple oxygen isotope measurements
serve as a unique tool that can determine equilibrium vs. non-equilibrium vs. alteration processes.
Triple oxygen isotope compositions allow for a definitive test for diagenesis. If a sample does not
lie on the equilibrium curve, it is not in equilibrium with modern seawater.
The silica–water triple oxygen isotope fractionation system appears to be well
constrained. Inconsistencies remain in published carbonate–water triple oxygen isotope
fractionation values, probably due to the different methods used to obtain the triple oxygen
isotope values. Nevertheless, application of the calcite–water or quartz–water triple oxygen
isotope fractionation curves to samples where the formation water is no longer present allows
us to constrain the temperatures of formation and conditions of diagenesis. A fluid–rock
mixing model can be used to ‘see through’ alteration and calculate the primary oxygen isotope
compositions of silicate and carbonate rocks. Applying this model to silicates and carbonates
from the geologic record shows that the low δ18O values of the ancient samples do not appear
to be related to changing ocean temperatures and/or ocean isotope values. Instead, Phanerozoic
carbonate and chert samples are better explained by diagenesis and Proterozoic and older chert
can be explained by either diagenesis or precipitation in hydrothermal fluid.
ACKNOWLEDGEMENTS
J.A.G.W and Z.S acknowledges support from NSF GRFP grant DGE-1418062, NSF
EAR 1551226, and NSF EAR 1747703. We thank Jim Jenks, Viorel Atudorei, and UNM
EPS collections for the Smithian and Spathian ammonite samples. The authors also thank
constructive reviews by Weifu Guo and Page Chamberlain and the patience of editors Ilya
Bindeman and Andreas Pack.
REFERENCES
Adkins JF, Boyle EA, Curry WB, Lutringer A (2003) Stable isotopes in deep-sea corals and a new mechanism for
“vital effects”. Geochim Cosmochim Acta 67:1129–1143
Adnew GA, Hofmann ME, Paul D, Laskar A, Surma J, Albrecht N, Pack A, Schwieters J, Koren G, Peters W,
Ahm A-SC, Bjerrum CJ, Blättler CL, Swart PK, Higgins JA (2018) Quantifying early marine diagenesis in shallow-
water carbonate sediments. Geochim Cosmochim Acta 236:140–159
Bajnai D, Fiebig J, Tomašových A, Milner Garcia S, Rollion-Bard C, Raddatz J, Löffler N, Primo-Ramos C, Brand U
(2018) Assessing kinetic fractionation in brachiopod calcite using clumped isotopes. Sci Rep 8:533
Banner JL, Hanson GN (1990) Calculation of simultaneous isotopic and trace element variations during water–rock
interaction with applications to carbonate diagenesis. Geochim Cosmochim Acta 54:3123–3137
Beck WC, Grossman EL, Morse JW (2005) Experimental studies of oxygen isotope fractionation in the carbonic acid
system at 15°, 25°, and 40°C. Geochim Cosmochim Acta 69:3493–3503
Acta 275:36–47
Bigeleisen J, Goeppert Mayer M (1947) Calculation of equilibrium constants for isotopic exchange reactions. J Chem
Phys 15:261–267
Bijma J, Spero HJ, Lea DW (1999) Reassessing foraminiferal stable isotope geochemistry: impact of the oceanic
carbonate system (experimental results). In: Use of Proxies in Paleoceanography: Examples from the South
Atlantic. Fischer G, Wefer G, (eds). Springer Berlin Heidelberg, p 489–512
Nature 557:545–548
Bottinga Y, Javoy M (1973) Comments on oxygen isotope geothermometry. Earth Planet Sci Lett 20:250–265
Brand U (2004) Carbon, oxygen and strontium isotopes in Paleozoic carbonate components: an evaluation of original
seawater-chemistry proxies. Chem Geol 204:23–44
Brand U, Bitner MA, Logan A, Azmy K, Crippa G, Angiolini L, Colin P, Griesshaber E, Harper EM, Ruggiero ET,
Häussermann V (2019) Brachiopod-based oxygen-isotope thermometer: Update and review. Rivista Italiana di
Paleontologia e Stratigrafia 125:775–787
Brandriss ME, O′Neil JR, Edlund MB, Stoermer EF (1998) Oxygen isotope fractionation between diatomaceous
Brock TD (1985) Life at high temperatures. Science 230:132–138
Carlson PE, Noronha AL, Banner JL, Jenson JW, Moore MW, Partin JW, Deininger M, Breecker DO, Bautista KK
(2020) Constraining speleothem oxygen isotope disequilibrium driven by rapid CO2 degassing and calcite
precipitation: Insights from monitoring and modeling. Geochim Cosmochim Acta 284:222–238
Chamberlain CP, Ibarra DE, Lloyd MK, Kukla T, Sjostrom D, Gao Y, Sharp ZD (2020) Triple oxygen isotope systematics
of meteoric hydrothermal systems and implications for paleoaltimetry. Geochem Perspect Lett in press
Clayton RN, Kieffer SW (1991) Oxygen isotopic thermometer calibrations. In: Stable Isotope Geochemistry: A Tribute
to Samuel Epstein. Vol 3. Taylor HPJ, O′Neil JR, Kaplan IR, (eds). Lancaster Press, Inc., San Antonio, p 3–10
Coplen TB (2007) Calibration of the calcite–water oxygen-isotope geothermometer at Devils Hole, Nevada, a natural
laboratory. Geochim Cosmochim Acta 71:3948–3957
Criss RE (1991) Temperature dependence of isotopic fractionation factors. In: Stable Isotope Geochemistry: A Tribute to
Samuel Epstein. Vol 3. Taylor HPJ, O′Neil JR, Kaplan IR, (eds). The Geochemical Society, San Antonio, p 11–16
Crundwell FK (2017) On the mechanism of the dissolution of quartz and silica in aqueous solutions. ACS Omega
2:1116–1127
Cummins RC, Finnegan S, Fike DA, Eiler JM, Fischer WW (2014) Carbonate clumped isotope constraints on Silurian
ocean temperature and seawater δ18O. Geochim Cosmochim Acta 140:241–258
Daëron M, Drysdale RN, Peral M, Huyghe D, Blamart D, Coplen TB, Lartaud F, Zanchetta G (2019) Most Earth-
surface calcites precipitate out of isotopic equilibrium. Nat Commun 10:429
De Ronde CEJ, Channer DMd, Faure K, Bray CJ, Spooner ETC (1997) Fluid chemistry of Archean seafloor hydrothermal
vents: Implications for the composition of circa 3.2 Ga seawater. Geochim Cosmochim Acta 61:4025–4042
Degens ET, Epstein S (1962) Relationship between O18/O16 ratios in coexisting carbonates, cherts, and diatomites 1:
Geological Notes. AAPG Bull 46:534–542
Devriendt LS, Watkins JM, McGregor HV (2017) Oxygen isotope fractionation in the CaCO3–DIC–H2O system.
Dietzel M, Tang J, Leis A, Köhler SJ (2009) Oxygen isotopic fractionation during inorganic calcite precipitation—
Effects of temperature, precipitation rate and pH. Chem Geol 268:107–115
Dodd JP, Sharp ZD (2010) A laser fluorination method for oxygen isotope analysis of biogenic silica and a new oxygen
isotope calibration of modern diatoms in freshwater environments. Geochim Cosmochim Acta 74:1381–1390
Dodd JP, Sharp ZD, Fawcett PJ, Brearley AJ, McCubbin FM (2012) Rapid post-mortem maturation of diatom silica
oxygen isotope values. Geochem Geophys Geosyst 13:Q09014
Eiler JM (2011) Paleoclimate reconstruction using carbonate clumped isotope thermometry. Quat Sci Rev 30:3575–3588
Epstein S, Buchsbaum R, Lowenstam HA, Urey HC (1951) Carbonate–water isotopic temperature scale. Geol Soc
Am Bull 62:417–426
Epstein S, Buchsbaum R, Lowenstam HA, Urey HC (1953) Revised carbonate–water isotopic temperature scale.
Geol Soc Am Bull 64:1315–1326
Farber K, Dziggel A, Meyer FM, Prochaska W, Hofmann A, Harris C (2015) Fluid inclusion analysis of silicified
Palaeoarchaean oceanic crust—A record of Archaean seawater? Precambrian Res 266:150–164
Finnegan S, Bergmann K, Eiler JM, Jones DS, Fike DA, Eisenman I, Hughes NC, Tripati AK, Fischer WW (2011)
The magnitude and duration of late Ordovician–early Silurian glaciation. Science 331:903
Fosu BR, Subba R, Peethambaran R, Bhattacharya SK, Ghosh P (2020) Technical note: Developments and
applications in triple oxygen isotope analysis of carbonates. ACS Earth Space Chem 4:702–710
Gao G, Land LS (1991) Geochemistry of Cambro–Ordovician Arbuckle limestone, Oklahoma: Implications for
diagenetic δ18O alteration and secular δ13C and 87Sr/86Sr variation. Geochim Cosmochim Acta 55:2911–2920
Garrels RM, Christ CL (1965) Solutions, Minerals, and Equilibria. Freeman, Cooper and Co, San Francisco, CA
Gonfiantini R, Panichi C, Tongiorgi E (1968) Isotopic disequilibrium in travertine deposition. Earth Planet Sci Lett 5:55–58
Gregory RT (1991) Oxygen isotope history of seawater revisited: Timescales for boundary event changes in the
oxygen isotope compostion of seawater. In: Stable Isotope Geochemistry: A Tribute to Samuel Epstein. H.P
Taylor J, O′Neil JR, Kaplan IR, (eds). The Geochemical Society, San Antonio, p 65–76
Gregory RT, Taylor HP (1981) An oxygen isotope profile in a section of Cretaceous oceanic crust, Samail Ophiolite,
Oman: Evidence for d18O buffering of the oceans by deep (>5 km) seawater-hydrothermal circulation at mid-
ocean ridges. J Geophys Res 86:2737–2755
Grossman El, Zhang C, Yancey TE (1991) Stable-isotope stratigraphy of brachiopods from Pennsylvanian shales in
Texas. GSA Bulletin 103:953–965
Guex J, Hungerbuhler A, Jenks J, O′Dogherty L, Atudorei V, Taylor D, Bucher H, Bartoloni A (2010) Spathian
(Lower Triassic) ammonoids from western USA (Idaho, California, Utah and Nevada). Memoires de Geologie
(Lausanne) 49:82
Hayles JA, Gao C, Cao X, Liu Y, Bao H (2018) Theoretical calibration of the triple oxygen isotope thermometer.
Herwartz D (2021) Triple oxygen isotope variations in Earth’s crust. Rev Mineral Geochem 86:291–322
Herwartz D, Pack A, Krylov D, Xiao YL, Muehlenbachs K, Sengupta S, Di Rocco T (2015) Revealing the climate of
snowball Earth from D17O systematics of hydrothermal rocks. PNAS 112:5337–5341
Hill PS, Tripati AK, Schauble EA (2014) Theoretical constraints on the effects of pH, salinity, and temperature on
clumped isotope signatures of dissolved inorganic carbon species and precipitating carbonate minerals. Geochim
Hren MT, Tice MM, Chamberlain CP (2009) Oxygen and hydrogen isotope evidence for a temperate climate 3.42
billion years ago. Nature 462:205–208
Isley AE (1995) Hydrothermal plumes and the delivery of iron to Banded Iron Formation. J Geol 103:169–185
controversy and an improved geological water cycle model for the past 3.4 billion years. Earth Sci Rev 83:83–122
Jenks J, Brayard A (2018) Smithian (Early Triassic ammonoids from Crittenden Springs, Elko County, Nevada:
Taphonomy, Biostratigraphy and Biogeography. New Mexico Museum of Natural Sciences Bulletin 78:175 pp.
Johnson BW, Wing BA (2020) Limited Archaean continental emergence reflected in an early Archaean 18O-enriched
ocean. Nat Geosci 13:243–248
Kah L (2000) Depositional δ18O signatures in Proterozoic dolostones: constraints on seawater chemistry and early
diagenesis. In: Carbonate Sedimentation and Diagenesis in the Evolving Precambrian World. Vol 67. Grotzinger
JP, James NP, (eds). SEPM Special Publication, Tulsa, OK, USA, p 345–360
Kanzaki Y (2020) Interpretation of oxygen isotopes in Phanerozoic ophiolites and sedimentary rocks. Geochemistry,
Geophysics, Geosystems 21:e2020GC009000
Karhu J, Epstein S (1986) The implication of the oxygen isotope records in coexisting cherts and phosphates. Geochim
Kasting JF, Tazewell Howard M, Wallmann K, Veizer J, Shields G, Jaffrés J (2006) Paleoclimates, ocean depth, and
the oxygen isotopic composition of seawater. Earth Planet Sci Lett 252:82–93
Keith ML, Weber JN (1964) Carbon and oxygen isotopic composition of selected limestones and fossils. Geochim
Kim ST, O′Neil JR (1997) Equilibrium and nonequilibrium oxygen isotope effects in synthetic carbonates. Geochim
Kita I, Taguchi S, Matsubaya O (1985) Oxygen isotope fractionation between amorphous silica and water at 34–93 °C.
Nature 314:83–84
Knauth LP (1973) Oxygen and hydrogen isotope ratios in cherts and related rocks. PhD California Institute of Technology
Knauth LP (1979) A model for the origin of chert in limestone. Geology 7:274–277
Knauth LP (2005) Temperature and salinity history of the Precambrian ocean: Implications for the course of microbial
evolution. Palaeogeogr Palaeoclimatol Palaeoecol 219:53–69
Knauth LP, Epstein S (1976) Hydrogen and oxygen isotope ratios in nodular and bedded cherts. Geochim Cosmochim
Acta 40:1095–1108
Knauth LP, Lowe DR (1978) Oxygen isotope geochemistry of cherts from the Onverwacht Group (3.4 Ga), Transvaal, South
Africa, with implication for secular variations in the isotopic composition of cherts. Earth Planet Sci Lett 41:209–222
Knauth LP, Lowe DR (2003) High Archean climatic temperature inferred from oxygen isotope geochemistry of cherts
in the 3.5 Ga Swaziland Supergroup, South Africa. GSA Bull 115:566–580
Krauskopf KB (1956) Dissolution and precipitation of silica at low temperatures. Geochim Cosmochim Acta 10:1–26
Leclerc AJ, Labeyrie L (1987) Temperature dependence of the oxygen isotopic fractionation between diatom silica
and water. Earth Planet Sci Lett 84:69–74
Lohmann KC, Walker JCG (1989) The δ18O record of Phanerozoic abiotic marine calcite cements. Geophys Res Lett
16:319–322
Longinelli A, Iacumin P, Ramigni M (2002) δ18O of carbonate, quartz and phosphate from belemnite guards:
implications for the isotopic record of old fossils and the isotopic composition of ancient seawater. Earth Planet
Sci Lett 203:445–459
Longinelli A, Wierzbowski H, Di Matteo A (2003) δ18O(PO43−) and δ18O(CO32−) from belemnite guards from Eastern
Europe: implications for palaeoceanographic reconstructions and for the preservation of pristine isotopic values.
Lowe DR (1983) Restricted shallow-water sedimentation of Early Archean stromatolitic and evaporitic strata of the
Strelley Pool Chert, Pilbara Block, Western Australia. Precambrian Res 19:239–283
Lowell RP, Keller SM (2003) High-temperature seafloor hydrothermal circulation over geologic time and archean
banded iron formations. Geophys Res Lett 30:1391
Lowenstam HA (1961) Mineralogy, O18/O16 ratios, and strontium and magnesium contents of recent and fossil
brachiopods and their bearing on the history of the oceans. J Geol 69:241–260
Loyd SJ, Sample J, Tripati RE, Defliese WF, Brooks K, Hovland M, Torres M, Marlow J, Hancock LG, Martin R,
Lyons T (2016) Methane seep carbonates yield clumped isotope signatures out of equilibrium with formation
temperatures. Nat Commun 7:12274
platinum: An innovative technique for measuring δ17O in CO2. Anal Chem 85:6894–6901
Maliva RG, Knoll AH, Siever R (1989) Secular change in chert distribution; a reflection of evolving biological
participation in the silica cycle. PALAIOS 4:519–532
Maliva RG, Knoll AH, Simonson BM (2005) Secular change in the Precambrian silica cycle: Insights from chert
petrology. GSA Bull 117:835–845
Marin-Carbonne J, Chaussidon M, Robert F (2012) Micrometer-scale chemical and isotopic criteria (O and Si) on
Marin-Carbonne J, Robert F, Chaussidon M (2014) The silicon and oxygen isotope compositions of Precambrian
cherts: A record of oceanic paleo-temperatures? Precambrian Res 247:223–234
Marshall JD (1992) Climatic and oceanographic isotopic signals from the carbonate rock record and their preservation.
Geol Mag 129:143–160
McConnaughey T (1989) 13C and 18O isotopic disequilibrium in biological carbonates: II. In vitro simulation of kinetic
isotope effects. Geochim Cosmochim Acta 53:163–171
Mickler PJ, Banner JL, Stern L, Asmerom Y, Edwards RL, Ito E (2004) Stable isotope variations in modern tropical
speleothems: Evaluating equilibrium vs. kinetic isotope effects Geochim Cosmochim Acta 68:4381–4393
Mickler PJ, Stern LA, Banner JL (2006) Large kinetic isotope effects in modern speleothems. GSA Bull 118:65–81
Mills GA, Urey HC (1940) The kinetics of isotopic exchange between carbon dioxide, bicarbonate ion, carbonate ion
and water 1. J Am Chem Soc 62:1019–1026
Muehlenbachs K (1986) Alteration of the oceanic crust and the 18O history of seawater. Rev Mineral Geochem 16:425–444
Muehlenbachs K (1998) The oxygen isotopic composition of the oceans, sediments and the seafloor. Chem Geol 145:263–273
Muehlenbachs K, Clayton RN (1976) Oxygen isotope composition of the oceanic crust and its bearing on seawater.
O′Neil JR, Clayton RN, Mayeda TK (1969) Oxygen isotope fractionation in divalent metal carbonates. J Chem Phys
51:5547–5558
implications for continental paleoclimate reconstruction. Rev Mineral Geochem 86:429–462
Perry EC (1967) The oxygen isotope chemistry of ancient cherts. Earth Planet Sci Lett 3:62–66
Perry EC, Lefticariu L (2007) 7.05—Formation and geochemistry of precambrian cherts. In: Treatise on Geochemistry.
Holland HD, Turekian KK, (eds). Pergamon, Oxford, p 1–21
Perry EC, Jr., Tan FC (1972) Significance of oxygen and carbon isotope variations in early Precambrian cherts and
carbonate rocks of southern Africa. GSA Bull 83:647–664
Perry EC, Ahmad SN, Swulius TM (1978) The Oxygen Isotope Composition of 3,800 M.Y. Old Metamorphosed
Chert and Iron Formation from Isukasia, West Greenland. J Geol 86:223–239
Peters STM, Szilas K, Sengupta S, Kirkland CL, Garbe-Schönberg D, Pack A (2020) >2.7 Ga metamorphic peridotites from
southeast Greenland record the oxygen isotope composition of Archean seawater. Earth Planet Sci Lett 544:116331
Popp BN, Anderson TF, Sandberg PA (1986) Brachiopods as indicators of original isotopic compositions in some
Paleozoic limestones. Geol Soc Am Bull 97:1262–1269
Price GD, Bajnai D, Fiebig J (2020) Carbonate clumped isotope evidence for latitudinal seawater temperature gradients
and the oxygen isotope composition of Early Cretaceous seas. Palaeogeogr Palaeoclimatol Palaeoecol 109777
Prokoph A, Shields GA, Veizer J (2008) Compilation and time-series analysis of a marine carbonate δ18O, δ13C,
87
Sr/86Sr and δ34S database through Earth history. Earth Sci Rev 87:113–133
Rosing MT, Bird DK, Sleep NH, Bjerrum CJ (2010) No climate paradox under the faint early Sun. Nature 464:744–747
Sagan C, Mullen G (1972) Earth and Mars: Evolution of atmospheres and surface temperatures. Science 177:52–56
Sakai S, Matsuda S, Hikida T, Shimono A, McManus JB, Zahniser M, Nelson D, Dettman DL, Yang D, Ohkouchi N
by tunable infrared laser absorption spectroscopy. Anal Chem 89:11846–11852
oxide, perchlorate, phosphate, silicate, and sulfate minerals. Rev Mineral Geochem 86
Schmidt M, Botz R, Stoffers P, Anders T, Bohrmann G (1997) Oxygen isotopes in marine diatoms: A comparative
study of analytical techniques and new results on the isotope composition of recent marine diatoms. Geochim
Schmidt G, Bigg G, Rohling E (1999) Global seawater oxygen-18 database. http://datagissnasagov/o18data
Schmidt M, Botz R, Rickert D, Bohrmann G, Hall SR, Mann S (2001) Oxygen isotopes of marine diatoms and
relations to opal-A maturation. Geochim Cosmochim Acta 65:201–211
Sengupta S, Pack A (2018) Triple oxygen isotope mass balance for the Earth′s oceans with application to Archean
Sengupta S, Peters STM, Reitner J, Duda J-P, Pack A (2020) Triple oxygen isotopes of cherts through time. Chem
Geol 554:119789
Shackleton NJ, Wiseman JDH, Buckley HA (1973) Non-equilibrium isotopic fractionation between seawater and
planktonic foraminiferal tests. Nature 242:177–179
Sharma T, Clayton RN (1965) Measurement of O18/O16 ratios of total oxygen of carbonates. Geochim Cosmochim
Acta 29:1347–1353
Sharp ZD (2017) Principles of Stable Isotope Geochemistry, 2nd Edition. Open Educational Resources, Albuquerque, NM
Sharp ZD, Wostbrock JAG (2021) Standardization for the triple oxygen isotope system: waters, silicates, carbonates,
air, and sulfates. Rev Mineral Geochem 86:179–196
Shemesh A, Charles CD, Fairbanks RG (1992) Oxygen isotopes in biogenic silica: global changes in ocean
temperature and isotopic composition. Science 256:1434–1436
Shemesh A, Burckle LH, Hays JD (1995) Late Pleistocene oxygen isotope records of biogenic silica from the Atlantic
sector of the Southern Ocean. Paleoceanography 10:179–196
Silverman DN (1973) Carbonic anhydrase catalyzed oxygen-18 exchange between bicarbonate and water. Arch
Biochem Biophys 155:452–457
Spero HJ, Bijma J, Lea DW, Bemis BE (1997) Effect of seawater carbonate concentration on foraminiferal carbon and
oxygen isotopes. Nature 390:497–500
Sun Y, Joachimski MM, Wignall PB, Yan C, Chen Y, Jiang H, Wang L, Lai X (2012) Lethally hot temperatures during
the early Triassic greenhouse. Science 338:366–370
Swart PK (2015) The geochemistry of carbonate diagenesis: The past, present and future. Sedimentology 62:1233–1304
Taylor HP (1978) Oxygen and hydrogen isotope studies of plutonic granitic rocks. Earth Planet Sci Lett 38:177–210
Thiemens MH (2006) History and applications of mass-independent isotope effects. Ann Rev Earth Planet Sci 34:217–262
Tice MM, Lowe DR (2006) The origin of carbonaceous matter in pre-3.0 Ga greenstone terrains: A review and new
evidence from the 3.42 Ga Buck Reef Chert. Earth Sci Rev 76:259–300
Tice M, Bostick B, Lowe D (2004) Thermal history of the 3.5–3.2 Ga Onverwacht and Fig Tree Groups, Barberton
greenstone belt, South Africa, inferred by Raman microspectroscopy of carbonaceous material. Geology 32:37–40
Tremaine DM, Froelich PN, Wang Y (2011) Speleothem calcite farmed in situ: Modern calibration of δ18O and δ13C
paleoclimate proxies in a continuously-monitored natural cave system. Geochim Cosmochim Acta 75:4929–4950
Tyler JJ, Sloane HJ, Rickaby REM, Cox EJ, Leng MJ (2017) Post-mortem oxygen isotope exchange within cultured
diatom silica. Rapid Commun Mass Spectrom 31:1749–1760
Uchikawa J, Zeebe RE (2012) The effect of carbonic anhydrase on the kinetics and equilibrium of the oxygen
isotope exchange in the CO2–H2O system: Implications for δ18O vital effects in biogenic carbonates. Geochim
Urey HC, Epstein S, McKinney CR (1951) Measurement of paleotemperatures and temperatures of the Upper Cretaceous
of England, Denmark, and the southeastern United States. Geol Soc Am Bull 62:399–416
Veizer J, Hoefs J (1976) The nature of O18/O16 and C13/C12 secular trends in sedimentary carbonate rocks. Geochim
Veizer J, Prokoph A (2015) Temperatures and oxygen isotopic composition of Phanerozoic oceans. Earth Sci Rev 146:92–104
Veizer J, Fritz P, Jones B (1986) Geochemistry of brachiopods: Oxygen and carbon isotopic records of Paleozoic
oceans. Geochim Cosmochim Acta 50:1679–1696
Veizer J, Hoefs J, Lowe DR, Thurston PC (1989) Geochemistry of Precambrian carbonates: II. Archean greenstone
belts and Archean sea water. Geochim Cosmochim Acta 53:859–871
Veizer J, Bruckschen P, Pawellek F, Diener A, Podlaha OG, Carden GA, Jasper T, Korte C, Strauss H, Azmy K, Ala D
(1997) Oxygen isotope evolution of Phanerozoic seawater. Palaeogeogr Palaeoclimatol Palaeoecol 132:159–172
Veizer J, Ala D, Azmy K, Bruckschen P, Buhl D, Bruhn F, Carden GA, Diener A, Ebneth S, Godderis Y, Jasper T (1999)
87
Sr/86Sr, δ13C and δ18O evolution of Phanerozoic seawater. Chem Geol 161:59–88
Wadleigh MA, Veizer J (1992) 18O16O and 13C12C in lower Paleozoic articulate brachiopods: Implications for the
isotopic composition of seawater. Geochim Cosmochim Acta 56:431–443
Wallmann K (2001) The geological water cycle and the evolution of marine δ18O values. Geochim Cosmochim Acta
65:2469–2485
Watkins JM, Nielsen LC, Ryerson FJ, DePaolo DJ (2013) The influence of kinetics on the oxygen isotope composition
of calcium carbonate. Earth Planet Sci Lett 375:349–360
Wenzel B, Lécuyer C, Joachimski MM (2000) Comparing oxygen isotope records of silurian calcite and phosphate—
δ18O compositions of brachiopods and conodonts. Geochim Cosmochim Acta 64:1859–1872
Winter BL, Knauth LP (1992) Stable isotope geochemistry of cherts and carbonates from the 2.0 Ga gunflint
iron formation: implications for the depositional setting, and the effects of diagenesis and metamorphism.
Wostbrock JAG, Brand U, Coplen TB, Swart PK, Carlson SJ, Sharp ZD (2020b) Calibration of carbonate–water triple oxygen
isotope fractionation: seeing through diagenesis in ancient carbonates. Geochim Cosmochim Acta 288:369–388
Yanchilina AG, Yam R, Kolodny Y, Shemesh A (2020) From diatom opal-A δ18O to chert δ18O in deep sea sediments.
Zakharov DO, Marin-Carbonne J, Alleon J, Bindeman IN (2021)Triple oxygen isotope trend recorded by Precambrian cherts:
a perspective from combined bulk and in situ secondary ion probe measurements. Rev Mineral Geochem 86:323–365
Zeebe RE (1999) An explanation of the effect of seawater carbonate concentration on foraminiferal oxygen isotopes.
Zeebe RE (2014) Kinetic fractionation of carbon and oxygen isotopes during hydration of carbon dioxide. Geochim
Zheng Y-F (1991) Calculation of oxygen isotope fractionation in metal oxides. Geochim Cosmochim Acta 55:2299–2307
Vol. 86 pp. 401–428, 2021 12
Triple Oxygen Isotope Systematics

in the Hydrologic Cycle
Jakub Surma
Institute of Geology and Mineralogy
University of Cologne
Zülpicher Straße 49b
Cologne, 50674
Germany
jakub.surma@uni-goettingen.de
presently at:
Geoscience Center
Georg August University
Göttingen, 37077
Germany
Sergey Assonov
formerly at:
Cologne, 50674
Germany
Michael Staubwasser
Cologne, 50674
Germany
INTRODUCTION
The analysis of hydrogen (δD) and oxygen (δ18O) isotope ratios of H2O are widely used
tools for studies of the hydrological cycle (Friedman 1953; Dansgaard 1954; Gonfiantini
1986; Gat 1996; Araguás-Araguás et al. 2000; Gat et al. 2000) and climate reconstruction
(Dansgaard 1964; Johnsen et al. 1989; Petit et al. 1999). Natural variations of δD and δ18O in
precipitation are well correlated and fall on a common global trend, the Global Meteoric Water
Line (GMWL, Craig 1961):
D 
8   18 O 10‰ (1)
The slope (  2GMWL ) of 8 in this equation is defined by mass-dependent equilibrium fractionation

during condensation of atmospheric vapor (Dansgaard 1964; Horita and Wesolowski 1994).

1943-2666/21/0086-0012$05.00 (online)
402 Surma et al.
The 10‰ positive offset in δD results from kinetic (diffusive) isotope fractionation during the
initial formation of atmospheric vapor over the ocean which follows a slightly lower δD/δ18O
slope than the GWML (Craig and Gordon 1965; Merlivat 1978). Typically, any evaporation
process in the hydrologic cycle results in waters that deviate from the average global  2GMWL.
This deviation is expressed with the d-excess parameter (e.g., Gat 1996):
d-excess D  8   18 O (2)
The process of isotopic fractionation during vapor formation is explained by the evaporation
model of Craig and Gordon (1965) (Fig. 1). The model describes the total isotopic fractionation
(*aL–V–evap) resulting from evaporation from a water body (*RW), which is controlled by relative
humidity (h), fractionation factors for equilibrium and diffusion, and the isotopic composition
of free atmospheric vapor (*RA). At lower h, the vapor concentration gradient between the
interface layer (vapor saturation) and the free atmosphere increases and amplifies diffusivity.
Hence, the magnitude of d-excess is mostly controlled by relative humidity (h) and becomes
larger when h decreases (Craig and Gordon 1965; Cappa et al. 2003; Steen-Larsen et al.
2014). Therefore, d-excess is traditionally used to reconstruct evaporation conditions at vapor
source regions and to quantify evaporation of continental water bodies (Gat and Bowser 1991;
Gat 1996; Masson-Delmotte et al. 2005; Pfahl and Wernli 2008).
Figure 1. The Craig and Gordon Model for evaporation of surface water (modified from Gat 1996).
However, the d-excess parameter is not exclusively controlled by h but also by the different
temperature sensitivity of equilibrium fractionation factors for D/H (2aL–V–eq) and 18O/16O
(18aL–V–eq). This prevents a unique interpretation by introducing uncertainty to h estimates
(Horita and Wesolowski 1994; Luz et al. 2009). Also, the diffusivity fractionation factors (2adiff
and 18adiff, respectively) of both isotope systems underlie a temperature dependent relationship
(Luz et al. 2009):
2
 diff  1.25  0.02  T    18

 diff  1  1 (3)
Due to the different temperature relationships of D/H and 18O/16O fractionation factors
for liquid–vapor and solid–vapor equilibrium (2aS–V–eq and 18aS–V–eq, respectively), d-excess is
additionally affected by precipitation temperature (Merlivat and Nief 1967; Majoube 1971; Ellehoj
et al. 2013). The additional combined analysis of 17O/16O and 18O/16O of H2O was suggested to be
a promising complement to reconstruct relative humidity at the vapor source (Angert et al. 2004).
Analysis
Due to methodical limitations and analytical uncertainties, the use of triple oxygen isotopes
in hydrological applications was impractical in the past. However, the temperature dependency
Systematics in the Hydrologic Cycle 403
of d-excess created the need for another second order parameter of water isotopologues.
In parallel, methodical advances over the past two decades provided the necessary analytical
resolution to identify mass-dependent anomalies in the 16O–17O–18O distribution in natural
waters (Meijer and Li, 1998; Baker et al. 2002). The latter method was greatly advanced by
Barkan and Luz (2005). An analytical uncertainty of ±0.005‰ for 17O-excess was achieved
based on liberation of oxygen gas from H2O by means of fluorination with a CoF3 reagent,
providing an adequate precision to resolve small variations in the hydrosphere and to
determine H217O/H216O equilibrium and diffusivity fractionation factors (17aL–V–eq and 17adiff,
respectively) with a sufficient accuracy (Barkan and Luz 2005, 2007). Over the past decade,
laser absorption instruments have been developed and also proven to achieve sufficiently
high precision to resolve 17O-excess variations in natural waters (Berman et al. 2013;
Steig et al. 2014; Affolter et al. 2015; Schauer et al. 2016; Tian et al. 2016).
Analytical requirements: Two-point calibration. Reported d18O in natural waters so far
ranges from −69.6‰ (East Antarctic snow) to 29.2‰ (Sistan Desert, Iran) and, thus, covers a total
range of approximately 100‰ (Surma et al. 2015; Touzeau et al. 2016). In order to account for
inter-laboratory scaling differences, a two-point calibration based on certified reference materials
is essential (e.g., Kaiser 2008; Kusakabe and Matsuhisa 2008). A commonly used normalization
in triple oxygen isotope studies is the one suggested by Schoenemann et al. (2013):
 * OSLAP/VSMOW
nom
 * O smp
 _ VSMOW  SLAP  * Osmp_VSMOW  (4)
 OSLAP/VSMOW
* meas
where d*Osmp_VSMOW is the measured value of a sample expressed against VSMOW,

 * OSLAP/VSMOW
nom
is the assigned distance between SLAP−2 and VSMOW−2, and  * OSLAP/VSMOW
meas
is the difference between SLAP−2 and VSMOW−2 in the respective laboratory. Assigned
values for SLAP−2 (vs. VSMOW−2) are  17 OSLAP/VSMOW nom
= −29.6986‰ and
 OSLAP/VSMOW = −55.5‰ ( O-excess = 0 per meg).
18 nom 17
New data presented in this work

This review is supplemented with new samples and data from one case study in the high-
altitude environment of the Southern German Alps. Sampling was carried out at Mt. Zugspitze
(2962 m above sea level, m a.s.l.), 10 km SW of Garmisch-Partenkirchen (720 m a.s.l.).
The temperate regional climate is characterized by a mean annual temperature of −4.3 °C.
Samples were collected from February to May 2016 in the local surrounding of Mt. Zugspitze—
at the Schneefernerhaus Research Station (UFS, 2,650 m a.s.l.) and the nearby Zugspitze plateau
(Zugspitzplatt, N 10.99°, E 47.41°; ~2,450 m a.s.l.). The sample set comprises 25 atmospheric
vapor samples (Table S1) that were collected in February and May 2016 by means of cryogenic
extraction at UFS station in LN2 at low pressure in a modified version of a high efficiency trap
(Brenninkmeijer and Röckmann 1996). We also obtained a total number of 25 precipitation/snow
samples and a larger number of samples from the seasonal local snow cover (Tables S2 and S3).
Isotope analysis. All samples were analyzed for their triple oxygen isotope and hydrogen
isotope composition at the Institute for Geology and Mineralogy, University of Cologne.
17
O/16O and 18O/16O analysis is performed with modified method based on Barkan and Luz
(2005). Details on the analytical procedure are given in Surma et al. (2015, 2018). Long-term
external reproducibility of standard waters and selected samples is approx. ±0.06‰ (δ17O),
±0.1‰ (δ18O), and ±8 per meg (17O-excess). D/H ratios were determined by continuous flow
analysis of H2 gas liberated by TC/EA carbon reduction of H2O at 1550 °C (also see Surma
et al. 2018). The average long-term external reproducibility is approximately ±0.3‰ for δD
and ± 2.1‰ for d-excess. Laboratory reference waters were analyzed between every 5 to 10
samples. All isotope data were normalized to the SMOW-SLAP scale following the procedure
suggested by Schoenemann et al. (2013) and standard-sample bracketing (Table S4).
404 Surma et al.
TRIPLE OXYGEN ISOTOPES IN WATER

Triple oxygen isotope fractionation
The triple oxygen isotope fractionation exponent for equilibrium liquid–vapor fractionation
(qL–V–eq) was determined by Barkan and Luz (2005), giving a constant qL–V–eq of 0.529 for the
entire experimental T range from 11.4 to 41.5 °C. Therefore, 17O-excess can be considered
temperature-independent for equilibrium fractionation of evaporation from water surfaces and
formation of rain. A difficulty arises from the fact that no experimental data are available for
triple oxygen isotope fractionation at solid–vapor equilibrium of H2O. Theoretical predictions
suggest qS–V–eq ≈ 0.528 (Van Hook 1968). Landais et al. (2012b) confirm this value with a set
of vapor and snowfall samples from Greenland.
Molecular diffusion of water vapor in air is characterized by a somewhat lower qdiff
of 0.5185 ± 0.0002 (Barkan and Luz 2007). However, the experiment has shown a small
temperature dependence in the range from 0.5183 ± 0.0001 (25 °C) to 0.5187 ± 0.0001 (40 °C).
It is unclear whether this qdiff/T trend is linear and whether a lower qdiff would apply for
diffusive fractionation at low temperatures (e.g., 0.5172 at −50 °C if the trend is extrapolated
linearly; Bao et al. 2016).
17
O-excess during the formation of vapor
In the field of hydrology 17O-excess is the commonly used term to describe deviations
of 17O/16O ratios from a reference slope with  GMWL = 0.528 and typically reported in per meg
17
with respect to VSMOW (Meijer and Li, 1998; Luz and Barkan 2010):
17
O-excess (per meg)  ( 17 O  0.528   18 O)  106 (5)
with δ′*O = ln(δ*O+1). 
17
GMWLhere is defined by the slope of average global precipitation—
represented by the Global Meteoric Water Line—in the δ′17O vs. δ′18O space. The GMWL
itself shows a positive offset of ~33 per meg with respect to seawater (Luz and Barkan 2010).
The exact magnitude—or slope (Fig. 2)—of *aL–V–evap is determined by h, respective
equilibrium and diffusion fractionation factors, and the isotopic composition of vapor in the
free atmosphere (cf. Fig. 1; Craig and Gordon 1965; Ehhalt and Knott 1965; Criss 1999;
Cappa et al. 2003):
*
 diff  *  L-V_eq  1  h 
*
 L-V_evap  (6)

1  *  L-Veq  h  * RA / * RW 
For a single moisture source (e.g., open ocean) and absence of vapor admixture, all
atmospheric vapor is generated from the evaporating water itself and *RV =*RV (closure
assumption). In that case Equation (6) reduces to:
*
 L-V_evap
 *

 L-V_eq  *  diff  1  h   h  (7)
The temperature of ambient air may deviate from the temperature of the water surface
in natural settings. Here, normalized humidity (hn) should be used in Equations (6) and (7)
instead of h. hn is calculated by:
qsat_a
hn  h  (8)
qsat_s
where qsat_a is the vapor concentration in the free air and qsat_s is the saturated vapor
concentration at water temperature (Barkan and Luz 2007; Uemura et al. 2010).
Figure 2. Fractionation of triple oxygen isotopes in the atmosphere (modified from Barkan and Luz 2007).
Initial atmospheric vapor (white diamond) forms from sea water (black diamond) at non-equilibrium
conditions. Evaporation results in positive lower δ′17O and δ′18O, and higher 17O-excess of vapor with
respect to the reference slope (blue). Transport, cooling, and rainout leads to steady isotopic depletion of
the residual vapor reservoir.
In addition, air turbulence—or wind—reduce the magnitude of diffusive isotope

fractionation during the evaporation process. In the Craig and Gordon Model, this effect is
described by the turbulently mixed sublayer, in which actual mixing of air with different
vapor concentration balances the h gradient without molecular diffusion (Fig. 1). However,
the derivation of a clear relationship between wind speed and lowering of molecular diffusion
is complex (Merlivat and Jouzel 1979). A recent study demonstrates that, we cite: ‘for
isotope water balance studies where winds are frequently above 2 m/s, the C–G model may
be inadequate without appropriate corrections for spray vaporization, or the introduction
of appropriate kinetic isotope fractionation factors’ (Gonfiantini et al. 2020). This would be
strongest felt over water bodies deep enough to sustain breaking waves. In order to account for
boundary layer turbulence, the effect is usually parametrized in the Craig and Gordon model
by correcting *adiff with the exponent n:
 
n
*
 diff  corr   *
 diff (9)
n is determined empirically and ranges from 1 (pure molecular diffusion) to 0.5 (rough
continental regime) (Dongmann et al. 1974; Mathieu and Bariac 1996; Haese et al. 2013).
Reduction of diffusive fractionation with n < 0.5 is observed for wind tunnel experiments and
evaporation above the ocean surface (Merlivat and Jouzel 1979; Uemura et al. 2010).
NATURAL VARIATIONS OF 17O-EXCESS IN WATER

Meaning and purpose of the Global Meteoric Water Line
δD and δ18O in meteoric waters around the world are tightly correlated and follow the
δD/δ18O trend (l2) with a slope of 8 (Eqn. 1), i.e., the GMWL (Craig 1961; Dansgaard 1964).
The +10‰ offset (d-excess) results from diffusive, non-equilibrium fractionation during
formation of pristine vapor above the ocean (Eqn. 6), resulting in relative preferential
enrichment of deuterium with respect to 18O in the vapor phase. Precipitation that forms from
that vapor follows the GMWL towards more depleted values with increasing distance from the
coast, higher latitude, and increasing altitude, whereby the underlying process is approximated
by Rayleigh distillation (Dansgaard 1964; Horita et al. 2008). The GMWL concept provides an
empirical reference frame for the field of stable isotope hydrology, but is in general agreement
with Rayleigh fractionation (Dansgaard 1964; Criss 1999).
406 Surma et al.
Meijer and Li (1998) analyzed a set of natural waters yielding 17

GMWL = 0.528, thus providing
an analogous reference frame for 17O/16O and 18O/16O. Including standard measurements of
GISP, SLAP, and Antarctic snow data (Barkan and Luz 2005; Landais et al. 2008) to their
larger dataset of natural waters (n = 52), Luz and Barkan (2010) confirmed this slope for the
GMWL with a positive offset of 0.033‰ with respect to VSMOW−2:
´ 17 O  0.528  ´ 18 O  0.033‰ (10)
Similar to d-excess, the deviation from this line is reported as 17O-excess (cf. Eqn. 5).
The value of  GMWL = 0.528 is close to θL–V_eq = 0.529, thus confirming Rayleigh equilibrium
17
condensation as the controlling mechanism of precipitate formation (Passey and Levin 2021,
this volume). Figure 3 shows a compilation of reported triple oxygen isotope data of global
precipitation, surface water bodies, hydrothermal waters, and seawater, that cover a total range
of ≈ 100‰ in δ18O. In the case of closure ( * RV0  * RA0 , cf. Eqn. 7), vapor isotopic composition
( * RV0 ) is calculated by (Craig and Gordon 1965):
1

*
RV0 *
RW0  *
 
 L-V_eq   diff  h  1  *  diff
*
 (11)
where * RW0 is the isotopic composition of ocean surface water. When the air parcel is advected
and cooled after initial vapor formation, first condensate forms in equilibrium with * RV0 :
*
R
P0
*
RV0  *  L-V_eq (12)
The residual fraction (fres) of initial oceanic vapor decreases with increasing distance from
the source. The isotopic composition of residual vapor is calculated by the Rayleigh equation:
*
 1

*
RA *
RV0  fres L-V_eq (13)
Precipitating water in equilibrium results in observed  2GMWL= 8 and 17GMWL = 0.528. For
the triple oxygen isotope space this slope is valid for δ18O ranges between −40‰ and −5‰
(Fig. 3), thus providing a reasonable reference frame for global precipitation that is mostly
provided by Rayleigh distillation of oceanic vapor (e.g., Trenberth et al. 2007).
The bell-shaped distribution of 17O-excess vs. δ18O indicates that water samples found in
the lower range of the δ18O scale (polar snow) as well as in highly enriched samples (surface
waters from arid regions) are systematically lower in 17O-excess with respect to the GMWL
(Fig. 3). We note that the apparent bell-shape to some degree may also result from a bias
towards samples that are studied precisely because of their inherent kinetic fractionation
effects found in polar and hyper-arid regions. The vast majority of continental waters not too
far removed from the ocean source will likely fall in the δ18O range from −30‰ to −5‰.
In the case of polar snow, the systematic deviation from the GMWL leads to pronounced
Local Meteoric Water Lines (LMWL) > 0.528—i.e., showing positive correlation of 17O-excess
and δ18O—that result from snow formation at vapor supersaturation, and thus kinetic
fractionation, at temperatures below −20 °C (Landais et al. 2012a,b; Casado et al. 2016).
Miller (2018) states that this pattern of snow with δ18O below −40‰ is also a result of
‘diamond dust’ (open-sky precipitation) contribution to central Antarctic snow cover, which
is characterized by a fractionation slope of l17 = 0.531 in this study. It is also suggested that
the LMWL in remote polar regions may be additionally affected by incorporation of mass-
independently, fractionated, largely 17O-enriched stratospheric oxygen in precipitation (Winkler
et al. 2013; Miller 2018). Evidence for stratospheric water intrusions is suggested by negative
correlation of 17O-excess and δ18O found in the study which contradict the general synoptic
Figure 3. Compilation of 17O-excess vs. δ18O in global waters. The gray horizontal line depicts the GMWL
as reported by Luz and Barkan (2010). Precipitation in polar regions (diamonds) show values below −24‰
in δ18O. 17O-excess ranges from 30 to 70 per meg in snow from Greenland and Alert, Canada, and shows
correlation with decreasing δ18O (below −40‰) for Antarctic precipitation (Landais et al. 2008, 2012a,b;
Lin et al. 2013a; Winkler et al. 2013; Pang et al. 2015, 2019; Touzeau et al. 2016). Fresh waters (gray
circles) comprise snowfall, rain, cave drip water, tap water, lakes, springs, and rivers (Landais et al. 2010;
Luz and Barkan 2010; Lin et al. 2013a; Affolter et al. 2015; Li et al. 2015, 2017; Gázquez et al. 2018; Tian
et al. 2018; Alexandre et al. 2019; Passey and Ji, 2019; Tian et al. 2019; Uechi and Uemura 2019; Bergel
et al. 2020; Bershaw et al. 2020; Sha et al. 2020; Voigt et al. 2020; this study). In general, those waters
range from −22‰ to ~0‰ in δ18O and show a large variability in 17O-excess, ranging from −40 to 120
per meg. Water bodies from arid, evaporative environments (yellow crosses) generally show δ18O values
higher than −8‰ and progressively lower 17O-excess with increasing δ18O. Lowest 17O-excess values are
found in highly evaporated water bodies from the Sistan Desert, Iran, the Atacama Desert, Chile, and Lake
Chichancanab, Mexico (Surma et al. 2015, 2018; Evans et al. 2018; Voigt et al. 2020). Hydrothermal waters
(blue triangles) are in good agreement with freshwaters and show a narrow range in δ18O from −12‰ to
−6‰ and from −20 to 45 per meg in 17O-excess (Sharp et al. 2016; Wostbrock et al. 2018; Zakharov et al.
2019b). Seawater measurements are indicated with asterisks (Luz awnd Barkan 2010).
positive correlation in Antarctic precipitation. We note that such effects at mid-latitudes, which
is the major focus of this review, are much less likely. First, stratospheric air is brought to the
troposphere mostly by the polar vortex. The stratosphere-troposphere exchange flux at mid-
latitudes is very limited. Second, the troposphere in the mid-latitudes has a much higher water
content than the polar troposphere, so the mass-balance may be too unfavorable to observe
stratospheric downdraft. Third, 17O-excess found in snow at polar regions demonstrates only
limited deviations, if any, from the range expected (Landais et al. 2012a; Touzeau et al. 2016).
Isotopically enriched samples from arid regions do not reflect fractionation effects during
precipitate formation but a systematic decrease of 17O-excess during intensive evaporation of
continental waters (Surma et al. 2015, 2018; Voigt et al. 2020). Here the negative correlation
of 17O-excess and δ18O in those waters is the result of excessive evaporation, where total
fractionation integrated over the course of a day is controlled by relative humidity, evaporation
degree, and evaporative loss with respect to recharge. Resulting trajectories in the δ′17O vs.
δ′18O space are flatter than  GMWL and analog to local evaporation lines (LEL) known in the
17
traditional δD vs. δ O system (Fontes and Gonfiantini 1967; Gat 1984; Gat and Bowser 1991).
18
However, it should be noted that triple oxygen isotopes in natural evaporative systems do not
depict straight lines but typically form curved trajectories.
408 Surma et al.
17
O-excess in polar snow precipitation. Ice cores from Greenland and Antarctica are
important records of past climate over glacial-interglacial cycles (e.g., Petit et al. 1999;
NorthGRIP members 2004). The d-excess parameter is widely used to reconstruct variations
in temperature and relative humidity at oceanic moisture sources (e.g., Vimeux et al. 1999;
Steen-Larsen et al. 2011). However, the use of 17O-excess shows promise as a novel and
more robust temperature independent tracer that provides additional constraints on source
region humidity (Uemura et al. 2010).
Landais et al. (2012b) investigated seasonal triple oxygen isotope variations in surface
snow at the NEEM drilling site, Greenland, where a slight anticorrelation of δ18O and
17
O-excess in snow is found. Maximum 17O-excess is present in samples with lowest δ18O
values. Measurements of 17O-excess and δ18O in atmospheric vapor and concomitant snow
at NEEM site show similar values in both phases that were also found to be in equilibrium
based on previous δ18O and δD measurements (Steen-Larsen et al. 2011; Landais et al. 2012b).
Modeling of air temperature and relative humidity in moisture source regions also reveals that
17
O-excess reflects h conditions at the vapor source. Triple oxygen isotope studies on Greenland
ice cores further provide valuable information on the coupling between high-latitude and low-
to mid-latitude climate variations in the northern hemisphere during abrupt climatic changes
over the course of the last glaciation (Guillevic et al. 2014; Landais et al. 2018).
For the case of Antarctica, Winkler et al. (2012) demonstrate that moisture source
humidity can be reliably reconstructed from coastal ice core records, whereas the isotopic
composition of snow in continental areas is dominated by local temperature effects. This is
also confirmed by a spatial gradient which is observed for 17O-excess in precipitation across
Antarctica (Fig. 3), showing elevated values in coastal areas that are dominated by marine
conditions and low 17O-excess in the Antarctic interior (Landais et al. 2012a; Schoenemann
et al. 2014; Touzeau et al. 2016). Low 17O-excess in continental Antarctic precipitation is
attributed to seasonal variations, variability in sea ice extent, and supersaturation of vapor over
ice at low temperatures (Risi et al. 2013; Schoenemann et al. 2014).
Supersaturation of vapor over ice. In order to interpret polar precipitation records, one needs
to account for the supersaturation effect at cloud temperatures < −20 °C (Jouzel and Merlivat 1984;
Angert et al. 2004). The effect considers diffusive fractionation that occurs due to supersaturation
of vapor over ice during precipitate formation which reduces the effective fractionation:

*
 eff *
 S-V_eq  *  kin (14)
where akin is the kinetic fractionation factor for vapor supersaturation over ice and given by
*
(Jouzel and Merlivat 1984; Angert et al. 2004):
S
*
 kin  (15)
*
S-V_eq   diff   S  1  1
*
The supersaturation coefficient (S) determines the relative contribution of diffusive and
equilibrium fractionation during ice formation from vapor. S is dimensionless and a linear function
of temperature, T (in °C), and the site-specifically adjusted factor m (Jouzel and Merlivat 1984):
S = 1 − m∙T (16)
Typical values for m in polar regions range from 0.001 to 0.008 (e.g., Masson-Delmotte
et al. 2005; Steen-Larsen et al. 2011; Landais et al. 2012a, b). Higher m values (and thus
higher S) increase the diffusive contribution, resulting in a steeper  GMWL as it is observed
17
in low δ O snow (Fig. 3). m = 0 (or S = 1) will accordingly result in pure equilibrium
18
fractionation. Even though Arctic precipitation shows positive 17O-excess with respect to
continental Antarctic precipitation, supersaturation of vapor over ice is also identified as a

driving mechanism for 17O-excess at the NEEM drilling site, Greenland (Landais et al.
2012b). On the other hand, elevated supersaturation leads to lower  2GMWL (d-excess) in polar
precipitation. This is explained by a much lower δD-δ18O slope of solid–vapor fractionation
at low temperatures compared to the average equilibrium value of ≈ 8 (Landais et al. 2012a).
17
O-excess during strong evaporation in arid environments. Since evaporation results
in a positive 17O-excess in the vapor phase, residual water that experiences evaporation is left
with a negative 17O-excess with respect to the initial water (Fig. 4). Based on the Craig and
Gordon Model and Equation (6), the isotopic evolution of an evaporating water body can be
predicted as a function of h and * RA. Two general pathways of evaporating water bodies can be
distinguished (Fig. 5), based on their hydrological setting that either excludes (Eqns. 17 to 19)
or includes (Eqns. 20 and 21) groundwater recharge.
Figure 4. Fractionation of triple oxygen isotopes in an evaporating water body. The initial water body
(black diamond) evolves along the red trend after experiencing evaporation. The resulting trajectory is
flatter than the reference slope (blue), causing a negative 17O-excess in the residual water (red diamonds)
with respect to the initial water. Secondary vapor (white diamonds) which is formed by continental evapo-
ration has a positive 17O-excess with respect to the liquid phase.
Figure 5. Schematics of oxygen isotope distribution during evaporation of a continental water body in dif-
ferent hydrological settings (modified from Surma et al. 2018). Simple (pan) evaporation (solid black line)
follows a distinct evaporation trajectory towards the isotopic end-point (* RSS, diamond).* RSS is balanced
by the evaporative isotope flux and equilibrium exchange (Criss 1999). The value of * RSS is determined by
*
RA and h (see Eqn. 18). If the water body is additionally balanced by groundwater recharge, its isotopic
composition will assume a steady state that falls below the pan evaporation trend (solid blue circles),
depending on the E/I value. Increasing E/I leads to higher δ18O and lower 17O-excess—up to E/I = 1 (i.e. a
stable terminal lake; open blue circle). Non-steady state admixture of initial water results in mixing lines
(dotted green line) below evaporation trends, see also Voigt et al. (2020).
410 Surma et al.
If the boundary parameters (* RA, wind turbulence, isotopic composition of initial water)
are well constrained, the isotopic compositions of water bodies in a non-recharged lake
system (* RW) provide information on average annual h of an arid region (Surma et al. 2015).
The isotopic composition of a non-recharged water body (* RW) is calculated by (Criss 1999):
*
RW  
fresu  * RWI  * RSS  * RSS (17)
where * RWL is the isotopic composition of the initial water and * RSS is the isotopic stationary
state (end-point) which is dictated by * RA and h:
*
 L-V_eq  h  * RA
*
RSS  (18)
1  *  0L-V_evap  1  h 
The exponent u describes the liquid–vapor fractionation factor as a function of h:
1  *  0L-V_evap  1  h 
u (19)
 L-V_evap  1  h 
* 0
where  L-V_evap is the total fractionation factor at a hypothetical relative humidity of h = 0.

* 0
For h = 0 Equation (7) reduces to: *  0

L-V_evap
*
 L-V_eq  *  diff . The isotopic composition of a
*
water body in steady state ( RWS) is calculated as a function of evaporation over inflow (E/I)
by (Criss 1999):
 0L-V_evap  1  h   * RWI  *  L-V_eq  h  E  * RA

*
*
RWS  I (20)
E  *  0L-V_evap  1  h   1  E
I I  
In the case of a terminal lake setting (E/I = 1), Equation (20) reduces to:
*
RWS *
 0L-V_evap  1  h   * RWI  *  L-V_eq  h  E  * RA (21)
I
A case study on saline ponds in the hyper-arid Atacama Desert demonstrates that isotopic
compositions of recharged water bodies (* RWS) can be accurately modeled for given boundary
conditions. If h and * RA are monitored independently, 17O-excess—in combination with
d-excess—is a useful tracer for the assessment of lake balances and h and parametrization
of wind effects (Surma et al. 2018). A recent study shows, that unlike in classic δD and δ18O
isotope hydrology, δ17O allows for the separation of mixing and evaporation (Voigt et al. 2020).
Triple oxygen isotopes in rainfall and mid-latitude snow. Li et al. (2015) show a strong
latitude effect in 17O-excess in tap waters across the US. The samples reflect the integrated average
isotopic composition of regional precipitation. Lowest 17O-excess values were found around the
Gulf of Mexico (–6 to 12 per meg), whereas highest values were found in Northwestern U.S.
(31 to 43 per meg). In general, these variations are mainly a product of rain amount effects, re-
evaporation of meteoric water, and mixing with groundwater. Humidity conditions at precipitation
moisture sources do not explain the observed variability, therefore the authors stress that local,
continental moisture recycling may indeed be responsible for elevated 17O-excess in precipitation,
as discussed by other authors (Landais et al. 2010; Li et al. 2015; Tian et al. 2019). Bershaw et
al. (2020) also observe slightly elevated 17O-excess in surface waters from Northwestern U.S.
and suggest an effect of analytical uncertainty and/or the admixture of stratospheric vapor as
a potential mechanism (the role of stratospheric intrusions in mid-latitudes is discussed in the
following section). Tap waters in the Central U.S. deviate from the latitudinal gradient and show
anomalously low values which are linked to different moisture sources. This is also confirmed
by higher resolution precipitation records for the region, even though subcloud re-evaporation
of rainfall in spring and summer is also identified as a seasonal controlling mechanism for
17
O-excess (Tian et al. 2018). This is shown by a large range of 17O-excess (–17 to 64 per meg)
and a flatter slope of the local meteoric water line (17 17
LMWL ) compared to GMWL . Landais et al.
(2010) show that subcloud re-evaporation of rainfall during African monsoon in tropical Niger
may alter 17O-excess in precipitation by ~40 per meg over the course of a single rain event.
Highest values are found in samples that precipitated at the beginning of the event, whereas
precipitate that was formed at the end shows a more negative 17O-excess. This shift is linked to
the decrease in relative humidity over the course of the rain event.
A two-year precipitation record on the maritime island of Okinawa, Japan, reveals
that 17O-excess of local rainfall varies with seasonal change of air mass trajectories. Thus,
17
O-excess in coastal regions is largely controlled by variable humidity conditions at different
source regions (Uechi and Uemura 2019). The authors also discuss that careful selection of a
representative reference line (e.g., λ17 = 0.527 instead of 0.528) may avoid artificially inflated
17
O-excess values which is also subject of discussion for the trend observed at Alert, Canada
(Lin et al. 2013a,b; Miller 2013).
Precipitation data from Indiana, U.S., show that local precipitation is in general agreement
with 17GMWL (Tian et al. 2018). However, the authors observe notably lower  LMWL = 0.5258 for
17
17
precipitation sampled in summer which hat lower average O-excess compared to winter
precipitation and is associated with re-evaporation. The corresponding δD vs. δ18O slopes ( 2LMWL)
are slightly lower than  2GMWL. A comparable 17LMWL (= 0.5264) was found in summer precipitation
samples from NW China. Since these samples were associated with relatively high 17O-excess, the
authors suggest a contribution from re-evaporated earlier rainfall events (Tian et al. 2019).
Isotope values of snow from Mt. Zugspitze (this work) range from −21.7 to −8.3‰ in
δ18OP, from 17 to 62 per meg in 17O-excessP, and from 4 to 16‰ in d-excessP (Figs. 3 and 6). A
slightly negative correlation of 17O-excessP with δ18OP of  LMWL = 0.5272 ± 0.0006 is observed
17
indicating a somewhat steeper 17O-excess-δ18O relation than the GMWL (Fig. 6). The local
 2LMWL of 8.1 ± 0.3 is in good agreement with global average and also with local precipitation
records reported for the Zugspitze plateau ( 2LMWL = 7.95; Hürkamp et al. (2019)) and for
Garmisch-Partenkirchen (  2LMWL = 8.12; Stumpp et al. (2014)). The range of isotope distribution
in snow cover of the Zugspitze plateau is also consistent with the isotope distribution in collected
snowfall (cf. Tables S3 and S7). Affolter et al. (2015) report comparable  LMWL = 0.527 and
17
 LMWL = 7.92 for precipitation collected across Switzerland.

2
Figure 6. Isotope data of vapor from February (open circles) and May (open diamonds), and freshly pre-
cipitated snow (solid squares) from Mt. Zugspitze. Gray lines represent the GMWL, dashed, red, and solid
black lines represent l17 and l2 of vapor in February and May and in total precipitation, respectively. 17 A
and  2A of vapor data are generally flatter than the GMWL, resulting in a decrease of 17O-excessA) and
d-excessA (  2A_Feb
 7.3  0.4,  2A_May 7.5  0.2 ) with increasing δ18OA (also see Table S5). This lower slope
is also visible in 17O-excessP of precipitation  (17
P 0.5272  0.0006, 17
P 0.5272  0.0006) but not in re-
spective d-excessP ( 2
P 8 . 1  0 .3 ). See Tables S6 and S7 for data.
412 Surma et al.
17
O-EXCESS DISTRIBUTION IN ATMOSPHERIC VAPOR
Only a few studies investigated the triple oxygen isotope composition of atmospheric
vapor so far. Figure 7 shows a compilation of isotope data from samples collected above the
Pacific Ocean, the NEEM drilling site, Greenland, in Alert, Canada, and at Mt. Zugspitze,
Germany (Uemura et al. 2010; Landais et al. 2012b; Lin et al. 2013; this study).
Lowest 17O-excess values (−6 to 46 per meg) are associated with most positive δ18O
values in the range of −23 to −11‰ in primary atmospheric vapor above the Pacific Ocean’s
surface (Uemura et al. 2010). A negative correlation of 17O-excess with relative humidity
was identified (−0.64 per meg/%). Assuming closure for vapor above the open ocean, the
authors built on the Craig and Gordon Model and the relationship of 17O-excess vs. relative
humidity (h) as described by (Barkan and Luz 2007):
17
O-excess  ln  17

 L-V_eq  ( 17  diff  1  h   h
(22)
0.528  ln  18
 L-Veq  (  diff  1  h   h
18

In order to fit the output of Equation (22) to actual measurement data, Uemura et al.
(2010) found the correction exponent for boundary layer turbulence (Eqn. 9) to be n = 0.29
(18 adiff(corr) = 1.008) above the open sea which is somewhat higher than a previously reported
correction by Merlivat and Jouzel (1979) that was based on wind tunnel experiments (n = 0.18,
18
adiff(corr) = 1.005). Luz and Barkan (2010) recalculated the correction by Uemura et al. (2010)
for a seawater value with 17O-excess of −5 per meg, yielding n = 0.33 (18 adiff(corr) = 1.0092).
This relation provides a basis to quantify h based on 17O-excess variations in remote vapor and
precipitation. Landais et al. (2012b) show that 17O-excess in vapor (15 to 48 per meg) at the
NEEM drilling site in Greenland varies in the same range as 17O-excess in contemporaneous
precipitation (23 to 43 per meg). This observation confirms that triple oxygen isotopes in polar
snow reflect the composition of atmospheric vapor when qS–V_eq = 0.528 is applied.
The highest values and largest variations of 17O-excess (23 to 190 per meg) in a single
study were reported for vapor samples from Alert, in arctic Canada (Lin et al. 2013a).
The positive anomalies were explained with stratospheric intrusions that contain mass-
independent oxygen isotope signatures, but this interpretation was not independently verified.
Atmospheric vapor samples from Mt. Zugspitze (Figs. 6 and 7) show elevated 17O-excess
between 30 and 82 per meg with decreasing δ18O (–34.4 to –20.4‰) and a δ′17O–δ′18O trend
of 17A = 0.5265 ± 0.0006. A difference is observed between vapor that was sampled in February
A = 0.5270 ± 0.0007) and vapor that was sampled in May (  A of 0.5245 ± 0.0005), the latter
( 17 17
showing a clearer correlation of O-excess with δ O (Fig. 6 a). Corresponding δDA vs. δ18OA
17 18
( 2A) for February and May are nearly identical within error (Fig. 6 b). In general, all slopes
are somewhat flatter than the GMWL ( GMWL = 0.528,  GMWL = 8; Dansgaard 1964; Luz and
17 2
Barkan 2010). The average isotopic composition of vapor is shifted by Δδ18OA–P = −12.9‰,
Δ17O-excessA-P = 10 per meg, and Δd-excessA-P = 18‰ with respect to the average isotopic
composition of local precipitation (Figs. 6 and 8). Regression analysis demonstrates, that local
variations in * RA (and also * RP) are largely unrelated to meteorological conditions (Ta, h),
except for δ18OA, which is positively correlated with Ta (Table 1).
Even though the regular occurrence of deep stratospheric intrusions is also evident at Mt.
Zugspitze (Trickl et al. 2010), we may probably exclude a mass-independent signal in 17O-excess
of ambient vapor, since vapor concentrations in stratospheric air are extremely low. High positive
17
O anomalies (1000 to 1800 per meg) were observed in the in the southern high-latitude
stratosphere at vapor concentrations below 10 ppm and more than 1 km above the tropopause
(Franz and Röckmann 2005). Therefore, exceptionally dry air in the lower troposphere would be
Figure 7. Compilation of 17O-excess vs. δ18O in atmospheric vapor samples and seawater (asterisks). At-
mospheric vapor that was sampled above the Pacific Ocean (crosses) shows most positive δ18O values and
covers the lower range in 17O-excess from −6 to 46 per meg (Uemura et al. 2010). Variations are attributed
to variable h above the evaporating sea surface. Vapor samples from Alert, Canada (open circles) show the
largest range and highest absolute values in 17O-excess (Lin et al. 2013a). However, data consistency of
that study has been questioned (Lin et al. 2013b; Miller 2013). Atmospheric vapor that was samples at the
NEEM drilling site, Greenland (diamonds), range from −45 to −38‰ in δ18O and from 20 to 50 per meg in
17
O-excess (Landais et al. 2012b). Samples were found to be in equilibrium with local precipitation and in
good agreement with a Rayleigh condensation model, also reflecting moisture source conditions. Ambient
vapor samples from Mt. Zugspitze, Germany (triangles), range from −34 to −20‰ in δ18O and from 30 to
80 per meg in 17O-excess, indicating a negative correlation of 17O-excess with δ18O (this study).
Figure 8. Record of Schneefernerhaus Research Station meteorological data: 2 m air temperature (Ta),
relative humidity (h), and dew point temperature (Td) from 2016−02−16 to 2016−05−27. Isotope data
(δ18O, 17O-excess, and d-excess) are shown for atmospheric vapor (open symbols, dashed lines) and pre-
cipitation (solid symbols and lines). Gray sequences indicate warmer periods. Note the scale contraction
for the period between 2016−02−28 and 2016−05−19. Error bars are ±1σ SE for 17O-excess and d-excess,
and smaller than symbol size for δ18O. Meteorological data for time of sample collection is provided in
Tables S1 and S2 in the Supporting Information.
414 Surma et al.
necessary in order to achieve low mixing ratios of tropospheric vapor and, thus, a conservation
of the stratospheric signal in the mass balance: i) vapor concentrations at Mt. Zugspitze did
not fall below 2000 ppmv during vapor extractions, so that photochemically induced anomalies
are unlikely to be evident in our samples (Hausmann et al. 2017). A mass-balance calculation
with local vapor (δ18O = –22.8‰, 17O-excess = 40 per meg 2000 ppmv vapor concentration)
and hypothetical stratospheric vapor (δ18O = –64.1‰, 17O-excess = 1300 per meg, 7.3 ppmv
concentration; average values obtained from high 17O-excess samples reported by Franz and
Röckmann 2005) would yield a mix with δ18O = –23.2 and 17O-excess = 43 per meg. ii) The
mid-latitude tropopause is considerably higher (7–10 km) compared to polar regions (~6 km),
limiting the contribution of stratospheric downdrafts to the lower troposphere. Nevertheless,
more data of ambient vapor and would hold the potential to prove the effect of stratospheric deep
intrusions on the oxygen isotope balance. The event of highest ozone of 101.9 ppbv registered at
Zugspitze (Hausmann et al. 2017) is 50 ppbv in excess above the tropospheric ozone background
of ~50 ppbv. In comparison, old stratospheric air in polar vortex contains at ~3 to 4 ppmv ozone
(e.g., El Amraoui et al. 2018) which is > 3000 ppbv above tropospheric ozone background.
Therefore, the isotopic balance of atmospheric vapor at mid-latitudes is likely dominated by
synoptic processes (i.e., Rayleigh distillation and potential continental recycling).
Rayleigh distillation. Several studies have demonstrated that 17O-excess in vapor and
precipitation reflects evaporation conditions at the initial moisture source (Uemura et al.
2010; Landais et al. 2012b; Uechi and Uemura 2019). However, as summarized above in the
discussion of mid-latitude rainfall data, there is some indication of subsequent modification
by evaporative recycling of continental moisture. With the Mt. Zugspitze data at hand,
we may test the respective primary and subsequent controls of water vapor isotopic composition.
*
RA in the troposphere at high altitude and—by inference—the continental interior. Rozanski
et al. (1982) have demonstrated that formation of condensate from vapor in westerly air
masses may be well approximated by a Rayleigh approach. δDA and d-excessA analyses by
Dütsch et al. (2018) demonstrate that the occurrence of mixed phase clouds is negligible for
westerly moisture pathways, thus, excluding complexity of vapor, water, and ice coexistence
during cloud formation (Ciais and Jouzel 1994). This is also supported by the liquid–vapor
precipitate formation in rapidly advected air masses at Mt. Zugspitze. Based on the analysis of
11 snow and vapor samples that were collected simultaneously (Table S2, Roman numerals),
we determined to overlying fractionation process for snow formation at Mt. Zugspitze.
Observed deviation of actual samples (18eA–P_meas from predicted fractionation (18eA–P_cal) is
calculated by Δ18εA–P (18eA–P_meas −18eA–P_cal), with 18eA–P_meas ≈ 18Op − 18OA and:
18

A-P_cal ( 18 S-V_eq  1)  1000 (23)
where aS–V_eq is the solid–vapor equilibrium fractionation factor (Majoube 1971). Δ εA–P
18 18
based on this calculation averages −4.4 ± 2.0‰ (Fig. 9). Replacing 18 aS–V_eq in Equation (23)
with 18 aL–V_eq to account for the occurrence of supercooled water droplets provides a better
approximation for 18eA–P_meas (Δ18εA–P = 1.9 ± 0.9‰) (Ciais and Jouzel 1994; Bolot et al. 2013),
thus confirming that liquid orographic clouds are the prevalent cloud species in strong updraft
regimes at temperatures between 0 and −15 °C, such as Mt. Zugspitze (Kneifel et al. 2014;
Lohmann et al. 2016; Lowenthal et al. 2016). Respective 17O-excessP_cal and d-excessP_cal
deviate by 19 ± 19 per meg and 11.1 ± 4.3‰. Due to their large scatter, these values are also
reasonably well reproduced at given uncertainty, even though d-excessP is systematically
lower, presumably due to the different T dependencies of 2 aL–V_eq and 18 aL–V_eq. Typically, the
orographic cloud basis is located either below or at the altitude of Mt. Zugspitze, hence we
assume that pairs of vapor–precipitation samples are closely related. However, vapor samples
may potentially not represent free atmospheric vapor but a mixture of atmospheric and snow
surface vapor. We also note that the determination of *αL–V_eq does not cover T below 0 °C and is
therefore extrapolated for ambient conditions at Mt. Zugspitze (Horita and Wesolowski 1994).
Figure 9. Difference between measurement data of precipitation samples and predicted isotopic composi-
tion of precipitation (Δ18εA–P). Values are calculated with 18αL–V_eq (solid black squares and line, Horita
and Wesolowski 1994) and 18αS–V_eq (solid red squares and line, Majoube 1971).
The evolution of * RA (Fig. 10) is based on Monte Carlo simulations (n = 500) for
Equations (11) and (13), using a * RW0 of ocean surface with dDW0 = 0‰, d17OW0= –0.005‰,
and d18OW0 = 0‰. Modeling parameters (source h, Ta, * adiff, *αL–V_eq) were approximated from
HYSPLIT air trajectory reanalysis (see Table S8 for details). * adiff is corrected for air turbulence
by Equation (9) with an average n = 0.26 (Merlivat and Jouzel 1979; Luz and Barkan 2010).
The range of observed δ18OA is well reproduced for fres between 0.5 (days of relatively warm
Ta) and 0.15 (days of lower Ta). Higher δ18OA is found in vapor from warmer air masses and
lower δ18OA in vapor provided by cold air (Fig. 8), indicating a distillation effect that is also in
agreement with the HYSPLIT output for specific humidity. This observation is also confirmed
by regression analysis of δ18OA vs. Ta (Table 1). A change from a warmer (25 °C) to a cooler
(10 °C) vapor source would shift δ18OA according to the change in equilibrium fractionation
during vapor formation (Fig. 10, blue arrows), which is only a minor effect compared to that
of relative condensation loss from the air parcel. The variability in 17O-excessA is controlled
by ambient h and wind turbulence (n) at the moisture source (e.g., Uemura et al. (2010); Uechi
and Uemura (2019)). The magnitude of initial d-excessA is, in contrast, additionally controlled
by surface temperature at the source, resulting from the different T dependency of 2αL–V_eq
Table 1. Regression analysis (slope, standard error (SE), and coefficient of determination (R2) of
δ18O, 17O-excess, and d-excess in vapor and precipitation samples vs. meteorological parameters (left
column): Air temperature (Ta) and relative humidity (h).
Parameter δ18O 17
O-excess d-excess
slope SE R2 slope SE R2 slope SE R2
Vapor (total)
Ta 0.52 0.11 0.47 −0.81 0.48 0.11 −0.01 0.20 0.00
h −1.46 4.36 0.00 14.53 13.79 0.05 −12.14 4.82 0.22
Vapor (February)
Ta 0.70 0.18 0.51 −1.50 0.70 0.25 0.05 0.40 0.00
h 5.27 5.31 0.07 2.66 5.31 0.00 −20.80 6.17 0.45
Vapor (May)
Ta 0.51 0.24 0.40 −1.78 0.98 0.32 −0.04 0.19 0.01
h −12.49 5.90 0.39 44.65 24.06 0.33 4.36 4.33 0.13
Precipitation (total)
Ta 0.21 0.17 0.06 0.03 0.50 0.00 −0.51 0.31 0.18
h 4.70 5.69 0.03 −27.08 15.23 0.12 −12.26 11.02 0.09
416 Surma et al.
Figure 10. Rayleigh fractionation model for initial atmospheric vapor at given source condition for 17O-
excessA vs. δ18OA (a), and for d-excessA vs. δ18OA (b). Simple Rayleigh distillation of initial oceanic vapor
(solid frame, fres = 1) at hs = 0.8 (white) and at hs = 0.6 (gray) results in steady isotopic depletion of * RA.
Numbers represent respective fres. Blue arrows indicate a change in Ts from 25 to 10 °C, red arrows the
change if a wind correction exponent of n = 0.29 instead of n = 0.23 is used. Ellipsoids represent probability
density functions for a 0.95 quantile of the Monte Carlo simulation. Black and red open symbols repre-
sent atmospheric vapor sampled in February and May at Mt. Zugspitze, respectively.
2
 diff
and 18αL–V_eq, and 18 (Horita and Wesolowski 1994; Luz et al. 2009). Although the range
 diff
of δ18OA may be successfully modeled by continuous vapor distillation to a reasonable degree,
high 17O-excessA and d-excessA values are not reproduced. Variation of from n = 0.26 to
n = 0.33 do not sufficiently increase 17O-excessA and d-excessA values (Fig. 10, red arrows).
Lower h over the ocean might also elevate initial 17O-excessA and d-excessA, but h < 0.6
is unlikely for given oceanic moisture sources (Pfahl and Sodemann 2014). Consequently,
Rayleigh distillation alone may not explain the observation.
Moisture pathways and air trajectory modeling. Reanalysis of air trajectories provides
important information on atmospheric pathways and potential moisture sources (Pfahl and
Wernli 2008; Vogelmann et al. 2015; Yu et al. 2015; Uechi and Uemura 2019). Atmospheric
transport to Mt. Zugspitze is determined by three patterns: (i) Long-range transport from
North America, (ii) long-range transport from Northern Africa, and (iii) stratospheric deep
intrusions (Hausmann et al. 2017). We modeled 240 h back-trajectories (including Ta and
specific humidity) using the NOAA HYSPLIT software package and GDAS reanalysis data
sets (available at ftp://arlftp.arlhq.noaa.gov/pub/archives/gdas1) (Stein et al. 2015). Since the
higher resolution record (0.5° grid) is incomplete, we use datasets of 1.0° spatial resolution for
modeling. According to previous studies we use altitudes of 1,900 to 2,200 m above ground
level (m a.g.l.) to approximate the low-level troposphere (3,000 m a.s.l.) at Mt. Zugspitze
(Trickl et al. 2010; Vogelmann et al. 2015) and identify two source patterns: The Central
Atlantic, between N 30° and N 50° for warmer air, and the Nordic Sea providing cold air
(Fig. 11). These patterns are also in agreement with general moisture pathways that were
determined for the Northern Alps by Sodemann and Zubler (2010). Analysis of specific
humidity indicates low continental evaporation rates in winter with the Central Atlantic being
identified as the predominant moisture source in February, whereas moisture uptake above
Western and Central Europe contributes to the moisture balance in late spring.
Continental moisture recycling. Triple oxygen isotope studies of mid-latitude waters
suggest that continental recycling may be an important local effect (Li et al. 2015; Tian et al.
2018; Tian et al. 2019) but a rigorous demonstration has not been provided yet. It is also known
from δD and δ18O analyses that moisture recycling plays an important role in the continental
moisture balance (Bastrikov et al. 2014; Wei and Lee 2019). The continental vapor budget (* RA)
is therefore balanced by the isotopic composition of initial vapor (* RA0) and the surface vapor flux
(* RV) with qR being the relative quantity of local, recycled moisture (Aemisegger et al. 2014):
Figure 11. HYSPLIT 240 h back trajectories (modelled for 2,200 m a.g.l. and 3 h resolution) of air masses
arriving at Mt. Zugspitze (asterisk). Modeling is performed for both field campaigns in February (a)
and May (b). Trajectories of warmer air masses (red) suggest moisture uptake above the Central Atlantic
Ocean, colder air masses (black) cross the Northern Atlantic and the Nordic Sea.
*
RA  1  qR   * RA 0
 qR  * RV (24)
The surface vapor flux is defined as:
*
RV  qT  * RVT  1  qT   * RVE (25)
where qR is the relative quantity of recycled vapor being contributed by plant transpiration and
*
RVT the corresponding isotopic ratio which is approximated by quantitative, non-fractionating
transfer from surface water (* RW) to vapor, so that * RVT = * RW (Wang and Yakir 2000; Peng et
al. 2005). It has been shown that moisture of the continental boundary layer is significantly
balanced by recycling of surface water (He and Smith 1999; Welp et al. 2012; Jasechko et al.
2013). * RVEis the isotopic ratio of the evaporative flux from surface waters and saturated soil
(* RWi). Figure 12 outlines the process of initial vapor formation and continental recycling.
Figure 12. Schematic representation showing isotopic fluxes during Rayleigh distillation with continental
recycling: I. Initial vapor forms above the ocean surface. II. First condensate formation and precipitation.
III (a and b). Evaporation of continental surface water, admixture of vapor, and subsequent precipitation.
Solid, black arrows indicate H2O fluxes, gray arrows indicate mixing of residual vapor with secondary vapor.
418 Surma et al.
The results from HYSPLIT air trajectory reanalysis (see above) and previous studies
suggest that moisture recycling above Western and Central Europe affects the balance of * RA0
(van der Ent et al. 2010; Sodemann and Zubler 2010). Based on the 1D-model (Fig. 12), we
here use Equations (11), (13), (24), and (25) to calculate the evolution of atmospheric vapor,
*
RAi* (Fig. 13 a and b). Average qT = 0.7 is suggested by different previous studies for Central
Europe (Lawrence et al. 2007; Aemisegger et al. 2014; Christner et al. 2017). Wind-induced
turbulence (Eqn. 9) for continental evaporation is parametrized with n = 0.5 (Dongmann et al.
1974; Gat 1996; Haese et al. 2013). * RWiis approximated with mean regional * Rp (February:
δ17O ≈ –5.525‰, δ18O ≈ –10.5‰, δD ≈ –74.0‰; May: δ17O ≈ –4.199‰, δ18O ≈ –8.0‰,
δD ≈ –54.0‰), using average precipitation over Western to Central Europe from the Online
Isotopes in Precipitation Calculator (OIPC) (Bowen and Revenaugh 2003; Bowen 2018) with
17
O-excess being expected to be close to the average GMWL value (≈ 33 per meg).
Continental recycling shows a small increase of 17O-excessA for higher qR (Fig. 13 a),
whereas no significant effect is visible for d-excessA (Fig. 13 b). Changing * RW of surface
water to lower values (February) results in an according shift in δ18OA which is visibly larger
for qR = 0.5 but does not affect both excess parameters. This may be explained by the fact
that average surface water’s d-excessW ≈ 10‰ and, thus, lower than d-excessA of arriving
moisture (≈ 10‰). Evaporation from surface water will elevate d-excessV in the local vapor
flux to a degree which is not significantly different from arriving d-excessA. 17O-excessW in
surface water, on the other hand, is already comparably high to 17O-excessA in the arriving air
parcel. Re-evaporation elevates 17O-excessV in the vapor flux and induces a positive shift in the
resulting mixed vapor. Since typical qR is below 0.2 during winter and spring in Western and
Central Europe, the uncertainty in * RW may be neglected here (Trenberth 1999; van der Ent et
al. 2010). However, an accurate estimate will be more critical when recycling is balanced for
summer months and intensive evaporation elevates qR and significantly alters * RW.
Vapor contribution from a local evaporation. Considering the above findings, recycling
over continental Europe may probably explain some increase in 17O-excessA and should be
accounted for in the total balance but is not sufficient to explain high 17O-excessA and d-excessA
found in atmospheric vapor at Mt. Zugspitze. Even at low continental h = 0.6 and qR = 0.5—as
used in this model—only a minor increase in 17O-excessA would result from this process.
The effect on d-excessA is even smaller. This is mainly due to the fact that transpiration by
vegetation (qT = 0.7) effectively diminishes fractionation of surface water during continental
vapor formation. Much larger effects may be expected when vapor is formed from open water
surfaces or saturated soil. However, the range of 17O-excessA modeled with low qR is in good
agreement with 17O-excess ranging from 14 to 27 per meg in natural water data reported
for low altitudes in Western and Central Europe (Luz and Barkan 2010; Affolter et al. 2015;
Alexandre et al. 2019). Our results also show that modeling with qR = 0.5 provides sufficiently
high 17O-excess (~40 per meg) to explain elevated values in precipitation as reported for
Northwestern U.S. (Li et al. 2015) or inland China (Tian et al. 2019).
Even though the recycling ratio (qR) is relatively small for Western and Central Europe at
lower altitudes, evaporative loss from local snow cover provides significant vapor quantities
in Alpine hydrological balances (Froehlich et al. 2008; van der Ent et al. 2010; Schlaepfer et
al. 2014). Evaporation may account for up to 90% of total mass loss from snow cover and is
especially high when turbulent transport of snow is effective, i.e., at wind-exposed locations and
high-altitude mountain ridges (Strasser et al. 2008). Ambient conditions at Mt. Zugspitze are
characterized by high snow accumulation and wind-induced re-deposition, providing a constant
source of fresh snow surfaces, that i) lower a potential fractionation limitation by self-diffusion
in stagnant snow cover (Friedman et al. 1991; Schlaepfer et al. 2014), and ii) can supply a
significant quantity of moisture from local snow cover (Kaiser et al. 2002; Froehlich et al. 2008).
Figure 13. Distillation model as shown in Figure 10 but accounting for moisture recycling over Central
Europe for 17O-excessA vs. δ18OA (a), and for d-excessA vs. δ18OA (b). Bold ellipsoids were modeled with
average surface * RW for May. Atmospheric vapor evolves along a Rayleigh condensation path (cf. Fig. 10)
and mixes with hypothetical quantities of qR = 0.1 (green) and qR = 0.5 (yellow) of recycled moisture that
evaporates from surface water (* RW). Light ellipsoids indicate same vapor trends but with average * RW for
February. Rayleigh fractionation model for atmospheric vapor including continental moisture recycling
with qR = 0.2 at given fres (gray ellipsoids, numbers) for for 17O-excessA vs. δ18OA (c) and d-excessA vs.
δ18OA (d). Solid black circles represent the average isotopic composition of local snow cover. The isotopic
compositions of local vapor fluxes at Mt. Zugspitze (* RV) for February (blue) and May (red) vary with
changes in h and Ta. Gray ellipsoids and open symbols represent the same Monte Carlo simulation and
data as described in Figure 10, respectively.
Local * RW input values for calculating the isotopic composition of the vapor flux are obtained
from the average composition of the local snow cover (δ17O = –8.729‰, δ18O = –16.555‰,
δD = –122.3‰). Both * RW calculated for February and May are identical within analytical
uncertainty. Turbulence in the diffusive boundary layer is parametrized with n = 0.5. A slightly
higher value of n = 0.58 is suggested for water droplets and snow particles suspended in air
(Stewart 1975; Froehlich et al. 2008). However, it is more likely that prevailing winds induce
steady turnover of surface snow but do not necessarily keep it in suspension.
Isotopic fluxes that were calculated for snow evaporation (* RV_Feb and * RV_May) only
show a small difference (Fig. 13 c and d) that can be attributed to the different average h
and Ta used in the calculation (February: h = 0.73 ± 0.05, Ta = −5.9 ± 3 °C; May: h = 0.67 ± 0.05,
Ta = 1.0 ± 3 °C). The isotopic composition of local evaporation, * RV, represents the isotopic
end-member in the local atmospheric vapor balance. Actual * RA of the free atmosphere results
from mixing of * RV and the isotopic composition of vapor in the arriving air mass (Fig. 13
c and d, gray ellipsoids). A higher recycling ratio will shift * RA towards the composition of
*
RV, whereas low qR will not change the isotopic composition of arriving vapor significantly.
This explains why peak values of 17O-excessA and d-excessA coincide with low Ta and δ18OA,
while minimum 17O-excessA and d-excessA are associated with higher Ta and δ18OA (Fig. 8).
420 Surma et al.
Cold air generally holds lower H2O concentrations which would, in turn, decrease the
contribution of arriving vapor in the local vapor balance, elevating local qR and, thus, the
imprint of the local vapor flux. Some uncertainty may arise from the fact that vapor samples
were extracted at ~2.5 m above ground and may—due to insufficient mixing of the diffusive
boundary layer—reflect the isotopic composition of the evaporative flux rather than free
atmospheric vapor.
FUTURE WORK AND CONCLUSION

Future work
Analytical challenges. It was demonstrated that fluorination of H2O using a CoF3 reagent
combined with dual-inlet mass spectrometry provides a reasonably good reproducibility
(±5 per meg) in 17O-excess to identify small natural variations (Barkan and Luz 2005).
Analytical errors (seen as data scatter) in δ17O and δ18O are well correlated, thus not affecting
the precision for 17O-excess (Landais et al. 2006). A sample memory effect occurring in CoF3
lines is evident but well corrigible (i.e., by four blind injections between samples with a 55‰
difference in δ18O (Barkan and Luz 2005)). Mass spectrometric analysis may be affected by
unrecognized blanks, leaks, and scale compression. So far, two-point data normalization using
VSMOW-2 and SLAP-2 provides an adequate correction (Schoenemann et al. 2013). Still,
this normalization has to be extrapolated for samples exceeding δ18O = 0‰ (e.g., waters from
arid regions), hence an isotopically enriched standard with a commonly accepted triple oxygen
isotope composition would improve inter-laboratory comparison.
In the past years, it has been shown that cavity ring-down spectroscopy provides
sufficiently low uncertainty for measuring natural variations in 17O-excess (Berman et al. 2013;
Schoenemann et al. 2014; Affolter et al. 2015; Tian et al. 2018). However, water mixing by the
autosampler syringe and water adhesion to internal surfaces of the instrument were identified
as causes for considerably high memory effects (Berman et al. 2013 and references therein).
Three approaches are suggested to minimize these effects: i) preconditioning with blind
injections, ii) a mathematical memory correction, and iii) avoiding large differences in isotopic
composition of adjacent samples (Schauer et al. 2016). In contrast to mass spectrometric analysis,
data scatter in δ17O and δ18O is not correlated, thus increasing the uncertainty for 17O-excess
(Berman et al. 2013). Still, the improvement of analytical protocols has demonstrated that
optical isotope analysis may reach ± 10 per meg uncertainty providing a sample throughput
comparable to that of isotope laboratories using the fluorination method (Schauer et al. 2016).
In the future, these developments will be crucial for the expansion of spatial coverage and
automated long-term observations that will be needed for a global monitoring network of triple
oxygen isotopes in precipitation and vapor. Precise knowledge of 17O-excess in atmospheric
vapor would e.g., be essential to constrain the evaporative fractionation in lake balance models
(Gázquez et al. 2018; Surma et al. 2018; Voigt et al. 2020) and to improve the performance of
general circulation models (Steen-Larsen et al. 2016).
Modeling of cloud processes and global transport. In order to improve the understanding
of 17O-excess distribution in precipitation and vapor across the globe, adequate prediction using
isotope-enabled general circulation models (GCM) is inevitable. The fractionation of water
isotopologues (including H217O) was implemented into the LMDZ GCM and extensively tested
for variations along latitudinal gradients and seasonal variations (Risi et al. 2013). Accounting
for evaporative conditions, Rayleigh distillation, tropospheric vapor mixing, and rain re-
evaporation the authors have shown that the sensitivity of 17O-excess to evaporative conditions
is apparently underestimated by the model. Schoenemann and Steig (2016) also demonstrated
that intermediate complexity modeling (ICM) which accounts for seasonal climatological cycles
(i.e., relative humidity, sea surface temperature, sea ice extent, evaporation, precipitation rate)
and diffusivity during low temperature precipitation, captures 17O-excess variations observed in
Antarctic precipitation. The authors show that variable moisture source conditions alone cannot
explain large seasonal at Vostok/East Antarctica, stressing the importance of local temperature
effects on snow formation. It is suggested to apply additional corrections for snow sublimation
and ‘diamond dust’ (clear-sky precipitation) input in interior Antarctica (Casado et al. 2018;
Miller 2018; Pang et al. 2019). Improving these parametrizations and the understanding of
advection processes during poleward transport of moisture will be critical for triple oxygen
isotope based GCM simulations of past climate (e.g., Cauquoin et al. 2019a,b).
In order to improve GCM estimates on spatio-temporal 17O-excess distribution in vapor
and precipitation, the coverage of measurements should be drastically improved to provide a
profound basis for GCM vs. observation evaluations (Risi et al. 2013; Steen-Larsen et al. 2016).
Better knowledge of equilibrium and diffusive solid–vapor fractionation at low temperatures
will be necessary to improve GCM capabilities in polar regions. Adequate integration of
evaporation from soils and plants will be essential to provide robust estimates on continental
moisture recycling (Gat and Matsui 1991; Aemisegger et al. 2014).
Mineral–water systems. 17O-excess studies of ice core records have shown that 17O-excess
is a valuable tracer for paleoclimate reconstruction (Guillevic et al. 2014; Landais et al. 2018).
However, ice records are mostly restricted to polar areas, emphasizing the need of other triple
oxygen isotope records in low- and mid-latitude regions. Recent publications demonstrate
a several mineral–water systems that show the capability to capture triple oxygen isotope
compositions of ambient water by direct equilibration with mineral-bound oxygen or in
hydration water and fluid inclusions.
Speleothems provide a unique climate archive with high temporal resolution for low- and
mid-latitudes since they are known to reflect the isotopic composition of regional precipitation
(e.g., Bar-Matthews et al. 1996). 17O-excess in meteoric water that was reconstructed from
speleothem carbonates is in good agreement with actual precipitation measurements and
show significant regional differences between sampling sites (South American and Asian
monsoon regions, eastern Mediterranean, China, and Central Asia). The authors explain
these variations to different moisture source conditions of regional precipitation (Sha et al.
2020). However, reconstructed moisture source humidity remains mostly constant in those
records during glacial-interglacial transitions. Affolter et al. (2015) also present a method for
direct 17O-excess measurements on fluid inclusions in speleothem samples, providing a more
direct estimate of paleo precipitation without requiring accurate constraints on the isotopic
fractionation between water and calcite (Bergel et al. 2020; Fosu et al. 2020; Voarintsoa et al.
2020). However, this approach requires larger sample amounts (0.5–1.5 g of carbonate).
Authigenic lake minerals that form in equilibrium with ambient waters provide another
valuable tracer for 17O-excess reconstruction. It was shown shown that isotopic compositions
of parent water bodies (i.e., lakes and ponds) can be precisely reconstructed by liberation and
analysis of gypsum hydration water, and accurate knowledge of triple oxygen fractionation
between the structurally bonded and the ambient water (Gázquez et al. 2015, 2017; Herwartz et
al. 2017). Quantitative estimates on local relative humidity changes were made by combining
this approach with d-excess measurements to natural samples from lakes in arid regions
(Evans et al. 2018; Gázquez et al. 2018). Recent studies also reveal that 17O-excess in lake
carbonates reflects the isotopic composition of parent water bodies and primary catchment
precipitation (Passey et al. 2014; Passey and Ji 2019; Passey and Levin 2021, this volume).
It is also suggested that additional analysis of ‘clumped isotopes’ (Δ47) is valuable to improve
back-projection modeling of parent waters by adding constraints on the carbonate formation
temperature (Ghosh et al. 2006; Passey and Ji 2019).
422 Surma et al.
Analysis of triple oxygen isotopes in fine-grained and clay-rich sediments by Bindeman et

al. (2019) has also shown that integrated isotope signals of regional precipitation are captured
in weathering products of bedrock. Using appropriate fractionation factors for water–clay
equilibrium (Bindeman et al. (2019) and references therein) and a model for back-projection
of water, mean annual temperatures can be reconstructed. Another application of this approach
is the reconstruction of a paleo-GMWL based on the analysis of shales and clay minerals
(Bindeman 2021, this volume).
Landais et al. (2006) have demonstrated that δ17O–δ18O variations in plant leaf water
follow a transpiration fractionation trend (λtransp) which is distinctively lower than equilibrium
fractionation of water and that λtransp is controlled by ambient relative humidity. Recent studies
have shown that phytoliths (amorphous silica microstructures in plants) reflect the triple oxygen
isotope composition of leaf water (Alexandre et al. 2018; Alexandre et al. 2019).Assuming
constant triple oxygen isotope fractionation between silica and leaf water (θPhyto–LW = 0.521),
this approach may provide a unique tracer for paleo-humidity. However, it has yet to be proven
that θPhyto–LW is climate independent and can thus be applied for reconstructing paleo-leaf
water. If so, reconstructed leaf water could provide e.g., insights into humidity conditions and
the triple oxygen isotopic composition of plant CO2 (Alexandre et al. 2019).
Water–rock interaction and bedrock re-equilibration also provide valuable hydrological
information for geological periods (i.e., the Precambrian) where other records are rare
(Herwartz 2021, this volume). Zakharov et al. (2019b) show that regional meteoric water can
be reconstructed from triple oxygen isotope mixing patterns found in Icelandic metamorphic
rocks. In studies this approach was used to reconstruct the average δ18O continental ice during
Paleoproterozoic global glaciations (Herwartz et al. 2015; Zakharov et al. 2019a). Modeling
the isotopic balance for water–rock interaction of Precambrian cherts provides key information
on the isotopic composition of the ancient ocean (Sengupta and Pack 2018; Liljestrand et al.
2020; Zakharov et al. 2021, this volume).
Conclusion
Significant improvement of analytical techniques during the last two decades has triggered the
use of triple oxygen isotopes in the hydrological cycle, the investigation of underlying fractionation
mechanisms, triple-O-based climate reconstructions, and the comprehension of hydrological
processes. A major achievement is understanding the global distribution of triple oxygen isotopes
which is described by δ′17O = 0.528 ⋅ δ′18O + 0.033‰ and referred to as the GMWL.
Essential isotope fractionation processes operating on the global scale include equilibrium
evaporation, diffusion of water vapor, and vapor condensation. Though, diffusion and the
Craig and Gordon model are well established for δD and δ18O, the accurate determination of
triple oxygen isotope fractionation factors describing the 17O/16O and 18O/16O distribution for
dominating processes marks a major breakthrough. Processes that involve diffusive fractionation,
such as evaporation or snow formation at low temperatures produce trends that deviate from this
relationship and which are found in extreme environments like remote polar regions (low δ18O)
or hyper-arid deserts (high δ18O). However, also waters that fall in the range of intermediate δ18O
values (e.g., mid-latitude precipitation) form local δ′17O–δ′18O trends that deviate from overall
17
GMWL = 0.528 and reflect re-evaporation of rain droplets or continental moisture recycling.
The temperature independent 17O-excess parameter is a useful, complementary tool in

addition to traditional d-excess for the stable isotope-based assessment of hydrological processes,
paleoclimatology and paleohydrology. Its sensitivity to diffusive fractionation makes it a valuable
tracer for (paleo-)humidity conditions and locations of moisture source regions. However, it has
been demonstrated that second order effects such as distillation at low temperatures and re-
evaporation of rain droplets may systematically lower the 17O-excess in the final precipitate.
Similar to the use of d-excess in arid regions, 17O-excess is used for hydrological balancing
of evaporative water bodies. Combining both parameters results in a better understanding and
parametrization of environmental conditions. The use of lake balance models and adequate
constraints of boundary conditions (isotopic composition of inflowing water and atmospheric
vapor, wind regime) provides valuable estimates on humidity conditions in arid regions.
Analysis of authigenic minerals from lake bodies also demonstrates the potential of 17O-excess
for paleoclimate applications.
Studies of precipitation in mid-latitudes suggest that elevated 17O-excess in continental
interior cannot be explained by the above-mentioned effects and result from moisture recycling
along air trajectories. In this work we demonstrate that 17O-excess of atmospheric vapor provides
valuable information on continental moisture recycling. However, no data on triple oxygen
isotopes in continental vapor from mid-latitudes were available so far to obtain systematic
constraints on this effect.
The investigation of moisture recycling would be greatly improved with larger datasets on both,
precipitation records and atmospheric vapor. Measurements of 17O-excess in atmospheric vapor in
a high temporal resolution will be necessary to improve the understanding of local meteorological
effects on local evaporative fluxes and, thus, the isotopic balance of H2O in the atmosphere.
Triple oxygen isotope analysis holds large potential to constrain isotopic estimates for land-locked
water reservoirs that are balanced by local evaporation and sublimation to a large extent.
ACKNOWLEDGEMENTS
The project was funded by the German Research Foundation (DFG, grant no. STA 936/8−1).
REFERENCES
Aemisegger F, Pfahl S, Sodemann H, Lehner I, Seneviratne SI, Wernli H (2014) Deuterium excess as a proxy for
continental moisture recycling and plant transpiration. Atmos Chem Phys 14:4029–4054
Affolter S, Häuselmann AD, Fleitmann D, Häuselmann P, Leuenberger M (2015) Triple isotope (δD, δ17O, δ18O)
study on precipitation, drip water and speleothem fluid inclusions for a Western Central European cave (NW
Switzerland). Quat Sci Rev 127:73–89
Alexandre A, Landais A, Vallet-Coulomb C, Piel C, Devidal S, Pauchet S, Sonzogni C, Couapel M, Pasturel M,
Cornuault P, Xin J, Mazur J-C, Prié F, Bentaleb I, Webb E, Chalié F, Roy J (2018) The triple oxygen isotope
composition of phytoliths as a proxy of continental atmospheric humidity: insights from climate chamber and
climate transect calibrations. Biogeosciences 15: 3223–3241
Alexandre A, Webb E, Landais A, Piel C, Devidal S, Sonzogni C, Couapel M, Mazur J-C, Pierre M, Prié F, Vallet-Coulomb
C, Outrequin C, Roy J (2019) Effects of leaf length and development stage on the triple oxygen isotope signature of
grass leaf water and phytoliths: insights for a proxy of continental atmospheric humidity. Biogeosciences 16:4613–4625
Araguás-Araguás L, Froehlich K, Rozanski K (2000) Deuterium and oxygen−18 isotope composition of precipitation
and atmospheric moisture. Hydrol Process 14:1341–1355
17 16
Bao H, Cao X, Hayles JA (2016) Triple oxygen isotopes: fundamental relationships and applications. Ann Rev Earth
Bar-Matthews M, Ayalon A, Matthews A, Sass E, Halicz L (1996) Carbon and oxygen isotope study of the active
water-carbonate system in a karstic Mediterranean cave: Implications for paleoclimateresearch in semiarid
regions. Geochim Cosmochim Acta 60:337–347
Spectrom 19:3737–3742
424 Surma et al.
Bastrikov V, Steen-Larsen HC, Masson-Delmotte V, Gribanov K, Cattani O, Jouzel J, Zakharov V (2014) Continuous
measurements of atmospheric water vapour isotopes in western Siberia (Kourovka). Atmos Meas Tech 7:1763–1776
Acta 275:36–47
Berman ESF, Levin NE, Landais A, Li S, Owano T (2013) Measurement of δ18O, δ17O, and 17O-excess in water by
off-axis integrated cavity output spectroscopy and isotope ratio mass spectrometry. Anal Chem 85:10392–10398
Bershaw J, Hansen DD, Schauer AJ (2020) Deuterium excess and 17O-excess variability in meteoric water across the
Pacific Northwest, USA. Tellus 72B:1773722
Bolot M, Legras B, Moyer EJ (2013) Modelling and interpreting the isotopic composition of water vapour in
convective updrafts. Atmos Chem Phys 13:7903–7935
Bowen G (2018) The Online Isotopes in Precipitation Calculator, version 3.1. available at: http://www.waterisotopes.org
Bowen GJ, Revenaugh J (2003) Interpolating the isotopic composition of modern meteoric precipitation. Water
Resour Res 39:1299
Brenninkmeijer CAM, Röckmann T (1996) Russian doll type cryogenic traps: improved design and isotope separation
effects. Anal Chem 68:3050–3053
Cappa CD, Hendricks MB, DePaolo DJ, Cohen RC (2003) Isotopic fractionation of water during evaporation. J
Geophys Res 108:4525
Casado M, Cauquoin A, Landais A, Israel D, Orsi A, Pangui E, Landsberg J, Kerstel E, Prie F, Doussin J-F (2016)
Experimental determination and theoretical framework of kinetic fractionation at the water vapour-ice interface
at low temperature. Geochim Cosmochim Acta 174:54–69
Cauquoin A, Risi C, Vignon É (2019a) Importance of the advection scheme for the simulation of water isotopes over
Antarctica by atmospheric general circulation models: A case study for present-day and Last Glacial Maximum
with LMDZ-iso. Earth Planet Sci Lett 524:115731
Cauquoin A, Werner M, Lohmann G (2019b) Water isotopes—Climate relationships for the mid-Holocene and
preindustrial period simulated with an isotope-enabled version of MPI-ESM. Clim Past 15:1913–1937
Christner E, Kohler M, Schneider M (2017) The influence of snow sublimation and meltwater evaporation on δD of
water vapor in the atmospheric boundary layer of central Europe. Atmos Chem Phys 17:1207–1225
Ciais P, Jouzel J (1994) Deuterium and oxygen 18 in precipitation: Isotopic model, including mixed cloud processes.
J Geophys Res 99:16703–16803
Craig H (1961) Isotopic Variations in Meteoric Waters. Science 133:1702–1703
Craig H, Gordon L (1965) Deuterium and oxygen 18 variations in the ocean and the marine atmosphere. In: Stable Isotopes
in Oceanographic Studies and Paleotemperatures. Tongiorgi E (ed). Laboratorio di Geologia Nucleare, Pisa, p 9–130
Criss RE (1999) Principles of Stable Isotope Distribution. Oxford University Press, New York
Dansgaard W (1954) The O18-abundance in fresh water. Geochim Cosmochim Acta 6:241–260
Dongmann G, Nürnberg HW, Förstel H, Wagener K (1974) On the enrichment of H218O in the leaves of transpiring
plants. Radiat Environ Biophys 11:41–52
Dütsch M, Pfahl S, Meyer M, Wernli H (2018) Lagrangian process attribution of isotopic variations in near-surface
water vapour in a 30-year regional climate simulation over Europe. Atmos Chem Phys 18:1653–669
Ehhalt D, Knott K (1965) Kinetische Isotopentrennung bei der Verdampfung von Wasser. Tellus 17:389–397
El Amraoui L, Peuch V-H, Ricaud P, Massart S, Semane N, Teyssèdre H, Cariolle D, Karcher F (2018) Ozone loss in
the 2002 - 2003 Arctic vortex deduced from the assimilation of Odin/SMR O3 and N2O measurements: N2O as
a dynamical tracer. Q J R Meteorol Soc 134:217–228
Ellehoj MD, Steen-Larsen HC, Johnsen SJ, Madsen MB (2013) Ice-vapor equilibrium fractionation factor of
hydrogen and oxygen isotopes: Experimental investigations and implications for stable water isotope studies.
Evans NP, Bauska TK, Gázquez-Sánchez F, Brenner M, Curtis JH, Hodell DA (2018) Quantification of drought
during the collapse of the classic Maya civilization. Science 361:498–501
Fontes JC, Gonfiantini R (1967) Comportement isotopique au cours de l’evaporation de deux bassins sahariens. Earth
Fosu BR, Subba R, Peethambaran R, Bhattacharya SK, Ghosh P (2020) Technical Note: Developments and
applications in triple oxygen isotope analysis of carbonates. ACS Earth Space Chem 4:702–710
Franz P, Röckmann T (2005) High-precision isotope measurements of H216O, H217O, H218O, and the Δ17O-anomaly of
water vapor in the southern lowermost stratosphere. Atmos Chem Phys 5:2949–2959
Friedman I (1953) Deuterium content of natural waters and other substances. Geochim Cosmochim Acta 4:89–103
Friedman I, Benson C, Gleason J (1991) Isotopic changes during snow metamorphism. In: Stable Isotope
Geochemistry: A Tribute to Samuel Epstein. Taylor Jr HP, O’Neil JR, Kaplan IR (eds) Geochemical Society,
San Antonio p 211–221
Froehlich K, Kralik M, Papesch W, Rank D, Scheifinger H, Stichler W (2008) Deuterium excess in precipitation of
Alpine regions - Moisture recycling. Isotopes Environ Health Stud 44:1–10
Gat JR (1984) The stable isotope composition of Dead Sea waters. Earth Planet Sci Lett 71:361–376
Gat JR (1996) Oxygen and hydrogen isotopes in the hydrologic cycle. Ann Rev Earth Planet Sci 24:225–262
Gat JR, Bowser C (1991) The heavy isotope enrichment of water in coupled evaporative systems. In: Stable Isotope
Geochemistry: A Tribute to Samuel Epstein. Taylor Jr HP, O’Neil JR, Kaplan, IR (eds) Geochemical Society,
San Antonio p 159–169.
Gat JR, Matsui E (1991) Atmospheric water balance in the Amazon Basin: an isotopic evapotranspiration model. J
Geophys Res 96:13,179–13,188
Gat JR, Mook WG, Meijer HAJ (2000) Volume II: Atmospheric water. In: Environmental Isotopes in the Hydrological
Cycle. Mook WG (ed) UNESCO, Paris
Gázquez F, Mather I, Rolfe J, Evans NP, Herwartz D, Staubwasser M, Hodell DA (2015) Simultaneous analysis of
17 16
O/ O, 18O/16O and 2H/1H of gypsum hydration water by cavity ring-down laser spectroscopy. Rapid Commun
Gázquez F, Evans NP, Hodell DA (2017) Precise and accurate isotope fractionation factors (α17O, α18O and αD) for
water and CaSO4·2H2O (gypsum). Geochim Cosmochim Acta 198:259–270
Ghosh P, Adkins J, Affek H, Balta B, Guo W, Schauble EA, Schrag D, Eiler JM (2006) 13C–18O bonds in carbonate
minerals: A new kind of paleothermometer. Geochim Cosmochim Acta 70:1439–1456
Gonfiantini R (1986) Environmental isotopes in lake studies. In: Handbook of Environmental Isotope Geochemistry.
Fritz P, Fontes J-Ch (eds) Elsevier, Amsterdam p 119–168
Gonfiantini R, Wassenaar LI, Araguas-Araguas L (2020) Stable isotope fractionations in the evaporation of water: The
wind effect. Hydrol Process 1–12
Guillevic M, Bazin L, Landais A, Stowasser C, Masson-Delmotte V, Blunier T, Eynaud F, Falourd S, Michel E,
Minster B, Popp T, Prié F, Vinther BM (2014) Evidence for a three-phase sequence during Heinrich Stadial 4
using a multiproxy approach based on Greenland ice core records. Clim Past 10:2115–2133
Haese B, Werner M, Lohmann G (2013) Stable water isotopes in the coupled atmosphere–land surface model
ECHAM5-JSBACH. Geosci Model Dev 6:1463–1480
Hausmann P, Sussmann R, Trickl T, Schneider M (2017) A decadal time series of water vapor and D/H isotope ratios
above Zugspitze: transport patterns to central Europe. Atmos Chem Phys 17:7635–7651
He H, Smith RB (1999) Stable isotope composition of water vapor in the atmospheric boundary layer above the
forests of New England. J. Geophys Res Atmos 104:11657–11673
Herwartz D (2021) Triple oxygen isotopes variations in Earth’s crust. Rev Mineral Geochem 86:291–322
Horita J, Rozanski K, Cohen S (2008) Isotopes in environmental and health studies isotope effects in the evaporation
of water: a status report of the Craig–Gordon model. Isotopes Environ Health Stud 44:23–49
Hürkamp K, Zentner N, Reckerth A, Weishaupt S, Wetzel KF, Tschiersch J, Stumpp C (2019) Spatial and temporal variability
of snow isotopic composition on Mt. Zugspitze, Bavarian Alps, Germany. J Hydrol Hydromechanics 67:49–58
Jasechko S, Sharp ZD, Gibson JJ, Birks SJ, Yi Y, Fawcett PJ (2013) Terrestrial water fluxes dominated by transpiration.
Nature 496:347–350
Johnsen SJ, Dansgaard W, White JWC (1989) The origin of Arctic precipitation under present and glacial conditions.
Tellus 41B:452–468
Jouzel J, Merlivat L (1984) Deuterium and oxygen 18 in precipitation: modeling of the isotopic effects during snow
formation. J Geophys Res 89:11749–11757
Kaiser A, Scheifinger H, Kralik M, Papesch W, Rank D, Stichler W (2002) Links between meteorological conditions
and spatial/temporal variations in long-term isotope records from the Austrian precipitation network.
International conference on study of environmental change using isotope techniques, IAEA-CN−80/63
Kaiser J (2008) Reformulated 17O correction of mass spectrometric stable isotope measurements in carbon dioxide and a
critical appraisal of historic ‘absolute’ carbon and oxygen isotope ratios. Geochim Cosmochim Acta 72:1312–1334
Kneifel S, Redl S, Orlandi E, Löhnert U, Cadeddu MP, Turner DD, Chen M-T (2014) Absorption properties of
supercooled liquid water between 31 and 225 GHz: evaluation of absorption models using ground-based
observations. J Appl Meteorol Climatol 53:1028–1045
Kusakabe M, Matsuhisa Y (2008) Oxygen three-isotope ratios of silicate reference materials determined by direct
comparison with VSMOW-oxygen. Geochem J 42:309–317
426 Surma et al.
Landais A, Risi C, Bony S, Vimeux F, Descroix L, Falourd S, Bouygues A (2010) Combined measurements of 17Oexcess
and d-excess in African monsoon precipitation: Implications for evaluating convective parameterizations. Earth
Landais A, Ekaykin A, Barkan E, Winkler R, Luz B (2012a) Seasonal variations of 17O-excess and d-excess in snow
precipitation at Vostok station, East Antarctica. J Glaciol 58:725–733
Landais A, Steen-Larsen HC, Guillevic M, Masson-Delmotte V, Vinther B, Winkler R (2012b) Triple isotopic composition of
oxygen in surface snow and water vapor at NEEM (Greenland). Geochim Cosmochim Acta 77:304–316
Landais A, Capron E, Toucanne S, Rhodes R, Popp T, Vinther B, Minster B, Prié F (2018) Ice core evidence for
decoupling between midlatitude atmospheric water cycle and Greenland temperature during the last deglaciation.
Clim Past 14:1405–1415
Lawrence DM, Thornton PE, Oleson KW, Bonan GB (2007) The partitioning of evapotranspiration into transpiration, soil
evaporation, and canopy evaporation in a GCM: Impacts on land–atmosphere interaction. J Hydrometeorol 8:862–880
Li S, Levin NE, Soderberg K, Dennis KJ, Caylor KK (2017) Triple oxygen isotope composition of leaf waters in
Mpala, central Kenya. Earth Planet Sci Lett 468:38–50
Lin Y, Clayton RN, Huang L, Nakamura N, Lyons JR (2013a) Oxygen isotope anomaly observed in water vapor from
Alert, Canada and the implication for the stratosphere. PNAS 110:15,608–15,613
Lin Y, Clayton RN, Huang L, Nakamura N, Lyons JR (2013b) Reply to Miller: Concerning the oxygen isotope
anomaly observed in water vapor from Alert, Canada, and its stratospheric source. PNAS 110:E4568
Lohmann U, Henneberger J, Henneberg O, Fugal JP, Bühl J, Kanji ZA (2016) Persistence of orographic mixed-phase
clouds. Geophys Res Lett 43:10,512–10,519
Lowenthal D, Hallar AG, McCubbin I, David R, Borys R, Blossey P, Muhlbauer A, Kuang Z, Moore M (2016)
Isotopic fractionation in wintertime orographic clouds. J Atmos Oceanic Technol 33:2663–2678
Luz B, Barkan E, Yam R, Shemesh A (2009) Fractionation of oxygen and hydrogen isotopes in evaporating water.
Majoube M (1971) Fractionnement en oxygène 18 entre la glace et la vapeur d’eau. J Clim Phys 68:625–636
Masson-Delmotte V, Jouzel J, Landais A, Stievenard M, Johnsen SJ, White JWC., Werner M, Sveinbjornsdottir A,
Fuhrer K (2005) GRIP deuterium excess reveals rapid and orbital-scale changes in Greenland moisture origin.
Science 309:118–121
Mathieu R, Bariac T (1996) A numerical model for the simulation of stable isotope profiles in drying soils. J Geophys
Res 101:12,685–12,696
Meijer HAJ, Li WJ (1998) The use of electrolysis for accurate δ17O and δ18O isotope measurements in water. Isotopes
Environ Health Stud 34:349–369
Merlivat L (1978) Molecular diffusivities of H216O, HD16O, and H218O in gases. J Chem Phys 69:2864–2871
Merlivat L, Nief G (1967) Fractionnement isotopique lors des changements d’état solide–vapeur et liquide–vapeur de
l’eau à des températures inférieures à 0°C. Tellus 19:122–127
Merlivat L, Jouzel J (1979) Global climatic interpretation of the deuterium–oxygen 18 relationship for precipitation.
Miller MF (2013) Oxygen isotope anomaly not present in water vapor from Alert, Canada. PNAS 110:E4567
Miller MF (2018) Precipitation regime influence on oxygen triple-isotope distributions in Antarctic precipitation and
ice cores. Earth Planet Sci Lett 481:316–327
NorthGRIP Members (2004) High-resolution record of Northern Hemisphere climate extending into the last
interglacial period. Nature 431:147–151
Pang H, Hou S, Landais A, Masson-Delmotte V, Jouzel J, Steen-Larsen HC, Risi C, Zhang W, Wu S, Li Y, An C,
Wang Y, Prie F, Minster B, Falourd S, Stenni B, Scarchilli C, Fujita K, Grigioni P (2019) Influence of summer
sublimation on δD, δ18O, and δ17O in precipitation, East Antarctica, and implications for climate reconstruction
from ice cores. J Geophys Res Atmos 124:7339–7358
implications for continental paleoclimate reconstruction. Rev Mineral Geochem 86:429–462
Peng H, Mayer B, Norman A-L, Krouse HR (2005) Modelling of hydrogen and oxygen isotope compositions for local
precipitation. Tellus 57B:273–282
Petit JR, Jouzel J, Raynaud D, Barkov NI, Barnola J-M, Basile I, Bender M, Chappellaz J, Davis M, Delaygue G, Delmotte
M, Kotlyakov VM, Legrand M, Lipenkov VY, Lorius C, Pépin L, Ritz C, Saltzman E, Stievenard M (1999) Climate and
atmospheric history of the past 420,000 years from the Vostok ice core, Antarctica. Nature 399:429–436
Pfahl S, Sodemann H (2014) What controls deuterium excess in global precipitation? Clim Past 10:771–781
Pfahl S, Wernli H (2008) Air parcel trajectory analysis of stable isotopes in water vapor in the eastern Mediterranean.
J Geophys Res Atmos 113:D20104
Risi C, Landais A, Winkler R, Vimeux F (2013) Can we determine what controls the spatio-temporal distribution of
d-excess and 17O-excess in precipitation using the LMDZ general circulation model? Clim Past 9:2173–2193
Rozanski BK, Sonntag C, Münnich K (1982) Factors controlling stable isotope composition of European precipitation.
Tellus 34:142–150
Schauer AJ, Schoenemann SW, Steig EJ (2016) Routine high-precision analysis of triple water-isotope ratios using
cavity ring-down spectroscopy. Rapid Commun Mass Spectrom 30:2059–2069
Schlaepfer DR, Ewers BE, Shuman BN, Williams DG, Frank JM, Massman WJ, Lauenroth WK (2014) Terrestrial
water fluxes dominated by transpiration: Comment. Ecosphere 5:61
Insights from an intermediate complexity isotope model. J Geophys Res Atmos 121:11,215–11,247
Schoenemann SW, Steig EJ, Ding Q, Markle BR, Schauer AJ (2014) Triple water-isotopologue record from WAIS Divide,
Antarctica: Controls on glacial-interglacial changes in 17Oexcess of precipitation. J Geophys Res Atmos 119:8741–8763
Sha L, Mahata S, Duan P, Luz B, Zhang P, Baker J, Zong B, Ning Y, Brahim YA, Zhang H, Edwards RL, Cheng H (2020)
A novel application of triple oxygen isotope ratios of speleothems. Geochim Cosmochim Acta 270:360–378
Sodemann H, Zubler E (2010) Seasonal and inter-annual variability of the moisture sources for Alpine precipitation
during 1995–2002. Int J Climatol 30:947–961
Steen-Larsen HC, Masson-Delmotte V, Sjolte J, Johnsen SJ, Vinther BM, Bréon FM, Clausen HB, Dahl-Jensen D, Falourd S,
Fettweis X, Gallée H, Jouzel J, Kageyama M, Lerche H, Minster B, Picard G, Punge HJ, Risi C, Salas D, Schwander
J, Steffen K, Sveinbjörnsdóttir AE, Svensson A, White J (2011) Understanding the climatic signal in the water stable
isotope records from the NEEM shallow firn/ice cores in northwest Greenland. J Geophys Res Atmos 116:D06108
Steen-Larsen HC, Risi C, Werner M, Yoshimura K, Masson-Delmotte V (2016) Evaluating the skills of isotope-enabled
general circulation models against in situ atmospheric water vapor isotope observations. J Geophys Res 122:246–263
Steen-Larsen HC, Sveinbjörnsdottir AE, Peters AJ, Masson-Delmotte V, Guishard MP, Hsiao G, Jouzel J, Noone D,
Warren JK, White JWC (2014) Climatic controls on water vapor deuterium excess in the marine boundary layer
of the North Atlantic based on 500 days of in situ, continuous measurements. Atmos Chem Phys 14:7741–7756
Stein AF, Draxler RR, Rolph GD, Stunder BJB, Cohen MD, Ngan F (2015) NOAA’s HYSPLIT Atmospheric
Transport and Dispersion Modeling System. Bull Am Meteorol Soc 90:2059–2077
Stewart MK (1975) Stable Isotope Fractionation Due to Evaporation and Isotopic Exchange of Falling Waterdrops:
Applications to Atmospheric Processes and Evaporation of Lakes. J Geophys Res 80:1133–1146
Strasser U, Bernhardt M, Weber M, Liston GE, Mauser W (2008) Is snow sublimation important in the alpine water
balance? Cryosphere 2:53–66
Stumpp C, Klaus J, Stichler W (2014) Analysis of long-term stable isotopic composition in German precipitation. J
Hydrol 517:351–361
Surma J, Assonov S, Bolourchi MJ, Staubwasser M (2015) Triple oxygen isotope signatures in evaporated water
snowfall in the central United States. Sci Rep 8:6712
Tian C, Wang L, Tian F, Zhao S, Jiao W (2019) Spatial and temporal variations of tap water 17O-excess in China.
428 Surma et al.
Touzeau A, Landais A, Stenni B, Uemura R, Fukui K, Fujita S, Guilbaud S, Ekaykin A, Casado M, Barkan E, Luz
B, Magand O, Teste G, Le Meur E, Baroni M, Savarino J, Bourgeois I, Risi C (2016) Acquisition of isotopic
composition for surface snow in East Antarctica and the links to climatic parameters. Cryosphere 10:837–852
Trenberth KE (1999) Atmospheric moisture recycling: role of advection and local evaporation. J Clim 12:1368–1381
Trenberth KE, Smith L, Qian T, Dai A, Fasullo J (2007) Estimates of the global water budget and its annual cycle using
observational and model data. J Hydrometeorol 8:758–769
Trickl T, Feldmann H, Kanter H-J, Scheel H-E, Sprenger M, Stohl A, Wernli H (2010) Forecasted deep stratospheric
intrusions over Central Europe: Case studies and climatologies. Atmos Chem Phys 10:499–524
precipitation on a subtropical island. Earth Planet Sci Lett 513:20–28
van der Ent RJ, Savenije HHG, Schaefli B, Steele-Dunne SC (2010) Origin and fate of atmospheric moisture over
continents. Water Resour Res 46:W09525
Van Hook WA (1968) Vapor pressures of the isotopic waters and ices. J Phys Chem 72:1234–1244
Vimeux F, Masson V, Jouzel J, Stievenard M, Petit JR (1999) Glacial–interglacial changes in ocean surface conditions
in the Southern Hemisphere. Nature 398:410–413
Vogelmann H, Sussmann R, Trickl T, Reichert A (2015) Spatiotemporal variability of water vapor investigated using
lidar and FTIR vertical soundings above the Zugspitze. Atmos Chem Phys 15:3135–3148
Voigt C, Herwartz D, Dorador C, Staubwasser M (2020) Triple oxygen isotope systematics of evaporation and mixing
processes in a dynamic desert lake system. Hydrol Earth Syst Sci Discuss in review
Wang XF, Yakir D (2000) Using stable isotopes of water in evapotranspiration studies. Hydrol Process 14:1407–1421
Wei Z, Lee X (2019) The utility of near-surface water vapor deuterium excess as an indicator of atmospheric moisture
source. J Hydrol 577:123923
Welp LR, Lee X, Griffis TJ, Wen X-F, Xiao W, Li S, Sun X, Hu Z, Val Martin M, Huang J (2012) A meta-analysis of water
vapor deuterium-excess in the midlatitude atmospheric surface layer. Global Biogeochem Cycles 26:GB3021
Winkler R, Landais A, Risi C, Baroni M, Ekaykin A, Jouzel J, Petit JR, Prie F, Minster B, Falourd S (2013) Interannual
variation of water isotopologues at Vostok indicates a contribution from stratospheric water vapor. PNAS
110:17674–17679
Winkler R, Landais A, Sodemann H, Dümbgen L, Prié F, Masson-Delmotte V, Stenni B, Jouzel J (2012) Deglaciation
records of 17O-excess in East Antarctica: reliable reconstruction of oceanic normalized relative humidity from
coastal sites. Clim Past 8:1–16
Yu W, Tian L, Ma Y, Xu B, Qu D (2015) Simultaneous monitoring of stable oxygen isotope composition in water
vapour and precipitation over the central Tibetan Plateau. Atmos Chem Phys 15:10,251–10,262
Zakharov DO, Bindeman IN, Serebryakov NS, Prave AR, Azimov PY, Babarina II (2019a) Low δ18O rocks in the
Zakharov DO, Bindeman IN, Tanaka R, Friðleifsson GÓ, Reed MH, Hampton RL (2019b) Triple oxygen isotope systematics
as a tracer of fluids in the crust: A study from modern geothermal systems of Iceland. Chem Geol 530:119312
Zakharov DO, Marin-Carbonne J, Alleon J, Bindeman, IN (2021) Temporal triple oxygen isotope trend recorded by
Precambrian cherts: A perspective from combined bulk and in situ secondary ion probe measurements. Rev
Vol. 86 pp. 429–462, 2021 13
Triple Oxygen Isotopes in Meteoric Waters, Carbonates,

and Biological Apatites: Implications for
Continental Paleoclimate Reconstruction
Benjamin H. Passey and Naomi E. Levin
Department of Earth and Environmental Sciences
University of Michigan
1100 North University Avenue
Ann Arbor, MI, 48103
USA
passey@umich.edu
nelevin@umich.edu
INTRODUCTION
The usefulness of triple isotope studies of natural systems is contingent on the
existence of resolvable differences in mass-dependent fractionation exponents θ [where
θ = ln(17/16α)/ln(18/16α)] among processes that are prevalent in the system(s) of interest, or
on the existence of resolvable non-mass-dependent fractionation in the system. Both such
contingencies are satisfied in continental hydroclimate systems. The process of evaporation of
water involves molecular diffusion of water vapor through air, which carries a θ value of 0.5185
(Barkan and Luz 2007) that differs strongly from the value of 0.529 for equilibrium exchange
between water vapor and liquid water (Barkan and Luz 2005). This means that waters that
have been isotopically modified by evaporation can, in many cases, be clearly identified on the
basis of triple oxygen isotope analysis (Landais et al. 2006; Surma et al. 2015, 2018; Gázquez
et al. 2018), and that it may even be possible to reconstruct the isotopic composition of the
unevaporated source waters (Passey and Ji 2019). Leaf waters are highly evaporated and have
distinctive triple oxygen isotope compositions (Landais et al. 2006; Li et al. 2017), which has
implications for the triple oxygen isotope compositions of atmospheric CO2 and O2 (Liang et al.
2017), plant materials such as cellulose and phytoliths (Alexandre et al. 2018), and animals that
ingest significant amounts of leaf water (Pack et al. 2013; Passey et al. 2014).
Molecular oxygen in the atmosphere is strongly fractionated in a non-mass-dependent way
due to photochemically-driven isotopic exchange in the stratosphere (Luz et al. 1999; Bao et al.
2008; Pack 2021, this volume), with the degree of fractionation dependent on pCO2, pO2, and
global gross primary productivity (GPP) (Cao and Bao 2013; Young et al. 2014). Through the
process of respiration, this oxygen with ∆′17O of approximately −430 per meg (Barkan and Luz
2005; Pack et al. 2017; Wostbrock et al. 2020) can become part of body water and hence biominerals
(Pack et al. 2013). Note that in this chapter, we calculate ∆′17O as: ∆′17O = δ′17O – 0.528 δ′18O,
where δ′1xO = ln(x/16Rsample / x/16Rstandard), where x is 17 or 18. We report all δ and δ′ values in
per mil (×103), and all ∆′17O values in per meg (×106). This definition of ∆′17O is equivalent to
the ‘17O-excess’ parameter commonly used in the hydrological literature.
Recently, Guo and Zhou (2019a,b) have predicted from theory that kinetic fractionation
may impart materials like speleothems and corals with distinct triple oxygen isotope
compositions that testify to the extent of isotopic disequilibrium during mineralization.
1943-2666/21/0086-0013$05.00 (online)
430 Passey & Levin
As such, the continents are emerging as a particularly fruitful area of application for
triple oxygen isotope studies, with ∆′17O values strongly influenced by evaporation, possibly
influenced by kinetic fractionation during mineralization, and, for organisms with waters of
respiration comprising a large fraction of their overall water budget, by the unique ∆′17O values
of atmospheric O2. Triple oxygen isotopes are potentially useful in systems ranging from lakes to
soils to caves to plants to animals. At the time of this writing, very little information exists for soils
and caves, so we focus our review on lakes, animal body waters, plant waters, and on the isotopic
compositions of carbonates and bioapatites forming in natural waters and animals, respectively.
Precipitation (rain, snow) is the source of water on the continents, and thus is the triple
oxygen isotope ‘starting point’ for all of these systems. The triple oxygen isotope compositions
of continental carbonates and biological apatites are, to first-order, controlled by the isotopic
compositions of their parent waters (e.g., lake water, animal body water). Therefore, much
of this chapter focuses on the triple oxygen isotope compositions of natural waters and body
water. The format for this discussion is to model the triple oxygen isotope evolution of water
beginning with evaporation from the oceans, followed by Rayleigh distillation and production
of continental precipitation, evaporative modification of this precipitation on the continents,
and the triple oxygen isotope compositions of animals and plants that are sustained by this
precipitation. All along this journey, we present the basic governing equations for modeling
isotopic compositions, and we present observed triple oxygen isotope data in the context of
these models. As much as possible, we present the model predictions and data in the form of
∆′17O versus d18O bivariate plots, keeping the axis ranges uniform to give the reader a better
context for the comparative absolute values and variability from one system to another.
Useful application of triple oxygen isotopes in paleoclimate depends on the ability to
determine the triple oxygen isotope compositions of common minerals such as carbonates and
apatites to high precision and accuracy. Following our discussion of natural and biological
waters, we review the exciting developments of the past decade with respect to the analysis
of carbonates and apatites and summarize the current state of inter-laboratory reproducibility.
We examine the existing knowledge of key fractionation exponents for carbonate–water
fractionation and fractionation during acid digestion.
We conclude the chapter by highlighting gaps in knowledge and opportunities for future
research. Throughout this chapter, we use isotopic nomenclature as described in Miller and
Pack (2021, this volume).
METEORIC WATERS
Evaporation from the oceans
The oceans are the dominant source of water vapor to the atmosphere. The triple oxygen
isotope composition of the oceans is d18O ~ 0‰, ∆′17O = −5 per meg (Luz and Barkan
2010), From this point of origin, there are four principle factors that determine the isotopic
composition RE of the evaporation flux reaching the free atmosphere: 1) the (temperature-
dependent) equilibrium isotope fractionation between liquid and vapor, aeq; 2) the relative
humidity of air, normalized to the surface temperature of the water, hn; 3) the relative
contributions of molecular diffusion and turbulence in transporting water vapor to the free
atmosphere (embodied in the adiff parameter), and 4) the isotopic composition of ambient
atmospheric water vapor, Ra. The isotopic composition of the evaporative flux is given by:
Rw
 hRa
 eq (1)
RE 
 diff 1  hn 
Implications for Continental Paleoclimate Reconstruction 431
For adaptation to triple oxygen isotopes, equations such as Equation (1) are used twice: first
in the 18O/16O space to determine 18/16RE, and then in the 17O/16O space to determine 17/16RE.
For the latter, 17/16α values are computed from 18/16α values using appropriate fractionation
exponents θ and the relation θ = ln(17/16α)/ln(18/16α). Then, d17O and d18O values are calculated
from the R values using the appropriate R values for VSMOW, and from these the ∆′17O value
is calculated after converting d1xO values to δ′1xO values.
As pointed out by Luz and Barkan (2010), the basic form of Equation (1) has been obtained
from the standpoints of diffusion theory (Ehhalt and Knott 1965), kinetic theory (Criss 1999),
and from a steady-state evaporation model (Cappa et al. 2003). We note that Equation (1)
is a key equation for describing the isotopic evolution of evaporating bodies of water on
land, as discussed in the next section. Equation (1) also arises from the Craig–Gordon linear-
resistance model (Craig and Gordon 1965; see also Gat 1996, and Horita et al. 2008), which
envisions an equilibrium vapor layer immediately overlying the liquid water surface, with
further fractionation resulting from transport of this equilibrium vapor to the free atmosphere
via a combination of molecular diffusion and turbulence. Pure molecular diffusion carries
fractionations of 18/16α = 1.028 and θ = 0.5185 (as determined experimentally by Merlivat 1978
and Barkan and Luz 2007, respectively; see discussion in the latter and also Horita et al. (2008)
and Cappa et al. (2003) regarding discrepancies between experimentally-derived 18/16α values
and the theoretical value, 1.032, and possible explanations for these discrepancies). Thus,
in the λ = 0.528 reference frame, diffusion-dominated evaporation involves a substantially
higher ∆′17O value for the vapor, and, by mass balance, a corresponding lowering of ∆′17O
of the residual liquid (Fig. 1), while turbulence-dominated evaporation will result in less
elevated ∆′17Ο of vapors and more modest lowering of ∆′17O in the residual liquid. To first-
order, turbulence scales with windiness: in stagnant air, which occurs inside of leaves and
soils, and can be achieved in laboratory experiments, 18/16adiff approaches the 1.028 limit for
A
free atmosphere
c d
boundary layer
ion
tu
hn < 1
rb
us
ul
diff
en
equilibrium layer b
ce
hn = 1 equilibrium
Figure 1. Schematic depiction of triple oxygen isotope fractionation
a
during evaporation. A. In the Craig–Gordon model, an equilibrium
B trajectory of residual liquid for: layer of water vapor (b) with hn = 1 lies in contact with the water
turbulent vapor transport surface (a). From here, vapor is transported to the atmosphere via a
combination of turbulent transport, which is non-fractionating, and
8)
molecular diffusion, which is fractionating. The vapor produced by
not to scale 52
=
0. each process are noted by (c) and (d) in the figure. Note that the
(λ
=
0 Craig–Gordon model, these two processes are linked in series and not
1 7 O a
Δ’ in parallel as this simplified endmember schematic might imply. The
δ17O 0.5185
29
schematic also does not depict isotopic exchange with ambient wa-
0.5
b,c
d
ter vapor. B, C. Evaporation in triple oxygen isotope space, showing
the relative trajectories for vapor and residual liquid in the turbulent
transport endmember (c and double arrow, respectively), and for va-
δ18O
por and residual liquid in the diffusion endmember (d, hand pointer,
respectively). Liquid a and vapor b are linked by the equilibrium frac-
C + tionation exponent qeq = 0.529 (Barkan and Luz 2005), and molecular
d 0.5 diffusion (b to d) has a θ value of 0.5185 (Barkan and Luz 2007).
18
5 Δ’17O = 0 ( λ = 0.528)
0
Δ’17O b,c a
(λ = 0.528) 0.529
δ18O
432 Passey & Levin
molecular diffusion. Combined with the qdiff value of 0.5185, this leads to highly-modified
∆′17O values of vapor and residual liquid. In windy conditions, turbulence develops, 18/16adiff
values become substantially smaller, and ∆′17O values are less extreme even though the qdiff
value is unchanged. This is because deviations in ∆′17O are dependent both on the magnitude of
isotopic fraction (18/16α) and the numerical difference between the θ value of the fractionation
and the reference frame slope (here 0.528) (Guo and Zhou 2019a):
17O
AB  A B  0.528   ln  18/16  A B  (2)
where ∆∆′17ΟA–B is the difference in ∆′17O between two phases that exchange isotopes, θA–B is
the fractionation exponent between these two phases, and 18/16αA–B is the fractionation factor
between these phases.
A simplifying assumption when treating evaporation of seawater is the ‘closure
assumption’ (Merlivat and Jouzel 1979), which assumes that the only source of water vapor
in the atmosphere is evaporation from the oceans, so that RE = Ra. Inserting this equivalency
into Equation (1) gives:
RE Rw  eq   diff 1  hn   hn  (3)
Uemura et al. (2010) determined a value of 18/16adiff = 1.008 for evaporation from the South
Indian and Southern Oceans, meaning that the fractionation during vapor transport is closer
to the turbulence endmember (18/16adiff = 1) than it is to the molecular diffusion endmember
(18/16adiff = 1.028). Using 18/16adiff = 1.008, we use Equation (3) to model the isotopic compositions
of marine vapors and the theoretical initial condensates of those vapors for hn of 0.5, 0.7 and 0.9
(Fig. 2). As can be seen, lower relative humidity results in higher ∆′17O values of the water vapor
flux, which ultimately will lead to higher ∆′17O values of condensates, including precipitation
on the continents. This arises from a combination of enhanced diffusion due to the greater water
vapor concentration gradient between the boundary layer and free atmosphere, as well as the
decreased influence of ambient atmospheric water vapor at lower relative humidity, in terms of
isotopic exchange. In general, the positive ∆′17O values of global precipitation can be attributed
to (1) the influence of molecular diffusion during evaporation from the oceans, and (2) the fact
that condensation has a θ value of 0.529, slightly steeper than the 0.528 reference slope, which
imparts condensates with a slightly higher ∆′17O value than the parent vapors (Fig. 2B).
Rayleigh distillation and the triple oxygen isotope meteoric water line
Rayleigh distillation theory has stood the test of time as the explanation for the first-order
features of the δD – d18O global meteoric water line (GMWL) (Craig 1961; Dansgaard 1964),
and Rayleigh theory also accurately predicts the general features of the triple oxygen isotope
GMWL (Luz and Barkan 2010). The Rayleigh equation for water vapor in a condensing parcel
of air can be given as:
R f  R0 f
(  eq 1) (4)
where f is fraction of water vapor that remains in a parcel of air relative to the initial
amount of water vapor, R0 is the initial isotopic composition of water vapor in the parcel,
Rf is the instantaneous isotopic composition of vapor when fraction f remains, and aeq is the
equilibrium fractionation between liquid water and water vapor. The adiff parameter does not
appear in Equation (4) because the condensation of liquid requires that relative humidity is
100%; hence there is no concentration gradient in water vapor, and therefore no net transport
by molecular diffusion. The only isotopic fractionation at play is equilibrium between liquid
water and water vapor, which carries a fractionation exponent of 0.529 (Barkan and Luz 2005).
Evaporation of seawater to form atmospheric water vapor

B
Δʹ17O (per meg, SMOW-SLAP, λ = 0.528)

A pure turbulently pure molecularly- 100
initial
transported diffused vapor initial condensates
vapor (αdiff = 1) (αdiff = 1.028) 50 vapors
free atmosphere δ18O = –9.3‰ δ18O = –17.6‰ hn = 0.5
h = 0.7 Δʹ17O = –14 p.m. Δʹ17O = 85 p.m. 0.7
0
0.9 seawater
boundary layer
tu
-50
rb
h<1
ul
en
equilibrium layer
ce
equilib. vapor -100

δ18O = –9.3‰
h=1
equilibrium Δʹ17O = –14 p.m. -150
αdiff = 1.008
seawater
-200
δ18O = 0‰
-30 -20 -10 0 10 20 30
Δʹ17O = –5 p.m.
δ18O (‰, SMOW-SLAP)
Figure 2. Evaporation of seawater to form atmospheric water vapor. A. As in Figure 1A, but showing triple
oxygen isotopic compositions for equilibrium vapors, the pure turbulently-transported vapor endmem-
ber, and the pure molecularly-diffused vapor endmember. These values were calculated for 25 °C using
18/16
aeq = 1.0093 (Majoube 1971), qeq = 0.529 (Barkan and Luz 2005), 18/16adiff = 1.028 (Merlivat and Jouzel
1978) and qdiff = 0.5185 (Barkan and Luz 2007). B. Triple oxygen isotope compositions of initial vapors
calculated using Equation (3), and initial condensates from those vapors for different humidities. Here we
use 18/16adiff = 1.008, the best-fit value for marine vapors in the South Indian and Southern Oceans observed
by Uemura et al. 2010. Abbreviation: p. m. : per meg.
This commonly leads to the mistaken assumption that the ‘theoretical’ slope of the triple
oxygen isotope GMWL is also 0.529. However, evaluation of Equation (4) (see Fig. 3) shows
that this is not the case: for constant values of 18/16aeq = 1.0093 (Majoube 1971; 25 °C) and
qeq = 0.529, the GMWL slope is 0.5278. If 18/16aeq is allowed to increase in a systematic manner
as f decreases (as must be the case, because decreasing f can only be driven by decreasing
temperatures, which drives saturation and hence condensation), then slightly lower slopes are
observed. Note that qeq is not observed to vary significantly over Earth surface temperatures
(Barkan and Luz 2005). Luz and Barkan (2010) showed that the slope γ of the GMWL in a
pure Rayleigh distillation model can be given by:
17
 eq  1 (5)
 18
 eq  1
with the slope ranging from 0.5275 to 0.5279 over the temperature range 0–30 °C.
The canonical slope of 0.528 for the GMWL historically arises from Equation (5) and
from the initial observations of global meteoric water triple oxygen isotope compositions
(Landais et al. 2008; Luz and Barkan 2010). However, considerable additional data have been
published since 2010, and it is now clear (e.g., Sharp et al. 2018) that the slope is lower in
tropical + temperate climates. Figure 4 presents an up-to-date compilation from the literature,
excluding polar regions where d18O < −30‰, and screened for waters subject to evaporation
(lakes and ponds) (i.e., it is a precipitation dataset, not a meteoric water data set). Additionally,
the data are weighted where possible for amount of precipitation, which includes the use of
weekly- or monthly-averages over individual precipitation event values. Such weighting is
important for understanding the water inputs for lakes, soil waters, and animal body waters,
because we are interested in the amount-weighted isotopic composition of waters available to
these systems. For this dataset, the slope is 0.5272, with a clear tendency to ‘curve’ towards
lower ∆′17O values when d18O is ~ –10‰ and higher. The mechanism(s) for lower ∆′17O values
at higher d18O are as yet resolved, but following theoretical considerations presented above,
434 Passey & Levin
Rayleigh distillation by progressive rainout of atmospheric vapor masses

A B

100 relative humidity during
formation of initial vapors
f = 0.3 f = 0.5 f = 0.99
50 h = 0.5
0.7
0 0.9
δ18O = –13.6 ‰ δ18O = –8.8 ‰ f = 0.1 0.3 0.5 1
δ18O = –2.5 ‰
Δʹ17O = 21 p.m. Δʹ17O = 20 p.m. (initial
Δʹ17O = 19 p.m. condensate)
-50
fraction remaining of
initial vapor mass
-100
-150
-200
-30 -20 -10 0 10 20 30
Figure 3. Rayleigh distillation in triple oxygen isotope space. A. Schematic view of the process, with
isotopic compositions given for fractions remaining of initial vapor mass f = 0.99, 0.5, and 0.3. B. The pro-
cess in triple oxygen isotope space, modeled using Equation (4) for each of the three initial vapors shown
in Figure 2B. Note that this figure (3B) portrays the compositions of the condensates (i.e., precipitation).
Calculated using 18/16aeq = 1.0093 and qeq = 0.529.
A B δ18O =
100 -29.9 to -15‰
50 h = 0.5
0.7
0.9 -14.9 to -10‰
0
-50
-9.9 to -5‰
-100
-150 -4.9 to +2‰
-200
-30 -20 -10 0 10 20 30 -50 0 50 100
δ18O (‰, SMOW-SLAP) Δʹ17O (per meg)
Figure 4. Observed triple oxygen isotope compositions of global precipitation. A. Precipitation values
(+ signs) plotted in the context of the Rayleigh distillation models presented in Figure 3. B. Relative fre-
quency of ∆′17O values in 10 per meg bins, grouped according to d18O range. This illustrates the clear shift
towards lower ∆′17Ο values with higher d18O. ∆′17O values of precipitation lower than 0 per meg and higher
than 50 per meg are uncommon. Data are from Landais et al. (2008, 2010, 2012), Luz and Barkan (2010),
Li et al. (2015), Surma et al. (2015), Tian et al. (2018), Passey and Ji (2019), and Uechi and Uemura (2019).
viable mechanisms include higher relative humidity or a greater degree of turbulent transport
of water vapor during evaporation from the oceans, or increased re-evaporation of rainfall and
surface waters in settings where d18O is high.
Evaporation from isolated bodies of water
Precipitation is subject to re-evaporation following condensation, which takes place
during the journey of water back to the oceans. For isolated bodies of water evaporating into
a humid atmosphere (e.g., pan evaporation, raindrop re-evaporation), the isotopic ratio of the
residual water can be modeled as Stewart (1975), Criss (1999), Surma et al. (2015):
Rw f u  Rw,initial  Rw,ss   Rw,ss (6)
where Rw is the isotope ratio of the water body, f is the fraction of water remaining, Rw,initial is
the initial isotope ratio of the water, and where the exponent u is given by:
1   eq  diff 1  h  (7)
u
 eq  diff 1  h 
and with Rw,ss, the steady-state isotopic composition that the water body may reach at low f,
given by:
 eq hn Ra
Rw,ss  (8)
1   eq  diff 1  hn 
where atmospheric vapor is commonly assumed to be in equilibrium with the initial,
unevaporated water (Ra = RI/aeq). These equations predict that ∆′17O of the water body decreases
as evaporation proceeds (Fig. 5B). At higher hn, the evaporation trajectory is ‘steeper’ in a d18O
vs ∆′17O plot (lower effective λ value), but the steady-state ∆′17O and d18O values are higher
and lower, respectively, than for low hn. In other words, the lowest ∆′17O values are obtained
when f is low, hn is low, and (not illustrated in Fig. 5B), when 18/16adiff values are high (i.e.,
towards the 1.028 endmember). The predictions of Equations (6–8) are supported by a number
of laboratory evaporation experiments (e.g., Luz and Barkan 2010; Surma et al. 2015; Li et al.
2017), and by data from natural evaporating ponds (Surma et al. 2015, 2018). Equations (6–8)
are also used as a basis for modeling the evolution of evaporating raindrops, which is a more
complicated scenario because Ra is itself influenced by evaporation from the raindrop, because
adiff may vary according to the size and velocity of the raindrop, and because temperature
may change as the raindrop falls. For detailed treatments of evaporating rainfall, the reader is
referred to Landais et al. (2010), Bony et al. (2008), and references therein.
Closed-basin and throughflow lakes
Lakes are generally not isolated bodies of water, but rather have inflows of water (rivers,
groundwater, direct precipitation), outflows (rivers, groundwater), and evaporation. Following
Criss (1999), the isotopic mass balance of a lake can be described by:
(VRIw ) (9)
IRI  ORO  ERE
t
where V is lake volume, Rlw is the isotope ratio of lake water, I is the inflow flux (which
combines influx from rivers, groundwater, and precipitation), RI is the inflow isotope ratio,
O is the outflow flux, RO is the outflow isotope ratio, E is the evaporation flux, and RE is the
isotope ratio of water lost by evaporation. For a well-mixed lake with a constant volume and
in isotopic steady state, Equation (9) becomes:
I
I RI ORO  ERE (10)
In a closed basin lake at isotopic steady state, the isotope ratio of water leaving the lake
by evaporation must be equivalent to the isotope ratio of the water flux into the lake (RE = RI).
Inserting thie equivalency into Equation (1) and solving for Rw (which we rename Rlw) gives:
Rlw  eq  diff RI 1  hn    eq hRa (11)

436 Passey & Levin
Evaporation of water bodies on land

A Transient evaporation of isolated B

100
water bodies (pan evaporation) fraction remaining
initial waters of initial water body
50
0.7 0.7
0 f= f=
0.5 0.5
RE -50 hn = 0.8
0.3 0.3
-100
hn
Rw
=0
-150 hn = 0.6 0.1
.3
0.1
αdiff = 1.0142
-200
-30 -20 -10 0 10 20 30
C Steady-State Evaporation D
(flow-through lakes, closed-basin lakes) 100
RE
50
initial waters
Rw RO 0
RI
hn = 0.8
-50 0.6
RE 0.3
-100
evaporated waters
-150
RI Rw XE = 1, αdiff = 1.0142
-200
-30 -20 -10 0 10 20 30
E F δ18O (‰, SMOW-SLAP)
100 100
50 50
initial waters initial waters
hn = 0.8 hn = 0.8
0 0 0.6
0.6
0.3
0.3
-50 -50
evaporated waters
evaporated waters
-100 -100
-150 -150
XE = 0.5, αdiff = 1.0142 XE = 0.25, αdiff = 1.0142
-200 -200
-30 -20 -10 0 10 20 30 -30 -20 -10 0 10 20 30
δ18O (‰, SMOW-SLAP) δ18O (‰, SMOW-SLAP)
Figure 5. Models of open pan and steady-state evaporation, starting with initial waters modeled in Figure 3
for hn = 0.7. A, B. Transient (open pan) evaporation, modeled using Equations (6–8) for a range of hn, and
constant 18/16adiff = 1.0142. Initial waters are shown by triangles, and isotopic compositions at different stages
in the progress of evaporation are shown by circles. Fractional extents of evaporation (f) are noted in italicized
text. The curved solid arrows show the evaporation trajectories for different values of hn, and the endpoints of
the arrows for hn = 0.8 and hn = 0.6 are the final, steady state compositions of the water (the steady-state com-
positions for hn = 0.3 fall off the plot). Sub-horizontal gray lines connect the common fractional extents of
evaporation at different hn. C–F. Steady-state isotopic compositions calculated using Equation (16) for closed
basin lakes (XE = 1; D, E) and a flow through lakes (XE = 0.5, 0.25). Note that in contrast to Figure 5B, the
solid angled lines are not evaporation trajectories, but rather connect the steady-state isotopic compositions
for evaporation of the same initial water compositions at different relative humidity. The models in D, E, and
F differ only in XE, illustrating the sensitivity to this parameter. For all models, 18/16aeq = 1.0093, qeq = 0.529,
qdiff = 0.5185, and Ra is assumed to be in isotopic equilibrium with the initial (inflow) waters.
For a throughflow lake at steady-state, we can denote the fraction of water leaving the lake
by evaporation as:
E (12)
XE 
I
Combining Equation (12) with Equation (10) gives:
X E RE  1  X E  Ro  RI (13)
For a well-mixed lake, RO = Rlw (i.e., the outflow composition is the same as the lake
composition). Inserting this equivalency into Equation (13) and solving for RE gives:
RI  1  X E  Rlw (14)
RE 
XE
Combining this equation with Equation (1) gives:
RI  1  X E  Rlw R   eq hn Ra (15)
 w
XE  eq  diff 1  hn 
Finally, multiplying both sides by the denominator on the right hand side and solving for
Rlw gives a general solution for both closed basin lakes (XE = 1) and flowthrough lakes (XE < 1)
(cf., Criss 1999, Eqn. 4.46c, where Criss’ a0evap = aeqadiff):
 eq  diff 1  hn  RI  X E hn  eq Ra (16)
Rlw 
 eq  diff 1  hn  1  X E   X E
Figures 5D–F show the predictions of this equation, evaluated for a range of initial
water oxygen isotopic compositions, relative humidities, and XE values, and assuming that
atmospheric vapor is in equilibrium with inflow waters (Ra = RI/aeq). Note that these are
steady-state isotopic compositions, and not transient evaporation trajectories as shown for pan
evaporation in Figure 5B. Figures 5D–F show that ∆′17O is most changed when hn is low, and
fraction of evaporation from the lake (XE) is high.
The models shown in Figure 5 all hold Ra and adiff constant, whereas lake water ∆′17O and
d O values are highly sensitive to these parameters (Fig. 6). The sensitivity to Ra is greatest
18
at high relative humidity (Fig. 6A), in part because humid air holds more water vapor that
can exchange with surface waters. The sensitivity to adiff is greatest at low relative humidity
(Fig. 6B) due to the steeper water vapor concentration gradient between the boundary layer
and the free atmosphere at low humidity.
A final factor is the influence of atmospheric water vapor derived from evaporation of the
the water body itself (e.g., Benson and White 1994; Gat et al. 1994; Gat 1996; Gibson et al.
2016). Here, Equation (16) can be modified to (Passey and Ji 2019):
 eq RI  diff 1  hn   hn 1  F     eq hn X E Ra F (17)

Rlw 
X E   eq 1  X E   diff 1  hn   hn 1  F  
where F is the fraction of water derived from regional sources upwind of the lake, and (1 − F)
is the fraction of moisture derived from the lake (note that this equation does not explicitly
438 Passey & Levin
A 150 B 150
Surma et al. 2020 non-stagnant stagnant
ΔΔʹ17O of lake water (per meg)

experiments experiments
100 100
Uemura et al. 2010
U
50 50
0 0
-50 -50
hn = 0.8 hn = 0.8
-100 hn = 0.6 -100 hn = 0.6
hn = 0.3 hn = 0.3
-150 -150
-40 -20 0 20 40 60 80 100 1.000 1.010 1.020 1.030
Δʹ17O of atmospheric vapor (per meg) 18/16αdiff
C 150
100 increasing vapor

from lake evaporation
50
-50
hn = 0.8
-100 hn = 0.6
hn = 0.3
-150
0 0.2 0.4 0.6 0.8 1.0
Fraction of regional atmospheric vapor
Figure 6. Sensitivity of closed-basin lake water ∆′17O values to changes in ∆′17O of atmospheric vapor
(A), 18/16adiff (B), and fraction of atmospheric vapor derived from regional sources in equilibrium with pre-
cipitation (C). In all figures, the vertical axis reflects the difference in ∆′17O compared to the base model in
Figure 5D. The horizontal brackets in (A) represent the range of atmospheric water vapor values observed
in the marine environment by Uemura et al. (2010), and in the European Alps by Surma et al. (2021, this
volume). The horizontal brackets in (B) represent the range of values for laboratory experiments where air
above the evaporating water was stagnant and non-stagnant, as compiled by Horita et al. (2008; Table 2). The
‘U’ shows the best-fit 18/16adiff value for marine evaporation determined by Uemura et al. (2010), and the star
show the ‘canonical’ 1.014 value commonly assumed for continental evaporation. In (C), a fractional value
of 0 means that all atmospheric vapor over a lake is derived from evaporation from the lake itself, while a val-
ue of 1 means that all atmospheric vapor is derived from regional sources in equilibrium with precipitation.
account for the influence of lake evaporation on hn). F can approach 0 for lakes in arid climates
(e.g., Pyramid Lake, Nevada; Benson and White 1994), while Gat et al. (1994) estimate that F
ranges between 0.84 and 0.95 for the more humid Great Lakes of North America. In general,
changes in F have a smaller influence on the isotopic composition of lake water than do
changes in Ra and adiff (Fig. 6C).
Special considerations and case studies
The collective results from modeling and field studies of modern lakes and ponds (Luz
and Barkan 2010; Surma et al. 2015, 2018; Gázquez et al. 2018; Passey and Ji 2019) suggest a
number of approaches and challenges for using triple oxygen isotopes in modern and past lakes.
The essential challenge to quantitative paleoclimate reconstruction in lacustrine environments
is that there are six variables in Equation (17) (RI, Ra, hn, XE, adiff, F), whereas only two
isotope ratios (17O/16O, 18O/16O) are measured. How then can we reliably reconstruct any of
the remaining variables of this underconstrained system? Fortunately, there is a hierarchy in

sensitivity to these variables (approximately XE > hn > adiff > Ra > RI > F)), so a lack of direct
constraints on e.g., RI and F may not greatly hinder interpretations in certain cases. Here we
discuss some additional considerations, not mutually exclusive, for sound interpretation of
triple oxygen isotope data, and do so in the context of published results where appropriate.
Qualitative indicator of evaporation. Authigenic minerals are commonly used to
reconstruct past climates or past surface elevation in tectonic settings (e.g. Rowley and Garzione
2007), and both approaches generally assume that the reconstructed d18O values are reflective
of primary precipitation, with recognition that evaporation can throw interpretations into serious
error. Despite the large number of variables in Equation (17), all scenarios of evaporation lead to
lowering of ∆′17O of surface waters, save perhaps for cases where ∆′17O of atmospheric vapor is
unusually high. ∆′17O values lower than 0 per meg are uncommon for unevaporated precipitation
(Fig. 4), whereas values lower than 0 per meg are common for evaporated waters (Surma et al.
2015, 2018; Gázquez et al. 2018; Passey and Ji 2019). Therefore, at the most basic level, triple
oxygen isotope analysis will reveal whether the mineral parent waters are likely to have been
modified by evaporation (∆′17O < ~ 0 per meg), and thus are unsuitable for interpretation in terms
of d18O of pristine precipitation. Low ∆′17O values will also qualitatively indicate hydrological
and climatic conditions conducive for appreciable evaporation (high XE, low hn).
Directional constraints. As ∆′17O values of lake water becomes lower, the parameter space
that can account for the values diminishes. For example, ∆′17O values of ~ 0 per meg are consistent
with very low relative humidity but little net evaporation (Figs. 5B, F), as well as high relative
humidity and appreciable evaporation (Figs. 5D, E). On the other hand, ∆′17O values of lower
than ~ –50 per meg only occur when net evaporation is > 50 %, and relative humidity is ~0.6 or
lower (Figs. 5B, D, E). Therefore, although ∆′17O data may not permit precise reconstruction of
parameters such as hn or XE, they can be used to rule out portions of the parameter space.
Independently-constrained variables. Gázquez et al. (2018) presented the first application
of lacustrine triple oxygen isotope systematics to the paleoclimate record. Following methods
described by Gázquez et al. (2017), they analyzed triple oxygen isotopes and hydrogen isotopes
in waters of hydration of lacustrine gypsum extending back ~15 kyr at Lake Estanya in Spain.
Given that gypsum formation requires closed-basin or nearly closed basin settings, gypsum-
based studies are able isolate XE to values near 1. Gázquez et al. (2018) then used a Monte-
Carlo approach to randomly vary the other parameters in an equation similar to Equation (17),
and then selected only the parameter sets that correctly predicted the measured triple oxygen
and deuterium isotopic compositions of the gypsum hydration waters. This approach allowed
them to constrain changes in normalized atmospheric relative humidity over this time interval,
and to robustly conclude that hn during the Younger Dryas was substantially lower than
during the late Holocene (by ~20% or more on the 0–100% hn scale; see their Fig. 5). Their
inferred absolute values of hn for the Younger Dryas and late Holocene are more dependent on
assumptions made about the 18/16adiff parameter (see their Fig. 5c).
Passey and Ji (2019) sought to examine the predictions of Equation (17) by studying
modern lakes where many of the free parameters in Equation (17) could be constrained. They
examined three closed-basin lakes (XE = 1) (Great Salt Lake, Mono Lake, Pyramid Lake), each
having three or fewer major inflowing rivers that account for the vast majority of water input into
the lake, hence allowing RI to be constrained. Additionally, detailed climatologies exist for this
region, including monthly relative humidity and potential evapotranspiration, which together
allow for estimates of the effective relative humidity weighted to the months with highest
evaporation (here, the summer months). For Great Salt Lake, Pyramid Lake, and Mono Lake, the
evaporation-weighted relative humidities are 43%, 42%, and 44%, respectively, based on data
from New et al. (2002). A compilation of monthly lake water and air temperature data from 88
lakes globally in Hren and Sheldon (2012) shows that summer water temperatures are generally
440 Passey & Levin
similar or slightly warmer (by a few degrees) than summer air temperature; thus the water
surface temperature normalized relative humidities (hn) are likely similar to, or lower than, the
free air relative humidities reported above. Consequently, the major unconstrained parameters
for the closed-basin lakes was 18/16adiff and Ra. 18/16adiff is commonly assumed to be ~ 1.014 for
continental lakes (e.g., Gat 1996; Horita et al. 2008; Jasechko et al. 2013; Gibson et al. 2016).
Figure 7 shows the results for these lakes in the context of models where 18/16adiff = 1.0142 and
1.0086. For the 18/16adiff = 1.0142 models, the relative humidities required to explain the observed
lake water compositions are unrealistically high, whereas the 18/16adiff = 1.0086 models achieve
a much better fit to the known relative humidities and isotopic compositions. Alternatively,
models where 18/16adiff = 1.0142, but Ra is higher (60 per meg instead of 12 per meg) also better
predict the observed data within the constraints of known hn and XE.
There is good evidence that both Ra and 18/16adiff may depart from commonly assumed
values. For example, Surma et al. (2021, this volume) show that continental vapors in the
European Alps have substantially higher ∆′17O than water vapors in equilibrium with local
surface waters. The parameter 18/16adiff can theoretially range by 28‰ (!) (1 to 1.028), and
A 100 B Δʹ17O (per meg, SMOW-SLAP, λ = 0.528) 100
50 50
initial waters initial waters
0 0 h = 0.8
h = 0.8
0.6
-50 0.6 0.3
-50
0.3
-100 -100
-150 -150
XE = 1, αdiff = 1.0142, Ra = 12 p.m. XE = 1, αdiff = 1.0086, Ra = 12 p.m.
-200 -200
-30 -20 -10 0 10 20 30 -30 -20 -10 0 10 20 30
C
100
50
h = 0.8
0
0.6
-50
0.3
-100
-150
XE = 1, αdiff = 1.0142, Ra = 60 p.m.
-200
-30 -20 -10 0 10 20 30
Figure 7. Western United States closed-basin lakes and inflow rivers (Mono Lake, Pyramid Lake, and
Great Salt Lake). A. Inflow rivers (triangles) and lake waters (gray circles), plotted in the context of the
model (Eqn. 16) from Figure 5D, where 18/16adiff = 1.0142. Solid lines connect waters with initial (unevapo-
rated) isotopic compositions shown in Figure 3b for hn = 0.7. B. As (A), but plotted in the context of models
for 18/16adiff = 1.0086. The latter models are more realistic, given that potential evapotranspiration-weighted
relative humidity for these lakes is ~40 – 45%. C. As (A), but with Ra = 60 per meg instead of 12 per meg.
The significance of the value of 60 per meg is that it is in the upper ~1/3 of the range of values reported for
the European Alps by Surma et al. (2021, this volume). Data are from Passey and Ji (2019).
the wide range of values observed in experimental studies (see Fig. 6), and the closed-basin
results from Passey and Ji (2019) suggest that variation for natural lakes may be greater than
commonly assumed. Moving forward, there is a need better characterization of Ra and 18/16adiff
parameter across the range of continental evaporation scenarios.
Exploiting parameter-insensitive trajectories. Surma et al. (2018) explored evaporation
trajectories in triple oxygen isotope space using a combination of modeling and observational
results from the Atacama Desert, Chile. They suggest that in many cases, the shape of the
evaporation trajectory for flow-through lakes in ∆′17O versus d18O is insensitive to parameters
such as hn, adiff, and temperature (their Figs. 4a–e), whereas the extent to which samples evolve
along the trajectory is highly sensitive to XE and hn. However, variation in Ra can cause these
trajectories to differ significantly (their Figs. 4f–h), and in cases of transient evaporation, the
shape of the trajectory is strongly dependent on hn (Fig. 5B of this chapter).
Alternatively, Passey and Ji (2019) used a Monte-Carlo approach to show that the
evaporation slope llake between unevaporated inflow waters and evaporated lake waters is
relatively conservative despite the large number of variables in Equation (17). Thus, this
slope is predictable and can be used to ‘back-project’ the composition of the evaporated lake
water to the intersection with an assumed unevaporated triple oxygen isotope MWL. In other
words, while the reconstructed triple oxygen isotope composition cannot be used to uniquely
reconstruct individual parameters such as XE and hn, the data can be used to estimate the d18O
value of unevaporated precipitation, which is an important goal in the isotopic reconstruction
of past climates. The uncertainty is not insubstantial (generally a few per mil), but given that
closed basin lakes can be elevated in d18O by 10–15‰ relative to unevaporated catchment
precipitation, this level of error may be acceptable in many applications.
ANIMAL AND PLANT WATERS

Carbon and oxygen isotope analysis of vertebrate bioapatites is a cornerstone of
reconstructing continental climate and paleoecology (Koch 1998; Kohn and Cerling 2002),
but the unequivocal interpretation of oxygen isotope data has been hampered by the numerous
competing factors that influence animal body water d18O, and hence animal bioapatite d18O
(e.g., Luz and Kolodny 1985; Ayliffe and Chivas 1990; Bryant and Froelich 1995; Kohn 1996;
Levin et al. 2006; Blumenthal 2017). Evaporation is a key part of the water balance of many
animals, in terms of the intake of evaporated waters (e.g., leaf waters, evaporated surface
waters), and in terms of evaporation of water from the animal. Furthermore, in many animals,
a substantial fraction of the O in body water comes from atmospheric O2 via basic metabolic
respiration (i.e., CH2O + O2 → CO2 + H2O). Atmospheric O2 has a highly distinctive ∆′17O
value of –434 ± 22 per meg (this is the average and standard deviation from studies analyzing
tropospheric O2 directly against the VSMOW2-SLAP scale: Barkan and Luz 2005, Pack et al.
2017; Wostbrock et al. 2020; cf., Pack 2021, this volume). Therefore, triple oxygen isotopes
are uniquely suited for application to terrestrial vertebrates.
Andreas Pack’s group at Georg August University of Göttingen have pioneered this
emerging field (Gehler et al. 2011), focusing in particular on the prospect of reconstructing
past carbon dioxide levels (Pack et al. 2013; Gehler et al. 2016), following on the sensitivity
of the ∆′17O of atmospheric O2 to pCO2 (Luz et al. 1999; Bao et al. 2008; Cao and Bao
2013; Young et al. 2014; Brinjikji and Lyons 2021, this volume; Pack 2021, this volume).
O2-derived oxygen in body water is diluted by other sources of oxygen to animals, and the
isotopic composition of body water is further modified by isotopic fractionations associated
with both inputs and outputs of oxygen-bearing species. Therefore, accurate isotope mass
balance body water models are necessary for meaningful interpretation of triple oxygen
isotope data in terms of reconstructing the ∆′17O of atmospheric O2, and also of reconstructing
442 Passey & Levin
the paleoclimatic and paleoecological significance of data. Pack et al. (2013) developed a
triple oxygen isotope model based on the d18O model of Bryant and Froelich (1995), where
the input and output fluxes are scaled to body mass. In the model, the relative proportion of
O2-derived oxygen decreases as body mass increases, which follows the observation that
smaller animals are often more independent of water (i.e., require less drinking water) than
larger animals. Whiteman et al. (2019) developed a model that similarly ties the relative
proportion of O2-derived oxygen to metabolic rate and hence body mass.
Figure 8 shows the currently available triple oxygen isotope data for modern vertebrates,
plotted as a function of body mass (Pack et al. 2013; Passey et al. 2014; Whiteman et al. 2019).
The data support a general trend with body mass, with smaller animals having lower ∆′17O
values, which is consistent with a higher fraction of O2-derived oxygen, given that the ∆′17O of
O2 is ~ –0.43‰. However, Figure 8 also shows that there is a large degree of variation in ∆′17O
that is independent of body mass, and that the degree of variation at a fixed body mass can
rival or exceed the difference in ∆′17O between the low and high ends of the regression lines.
This is not surprising, given that previous studies have shown that d18O may vary greatly
according to factors unrelated to body mass, such as relative humidity (Ayliffe and Chivas
1990; Luz et al. 1990) and leaf-water intake (Levin et al. 2006; Blumenthal et al. 2017).
Here we explore such “mass independent” influences on the triple oxygen isotope
compositions of animals. We make use of a triple oxygen isotope version (Hu 2016) of the
d18O model developed by Kohn (1996), which allows for detailed adjustments in fluxes such
as plant leaf water (fractionated by evaporation), stem/root water (unfractionated relative
to soil water), the carbohydrate / protein / fat composition of diet (as each has a different
stoichiometric demand for O2 during respiration), and the relative effluxes of urinary, fecal, and
sweat water (unfractionated relative to body water) and breath / panting vapor / transcutaneous
vapor (fractionated relative to body water). The Kohn model, converted to the form of isotope
ratios and fractionation factors, is given by:
A  f  i i  mw  Rmw  fO2   in O2B RO2  f o   o  bw  Rbw (18)

50 1.0
i o
All Data: Birds: y=0.0178ln(x) + 0.081 (R2 = 0.17)
WEI (ml H2O / kJ metabolism)
y = 0.007ln(x) - 0.09 (R2 = 0.245) Eutherian Mammals:

0 y=0.0147ln(x) + 0.108 (R2 = 0.22)
0.8 Marsupial Mammals:
y=0.0063ln(x) + 0.177 (R2 = 0.042)
-50 Reptiles: y=-0.0001ln(x) + 0.25 (R2 = 0.000)
-100 0.6
-150 0.4
-200 Passey et al. 2014, this study
y = -0.004ln(x) - 0.043 (R² = 0.06)
Pack et al. (2013):
0.2
-250 y = 0.016ln(x) - 0.073 (R² = 0.61)
Whiteman et al. (2019):
y = 0.007ln(x) - 0.123 (R² = 0.23)
-300 0.0
10-3 10-2 10-1 1 10 102 103 104 0 10-3 10-2 10-1 1 10 102
Mass (kg) Mass (kg)
Figure 8. Mammal and bird body water ∆′ O values and water economy index (WEI) as a function of body
17
mass. A. Body water ∆′17O versus body mass. Squares are data from Pack et al. (2013), circles from Passey
et al. (2014), and diamonds from Whiteman et al. (2019). The apatite phosphate data from Pack et al. (2013)
were converted to the λ = 0.528 reference frame, and equivalent body water compositions were calculated
using 18/16aap-bw = 1.0173 and qap-bw = 0.523 (Pack et al. 2013). The apatite carbonate acid-digestion CO2 data
from Passey et al. (2014) were converted to body water using 18/16aap,carb-bw = 1.0332 and qap,carb-bw = 0.5245.
The eggshell carbonate data from Passey et al. (2014) were converted to body water using 18/16acarb-bw = 1.0380
and qcarb-bw = 0.5245 (see Passey et al. 2014; Table 3 footnotes for details). The Whiteman et al. (2019) data
were reported as body water, so no conversion was necessary. B. WEI versus body mass for birds, eutherian
mammals, marsupial mammals, and reptiles. Data are from Nagy and Petersen 1988.
where fi and fo are the fractional inputs and outputs of oxygen from each source, ai-mw are the
fractionations of the oxygen inputs relative to meteoric water (Rmw), ao–bw are the fractionations
of oxygen outputs relative to body water (Rbw), fO2 and RO2 are the fractional contribution and
isotopic composition of atmospheric oxygen, and ain–O2 is the fractionation of oxygen during
uptake in the lungs.
Figure 9 shows the different fluxes considered in the Kohn model, and the associated
isotopic fractionations. A central aspect of the Kohn model is the use of taxon-specific water
economy index values (WEI): this is the water use efficiency of the animal, measured in terms
of volume of liquid water per unit metabolic energy (e.g., ml/kJ). In essence, WEI is the water
usage “per unit of living” (Nagy and Petersen 1998), and relates to adaptations for water
conservation such as the urea-concentrating abilities of the kidneys, the ability of animals to
tolerate changes in body temperature (and thus reduce the need for evaporative cooling, an
adaptation termed ‘adaptive heterothermy’), and the ability to avoid heat (e.g., burrowing).
Animal Body Water and Biominerals
A Outputs
Inputs
carbon dioxide: αCO2–bwRbw
*drinking water: αdw–mwRmw
*fecal water: αfcw–bwRbw
leaf water: αlw–mwRmw
*urine water & urea/uric-O: αurw–bwRbw
atmospheric O2: αin–O2RO2
*sweat water: αsww–bwRbw
atmospheric H2Ov: αvap–mwRmw
transcutaneous water vapor: αtcw–bwRbw
food bound O: αfo–mwRmw nasal water vapor: αnsw–bwRbw
*unfract. food water: αufw–mwRmw breath water vapor: αbtw–bwRbw
B Present-day Δʹ17Oatm-O2 (-410 per meg) C Δʹ17Oatm-O2 = -1000 per meg

100 meteoric waters

max water
0 model
50 meteoric waters max water h=0.8
model -100 0.2
0 h=0.8 h=
0.2 0.8
max evap 0.6
0.4
-50 h= model -300
0.8 0.2
-400 max evap
model
-100 max ox 0.6 max ox
model -500 model
0.4
-150 -600
0.2
-200 -700
-30 -20 -10 0 10 20 30 -30 -20 -10 0 10 20 30
Figure 9. Triple oxygen isotope body water model (Eqn. 18) based on the Kohn (1996) model. A. Graphi-
cal depiction of the isotopic inputs and outputs considered in the model. For uniformity, all fluxes have an
assigned fractionation factor (α), but those marked with an asterisk are unity (i.e., α = 1, no isotopic fraction-
ation). 18/16α values are as reported by Kohn (1996), except for leaf water, for which we use a modified version
of the Flanagan et al. (1991) model (Roden et al. 2000). θ values are as follows: leaf water: θ = −0.008h +
0.522 (Landais et al. 2006, where h is relative humidity); atmospheric O2: θ = 0.5179 (Pack et al. 2013; Luz
and Barkan 2005); atmospheric water vapor: θ = 0.529 (Barkan and Luz 2005); food bound O: 0.5275 (best
guess by Pack et al. 2013; no data are available); carbon dioxide: θ = 0.5248 (Cao and Liu 2011, a theoretical
prediction in close agreement with our unpublished experimental value of 0.5243); transcutaneous water:
θ = 0.5235 (no data available; this value is intermediate between molecular diffusion of water vapor in air
(0.5185) and equilibrium liquid–vapor exchange (0.529); nasal water vapor and breath water vapor: θ = 0.529
(Barkan and Luz 2005). B. Evaluation of the model for max water, max evap, and max ox physiologies (see
text), modeled with ∆′17O(O2) = −410 per meg (the value for modern atmosphere measured by Pack et al.
2017). C. As (B), but modeled with ∆′17O(O2) = −1000 per meg, which corresponds to pCO2 = 940 ppm in
the Cao and Bao (2013) model under present-day global gross primary productivity (GPP) and pO2.
444 Passey & Levin
Compilations of WEI for different animals show only weak correlation with body mass
(Fig. 8B). Contrary to common expectations, animals with high metabolic rates (and hence high
absolute rates of O2 intake) do not consistently have higher relative intakes of O2 compared to
other sources of oxygen (e.g., drinking water, plant water). For isotopic mass balance, it is the
relative sizes of different fluxes (i.e., fractional contributions) that are important.
The WEI value becomes important in the Kohn model because it determines the amount
of liquid drinking water that the animal must consume. The total energy requirement of the
animal is determined based on body mass using allometric scaling equations, and this value
along with WEI determines the water requirement of the animal. If other sources of water to
the animal (plant water, metabolic water) are less than this requirement, then the drinking
water influx is used to make up the difference. If other sources of water are in excess of the
animal’s water requirement, then no drinking water is consumed, and the excess water exits
the animal via nonfractionating mechanisms (e.g., urinary and fecal water).
Figure 9 shows predictions of the Kohn model for three endmember animal physiologies.
In the ‘maximum water’ (max water) scenario, the animal is highly dependent on drinking water
(WEI = 0.6 ml/kJ), consumes very little leaf water (which is strongly modified by evaporation;
see Fig. 12), has relatively low digestibility and high fecal water content (more fecal water
means more water loss per unit of energy metabolized) and has non-fractionating water effluxes
(e.g., sweat versus panting). The hippopotamus is an example of a ‘max water’ physiology.
In the ‘maximum evaporation’ (max evap) scenario, the animal obtains most of its water in
the form of leaf water, which is strongly modified by evaporation). The animal is extremely
water-efficient (WEI = 0.05 ml/kJ) and hence drinks little or no water, has high digestibility
and low fecal water content, and loses water via fractionating mechanisms (panting versus
sweating). Selective browsers such as giraffe, deer, and ostrich are examples of ‘max evap’
physiologies. Finally, in the ‘max oxygen’ (max ox) scenario, the degree of O2-sourced
oxygen is maximized by imparting the animal with low WEI (0.05 ml/kJ), a diet high in fats
and low in free watercontent (which increases the fraction of metabolic water and hence usage
of O2), high digestibility, and low fecal water content. Kangaroo rats and other highly-water-
independent desert animals are examples of max ox physiologies.
Figure 10 shows the published data in the context of these model predictions. There is
general agreement between models and data. For example, highly water-dependent species
such as river otters, elephants, hippos, and humans tend to have high ∆′17O values. Of special
interest are domestic birds (Fig. 10B), including ostrich, which is normally a water-independent
species. The domestic birds all have high ∆′17O, consistent with a dry pellet diet that necessitates
substantial drinking, which ties the triple isotope composition of these animals close to the
composition of meteoric waters. Animals known to be selective leaf consumers (browsers)
generally have low ∆′17O, reflective of the signature of leaf water; this includes giraffe, deer,
and ostrich. Finally, most of the rodents plot in the direction of the max ox and max evap
models. The most conspicuous inconsistency between the model predictions and observations
is for the desert rodents in the dataset of Whiteman et al. (2019): many of these data are 50 to
100 per meg lower in ∆′17O than predicted by the model! It is unclear at this point why the data
/ model discrepancy exists. In terms of the modeling, a major uncertainty is the fractionation
between atmospheric oxygen and the oxygen that is absorbed by the lungs. It is known that
the lungs discriminate against 18O by several per mil (Epstein and Zeiri 1988), with values
for free-ranging animals poorly known (Kohn 1996). No measurements of θ values for this
process have been made, and if the true value is substantially higher than the estimate by Pack
et al. (2013) used here (θ = 0.5179; Luz and Barkan 2005), then modeled body water values
would be lower in ∆′17O. Furthermore, there may be aspects of the burrow enviroment of these
desert rodents that influence isotopic compositions (i.e., the composition of soil water vapor,
A 100
Pack et al. 2013
max water
50 model
pig
african elephant
0 h=0.8 porpoise
0.2 human max evap
-50 red deer model
h= shrew
0.8
-100 max ox 0.6
model kangaroo rat
0.4
-150 squirrel
0.2
wood mouse
-200
rodents
-250 other mammals
-300
-30 -20 -10 0 10 20 30
B 100 C 100
Passey et al. 2014 Whiteman et al. 2019
max water max water

50 model 50 model
domestic chickens,
duck, emu, ostrich
0 h=0.8 0 h=0.8
hippo zoo elephants
0.2 max evap 0.2 max evap
-50 rhino model -50 river otters model
h= bison
0.8 ostrich deer mice and
-100 deer
-100 max ox 0.6 0.6 house mice
model giraffe
0.4 0.4
-150 deer -150 hibernating
0.2 bears 0.2
kangaroo rats
-200 -200 and deer mouse
domestic birds
-250 wild birds -250 pocket mice
rodents
wild mammals
other mammals
-300 -300
-30 -20 -10 0 10 20 30 -30 -20 -10 0 10 20 30
Figure 10. Measured triple oxygen isotope compositions of animals. All datasets are presented in the
context of the max water, max evap, and max ox models shown in Figure 9B. A. Dataset of Pack et al.
(2013). B. Dataset of Passey et al. 2014, with the addition of three new data points: a modern giraffe
from Tsavo National Park, Kenya (d18O = 13.5‰, ∆′17O = −122 per meg), a modern deer from Parowan,
Utah, USA (d18O = 1.9‰, ∆′17O = −142 per meg), and a modern bison from Antelope Island, Utah, USA
(d18O = −1.6‰, ∆′17O = −64 per meg). C. Dataset of Whiteman et al. (2019). See Figure 8 caption for
explanation of how the bioapatite phosphate data of Pack et al. (2013) and the bioapatite carbonate and
eggshell carbonate data of Passey et al. (2014) were converted into equivalent body water values.
and even O2, which may be isotopically fractionated by respiration in soils), and the soil water
available to plants (and hence food) may be modified by evaporation. These discrepancies
highlight the need for ongoing research both in terms of collecting data from modern animals
and environments, determining key fractionation exponents, and refining models.
Stepping back, the results presented in Figure 10 are very encouraging for application
of triple oxygen isotopes to animals: The observed ranges in ∆′17O are the largest out of any
system discussed in this chapter, and there is clear ecological patterning that makes sense
446 Passey & Levin
in the context of the current state of modeling. As yet published datasets developed by our
group for modern mammals (Lehmann 2016) and dinosaurs (Hu 2016) point to clear signals
of aridity and changes in ∆′17O of past atmospheric O2, respectively. A such, studies of triple
oxygen isotopes in animals promise to reveal information not only about past climates, but
also about paleoecology and past global biogeochemical cycles.
Prospects for reconstructing ∆′17O of past atmospheric O2
The wide variation in ∆′17O of extant animals—and therefore certainly for extinct animals
as well—may appear to portend badly for the prospects of precisely reconstructing ∆′17O of past
atmospheric O2 based on analyses of fossil teeth, bones, and eggshell. However, this variability
can potentially be embraced and used to good effect towards reconstructing both ∆′17O(O2) and
aspects of paleoecology and paleoclimate. As Figure 9B predicts, animal communities living in
times of high pCO2 and hence anomalously-low ∆′17O(O2) (here, −1000 per meg) will be shifted
downwards in ∆′17O and will have a greater range in ∆′17O. Therefore, in an approach where
data are generated for specimens from multiple taxa, the utility of the ∆′17O(O2) signal lies not
in the absolute value of any particular specimen, but in the ∆′17O values of the upper and lower
limits of the population (i.e., position and range in ∆′17O space). Figure 11 shows the max water,
max evap, and max ox models calculated as a function of ∆′17O(O2), evaluated for a relative
humidity of 0.2 (i.e., an extreme low limit). Figure 11 also shows the highest 10 and lowest
10 ∆′17O values observed for modern animals, plotted as horizontal bars, corresponding on the
vertical axis to the observed ∆′17O value of each sample, and on the x-axis to range in ∆′17O(O2)
that is permissible for the sample. Note that this approach requires no information about the
water balance physiology of specific animals. For the community as a whole, there should be
a limited, concordant range in ∆′17O(O2) values that can explain all of the data. We term this
the environmental physiology isotope concordance (EPIC) approach (Hu 2016): If the body
water endmember models are appropriate for extinct organisms, and if all animals analyzed in
an assemblage existed under a uniform value of ∆′17O(O2), and if diagenesis is not a factor, then
there should be a concordant range in ∆′17O(O2) values that can explain all of the data.
Clearly, reliable implementation of this EPIC approach will be a challenging task. It will
require highly-vetted body water models that reliably bracket the range of common water balance
physiologies for terrestrial animals. The data in Figure 8B suggests that this may be tractable:
water economy relative to metabolic demand for O2 falls within fairly narrow limits, for the
vast majority of organisms, with no systematic differences among reptiles, birds, marsupial
mammals, and eutherian mammals. Given that the present-day manifestation of the extant range
of water balance physiologies is the product of more than 400 million years of evolution of
separate lineages animals on the continents, it is probably safe to assume that ranges were not
hugely dissimilar during e.g., the reign of the dinosaurs in the Mesozoic, or during radiation
of mammals in the Cenozoic. The EPIC approach will require the analysis of a wide range of
species that existed more or less at the same time (from contemporaneous fossil assemblages),
which is challenging both in terms of finding suitable assemblages, and in the laboratory
challenge of analyzing often very small fossils that make up the majority of the diversity (e.g.,
rodents, early mammals, dinosaur teeth with extremely thin tooth enamel). Diagenesis will need
to be addressed through both microanalytical methods and stable isotope approaches. All of
these challenges are considerable, but the scientific products—records of pCO2, paleoecology of
extinct species, and paleoclimate—are also considerable, and will make the effort worthwhile.
Plant waters
Actively transpiring leaves are evaporatively-enriched in 18O (Gonfiantini et al. 1965),
whereas there is no measurable fractionation during soil water uptake by the roots, nor during
transport in the xylem (White 1989; Ehleringer and Dawson 1992). Although more recent
pCO2 (ppm)
0 500 1000 1500 2000
200
concordant Δʹ17O (O2)
100
(per meg, SMOW-SLAP, λ = 0.528)
Δʹ17O of animal body water
0
ma
xw
ate
rm
-100 ode
l
-200
-300 m
ax
max
ev
ap
ox m
-400 m
od
el
ode
-500
l
0 -500 -1000 -1500 -2000

Δʹ17O of atmospheric O2
(per meg, SMOW-SLAP, λ = 0.528)
Figure 11. Illustration of the environmental physiology isotopic concordance (EPIC) approach for deter-
mining the ∆′17O of atmospheric O2 (∆′17O(O2)) based on triple oxygen isotope analysis of vertebrate bio-
genic carbonates. Bold dashed lines show the max water, max evap, and max ox body water models evalu-
ated as a function of ∆′17O(O2). Thin horizontal lines mark the ∆′17O values (y-axis) of the highest 10 and
lowest 10 observed values in the combined datasets of Pack et al. (2013), Passey et al. (2014), and White-
man et al. (2019). The horizontal extent of each line is bounded by the intersections with the body water
model lines. In other words, the extent of each horizontal line shows the range of ∆′17O(O2) values that
could give rise to the measured body water ∆′17O value. Omitted for clarity are horizontal lines for the
76 additional samples that fall between the highest 10 and lowest 10. The gray field shows the range of
∆′17O(O2) values that are concordant with all of the observed 96 data points, excluding the highest five and
lowest five values as a measure of outlier removal. The resulting field overlaps the present-day ∆′17O(O2)
of −430 per meg, and is consistent with the pre-industrial CO2 level of 280 ppm. Note that because the
residence time of atmospheric O2 is of order 103 years, the ∆′17O(O2) value has not changed substantially
during the past 150 years, and therefore reflects an atmosphere with ~280 ppm CO2. The relationship be-
tween pCO2 (top axis) and ∆′17O(O2) is from the model of Cao and Bao (2013).
and detailed models exist (see the review by Cernusak et al. 2016), the leaf water evaporation
model presented by Flanagan et al. (1991) is a straightforward steady-state model based
on the Craig–Gordon evaporation model, and a good entry point for understanding isotopic
enrichment in leaves:
  e e   e e   e 
Rleaf   eq   k Rs  i s    kb Rs  s a   Ra  a   (19)
  ei   ei   ei  
where Rleaf, Rs, and Ra are the isotope ratios of leaf water, soil water, and atmospheric water
vapor, respectively, ak is the fraction factor for molecular diffusion of water vapor in air, akb is the
effective fractionation factor for water vapor transport through the boundary layer adjacent to the
448 Passey & Levin
leaf surface (analogous to adiff), and ei, es, and ea are the water vapor pressures in the intercellular
air spaces inside the leaf, at surface of the leaf, and in the free atmosphere, respectively.
For adaptation to 17O, the following θ values are used: qk = qkb = 0.5185, qeq = 0.529.
Figure 12 shows the predictions of this model for a range of relative humidities, along with
measured values for leaves, stems/roots, and soil water presented by Landais et al. (2006; for
plants from Israel and some locations in Europe) and Li et al. (2017; for plants from Kenya).
We omit growth-chamber data from Alexandre et al. (2018), which show similar trends.
Why do leaf waters have such low ∆′17O values compared to e.g., evaporating pans of water and
closed-basin lakes? The answer lies in the stagnant environment of the intercellular air spaces
inside of the leaf, which allows for full expression of the isotope fractionation associated
with molecular diffusion of water vapor down the concentration gradient from 100% rh inside
of the leaf to <100% rh at the leaf surface. This fractionation is not fully expressed during
evaporation from open water surfaces (such as lakes) owing to turbulence.
Evaporation from leaves

A B
100
50 initial waters
RLW
0 h
= 0.9
-50
0.8
ei , αk -100 0.7
0.6
es , αkb boundary layer -150 0.5
Rs 0.4
ea free atmosphere
-200
-30 -20 -10 0 10 20 30
Figure 12. Triple oxygen isotope compositions of leaf waters. A. Schematic diagram showing the intercel-
lular, boundary layer, and free atmosphere regions considered in the leaf water model (Eqn. 19). B. Mod-
eled steady-state compositions of leaf water, modeled as a function of relative humidity. Gray circles show
steady state compositions for the indicated relative humidities and source waters (i.e., stem waters) indi-
cated by the triangles. Small crosses and open circles show stem water and leaf water data from Landais
et al. (2006). Small plus signs and open diamonds show stem and leaf water data from Li et al. (2017).
The fact that leaf water has distinctive ∆′17O is not only of consequence for leaves or the
animals that eat them. Leaf water influences the isotopic composition of atmospheric CO2 and
O2, so a thorough understanding of global leaf water triple oxygen isotope compositions will be
essential for understanding the triple oxygen isotope global biogeochemistry of these species
(Liang et al. 2017). The isotopic composition of plant macromolecules such as cellulose, of
which tree rings are largely composed, are strongly influenced by the leaf water (Roden et al.
2000), meaning that there should be signals of past climate (e.g., relative humidity) preserved
in ∆′17Ο of tree rings. Phytoliths preserve lasting records of plant ecophysiology, and the triple
oxygen isotope composition of these correlates strongly with relative humidity (Alexandre
et al. 2018). Much important research into the triple oxygen isotope systematics of plants
remains to be done, given that only three detailed studies have been published so far (Landais
et al. 2006; Li et al. 2017; Alexandre et al. 2018).
ANALYSIS OF CARBONATES AND BIOAPATITES

The modern era of triple oxygen isotope hydroclimate begins with Barkan and Luz
(2005), who improved the water CoF3 fluorination methods of Baker et al. (2002) to achieve
precisions in ∆′17O of better than 10 per meg. Laser-based methods are increasingly employed
to analyze waters, although not always with a substantial savings in time or effort, given the
intensive sample – standard bracketing that is necessary to achieve accuracy and precision at
the 10 per meg level (Berman et al. 2013; Steig et al. 2014; Tian et al. 2018).
The high-precision analysis of carbonates has been hampered by the fact that carbon dioxide
released by phosphoric acid digestion contains 13C. This means that carbon dioxide of nominal
mass 45 consists both of 13C16O2 (44.9932 u), and 12C17O16O (44.9940 u). The mass difference
between these isotopologues is 0.0008 u, which is well under the mass resolving power of
typical isotope ratio mass spectrometers, and is even beyond the capabilities of the Thermo 253
Ultra and Nu Panorama instruments. Therefore, the O in carbonate minerals or CO2 derived
from acid digestion of those minerals needs to be converted to, or equilibrated with, a gas that
has no interfering isobars for the 17O-containing isotopologues. This gas is molecular oxygen,
where 17O16O has a mass of 32.9940 u. In general, the approaches that have been developed can
be categorized as ‘conversion methods’, where nominally all of the O in CaCO3 (or CO2) are
converted to O2, or ‘exchange methods’, where CO2 is exchanged with subequal quantities of O2
or another phase (H2O, CeO2) that can be more easily fluorinated to produce O2, or where CO2
is analyzed before and after exchange with a large, uniform oxygen reservoir of known isotopic
composition, which permits back-calculation of the d17O of CO2. A third emerging category
bypasses the problem altogether by performing mass spectrometry on O+ ion fragments from
electron impact with CO2 (Adnew et al. 2019), or by using laser spectroscopy (Sakai et al. 2017).
All of the methods achieving high precision (~10 per meg or better for ∆′17O) employ
extended collection time dual inlet mass spectrometry. Ion currents are typically in the range of a
few tens of nA for the major m/z 32+ ion beam, and collection times are commonly ~1000–2000 s,
distributed over several tens of sample gas / reference gas cycles. m/z 33 is typically registered
through ~1011 to ~1012 Ω resistors. Shot-noise limited precision for ∆′17O under such analytical
conditions is on the order of 5 per meg, as calculated using Merritt and Hayes (1994, Eqn. 15).
Conversion methods
Direct fluorination. Compared to other methods discussed here, there is a fairly long history
of directly fluorinating carbonate minerals and carbon dioxide to produce O2 for mass spectrometry
(Sharma and Clayton 1965; Miller et al. 2002; Benedix et al. 2003). Nevertheless, the analysis
has been uncommon, in part because d18O could be more readily analyzed by phosphoric acid
digestion and analysis of CO2, and in part because CaCO3 and CO2 are highly refractory with
respect to F2 / BrF5, requiring high temperatures (~ > 700 °C) and very long reaction times
(~4 days) for quantitative conversion to O2. Wostbrock et al. (2020) have recently revived this
method and adapted it for high-precision triple oxygen isotope analysis. A key aspect of their
method, and indeed one that is common amongst most of the successful methods achieving
high precision in ∆′17O, is the careful post-fluorination clean-up of O2 using a molecular sieve
GC column (e.g., Pack and Herwartz 2014). Following fluorination of CaCO3 or CO2 for 96
hours at 750 °C, Wostbrock et al. (2020) pass the O2 gas over two lN2-temperature traps, through
a 100 °C NaCl trap to remove traces of F2, through an additional lN2 temperature trap, and then
absorb the gas onto a 5Å molecular sieve. The O2 is then released into a helium stream that
carries it through a 6’ long, 1/8” diameter molecular sieve gas chromatography column held at
room temperature. The O2 is then recollected on molecular sieve, and finally introduced into the
mass spectrometer. All of this clean-up is to ensure that the O2 gas is free of isobars, including
NF3, which can be fragmented into the m/z 33 NF+ ion in the ion source of the mass spectrometer.
450 Passey & Levin
Although the method of Wostbrock et al. (2020) is the ‘newest kid on the block’ with
respect to the modern era of triple oxygen isotope analysis of carbonates, we are of the opinion
that their method is currently the best available for accurate and relatively assumption-free
determination of ∆′17O values of carbonates. This is because the method features verifiable
100% conversion of CaCO3 oxygen to O2 oxygen, and hence there can be no isotope
fractionation, barring exchange with O inside of the extraction apparatus, the possibility of
which can be monitored using d18O values, and mitigated by running several replicates in
series. As described below, other methods involve measurable blank and fractionation effects,
and oftentimes require bespoke calculations to arrive at the final ∆′17O value of the carbonate.
No such steps are required for the fluorination analysis aside from standardization within the
SMOW-SLAP reference frame: the ∆′17O of the O2 gas measured by the mass spectrometer is
the ∆′17O of the CaCO3 (or CO2), with no calculations or corrections required. Additionally,
water (including VSMOW and SLAP) are analyzed on the same extraction line using nearly
identical procedures, and furthermore the silicate oxygen isotope scale is calibrated using
the same methods. For these reasons, we suggest that direct fluorination, when conducted
properly, be considered as the ‘gold standard’ method that is used to standardize the ∆′17O
values of carbonate minerals in the VSMOW-SLAP reference frame. These standardized ∆′17O
values can then be used to normalize data from other methods into a common reference frame.
Direct laser fluorination is used by Gehler et al. (2011, 2016) and Pack et al. (2013) to
analyze the phosphate component of bioapatite (primarily tooth enamel). The methods are
similar to those developed for laser fluorination of silicates, except that the bioapatite sample
is first fused under argon at 1000 °C to drive off organic matter and oxygen from the carbonate
and hydroxide components of bioapatite.
Acid digestion–reduction–fluorination. Passey et al. (2014) developed a method similar
to that conceived by Brenninkmeijer and Röckmann (1998), whereby carbon dioxide generated
by phosphoric acid digestion of carbonate or bioapatite is reduced in the presence of excess
molecular hydrogen and Fe catalyst to produce H2O, with the carbon exiting as methane. The
water is then fluorinated over CoF3 to produce O2, largely following the methods of Barkan and
Luz (2005). The sluggish reaction kinetics of the reduction reaction (CO2 + 4H2 → CH4 + 2H2O)
are addressed by using a closed circulating loop reactor design, which ensures that unreacted
CO2 can pass through the reactor multiple times for complete conversion to CH4 + H2O. The
reaction takes place at 560 °C inside of a glassy carbon tube, which is essential to minimize
exchange with O that might occur inside of a fused silica reactor, and to avoid unwanted catalytic
activity and exchange with oxides that might occur inside of an iron- or nickel- based reactor
tube. Observations using a Thermo ISQ7000 quadrupole mass spectrometer in our laboratory at
Michigan have verified that CO is an intermediate product of the reaction. During several passes
through the reactor, the CO is converted to CH4 + H2O. It also appears likely that C is deposited
onto the Fe catalyst, and subsequently reduced to CH4 by reaction with the excess H2. At any rate,
the product of interest in the reaction is H2O, which is trapped from the circulating loop at −78 °C
(dry ice temperature), and is then purged through a CoF3 reactor via a helium stream, according
to the basic procedures of Barkan and Luz (2005). The produced O2 is cleaned-up by passage
through a multi-loop open nickel tube immersed in liquid nitrogen (to remove HF produced in the
fluorination reaction), and then through a ~1.2 m, 1/8” o.d. 5Å molecular sieve GC held at −78 °C.
The O2 leaving the GC is trapped onto a 5Å molecular sieve held at liquid-N2 temperature, then
expanded at 90 °C for 10 minutes, and finally admitted into the mass spectrometer.
The precision for this analysis is ~10 per meg. The method achieves nominally 100% yield,
but has clear exchange and fractionation effects, which are revealed by comparing known d18O
values of the carbonates (measured as CO2) to the d18O of O2 generated by the method (Passey
et al. 2014). Slopes of this regression over a range of ~30‰ in d18O are typically ~0.95, meaning
that the exchange effect is about 5%. Fractionation effects result in mediocre precision in d17O
and d18O values (~0.2‰ and 0.4‰ 1σ, respectively), but these errors are correlated and mass-
dependent with a slope near 0.528, which means that ∆′17O values are virtually unaffected.
In our new system at Michigan, we have observed drift in ∆′17O values within some of our
analytical sessions. This drift can have a magnitude of ~10–30 per meg, and is systematic across
the ~ 4–5 different carbonate reference materials (NBS−18, NBS−19 / IAEA 603, and some
in-house standards) that we regularly analyze throughout our sessions. Therefore, the drift is
correctable by normalizing to long-term mean values for our standards (or, ideally in the future,
to community consensus ∆′17O values for standards such as NBS−18 and IAEA 603), in the
manner that is common for e.g., continuous flow methods. Because of the blank, fractionation,
and drift effects, the method is less appropriate than direct fluorination for defining absolute
∆′17Ο values. Merits of the method are that it is relatively safe, that the entire procedure takes
only ~2.5 hours per sample (i.e., from dropping the sample in phosphoric acid to admitting the
final O2 product in the mass spectrometer), and, compared to exchange methods, that it requires
only one kind of mass spectrometric determination (i.e., analysis of sample O2).
Exchange methods
Several methods have been developed based on isotopic exchange between CO2 and another
phase that is more amenable to analysis. These include analyzing CO2 before and after exchange
with excess CeO2 or CuO with known oxygen isotopic composition, which permits back-
calculation of the ∆′17O of the CO2 (Assonov and Brenninkmeijer 2001; Kawagucci et al. 2005;
Mahata et al. 2012; Mrozek et al. 2015), exchange of an excess of CO2 with CeO2, which is then
fluorinated to produce O2 (Hoffman and Pack 2010), exchange of subequal amounts of CO2
and H2O, fluorination of the H2O to produce O2, and back-calculation of the CO2 composition
(Barkan and Luz 2012), and Pt-catalyzed exchange between subequal amounts CO2 and O2,
analysis of the O2, and back-calculation of the CO2 composition (Mahata et al. 2013, 2016).
Of these methods, only the latter has become widely-adopted for high-precision analysis (e.g.,
Barkan et al. 2015, 2019; Liang et al. 2017; Prasanna et al. 2016; Sha et al. 2020; Fosu et al.
2020; Bergel et al. 2020; Voarintsoa et al. 2020), so we will focus our discussion on this method.
Pt-catalyzed CO2–O2 exchange. In this method, subequal quantities of CO2 and O2 are
exchanged over a hot platinum wire or sponge (~750 °C) for ~ 1 h. The isotopic compositions
of both gases are analyzed before the exchange. Following the exchange, the CO2 and O2 are
separated, collected, and reanalyzed separately (either using two mass spectrometers, or by
analyzing them in series after switching the configuration of a single mass spectrometer).
The d17O value of the CO2 can then be calculated according to (Barkan and Luz 2015):
1  17
17O in  CO2  

    
 Of  O2   1 1  17  Pt    17O in  O2   1  11 (20)
where
 18 O in  O2    18 Of  O2  (21)

 18 Of  CO2    18 O in  CO2 
and where d18Of(CO2) can be calculated as:

18Of  CO2   18  Pt 18Of  O2   1  (22)
In Equations (20–22), the subscript ‘in’ refers to the initial (pre-equilibration) isotopic
composition, the subscript ‘f’ refers to the final (post-equilibration) composition, and 17/16aPt
and 18/16aPt are the isotopic fractionations between CO2 and O2 when the exchange reaction
has reached steady-state.
452 Passey & Levin
Of key importance for this method are knowledge of 17/16aPt and 18/16aPt values,
and awareness of potential variation in these values. The 18/16aPt and qPt values [where
qPt = ln(17/16aPt)/ln(18/16aPt)] values can only be determined using pairs of CO2 and O2 with
known D17O. This presents a challenge, because the ∆′17O of CO2 cannot be measured directly,
which is the reason for the development of these methods in the first place. In most laboratories,
this problem has been addressed by reacting O2 with known ∆′17Ο over hot graphite wound
with Pt wire to produce CO2 (Barkan and Luz 1996). Here it is assumed that if there is no
exchangeable oxygen in the system, and if 100% of the O2 is converted to CO2, then the ∆′17O
value of the product CO2 must be the same as that of the O2. This assumption can be evaluated
by measuring the d18O of the CO2 and comparing this to the d18O of the O2 (they should be
identical), but it cannot be strictly evaluated in the case of d17O.
Based on results from different laboratories, there is no single set of 18/16aPt and qPt
values that characterize the exchange reaction. 18/16aPt values are typically 0.998–1.002 (i.e.,
± ~2 per mil or smaller d18O difference between exchanged CO2 and O2), and qPt values are
typically ~0.55–0.62. These θ values are well beyond the theoretical high-temperature limit for
mass dependent fractionation (0.5305; Young et al. 2002), implying kinetic fraction during the
exchange reaction. 18/16aPt and qPt values apparently depend on the precise physical configuration
of the exchange reactor, and must be determined for each system (Fosu et al. 2020).
These findings underscore the need for (1) agreed upon ∆′17O values for widely-available
carbonate standards that can be used to develop 18/16aPt and qPt values for each system, thus
obviating the need for each laboratory to develop its own CO2 gas of ‘known’ ∆′17O based on
combustion of O2, given potential errors in this approach, and (2) the routine analysis of (ideally)
carbonate standards with known ∆′17O to monitor for systematic changes in in analytical results.
Of course, (2) applies for any analytical method, whereas (1) is uniquely necessary to exchange
methods that depend on accurate knowledge of exchange fractionation factors. We suggest that
the ∆′17O values for NBS−19, NBS−18, and IAEA 603 reported by Wostbrock et al. (2020) be
provisionally used for these purposes, as these values were determined by the direct fluorination
method and normalized to analyses of VSMOW2-SLAP made using the same analytical
procedures and equipment. These calibration issues notwithstanding, it appears that several
different laboratories are able to produce internally consistent data with high precision, and are
able to resolve clear environmental signals when analyzing natural materials (e.g., Liang et al.
2017; Sha et al. 2020; Bergel et al. 2020; Voarintsoa et al. 2020; Fosu et al. 2020).
Other methods
Adnew et al. (2019) explored the measurement of O+ ion fragments produced from electron
bombardment of CO2 in the ion source of the Thermo 253 Ultra mass spectrometer. The benefit
of such a method is the ability to measure CO2 directly, sidestepping the need for lengthy
chemical conversion or exchange procedures. Of particular challenge with this method is the
relatively inefficient production of these ions, and the interferences of 16OH+ and 16OH2+ on the
17 +
O and 18O+ signals, respectively. The high mass resolving power of the 253 Ultra, however,
permits clear resolution of these ions on the low-mass shoulders of the larger 16OH+ and 16OH2+
peaks. Precisions for ∆′17O of 14 ppm (SEM) can be achieved with counting times of 20 hours.
Tunable Infrared Laser Direct Absorption Spectroscopy (TILDAS) shows great promise
for high precision analysis of ∆′17O in CO2 (Sakai et al. 2017; Prokhorov et al. 2019). In the
region of wavenumber 2310 cm−1, the four most abundant isotopologues of CO2 are clearly
resolved, with only minor overlap between 12C17O16O at 2309.98 cm−1, and 12C16O2 at 2310.00
cm−1. Sakai et al. (2017) report standard errors of less than 0.04‰ and 0.03‰ for 18O/16O and
17
O/16O, respectively, on CO2 released from sub-100 mg CaCO3 samples (i.e., about 50× less
material than is required for most of the methods described above). Both Sakai et al. (2017)
and Prokhorov et al. (2019) show tight correlations between d17O and d18O measured for
suites of samples spanning a large range in isotopic composition. These papers stop short of
presenting ∆′17O values. However, Wang et al. 2020, achieved precisions of 10 per meg on the
13 18 16
C O O ‘clumped’ isotopologue using a similar instrument from the same manufacturer as
the TILDAS used in Sakai et al. (2017; Aerodyne Research, Inc.). Given that this isotopologue
is 16 times less abundant than 12C17O16O, it seems likely that TILDAS will eventually achieve
the 10 per meg precision benchmark for ∆′17O.
Interlaboratory reproducibility
Several labs have reported ∆′17O values for the carbonate reference materials NBS18,
NBS19, and IAEA 603 (Table 1). Most groups report values for CO2 released from acid
digestion of these materials at 25 °C, while the Johns Hopkins / Michigan labs report
values for 90 °C acid digestion, and Wostbrock et al. (2020) report values for the both the
carbonate mineral and for the CO2 liberated by 25 °C acid digestion. For calculation of the
equivalent CaCO3 composition from CO2 values, we use the qacid value of 0.5230 determined
by Wostbrock et al. (2020). This value was determined for 25 °C acid digestion. Guo et al.
(2009) predict very little temperature dependence in qacid (0.00003) between 25 °C and 90
°C, so we provisionally use Wostbrock’s value for 90 °C acid digestion data given a lack of
qacid values at this temperature. In Table 1, we present new data from the Michigan lab for all
of the carbonate standards that have been analyzed since the measurement was established in
May 2008, uncorrected for the intra-session drift noted above. Note that the Michigan lab is
effectively a ‘different’ lab compared to the Johns Hopkins lab (Passey et al. 2014, Passey and
Ji 2019), given that the extraction line and mass spectrometer are different.
Table 1 shows that there is very close agreement in ∆′17O(CaCO3) values amongst the labs
using conversion methods, with standard deviations of 5 per meg or less for the populations
of ∆′17O values for each material. Part of this agreement could be fortuitous given our lack of
knowledge of qacid at 90 °C. Alternatively, it is possible that qacid has very little temperature
sensitivity, as predicted by Guo et al. (2009), in which case the interlaboratory agreement is real.
Another measure of methodological agreement is the magnitude of ∆′17O difference between
different materials. The final column in Table 1 shows the difference in ∆′17O between NBS18 and
NBS19. The magnitude of this difference (~50 per meg) is essentially the same across the three
labs from which conversion data are reported, with the value from Passey et al. (2014; 37 per meg)
showing the most departure from the mean. This agreement lends confidence to these methods.
The results from the exchange methods are more variable. Barkan et al. (2019) attributed
part of the difference in their values compared to those reported in Barkan et al. (2015) to
the possibility of inhomogeneity between different bottles of reference material. The Sha et
al. (2020) values appear to be systematically shifted downward, but the difference between
NBS18 and IAEA 603 (which is identical within error to NBS19) is 42 per meg, which is
similar to the values derived by conversion methods and the value reported by Fosu et al.
(2020). The values of Fosu et al. (2020) are within 20 per meg of the values reported by
Wostbrock et al. (2020). Note that Fosu et al. (2020) also used NBS18-CO2 and NBS19-CO2
to derive 18/16aPt and qPt values, although the ∆′17O values they report for NBS18 and NBS19
are independent of those determinations. Fosu et al. (2020) to-date have presented the most
thorough exploration of variation in 18/16aPt and qPt in a single extraction line.
Overall, the results in Table 1 are both encouraging and concerning. It is not clear precisely
why the considerable lack of agreement exists for the Pt-exchange method, but the variability
in 18/16aPt and qPt values observed across different laboratories, the exotic absolute values of
qPt, and the fact that 18/16aPt values deviate from the value expected for CO2–O2 equilibrium at
700−800 °C by several per mil (Fosu et al. 2020), collectively suggest that there are important
processes at play that are not fully understood, and that uncharacterized variability in these
parameters may be a factor. Additionally, these values are generally determined using CO2
prepared by Pt-catalyzed combustion of graphite with O2 of known ∆′17O. If unusual isotope
454 Passey & Levin
Table 1. Interlaboratory reproducibility for D′17O of carbonate reference materials.

Source NBS19 NBS18 IAEA 603 DD′17O
D′17O D′17O D′17O D′17O D′17O D′17O NBS18 – NBS19
CO2 CaCO3 CO2 CaCO3 CO2 CaCO3
Conversion Methods
a
Passey et al. (2014) −135 −95 −98 −58 − − 37
Passey and Ji (2019)a −143 −103 −88 −48 − − 55
Wostbrock et al. (2020)b −155 −102 −100 −48 −147 −100 54
a
Michigan IsoPaleoLab −142 −102 −91 −51 −148 −108 51
Avg. −100 Avg. −51 Avg. −104 49
Std. 4 Std. 5 Std. 5 8
Exchange Methods
Barkan et al. (2015)c −227 −176 3 54 − − 230
c
Barkan et al. (2019) −182 −131 −163 −112 −187 −136 19
Sha et al. (2020) c − − −225 −174 −267 −216 42d
Fosu et al. (2020)c −169 −118 −119 −67 − − 50
Avg. −141 Avg. −75 Avg. −176 100
Std. 30 Std. 96 Std. 57 114
Notes: D′17O values are reported in per meg relative to the VSMOW-SLAP scale, l = 0.528. CaCO3 values are
calculated based on CO2 values using qacid = 0.5230 (Wostbrock et al. 2020), 18aacid = 1.0103 (25 °C reactions)
and 18aacid = 1.00812 (90 °C reactions; Kim et al. 2007), except for the CaCO3 values for Wostbrock et al.
(2020), which are directly measured.
a
. Reduction of CO2 produced by 90 °C acid digestion, and CoF3 fluorination of the resulting water.
b
. Direct BrF5 fluorination carbonate, and of CO2 produced by 25 °C acid digestion.
c
. Pt-catalyzed exchange with O2 of CO2 produced by 25 °C acid digestion.
d
. D′17O difference between NBS18 and IAEA 603.
effects characterize this reaction in the same way they have been demonstrated to characterize
the Pt-catalyzed CO2–O2 exchange reaction, then the ∆′17O values of the produced CO2 may
deviate from those of the combustion O2. For all of these reasons, a logical first step would be
to use NBS18-CO2, NBS19-CO2, and IAEA 603-CO2 (or other materials calibrated by direct
fluorination methods) to develop or validate 18/16aPt and qPt values (e.g., Fosu et al. 2020).
Fractionation exponents for carbonate–water equilibrium, and acid digestion
Calculation of the triple oxygen isotope composition of carbonate and apatite parent waters
requires knowledge of the fractionation exponents for mineral–water equilibrium, and for acid
digestion. Theoretical predictions by Cao and Liu (2011) for qcc–H2O are 0.5235 at 25 °C, and they
predict weak temperature dependence over the range 0–100 °C (0.5233–0.5239). More recently,
Hayles et al. (2018) calculated a value of 0.5257 at 25 °C, and, Guo and Zhou (2019a) predicted
values of 0.5253 to 0.5263 over the 0–100 °C temperature range, with a value of 0.5256 at 25 °C.
Using an empirical approach and the reduction–fluorination conversion method, Passey
et al. (2014) reported a value of θ = 0.5245 for combined mineral–water fractionation and
acid digestion (90 °C). This value is based on analysis of fresh avian eggshell carbonate and
water extracted from the same eggs (egg whites), and from analysis of an estuarine oyster and
a marine coral and the waters they were found in at the time of sampling. Unpublished data
from slow calcite precipitations in our laboratory (passive degassing) over a temperature range
of 0–60 °C give a mean value of θ = 0.5242 for combined mineral–water fractionation and acid
digestion, with no resolvable relationship with temperature. More recently, Sha et al. (2020)
studied cave dripwater / speleothem pairs and obtained a value of 0.5235 combined mineral–
water fractionation and acid digestion at 25 °C using the Pt-exchange method. Bergel et al.
(2020) used the Pt-exchange method to derive a value of θ = 0.5231 for combined mineral–
water fractionation and acid digestion at 25 °C, based on analyses of freshwater mollusks
from well-constrained spring environments. Voarintsoa et al. (2020) of the same laboratory
observed a value of θ = 0.5225 for laboratory-precipitated carbonates in the temperature range
10 to 35 °C. Neither study observed clear temperature dependence of the θ value, although
the 35 °C values were distinctly lower (θ = 0.5220), which may reflect disequilibrium during
very rapid precipitation of this material. Voarintsoa et al. (2020) observe no clear influence
of carbonate polymorph (calcite vs. aragonite) or parent solution concentration on θ values.
The only direct measurement of the qacid value is that of Wostbrock et al. (2020), who
report a value of 0.5230 for 25 °C reactions, which is substantially lower than an early
theoretical prediction of 0.5283 by Guo et al. (2009).
The variability in θ values described above probably arises from a combination of
methodological differences, differences in the exact nature of the θ value (i.e., mineral–water,
or acid-digestion-produced CO2–water, with two different acid digestion temperatures used).
Even if more reliable experimental estimates are obtained (i.e., from direct fluorination of
laboratory-precipitated carbonates), it may still be the case that each laboratory would be
better to use its own laboratory-derived θ values, given that these are influenced by the same
analytical artifacts as samples processed in those laboratories.
FUTURE DIRECTIONS AND CONCLUDING REMARKS

We have summarized the considerable progress that has been made during the past 15
years in the exploration and application of triple oxygen isotopes to continental hydroclimate,
and have hopefully imbued the reader with an appreciation of the great potential for using triple
oxygen isotopes in mineral records to unravel aspects of past climate and ecology. Despite the
progress that has been made, the number of detailed studies into many systems such as lakes,
plants, and animals can still be counted on one hand (or maybe two!), and many systems are
just beginning to be studied or remain unstudied.
Soil carbonates. Triple oxygen isotope studies of soil carbonates promise to reveal
the extent to which soil waters from which the carbonates precipitated were evaporatively
enriched in heavy isotopes, as well as changes in the ∆′17O of precipitation that may result
from changes in regional atmospheric moisture sources. To our knowledge, the only published
data for soil carbonates are the handful of data points in Passey et al. (2014) for modern soils
in Inner Mongolia, California, and Ethiopia. Based on that dataset, and emerging data from
our lab (Ji 2016; Emily Beverly, Julia Kelson, and Tyler Huth, pers. comm.), ∆′17O values
for soil carbonate parent waters typically fall somewhere between values typical of meteoric
waters, and values typical of closed-basin lakes. In other words, many samples show signs of
evaporative modification, but the degree of lowering of ∆′17O is typically less than that observed
for closed basin lakes. As was the case in the study of clumped isotopes in soil carbonates, it is
becoming clear that triple oxygen isotopes will tell us as much about the seasonal timing and
circumstances of soil carbonate formation as they will about prevailing climatic conditions.
Speleothems. Sha et al. (2020) have presented the first triple oxygen isotope data for
speleothems, studying a variety of samples from several caves around the world. Based on the
present knowledge of triple oxygen isotope systematics, we can expect that cave drip water triple
oxygen isotope compositions will vary with: (1) precipitation compositions, which are sensitive
both to the effective relative humidity and turbulence over the oceans during the production of
456 Passey & Levin
atmospheric vapor bodies (e.g., Fig. 2), and (2) evaporative modification of precipitation before
it infiltrates into the subsurface or, in the case of rh < 1 caves, within the cave environments
(Fig. 5). The speleothem mineral will record these factors, as well as (3) any kinetic effects
that characterize the mineral growth process. Guo and Zhou (2019a,b) have explored the triple
oxygen isotope systematics of the system DIC–H2O–CaCO3, and have developed a reaction–
diffusion model to incorporate their predictions of triple isotope fractionation exponents into
the transient process of mineralization. They predict kinetically-driven isotope trajectories with
initial slopes of ~0.535, meaning that an array of kinetically-influenced sample would plot with
a positive slope on a plot of ∆′17O (λ = 0.528) versus d18O. This is a distinctive slope compared
to the universally negative slopes for the different systems that we have described in this review.
All of the expected effects for ∆′17O in speleothems (moisture source effect, evaporation effect,
kinetic effect) are subtle (Sha et al. 2020; Tyler Huth, pers. comm.), ranging by a few tens of per
meg at most, so improving analytical precision beyond the 10 per meg level will be expedient.
Tree rings and other organics. As far as we are aware, triple oxygen isotopes have not
been measured to high precision in organic molecules. This should be possible using direct
fluorination or high-temperature reduction to produce CO that could then be reduced to water
and fluorinated using the approach of Passey et al. (2014). As shown in Figure 12, leaf water
∆′17O shows high sensitivity to relative humidity. Plant cellulose derives ~40% (Roden et
al. 2000) of its oxygen from xylem water (which is not fractionated relative to soil water),
meaning that the evaporation signal will be damped, but that it will still be highly resolvable.
Thus, application to tree-ring records could help disentangle variations in d18O that are caused
by changes in the isotopic composition of precipitation, from those that are caused by changes
in evaporative enrichment. Biosynthesis of organic molecules involves isotopic fractionation,
the mechanisms of which could possibly be probed using triple oxygen isotopes. Oxygen-
containing molecules formed in waters of respiration should carry some of the anomalous
triple oxygen isotope composition of atmospheric oxygen.
As for the more ‘well studied’ systems (precipitation, lakes, plants, animals), the existing
datasets are only a beginning. Most of the precipitation data are based on opportunistic
sampling, and only the studies of Tian et al. (2018) and Uechi and Uemura (2019) involve
the kind of continuous, systematic sampling that will reveal the mechanisms of e.g., seasonal
variation in ∆′17O, and allow for determination of amount-weighted ∆′17Ο of precipitation.
The ∆′17Ο of atmospheric water vapor is a fundamental parameter for evaporation, but only a
handful of studies have conducted measurements (Uemura et al. 2010; Landais et al. 2012; Lin
et al. 2013; Surma et al. 2021, this volume), and of these only the Surma et al. study is from a
non-polar continental setting. Lake systems are tremendously understudied, especially in more
humid settings that may be less ‘exciting’ in terms of large evaporative ∆′17Ο signals, but that
are nonetheless important for understanding the full scope of the system. Back-calculation of
isotopic compositions of unevaporated waters (Passey and Ji 2019) could be improved with
more sophisticated methods such as Bayesian analysis (e.g., Bowen et al. 2018). Leaves come
in a range of physiologies (i.e., blades of grass, conifer needles, typical hardwood leaves)
for which detailed d18O–δD models have been developed (Cernusak et al. 2016). It will be
important to expand these models and observations to triple oxygen isotopes, both in terms of
understanding biogeochemical fluxes of O2 and CO2 (which are influenced by leaf water), and
understanding the compositions of water available to herbivores. Very little of the taxonomic
diversity of land vertebrates has been explored (e.g., reptiles, birds), and the work on mammals
summarized above is only a starting point. If we are to use triple oxygen isotopes of vertebrate
biominerals to reconstruct paleoecology, paleoclimate, and paleo-atmospheric composition,
then we will need exhaustive study of modern organisms in relation to their environments and
physiologies, and the further development and refinement of body water models.
Finally, improvement in inter-laboratory reproducibility is urgent and imperative. Yeung

et al. (2018) show that pressure-baseline corrections or continual referencing to well-defined,
isotopically-disparate materials (i.e., VSMOW2 and SLAP2) is necessary for accuracy at the
10 per meg level. In terms of standardization protocols, the field is still in the ‘wild west’ stage
compared to more mature fields with similar analytical challenges. Taking carbonate clumped
isotopes as an example, a key paper (Dennis et al. 2011) was published 5 years after the first
clumped isotope papers were published, outlining a rigorous protocol for standardization. This
protocol involved routine analysis of reference materials spanning a wide range in both bulk
isotopic composition (d13C and d18O), and in magnitude of the ∆47 clumping signal. This protocol
was almost universally-adopted, and as a result there is remarkable interlaboratory agreement in
the field (e.g., Petersen et al. 2019). The analogous approach for ∆′17Ο in carbonates would be
to employ a suite of at least three, and preferably four carbonate materials, two of which span a
wide range in d18O and have fairly high ∆′17O, and one (or two) of which span a similar range in
d18O and have fairly low ∆′17O. Routine analysis of such materials would allow for monitoring
of, correction for, issues of scale compression (in the ∆′17Ο space), as well as accuracy.
ACKNOWLEDGEMENTS
First and foremost, we express gratitude to the talented group of students and postdocs
with whom we’ve journeyed into the world of triple oxygen isotopes, including our ‘founding’
crew at Johns Hopkins University (Shuning Li, Huanting Hu, Haoyuan Ji Sophie Lehmann,
Nicole DeLuca, and Jessica Moerman), and those who have carried the torch at Michigan (Ian
Winkelstern, Emily Beverly, Drake Yarian, Phoebe Aron, Elise Pelletier, Sarah Katz, Natalie
Packard, Joonas Wasiljeff, Tyler Huth, Nick Ellis, and Julia Kelson). We thank Andreas Pack
and Ilya Bindeman for giving us the opportunity to write this chapter, and Jordan Wostbrock
and an anonymous reviewer for insightful reviews.
REFERENCES
Adnew GA, Hofmann MEG, Paul D, Laskar A, Surma J, Albrecht N, Pack A, Schwieters J, Koren G, Peters W,
Comm Mass Spectrom 33:1363−1380
Alexandre A, Landais A, Vallet-Coulomb C, Piel C, Devidal S, Pauchet S, Sonzogni C, Couapel M, Pasturel M,
Cornuault P, Xin J, Mazur J-C, Prié F, Bentaleb I, Webb E, Chalié F, Roy J (2018) The triple oxygen isotope
composition of phytoliths as a proxy of continental atmospheric humidity: insights from climate chamber and
climate transect calibrations. Biogeosciences 15:3223−3241
Assonov SS, Brenninkmeijer CAM (2001) A new method to determine the 17O isotopic abundance in CO2 using
oxygen isotope exchange with a solid oxide. Rapid Commum Mass Spectrom 15:2426−2437
Ayliffe LK, Chivas AR (1990) Oxygen isotope composition of the bone phosphate of Australian kangaroos: Potential
as a paleoenvironmental recorder. Geochim Cosmochim Acta 54:2603–2609
Baker L, Franchi IA, Maynard J, Wright IP, Pilinger CT (2002) A technique for the determination of 18O/16O and
17 16
O/ O isotopic ratios from small liquid and solid samples. Anal Chem 74:1665−1773
Barkan E, Luz B (1996) Conversion of O2 into CO2 for high-precision oxygen isotope measurements. Anal Chem
68:3507–3510
Spectrom 19:3737–3742
Barkan E, Luz B (2007) Diffusivity fractionations of H216O/H217O and H216O/H218O and their implications for isotope
hydrology. Rapid Commun Mass Spectrom 21: 2999– 3005
Barkan E, Luz B (2012) High-precision measurements of 17O/16O and 18O/16O ratios in CO2. Rapid Commun Mass
Spectrom 26:2733–2738
Barkan E, Musan I, Luz B (2015) High-precision measurements of δ17O and 17O-excess of NBS−19 and NBS−18.
458 Passey & Levin
Barkan E, Affek HP, Luz B, Bergel SJ, Voarintsoa NRG, Musan I (2019) Calibration of d17O and 17Oexcess values
of three international standards: IAEA−603, NBS19 and NBS18. Rapid Commun Mass Spectrom 33:737−740
glaciation. Nature 453:504−506
Benedix GK, Leshin LA, Farquhar J, Jackson T, Thiemens MH (2003) Carbonates in CM2 chondrites: Constraints on
alteration conditions from oxygen isotopic compositions and petrographic observations. Geochim Cosmochim
Acta 67:1577–1588
Benson LV, White JWC (1994) Stable isotopes of oxygen and hydrogen in the Truckee River-Pyramid Lake surface-
water system. 3. Source of water vapor overlying Pyramid Lake. Limnol Oceanogr 39:1945−1958
fractionations between H2O and biogenic aragonite, derived from freshwater mollusks. Geochim Cosmochim
Acta, in press
Berman ESF, Levin NE, Landais A, Li S, Owano T (2013) Measurement of d18O, d17O, and 17O-excess by off-axis
integrated cavity output spectroscopy and isotope ratio mass spectrometry. Anal Chem 85:10392−10398
Blumenthal SA, Levin NE, Brown FH, Brugal JP, Chritz KL, Harris JM, Jelhe GE, Cerling TE (2017) Aridity and
hominin environments. PNAS 114:7331−7336
Bony S, Risi C, Vimeux F (2008) Influence of convective processes on the isotopic composition (d17O, dD) of
precipitation and water vapor: 1. Radiative-convective equilibrium and Tropical Ocean-Global Atmosphere-
Coupled Ocean-Atmosphere Response Experiment (TOGA-COARE) simulations. J Geophys Res 113:D19305
Bowen GJ, Putman A, Brooks JR, Bowling DR, Oerter EJ, Good SP (2018) Inferring the source of evaporated waters
using stable H and O isotopes. Oecologia 187:1025–1039
Brenninkmeijer CAM, Röckmann T (1998) A rapid method for the preparation of O2 from CO2 for mass spectrometric
measurement of 17O/16O ratios. Rapid Comm Mass Spectrom 12:479–483
Brinjikji M, Lyons JR (2021) Mass-independent fractionation of oxygen isotopes in the atmosphere. Rev Mineral
Geochem 86:197–216
Bryant JD, Froelich PN (1995) A model of oxygen isotope fractionation in body water of large mammals. Geochim
Cao X, Bao H (2013) Dynamic model constraints on oxygen−17 depletion in atmospheric O2 after a snowball Earth.
PNAS 110:14546–14550
Cosmochim Acta 75:7435−7445
Cappa CD, Hendricks MB, DePalo DJ, Cohen RC (2003) Isotopic fractionation of water during evaporation. J
Geophys Res 108(D16):4525
Cernusak LA, Barbour MM, Arndt SK, Cheesman AW, English NB, Field TS, Helliker BR, Holloway-Phillips MM,
Holtum JAM, Kahmen A, McInerney FA, Munksgaard NC, Simonin KA, Song X, Stuart-Williams H, West JB,
Farquhar GD (2016) Stable isotopes in leaf water of terrestrial plants. Plant Cell Envir 39:1087−1102
Craig H, Gordon LI (1965) Deuterium and oxygen 18 variations in the ocean and the marine atmosphere. In E
Tongiorgi, ed. Proceedings of a conference on stable isotopes in oceanographic studies and paleotemperatures,
Spoleto, Italy, p 9−130
Criss REC (1999) Principles of stable isotope distribution. Oxford University Press
Dennis KJ, Affek HP, Passey BH, Schrag DP, Eiler JM (2011) Defining and absolute reference frame for mass
spectrometry. Geochim Cosmochim Acta 75:7117−7131
Ehhalt D, Knott K (1965) Kinetische isotopentrennung bei der verdampfung von wasser. Tellus 17:389−397
Ehleringer JR, Dawson TE (1992) Water-uptake by plants—perspectives from stable isotope composition. Plant Cell
Envir 15:1073−1082
Epstein S, Zeiri L (1988) Oxygen and carbon isotopic compositions of gases respired by humans. PNAS 85:1727−1731
Flanagan LB, Comstock JP, Ehleringer JR (1991) Comparison of modeled and observed environmental influences on the
stable oxygen and hydrogen isotope composition of leaf water in Phaseolus vulgaris L Plant Physiol 96:588−596
Fosu BR, Subba R, Peethambaran R, Ghosh P (2020) Technical note: Developments and applications in triple oxygen
isotope analysis of carbonates. ACS Earth Space Chem, in press
Gat JR (1996) Oxygen and hydrogen isotopes in the hydrologic cycle. Annu Rev Earth Planet Sci 24:225–262
Gat JR, Bowser CJ, Kendall C (1994) The contribution of evaporation from the Great Lakes to the continental
atmosphere: estimate based on stable isotope data. Geophys Res Lett 21:557−560
Gázquez F, Evans NP, Hodell DA (2017) Precise, accurate fractionation factors (α17O, α18O, αD) for water and
CaSO4∙2H2O (gypsum). Geochim Cosmochim Acta 198:259−270
Gázquez F, Morellón M, Bauska T, Herwartz D, Surma J, Moreno A, Staubwasser M, Valero-Garces B, Delgado-
paleo-humidity reconstruction. Earth Planet Sci Lett 481:177−188
Gehler A, Tütken T, Pack A (2011) Triple oxygen isotope analysis of bioapatite as a tracer for diagenetic alteration of
bones and teeth. Paleaogr Palaeoclim Palaeoecol 310:84−91
PETM using triple oxygen isotope analysis of mammalian bioapatite. PNAS 113:7739−7744
Gibson JJ, Birks SJ, Yi Y (2016) Stable isotope mass balance of lakes: a contemporary perspective. Quat Sci Rev
131B:316−328
Gonfiantini R (1978) Standards for stable isotope measurements in natural compounds. Nature 271:534−536
Guo W, Zhou C (2019a) Triple oxygen isotope fractionation in the DIC–H2O–CO2 system: A numerical framework
and its implications. Geochim Cosmochim Acta 246:541−564
Guo W, Zhou C (2019b) Patterns and controls of disequilibrium isotope effects in speleothems: Insights from an
isotope-enabled diffusion-reaction model and implications for quantitative thermometry. Geochim Cosmochim
Acta 267:196–226
Guo W, Mosenfelder JL, Goddard III WA, Eiler JM (2009) Isotopic fractionations associated with phosphoric
acid digestion of carbonate minerals: Insights from first-principles theoretical modeling and clumped isotope
measurements. Geochim Cosmochim Acta 73:7203−7225
Geochim Cosmochim Acta 235, 237−245
Hofmann MEG, Pack A (2010) Technique for high-precision analysis of triple oxygen isotope ratios in carbon
dioxide. Anal Chem 82:4357−4361
Horita J, Rozanski K, Cohen S (2008) Isotope effects in the evaporation of water: a status report of the Craig–Gordon
model. Isot Environ Health Stud 44:23–49
Hren MT, Sheldon ND (2012) Temporal variations in lake water temperature: Paleoenvironmental implications of
lake carbonate d18O and temperature records. Earth Planet Sci Lett 337−338:77−84
Hu H (2016) Triple oxygen isotopes of biominerals: A new proxy for reconstructing paleoaridity, paleoecophysiology,
paleo-carbon-cycling. PhD dissertation, Johns Hopkins University (143p)
Jasechko S, Sharp ZD, Gibson JJ, Birks SJ, Yi Y, Fawcett PJ (2013) Terrestrial water fluxes dominated by transpiration.
Nature 496:347−350
Ji H (2016) Triple oxygen isotopes in lake waters, lacustrine carbonates, pedogenic carbonates: An indicator for
evaporation. PhD dissertation, Johns Hopkins University
Kawagucci S, Tsunogai U, Kudo S, Nakagawa F, Honda H, Aoki S, Nakazawa T, Gamo T (2005) An analytical
system for determining d17O in CO2 using continuous flow-isotope ratio MS Anal Chem 77:4509−4514
Koch PL (1998) Isotopic reconstruction of past continental environments. Annu Rev Earth Planet Sci 26:573−613
Kohn MJ (1996) Predicting animal d18O: Accounting for diet and physiological adaptation. Geochim Cosmochim
Acta 60:4811−4829
Kohn MJ, Cerling TE (2002) Stable isotope compositions of biological apatite. Rev Mineral Geochem 48:455−488
Landais A, Barkan E, Luz B (2008) Record of δ18O and δ17O-excess in ice from Antarctica during the last 150,000
years. Geophys Res Lett 35:L02709
Landais A, Risi C, Bony S, Vimeux F, Descroix L, Falourd S, Bouygues A (2010) Combined measurement of 17O-excess
and d-excess in African monsoon precipitation: Implication for evaluating convective parameterizations. Earth
Planet Sci Lett 298:104−112
Landais A, Steen-Larsen HC, Guillevic M, Masson-Delmotte V, Vinther B, Winkler R (2012) Triple isotopic composition
of oxygen in surface snow, water vapor at NEEM (Greenland). Geochim Cosmochim Acta 77:304–316
Lehmann SB (2016) Studies of carbon, oxygen,, strontium isotopes in tooth enamel: Evaluating paleoenvironmental
change in South Africa and expanding the paleoclimate took kit. PhD dissertation, Johns Hopkins University
Levin NE, Cerling TE, Passey BH, Harris JM, Ehleringer JR (2006) A stable isotope aridity index for terrestrial
environments. PNAS 103:11201−11205
Geochim Cosmochim Acta 164:110−126
Li S, Levin NE, Soderberg K, Dennis KJ, Caylor KK (2017) Triple oxygen isotope composition of leaf waters from
Mpala, central Kenya. Earth Planet Sci Lett 468:38−50
Liang M-C, Mahata S, Laskar AH, Theimens MH, Newman S (2017) Oxygen isotope anomaly in tropospheric CO2
and implications for CO2 residence time in the atmosphere and gross primary productivity. Sci Report 7:13180
Lin Y, Clayton RN, Huang L, Nakamura N, Lyons JR (2013) Oxygen isotope anomaly observed in water vapor from
Alert, Canada and the implication for the stratosphere. PNAS 110:15608–15613
Luz B, Kolodny Y (1985) Oxygen isotope variations in phosphate of biogenic apatites, IV Mammal teeth and bones.
Earth Planet Sci Lett 75:29−36
460 Passey & Levin
biogeochemistry. Geochim Cosmochim Acta 69:1099−1110
Luz B, Barkan E (2010) Variations of 17O/16O and 18O/16O in meteoric waters. Geochim Cosmochim Acta 74:6276−6286
Luz B, Cormie AB, Schwarcz HP (1990) Oxygen isotope variations in phosphate of deer bones. Geochim Cosmochim
Acta 54:1723−1728
as a tracer of biospheric productivity. Nature 400:547−550
Mahata S, Bhattacharya SK, Wang C-H, Liang M-C (2012) An improved CeO2 method for high-precision
measurements of 17O/16O ratios for atmospheric carbon dioxide. Rapid Commun Mass Spectrom 26:1909−1922
platinum: An innovative technique for measuring Δ17O in CO2. Anal Chem 85:6894−6901
Mahata S, Bhattacharya S, Liang MC (2016) An improved method of high-precision determination of Δ17O of CO2 by
catalyzed exchange with O2 using hot platinum. Rapid Commun Mass Spectrom 30:119–131
Majoube MF (1971) Fractionnement en oxygène−18 et en deuterium entre l’eau et sa vapeur. J Chem Phys
58:1423−1436
Merlivat L (1978) Molecular diffusivities of H216O, HD16O, H218O in gases. J Chem Phys 69(6):2864−2871
Merlivat L, Jouzel J (1979) Global climatic interpretation of deuterium–oxygen-18 Relationship for precipitation. J
Geophys Res 84:5029−5033
Merrit DA, Hayes JM (1994) Factors controlling precision and accuracy in isotope-ratio-monitoring mass
spectrometry. Anal Chem 66:2336−2347
Miller MF, Franchi IA, Thiemens MH, Jackson TL, Brack A, Kurat G, Pilinger CT (2002) Mass-independent fractionation
of oxygen isotopes during thermal decomposition of carbonates. PNAS 99:10988−10993
Mrozek DJ, van der Veen C, Kliphuis M, Kaiser J, Weigel AA, Röckmann T (2015) Continuous-flow IRMS technique
for determining the 17O excess of CO2 using complete oxygen isotope exchange with cerium oxide. Atmos Meas
Tech 8:811:822
Nagy KA, Peterson CC (1988) Scaling of water flux rate in animals. University of California Publications in Zoology
120:1−172. University of California Press
New M, Lister D, Hulme M, Makin I (2002) A high-resolution data set of surface climate over global land areas. Clim
Res 21:1−25
Pack (2021) Isotopic traces of atmospheric O2 in rocks, minerals, and melts. Rev Mineral Geochem 86:217–240
variations in terrestrial rocks and minerals. Earth Planet Sci Lett 390:138−145
as a paleo-CO2-barometer. Geochim Cosmochim Acta 102:306−317
Pack A, Höweling A, Hezel DC, Stefanak MT, Beck A-K, Peters STM, Sengupta S, Herwartz D, Folco L (2017) Tracing
the oxygen isotope composition of the upper Earth’s atmosphere using cosmic spherules. Nat Commun 8:15702
from the western United States. Earth Planet Sci Lett 518:1−12
Passey BH, Hu H, Ji H, Montanari S, Li S, Henkes G, Levin NE (2014) Triple oxygen isotopes in biogenic and
sedimentary carbonates. Geochim Cosmochim Acta 141:1−25
Petersen SV,, 29 others (2019). Effects of improved 17O correction on interlaboratory agreement in clumped isotope
calibrations, estimates of mineral-specific offsets, and temperature dependence of acid digestion fractionation.
Geochem Geophys Geosyst 20:3495–3519
Prasanna K, Bhattacharya SK, Ghosh P, Mahata S, Liang M-C (2016) Isotopic homogenization, scrambling
associated with oxygen isotopic exchange on hot platinum: Studies on gas pairs (O2, CO2), (CO, CO2). RSC
Adv 6:51296−51303
Prokhorov I, Kluge T, Janssen C (2019) Laser absorption spectroscopy of rare and doubly substituted carbon dioxide
isotopologues. Anal Chem 91:15491−15499
Roden JS, Lin G, Ehleringer JR (2000) A mechanistic model for interpreting hydrogen and oxygen isotope ratios in
tree-ring cellulose. Geochim Cosmochim Acta 64:21−35
Rowley DB, Garzione CN (2007) Stable isotope-based paleoaltimetry. Annu Rev Earth Planet Sci 35:463−508
Sakai S, Matsuda S, Hikida T, Shimono A, McManus JB, Zahniser M, Nelson D, Dettman DL, Yang D, Ohkouchi N
by tuneable infrared laser absorption spectroscopy. Anal Chem 89:11846−11852
Sha L, Mahata S, Duan P, Luz B, Zhang P, Baker J, Zong B, Ning Y, Ait Brahim Y, Zhang H, Edwards RL, Cheng H
(2020) A novel application of triple oxygen isotope ratios of speleothems. Geochim Cosmochim Acta 270:360−378
Sharma T, Clayton RN (1965) Measurement of O18/O16 ratios of total oxygen in carbonates. Geochim Cosmochim
Acta 29:1347−1353
Geochem Perspec Lett 7:27−31
Steig EJ, Gkinis V, Schauer AJ, Shoenemann SW, Samek K, Hoffnagle J, Dennis KJ, Tan SM (2014) Calibrated
resonance. Atmos Meas Tech 7:2421
Stewart MK (1975) Stable isotope fractionation due to evaporation and isotopic exchange of falling waterdrops:
Applications to atmospheric processes and evaporation of lakes. J Geophys Res 80:1133−1146
bodies from the Sistan Oasis, Iran. Geophys Res Lett 42:8456−8462
Surma J, Assonov S, Staubwasser M (2021) Triple oxygen isotope systematics in the hydrologic cycle. Rev Mineral
Geochem 86:401–428
Tian C, Wang L, Kaseke KF, Bird BW (2018) Stable isotope compositions (d2H, d18O, d17O) of rainfall and snowfall
in the central United States. Sci Rep 8:6712
precipitation on a subtropical island. Earth Planet Sci Lett 513:20−28
and H2O in inorganically-precipitated calcite and aragonite. Chem Geol, in press
Wang Z, Nelson DD, Dettman DL, McManus JB, Quade J, Huntington KW, Schauer AJ, Sakai S (2020) Rapid
and precise analysis of carbon dioxide clumped isotopic composition by tunable infrared laser differential
spectroscopy. Anal Chem 92:2034−2042
White JWC (1989) Stable hydrogen isotope ratios in plants: a review of current theory, some potential applications.
In: Applications of Stable Isotopes in Ecological Research. PW Rundel, JR Ehleringer, KA Nagy (eds) Springer-
Verlag, New York, p 142−160
Whiteman JP, Sharp ZD, Gerson AR, Newsome SD (2019) Relating D17Ο values of animal body water to exogenous
water inputs and metabolism. BioSci 69:658−688
in triple-isotope measurements. Rapid Commun Mass Spec 32:1811−1821
Young ED, Galy A, Nagahara H (2002) Kinetic and equilibrium mass-dependent fractionation laws in nature and their
geochemical and cosmochemical significance. Geochim Cosmochim Acta 66:1095−1104
Young ED, Yeung LY, Kohl IE (2014) On the D17Ο budget of atmospheric O2. Geochim Cosmochim Acta 135:102−125
Vol. 86 pp. 463–488, 2021 14
Small Triple Oxygen Isotope Variations in Sulfate:

Mechanisms and Applications
Xiaobin Cao
International Center for Isotope Effects Research
School of Earth Sciences and Engineering
Nanjing University
Nanjing 210023
PR China
xiaobincao@nju.edu.cn
Huiming Bao
International Center for Isotope Effects Research
School of Earth Sciences and Engineering
Nanjing University
Nanjing 210023
PR China
and
Department of Geology & Geophysics
Louisiana State University,
Baton Rouge, LA 70803
USA
bao@lsu.edu
INTRODUCTION
Sulfate is the most abundant electron acceptor in the ocean today. A large fraction of
the buried organic matter in marine sediments is re-mineralized through microbial sulfate
reduction (MSR) during which the sulfate is reduced to H2S (Jørgensen 1982; Kasten and
Jørgensen 2000). The H2S can be re-oxidized to sulfate or buried as pyrite in sediments
(Jørgensen 1977). The burial of pyrite ultimately contributes to the rising of atmosphere O2
concentration (Berner and Canfield 1989). Sulfate, meanwhile, can be buried as gypsum and
anhydrite in evaporites (Claypool et al. 1980; Crockford et al. 2019; Spencer 2000) or as barite
in sedimentary rocks (Hanor 2000; Bao et al. 2008; Peng et al. 2011; Griffith and Paytan 2012;
Crockford et al. 2016).
Sedimentary rocks can be uplifted and weathered with or without being subducted, melted,
or metamorphosed. Thus, the initially buried sulfur minerals are transformed and eventually
turned to sulfate under oxidizing atmosphere through pyrite oxidation and evaporite dissolution
(Bottrell and Newton 2006). Any sulfur-bearing minerals in igneous and metamorphic rocks
will also eventually be released as sulfate upon oxidative weathering and carried to the oceans.
The oxygen isotope composition of sulfate reveals the chemical pathways sulfate has
experienced during its formation and consumption in sulfur cycling (Fig. 1). Secondary
atmospheric sulfate can carry atmospheric O3 and/or O2 signature (Savarino et al. 2000;
Harris et al. 2013; Bao 2015). At the surface, sulfate formed through sulfide mineral oxidation
carries atmosphere O2 and ambient water oxygen isotope signatures (Bao 2015). During MSR,

1943-2666/21/0086-0014$05.00 (online)
464 Cao & Bao
Figure 1. A global sulfur cycle in light of sulfate oxygen sources and sinks. Grey texts are sulfur reser-
voirs, bold black texts list the potential sources of oxygen for product sulfate during sulfur oxidation, MSR
refers to microbial sulfate reduction, and arrows mark processes.
sulfate can exchange its oxygen isotope composition with ambient water via intermediates
toward thermodynamic equilibrium (Wortmann et al. 2007; Zeebe 2010), erasing some or all
the O2 and water oxygen isotope signatures the sulfate may have acquired initially (Mizutani
and Rafter 1973; Fritz et al. 1989; Brunner et al. 2005). Evaporite sulfate represents well the
contemporary seawater sulfate of geological times (Claypool et al. 1980; Crockford et al. 2019).
Sulfate’s oxygen isotope composition has received less attention than sulfate’s sulfur
isotope composition, due largely to oxygen’s multiple sources and variable non-equilibrium
isotope signatures. The normalized 18O/16O ratio or the δ18O value is traditionally measured
(Lloyd 1967, 1968). In the last 20 years, data on triple oxygen isotope composition (i.e.,
Δ′17O ≡ δ′17O − 0.5305 × δ′18O) of sulfate has been accumulating. Distinctly large positive and
negative 17O anomalies have been found in the atmosphere and/or geological sulfate deposits
(see review papers Thiemens 2006; Bao 2015; Crockford et al. 2019). These discoveries have
provided exciting new insights into past atmospheric processes associated with volcanism (Bao
et al. 2010), desert salt deposits (Bao et al. 2000a,b), snowball Earth (Bao et al. 2008, 2009),
and gross primary productivity (Crockford et al. 2018; Hodgskiss et al. 2019).
Over the years, researchers have discovered that there are analytically resolvable differences
in the Δ′17O of sulfate produced by entirely mass-dependent reactions that do not involve
O3, an oxidant bearing a large 17O anomaly (Bao et al. 2008; Sun et al. 2015; Killingsworth
et al. 2018; Waldeck et al. 2019; Hemingway et al. 2020). We call these differences small
triple oxygen isotope variations or small 17O deviations (Bao 2019). Apparently intriguing
patterns have been reported and geological and environmental significances inferred.
However, interpreting small sulfate Δ′17O data is not a trivial matter.
In this chapter, we will explore the origins of small 17O deviations or small Δ′17O values
in sulfate. Large positive or negative sulfate 17O anomalies will, therefore, not be covered
here, and the readers can refer to recent reviews (Thiemens 2006; Bao 2015; Crockford et
al. 2019) for details. Small sulfate Δ′17O values are sensitive not only to source of oxygen
but also to reaction mechanisms because equilibrium and kinetic processes generate different
small non-zero Δ′17O values (Young et al. 2002; Angert et al. 2004; Barkan and Luz 2007;
Pack and Herwartz 2014; Bao et al. 2015). Sulfate reduction drives the remaining sulfate
oxygen toward isotope equilibrium with ambient water, resulting also in a change of the small
sulfate Δ′17O value. Since the change during sulfate reduction process is largely controlled
by ambient water isotope composition, this review will focus more on reaction mechanism
and associated oxygen isotope effects on sulfate formed via pyrite oxidation. We adopt the
approach of isotopologue-specific kinetic analysis (Cao and Bao 2017; Cao et al. 2019),
Small 17O Variations in Sulfate: Mechanisms and Applications 465
which helps to identify and subsequently estimate the most important parameters in determining
sulfate’s small Δ′17O values. Isotopologue specific kinetic details during sulfate redox reactions
are sketchy at this time and we will approach the problem using endmember scenarios.
The intrinsic triple isotope parameters determined will then be used to construct Δ′17O − δ18O
space for sulfates derived from different endmember scenarios. Such Δ′17O − δ18O space
should be applicable not only to small Δ′17O but also to the δ18O or large Δ′17O values, and
therefore can be tested and further revised. Specific examples on riverine and lake sulfate data
will be analyzed to show potential applications. Analytical methods and issues in measuring
small sulfate Δ′17O and future research opportunities are outlined in the end.
TRIPLE OXYGEN ISOTOPE SYSTEM

Oxygen has three stable isotopes, i.e., 16O, 17O, and 18O. The δ notation is introduced to describe
their small relative abundance variation in nature, and it is defined as (McKinney et al. 1950)
 17,18 R 
 17,18 O   17,18 sample  1   1000 ‰ (1)
 Rref 
where 17,18R is the mole ratio of 17,18O/16O; Rsample and Rref refer to R value for samples of
interest and reference, respectively. Standard Mean Ocean Water (SMOW) (Craig 1961) (later
Vienna-SMOW) is the reference material in most oxygen isotope studies. The notion of δ′ is
often used in triple oxygen isotope community for its many advantages (Miller 2002; Young et
al. 2002). Here (Hulston and Thode 1965)
 17,18 R 
 17,18 O  ln  17,18 sample   1000 ‰ (2)
 Rref 
When oxygen isotopes fractionate in a defined process, the corresponding fractionation

factor is defined as (McCrea 1950)
17,18
RA (3)
17,18
 AB  17,18
RB
where A and B are reactant and product or the transition state of a reaction path and the
reactant, respectively. When A and B reach isotope equilibrium, α is the equilibrium isotope
effect (EIE). When A and B is the transition state and the reactant, respectively, α is the kinetic
isotope effect (KIE) (Bigeleisen and Wolfsberg 1958; Bao et al. 2015). EIE and KIE are two
fundamental parameters of isotope fractionation. When we venture into the high-dimensional
triple oxygen isotope relationship between 17αAB and 18αAB, the community has invented a
designated Greek symbol. This is the θ value or the triple isotope exponent, defined as (Mook
2000; Angert et al. 2003; Barkan and Luz 2005, 2007; Cao and Liu 2011)
ln 17  AB (4)

ln 18  AB
For mass-dependent processes that have fractionation larger than a few per mil, the θ normally
varies between 0.5 and 0.5305 (Bao et al. 2015; Dauphas and Schauble 2016; Hayles et al.
2017). Often, a defined process cannot be an elementary process. In that case, the θ is apparent
or diagnostic for that defined process or processes. The θ value only exists when a process is
specified, but any oxygen-bearing compound can have its δ17O and δ18O values, and thus, its
small 17O deviation, i.e., the Δ′17O value. The Δ′17O is calculated once a reference slope C is
given (Angert et al. 2003; Pack and Herwartz 2014),
466 Cao & Bao
 17 O   17 O  C   18 O (5)
We recommend a C value of 0.5305 mainly because this value is the triple oxygen isotope
exponent at high-temperature limit for all equilibrium processes (Cao and Liu 2011; Pack and
Herwartz 2014). Detailed arguments can be found in Bao et al. (2016).
SULFOXYANIONS–WATER OXYGEN ISOTOPE EXCHANGE

Sulfur has multiple sulfoxyanion species. Among them, sulfate (SO42−) is the final stable
form during the oxidation of sulfur-bearing compounds while sulfite (SO32−) is arguably the most
important intermediate with thiosulfate (S2O32−) being somewhat important during pyrite oxidation.
The kinetic and equilibrium oxygen isotope exchange between the three sulfoxyanions and water
are crucial to interpreting oxygen isotope compositions in sulfate. They will be briefly reviewed here.
Sulfate–water system
Sulfate is a non-labile oxyanion. Experimental results indicate its oxygen isotope
composition remains unchanged for 109 years at most Earth surface conditions (Zak et al. 1980;
Chiba and Sakai 1985; Rennie and Turchyn 2014). The preservation of large positive and negative
non-mass-dependent 17O anomalies from ~30 Ma (Bao et al. 2010), 635 Ma (Bao et al. 2008,
2009), and from the mid- and early Proterozoic samples (Crockford et al. 2018; Hodgskiss et al.
2019) attests to sulfate oxygen’s endurance. When microbial sulfate reduction occurs, however,
the sulfate in solution can exchange oxygen isotopes with water via intermediate sulfite due to
reversibility of enzymatic reactions (see section Microbial Sulfate Reduction).
Sulfate δ18O has been observed to be 14.8‰ to 28‰ higher than that of ambient water during
MSR (Zeebe 2010; Brunner et al. 2012; Antler et al. 2017; Bertran et al. 2020). The variation
may reflect different degrees of reversibility. At 0 °C to 150 °C, the equilibrium oxygen isotope
fractionation between sulfate and water is predicted computationally to follow (Zeebe 2010)
1000  ln 18 SO
EQ
2  2.68  106  T 2  7.45 (6)
4 H2 O
where T is the temperature in Kelvin. At 25 °C, 1000 ln 

EQ
SO24  H 2 O is at ~ 23‰.
Currently, there is no experimental or theoretical calibration of the equilibrium triple
oxygen isotope exponent for sulfate–water system, i.e. SO EQ
2
4 H2 O
. In this volume, equilibrium
θ values for a series of sulfate minerals and water, i.e. SM EQ
 H 2 O were estimated theoretically
,
to range from 0.5242 to 0.5246 (Schauble and Young 2021, this volume). Because CO EQ
2
3 H2 O
is
determined to be slightly smaller than calcite H2 O (Guo and Zhou 2019), SO24 H2O is expected to
EQ EQ
be smaller than SM

EQ
 H 2 O , and assumed to be 0.524 here. This value will be applied to construct
endmember sulfates in Δ′17O–δ18O space.
Sulfite–water system
Here we use sulfite to represent all the dissolved S(IV) species, including dissolved sulfur
dioxide (SO2(aq)), bisulfite (SHO3−), sulfite (SO32−), and pyrosulfite (S2O52−) (Horner and Connick
2003). Bisulfite has two isomers, HSO3− where the hydrogen is bonding to sulfur and SO3H−
where the hydrogen is bonding to oxygen. Oxygen isotope exchange between sulfite and water
occurs via three proposed chemical reactions (Betts and Voss 1970; Horner and Connick 2003).
k7 a

SO3H   H   
 SO2  H 2 O (7a)
k 7 a
SO H    SO2    H 2 O   SO3

* k7 b * *
3


 SHO3   2
(7b)
k 7 b
SO H   S2  O2  O3   H 2O 
(7c)
* k7 c * *
3


 SHO3   2
k 7 c
where ‘*’ denotes oxygen from the SO3H− species prior to exchange. When bisulfite is the
dominant species, the exchange rate is determined by (Horner and Connick 2003)
rex 
1
3
 
k7 a H    k7 bc SHO3   (8)
where at 25 °C k7a and k7bc (k7bc = k7b + k7c) are 1.4 × 108 M−1s−1 and 8.0 × 103 M−1s−1, respectively
(Horner and Connick 2003). When sulfite is the dominant species, the exchange rate is (Horner
and Connick 2003)
2
1 SO32     H   Q4

rex  k7 a  k7 bc   (9)
3 Q2  Q2 1  Q4
 
where
SO32    H   SO3H  
Q2   Q4 
SHO3   HSO3 
,
The values of Q2 and Q4 are 10−6.34 and 4.9, respectively (Horner and Connick 2003).
Considering internal consistency, we adopt the value 4.9 instead of the newly determined
result 2.7 (Eldridge et al. 2018) for Q4 because k7a and k7bc were initially determined using Q4
of 4.9. Using the values of k7a, k7bc, Q2, and Q4 given above and 0.75 for the activity coefficient
of hydrogen ion [H + ] at pH 8.9 and [SO32−] at 0.3 M, we estimated the half-life of sulfite–
water oxygen exchange to be 75 s, which is consistent with the experimentally obtained 78 s
at the given chemical condition (Betts and Voss 1970). At pH 9.8, the exchange half-life is
estimated to be 79 min, being consistent with the 82 min determined by Betts and Voss (1970)
but is inconsistent with the 24.3 min determined by Wankel et al. (2014). The cause for the
discrepancy is unclear but may have to do with the pH buffer glycine used by Wankel et al.
(2014) because glycine, a potentially general acid catalyst (Horner and Connick 2003), could
have catalyzed the exchange reactions. At pH of 7, the half-life is estimated to be less than
1s for dilute solutions (e.g. 0.1 mM). In addition, exchange rate is dependent on [SO32−], as
shown in Equations (8) and (9). These analyses demonstrate that oxygen isotope exchange
between sulfite and water is rapid, especially at low pH conditions.
Although the equilibrium isotope fractionation between individual sulfite species and
water varies only with temperature, the fractionation between total dissolved S(IV) and water,
i.e. 18 SO
EQ
2
3 H2 O
, is pH dependent because sulfite species partition is pH dependent (Müller et
al. 2013b). The value of 1000 ln 18 SO EQ
2
3 H2 O
at 23 °C was determined to be 11.5‰ and 7.9‰
at pH 7.2 and 8, respectively (Brunner et al. 2006). The pH and temperature dependence of
1000 ln 18 SO EQ
2
3 H2 O
was also observed subsequently and determined to be (Wankel et al. 2014),
 13.61  0.299  pH  0.081  t  (10)

1000  ln 18 SO
EQ
2
3 H2 O

1000  ln   1
 1000 
where t is temperature in Celsius in the range of 2 °C to 95 °C and pH in the range of 4.5
to 9.8. According to Equation (10), the 1000 ln 18SO EQ
2
3 H2 O
at 23 °C should be 9.5‰ and
9.3‰ at pH 7.2 and 8, respectively. This is different from an experimentally determined
1000 ln 18 SO
EQ
2
3 H2 O
value of 15.2‰ at 22 °C, a value displaying no pH dependence in the range
of 6.3 to 9.7 (Müller et al. 2013b). The difference between the three experimental studies is
468 Cao & Bao
as large as 7.3‰. As of now, the value of 1000 ln 18 SO

EQ
2
3 H2 O
lies between 7.9‰ and 15.2‰ at
22 °C and further calibration effort is warranted. This issue will be revisited later.
Triple oxygen isotope exponent for equilibrium sulfite–water exchange, i.e., SO EQ
2
3 H2 O
, has
not been calibrated. However, due to the overall S–O bonding similarity between SO42− and
SO32−, we assume that SOEQ
2
3 H2 O
≈ SO
EQ
2
4 H2 O
and thus the value 0.524 is adopted for SO
EQ
2
3 H2 O
to
construct endmember sulfates in Δ′17O–δ18O space.
Thiosulfate–water system
Although thiosulfate (SSO32−) has a chemical structure similar to that of sulfate,
experimental results show that thiosulfate readily exchange oxygen isotopes with water (Pryor
and Tonellat 1967; Betts and Libich 1971). At pH > 5, oxygen exchange between thiosulfate
and water proceeds mainly through thiosulfate–sulfite–water exchange pathway in which
sulfite acts as a catalyst (Betts and Libich 1971). Two chemical reactions are responsible for
the oxygen exchange (Betts and Libich 1971)

SO32   S * SO32  
k

11 a
 2 * 2
 SSO3  SO3 (11a)
k
11 a

SO32   H 2 * O 
k

11 b
 * 2
 S O3  H 2 O (11b)
k11 b
where ‘*’ denotes the sulfur and oxygen prior to the change in thiosulfate and water,
respectively. Reaction (11b) refers to an overall oxygen exchange reaction represented by
chemical reactions (7a–7c). When pH is in the range of 5 to 10, the oxygen exchange rate
between thiosulfate and sulfite is far slower than the one between sulfite and water, hence the
overall exchange rate between thiosulfate and water can be approximated by the slower step,
i.e., the exchange rate between thiosulfate and sulfite (Betts and Libich 1971)
rex  3k11a SSO32   SO32   (12)
and at 25 °C k11a = 2.07 × 10−4 M−1s−1. When pH goes up to the range of 10 to 11, the two
rates are comparable, therefore, the overall oxygen exchange rate is determined by both
thiosulfate–sulfite and sulfite–water exchange rates. At even higher pH, e.g., above 11, the rate
of thiosulfate–sulfite exchange is faster than that of sulfite–water exchange, thus, the overall
exchange rate can be estimated by Equation (9). In acidic condition (pH < 4), thiosulfate is
unstable (Xu and Schoonen 1995).
No experimental or theoretical thiosulfate–water equilibrium α value is available in
literature at this time. According to Betts and Libich’s (1971) experiments, thiosulfate may
be enriched in heavy oxygen isotopes relative to sulfite, and the 18O enrichment is about 1‰
at pH of 10.8 and t of 50.3 °C. At lower temperature, the enrichment is expected to be larger.
Similarly, no θ calibration has been conducted for thiosulfate–water equilibrium. For now, we
will adopt the value 0.524 to construct endmember sulfates in Δ′17O–δ18O space.
MICROBIAL SULFATE REDUCTION (MSR)

Early on, researchers found that sulfate exchanges its oxygen isotopes with water through
intermediates during the reversible microbial sulfate reduction (MSR) processes (Mizutani
and Rafter 1973; Fritz et al. 1989) (Fig. 2). Several models have been proposed to calculate
sulfate δ18OSO4 during MSR (Brunner et al. 2005, 2012; Turchyn et al. 2010; Antler et al.
2013, 2017; Wankel et al. 2014; Bertran et al. 2020). Generally, the δ18OSO4 is a function of the
initial δ18OSO4, δ18OSO3, δ18OAMP (AMP: adenosine monophosphate), δ18OH2O, the rate and the
reversibility of the involved individual steps between sulfate and sulfite (Brunner et al. 2012;
Bertran et al. 2020) (Fig. 2). In the meantime, the δ18OSO3 and δ18OAMP themselves vary with
oxygen exchange rates between sulfite, AMP, and water, as well as their respective oxidation
rates to APS (Adenosine−5′-phosphosulfate). All these parameters are case specific and have
so far been under-constrained. Therefore, we are not attempting to explore all the possibilities
here. Instead, only the endmember case when the MSR processes approaching thermodynamic
equilibrium caused by the high level of MSR reversibility (Zeebe 2010) is considered. For this
case, sulfate oxygen isotope composition is calculated by,
17,18
RSO24  17,18 SO
EQ
2
17,18
RH2O (13)
4 H2 O
and 17,18αSO4-H2OEQ estimated by Equation (6), and the θ value is at 0.524.
Figure 2. Sketch of a model for microbial sulfate reduction. The grey dashed circle marks the cell mem-
brane; oxygen exchange between sulfate and water is achieved by sulfite and AMP exchange with ambient
water and oxidized back to sulfate internally; APS and AMP are adenosine-5′-phosphosulfate and adenos-
ine monophosphate, respectively.
SULFIDE OXIDATION MECHANISMS

The oxidation of mineral sulfides to sulfate on the Earth surface occurs mostly in aqueous
conditions. The process involves multiple reaction steps and multiple sulfur intermediates.
Oxygen from dissolved O2 and water can both be incorporated into sulfate. Atmospheric O2
(23.88‰ in δ18O and −0.553‰ in Δ′17O according to Barkan and Luz 2011) and water have
very different δ18O and Δ′17O. In addition to this source difference, the relative proportion of
these two sources depends on the oxidation pathway, which differs in equilibrium and kinetic
isotope effects. All these factors should be considered in interpreting sulfate δ18O value as
summarized in van Stempvoort and Krouse (1993) and must be considered in interpreting
sulfate’s small Δ′17O as well. There are uncalibrated parameters for δ18O and even more
uncalibrated ones for triple oxygen isotope behaviors during sulfide oxidation processes. In this
section, we begin to unravel factors that matter to sulfate triple oxygen isotope compositions
by examining endmember cases.
Thiosulfate oxidation on pyrite surface
Pyrite oxidation has been extensively studied due to its detrimental environmental impact
and connection to the mineral resource. Electrochemical oxidation is the widely accepted
mechanism (Luther 1987; Moses et al. 1987; Moses and Herman 1991; Rimstidt and Vaughan
2003). Previous studies show that factors, such as Eh, pH, oxidant type and concentration, grain
size, can affect the oxidation process. Readers should refer to (Rosso and Vaughan 2006; Chandra
and Gerson 2010) for comprehensive reviews on these topics. Here we focus on the oxygen
isotope compositions in sulfate formed via pyrite oxidation in responding to these factors.
Pyrite is a semiconductor, and its oxidation and reduction happen at different sites on
the pyrite surface. According to the proposed electrochemical model (Rimstidt and Vaughan
2003), three distinct reaction steps are involved. First, O2 or Fe3 + acquires electrons from an
Fe2 + site, i.e., the cathodic site, and Fe2 + is oxidized to Fe3 + . Second, electrons are transferred
from the sulfur site, i.e., the anodic site, to the Fe3 + at the cathodic site. The third step is
470 Cao & Bao
water molecule reacting with sulfur to form sulfoxyanion at the anodic site. These reactions
are repeated until thiosulfate is formed. At low pH conditions, thiosulfate is further oxidized
to sulfate on the surface (Rimstidt and Vaughan 2003). All the oxidation steps on the sulfur
(anodic) sites involve only water, no O2 bonds directly with S. We can, therefore, conclude that
oxygens in sulfate formed by thiosulfate oxidation on pyrite surface at low pH conditions all
come from water. Sulfate of this origin serves as an endmember case. Further elaboration of
this endmember is given below.
(1) Pyrite oxidation via Fe3 + happens via thiosulfate at surface (not in solution) at
low pH. Observed apparent sulfate–water oxygen isotope fractionation, Δδ18OSO4–H2O, during
pyrite oxidation by Fe3 + is in the range of 2.3‰ to 2.9‰ (Balci et al. 2007; Mazumdar et
al. 2008; Heidel and Tichomirowa 2011). However, observed apparent Δδ18OSO4–H2O during
sulfite oxidation by Fe3 + in solution is ~ 5.9‰ (Müller et al. 2013a) or 8.2‰ (Balci et al. 2012
and section Sulfite oxidation by Fe3 + in solution). A likely cause of this discrepancy is that
thiosulfate on pyrite surface does not enter solution but is oxidized by Fe3 + on pyrite surface
straight to sulfate at acidic conditions. This conclusion is also supported by the experiment
that showed sulfate was the first dominant sulfur species detected in solution (Borilova et al.
2018). Indeed, thiosulfate is unstable in solution and will be converted to sulfite at this low pH
conditions (see the next section for details).
One may argue that this discrepancy is caused by sulfite being out of oxygen isotope
equilibrium with water when sulfite is oxidized to sulfate by Fe3 + in solution. Müller et
al. (2013a) do conclude that sulfite–water was not at oxygen isotope equilibrium for their
sulfite oxidation by Fe3 + experiments. However, if this were the case, the rate of sulfite to
sulfate oxidation by Fe3 + would have to be faster than the rate of sulfite–water oxygen isotope
exchange, which is certainly not the case. The oxidation rate is ~ 3.2 × 10−2 M−1s−1 for their
experiments at pH 1.0 (Müller et al. 2013a). The sulfite–water exchange rate is, however,
~1.4 × 108 M−1s−1, i.e., k7a for reaction (7a), at pH 1, which is ten orders of magnitude faster
than the oxidation rate. Therefore, the observed small apparent disequilibrium in Muller et al.’s
experiment might be caused by other factors. We suspect that the high concentrations of sulfite
and Fe3 + used in the experiment may have slowed down sulfite–water isotope exchange due to
the formation of FeSO3 + and Fe2(OH)SO33 + complexes in solution (Lente and Fábián 2002).
At low sulfite and Fe3 + concentrations as in most natural solutions, however, such complexes
are minimal. At low pH conditions, sulfite is in oxygen isotope equilibrium with water at all
time during Fe3 + oxidation in the solution, especially at low sulfite and Fe3 + concentrations.
(2) Kinetic isotope effects are responsible for the apparently smaller Δδ18OSO4–H2O
during pyrite oxidation by Fe3 + . If indeed sulfate–water oxygen isotope fractionation is
smaller when going through thiosulfate oxidation on pyrite surface than going through sulfite
oxidation in solution, the underlying mechanism must be explored. At this time, no study has
been done on the subject. We can speculate that thiosulfate oxidation on the surface would
have to break the S–S bond in addition to the formation of S–O bond, and this additional bond-
breaking process adds a corresponding kinetic oxygen isotope effect to SO32− in SSO32− on the
surface, which reduces the level of 18O enrichment in final product sulfate.
(3) Thiosulfate on pyrite surface is likely at oxygen isotope equilibrium with water
before being oxidized to sulfate. At low pH, the rate of oxygen isotope exchange between
thiosulfate and water is high (Pryor and Tonellat 1967; Betts and Libich 1971), but the rate of
thiosulfate oxidation to sulfate on pyrite surface is absent, which renders a direct rate comparison
unattainable for now. However, two indirect lines of evidence suggest that the exchange rate
is much higher. First, different research groups have obtained similar apparent Δδ18OSO4–H2O
value for this oxidation process, ranging from 2.3‰ to 2.9‰, at variable overall oxidation rates
(Balci et al. 2007; Mazumdar et al. 2008; Heidel and Tichomirowa 2011). If isotope exchange
between sulfoxyanions and water were relatively slow, the sulfoxyanions-water system would be
out of equilibrium at different degrees, and the Δδ18OSO4–H2O would be highly variable. Second,
the δ18O of product sulfate via thiosulfate oxidation on the pyrite surface can be calculated by
3 1  (14a)
 RSO24  18 KIESSO32 SO24 18 SSO32 H2O  18 KIEH2OSO24  18 RH2O
18 py
4 4 
whereas the δ18O of product sulfate via sulfite oxidation by Fe3 + in the solution is given by
3 1  (14b)
 RSO24  18 KIESO32 / Fe3 SO24 18 SO
18 aq EQ
2
3 H2 O
 18 KIEH2OSO24  18 RH2O
4 4 
If 18KIEH2OSO4 is assumed to be the same for these two alternative processes, the value
of 18KIESSO3SO4 × 18 SSO EQ
2 is roughly estimated to be ~0.994 times of that of
3 H2 O
18
KIESO3/Fe3 + SO4 × SO32 H2O in order to fit the observations. Since 18KIESSO3SO4 is smaller
18 EQ
than 18KIESO3/Fe3 + SO4 (see analysis in this subsection (2) above) and 18 SSO EQ
2 and
3 H2 O
18 EQ
2
SO3  H 2 O
can be treated as the same in value (see section Thiosulfate–water system), the
simplest explanation is that the two αEQs in Equations (14a, 14b) are both fully expressed,
i.e., rapid exchange equilibrium with water, and 18KIESSO3SO4 is 0.994 times of the
18
KIESO3/Fe3 + SO4 in this case.
Therefore, the δ18O for sulfate derived from thiosulfate oxidation by Fe3 + on pyrite surface
is determined by
3 1 
 RSO24  17,18 KIESSO32 SO42 17,18 SSO
17,18 surf EQ
2
3 H2 O
 17,18 KIEH2OSO24  17,18 RH2 O (15)
 4 4 
The related kinetic and equilibrium triple isotope effects will be constrained by Monte Carlo
technique later.
Sulfite oxidation by O2 in solution
The electrochemical model of pyrite oxidation (Rimstidt and Vaughan 2003) precludes
any O2 being incorporated into sulfate if all the oxidation steps occur on the pyrite surface.
However, experimental results (Balci et al. 2007; Heidel et al. 2009; Tichomirowa and
Junghans 2009; Heidel and Tichomirowa 2010; Kohl and Bao 2011) and field observations
(Bao et al. 2008; Crockford et al. 2018; Killingsworth et al. 2018; Hodgskiss et al. 2019)
demonstrate to variable degrees that O2 isotope signal is incorporated in sulfate during pyrite
oxidation in aerated solution, which means that thiosulfate can be released from pyrite surface
to solution, even at low pH conditions. Thus, sulfite oxidation by O2 at low pH can serve as
another endmember case of sulfate formation. We elaborate on our argument below.
(1) Sulfite is an important intermediate for sulfur oxidation to sulfate. Thiosulfate
decomposes to S0 and sulfite at low pH conditions or tetrathionate (S4O62−) in the presence of
pyrite (Xu and Schoonen 1995). At pH 6, tetrathionate is observed to be the most abundant
intermediate (Goldhaber 1983). When oxidized to sulfate, tetrathionate is converted to sulfite
first (Druschel et al. 2003). Sulfur chain shortening is suggested to be one possible mechanism
(Druschel et al. 2003), although how exactly the conversion takes place is not clear (Moses
et al. 1987). Regardless of the pathway thiosulfate decomposes, sulfite is the inevitable
intermediate to the final oxidation product sulfate.
(2) Fe3 + loses out to O2 as a competitive oxidant at low pH. While sulfite oxidation has
a higher rate via Fe3 + than via O2 at low pH conditions, e.g., about two orders of magnitude
faster at pH 1 (Müller et al. 2013a), the Fe3 + oxidation pathway cannot be sustained if its
reduction product Fe2 + is not oxidized to Fe3 + by O2 quickly. At low pH conditions, the rate of
Fe2 + oxidation by O2 is sufficiently slow, e.g., 10−7 min−1 at pH 2 and pO2 of 0.2 Atm (Singer
and Stumm 1970), so that sulfite oxidation by O2 should dominate the oxidation process,
especially for experiments with no Fe3 + added initially.
472 Cao & Bao
(3) Sulfite maintains its oxygen isotope equilibrium with water during its oxidation
by O2 at low pH. This is logical because, as shown in the above subsection (1), sulfite–water
oxygen isotope exchange rate is several orders of magnitude higher than the oxidation rate by
Fe3 + , and the oxidation rate by Fe3 + is higher than by O2.
(4) The final product sulfate has 1/4 of its oxygen coming from O2 and 3/4 from
water. It has been proposed that sulfite oxidation by O2 is a radical chain reaction as described
below (Zhang and Millero 1991; Kuo et al. 2006)
M n   SO32   M 
Initiation n 1 
 SO•3
Propagation SO•3  O2  SO3OO• 
SO3OO•   SO32   SO3OO2   SO3• 
Oxidation SO3OO2   SO32   2SO3O2 
Termination SO3OO•   SO•3  S2O62   O2
SO3OO•   SO3OO•   S2 O62   2O2

in which oxygen sourced from O2 is in bold. Because sulfite–water oxygen isotope exchange is
rapid at low pH conditions, sulfite always carries equilibrated water oxygen isotope composition
before being oxidized to sulfate. In the oxidation step, the product sulfate obtains its 1/4 oxygen
from O2. This mechanism is supported by sulfite (Holt et al. 1981; Müller et al. 2013a) and pyrite
oxidation experiments (Tichomirowa and Junghans 2009; Kohl and Bao 2011). O2 does not
exchange oxygen isotopes with sulfite or water in the processes (Krouse et al. 1991).
Therefore, for this endmember case, i.e., sulfite oxidation by O2 in low pH solutions, the
sulfate oxygen isotope composition is determined by,
3 17,18 1 17,18
2  KIESO32 / O2 SO24 17,18 SO RH2O 
17,18 D.O. EQ 17,18
RSO 2
3 H2 O
KIEO2 SO24 17,18 RO2 (16)
4
4 4
technique later.
Sulfite oxidation by Fe3 + in solution
As discussed in the above section, thiosulfate can be released into solution and converted
to sulfite before being oxidized to sulfate at low pH conditions. If this sulfite is oxidized by
Fe3 + rather than O2 in solution, the product sulfate will have all of its oxygens sourced from
water. Here, we place sulfite oxidation by Fe3 + in solution as another endmember case.
In nature, this endmember case may be rare. One likely example is pure ZnS oxidation
by Fe3 + . Pure ZnS is acid-soluble, and its oxidation process was proposed to proceed entirely
in solution (Moses et al. 1987). The Δδ18OSO4–H2O obtained in pure ZnS oxidation via Fe3 +
experiments is 8.2‰ (Balci et al. 2012), which is close to the 5.9‰ obtained from sulfite
oxidation by Fe3 + in solution (Müller et al. 2013a). The 2.3‰ difference could arise from
the different sulfite and Fe3 + concentrations and Fe3 + /sulfite ratios in the two experiments.
The concentration difference may affect the oxidation kinetics (Lente and Fábián 2002) and
consequently, the oxygen isotope fractionations.
As discussed in the section above, sulfite should have maintained its oxygen isotope
equilibrium with water at low pH conditions. Thus, the oxygen isotope composition of sulfate
derived from sulfite oxidation by Fe3 + in solution can be calculated by,
 3 17,18 1 
2   KIESO32 / Fe3 SO24 17,18 SO  17,18 KIEH2OSO24  17,18 RH2 O
17,18 D.F.I. EQ
RSO 2
3 H2 O
(17)
4
 4 4 
technique later.
The role of microbes on oxygen isotope composition of sulfate derived from sulfur
oxidation
Microbial enzymatic processes affect sulfate oxygen isotopes during sulfur oxidation.
Two metabolic mechanisms can be identified (Fig. 3). One catalyzes the Fe2 + oxidation to
Fe3 + with O2 as the ultimate electron acceptor at low pH conditions and the product Fe3 + in
turn oxidizes sulfur abiotically (Fig. 3a) (Singer and Stumm 1970; Balci et al. 2007; Brunner
et al. 2008). The other catalyzes direct sulfur oxidation in which both Fe3 + and O2 can act as
oxidants (Fig. 3b) (McCready and Krouse 1982; Sand et al. 1995; Balci et al. 2012). Note that
these two mechanisms often work jointly (Schippers et al. 1996; Brunner et al. 2008; Vera et al.
2013), and no clear boundary exists for this division. In fact, even for enzyme-mediated direct
sulfur oxidation, there are different enzymatic pathways involved (Fig. 4), including sulfite
oxidation through sulfite dehydrogenase (Feng et al. 2007) or APS (Kelly 2003), tetrathionate
hydrolysis (Ghosh and Dam 2009), and Sox reactions (Friedrich et al. 2001). Therefore,
multiple oxidation pathways compete when microbes are participating in sulfur oxidation
processes. Oxygen isotope exchange between intermediates and water, substrates used by
enzymes, and the reversibility of enzymatic reactions could all influence the oxygen isotope
composition of the final sulfate. Experiments targeted at individual variables are scarce at this
time. As a result, these multiple, interacting factors make a quantitative prediction difficult
on the oxygen isotopes of sulfate derived from microbial oxidation. However, some general
conclusions on the role of microbes in sulfate oxygen isotopes during oxidation can be drawn.
Balci et al. (2007) found that their microbial anaerobic (Fe3 + as the oxidant) and aerobic (O2
as the oxidant) long-term pyrite oxidation experiments resulted in similar Δδ18OSO4–H2O. Their
interpretation is that Fe3 + is the direct oxidant in their experiments, even when O2 is the ultimate
electron acceptor in the aerobic experiments. Their later ZnS and S oxidation experiments (Balci
et al. 2012) support the interpretation. If correct, microbes appear to catalyze Fe2 + oxidation by
O2, which diminishes the fraction of O2 signature being incorporated in product sulfate (Fig. 3a).
When sulfur is oxidized directly via microbial enzymatic reactions and sulfite–water
oxygen isotope equilibrium is maintained, sulfite dehydrogenase pathway may have a similar
oxygen isotope effect as does the sulfite oxidation by Fe3 + (Fig. 4), in which the Δδ18OSO4–H2O
is ~ 8‰. Sulfite oxidation through APS (Fig. 4) is expected to be similar to microbial sulfate
reduction (Fig. 2), which can result in a Δδ18OSO4–H2O value at 14.8‰~28‰ depending
on reaction reversibility. However, this pathway may not be highly expressed during sulfur
oxidation (Klatt and Polerecky 2015). If tetrathionate can reach oxygen isotope equilibrium
with water, tetrathionate hydrolysis (Fig. 4) will have a similar isotope effect, as does thiosulfate
oxidation on pyrite surface i.e., about 2.6‰. However, existing experimental results indicate
that the rate of oxygen isotope exchange between tetrathionate and water is slow compared
with the rate of tetrathionate hydrolysis (Balci et al. 2017). Sox pathway has variable substrates,
e.g., SO32−, S2−, and S2O32− (Fig. 4). When sulfite is the substrate, the corresponding isotope
effect is expected to be similar to that for thiosulfate oxidation on the pyrite surface, i.e., about
2.6‰. When S2− is the substrate, four oxygen addition steps are required to produce sulfate.
474 Cao & Bao
Figure 3 Two endmember metabolic mechanisms (a and b), yet often working jointly, affect oxygen iso-
tope composition of sulfate derived from sulfur oxidation. The grey dashed circles mark the cell mem-
brane. Black arrows indicate the reaction path from reactants to products.
Figure 4. Enzyme-catalyzed sulfur oxidation pathways. The substrates can be SO32− (Kelly 2003; Feng
et al. 2007), S4O62− (Ghosh and Dam 2009), S2− (Friedrich et al. 2001), or S2O32− (Friedrich et al. 2001),
respectively; these substrates may or may not readily exchange oxygen with water during oxidation; the
added oxygens to these substrates are exclusively from water. Chemical compounds marked with bold
black are sulfur and oxygen sources in product sulfate.
If these steps are irreversible and oxygen isotope exchange rates between the intermediates
and water are relatively slow, a negative Δδ18OSO4–H2O value is expected due to kinetic isotope
effects. Negative Δδ18OSO4–H2O observed in elemental sulfur oxidation by A. thiooxidans (Smith
et al. 2012; Balci et al. 2017) may indicate the operation of a Sox pathway. When SSO32− is the
substrate, the Δδ18OSO4–H2O is expected to have a value in between the ones where SO32− and
S2− are substrates, respectively, since the oxidation of SSO32− can be regarded as the oxidation
of SO32− and S2− continuously via Sox pathway (Fig. 4). However, we should note that these
rough estimations are only for specific oxidation pathway instead of for specific microbial
community because one community may operate multiple oxidation pathways at the same time
(Bobadilla Fazzini et al. 2013). Because most of these microbial mediated sulfur oxidation
steps are physically separated from O2 reduction steps by specific enzymes (Kelly 1982; Balci
et al. 2017), the chance for O2 to be directly bonded to S and eventually to product sulfate is low.
In summary, the overall impact of microbial involvement in sulfate production via

sulfur oxidation is a reduction of the proportion of O2 isotope signal that could otherwise be
incorporated in sulfate if oxidation occurs abiotically.
Comments on laboratory experiments
While the results on oxygen isotope effects from the published sulfur oxidation
experiments generally converge, discrepancies and uncertainties exist. Knowing these caveats
help to scrutinize data for our prediction for sulfate endmembers in Δ′17O–δ18O space later.
Most of abiotic, aerated pyrite oxidation experiments show that less than 15% of the
oxygen in product sulfate is sourced from O2 (Taylor et al. 1984a; Balci et al. 2007; Heidel et
al. 2009; Tichomirowa and Junghans 2009; Heidel and Tichomirowa 2010), while Kohl and
Bao (2011)’s experiments show that more than 20% oxygen in sulfate is derived from O2 even
with Fe3 + added. We infer that strong hydrodynamics brought about by constant shaking in
Kohl and Bao’s experiments may have enhanced the release of thiosulfate into solutions, which
increases O2 incorporation in product sulfate. Also, Tichomirowa and Junghans (2009)’s and
Heidel et al. (2009)’s experiments indicate that smaller pyrite grain size (e.g., smaller than
63 μm) increases the release of thiosulfate and therefore also the O2 fraction in sulfate.
Pyrite oxidation experiments show that the duration of aerobic experiments
(Tichomirowa and Junghans 2009) and Fe3 + /surface ratio of anaerobic experiments (Heidel
and Tichomirowa 2011) affect sulfate oxygen isotope composition. Both factors point to the
importance of thoroughly washing clean pyrite grains before the experiment. Sulfate initially
presented on sulfide mineral surface is often difficult to eliminate, and a long experimental
duration may be required to overwhelm this “background” sulfate. A higher fraction of O2
signal in sulfate collected at the beginning of an experiment (Balci et al. 2007; Heidel et al.
2009; Tichomirowa and Junghans 2009; Heidel and Tichomirowa 2010; Ziegler et al. 2010)
may be explained by this high “background”.
Most laboratory experiments on pyrite oxidation were conducted at pH ~ 2 (Taylor et al.
1984a; Balci et al. 2007; Mazumdar et al. 2008; Tichomirowa and Junghans 2009; Heidel and
Tichomirowa 2011). In Kohl and Bao (2011) some of the pyrite oxidation experiments were
carried out in solutions buffered at higher pH conditions in which negative Δδ18OSO4–H2O were
observed, indicating that sulfite, the assumed intermediate, probably did not reach oxygen isotope
equilibrium with the water. We calculated, based on Equation (9), that sulfite–water oxygen
isotope exchange rate is at ~9 × 10−5, 9 × 10−7, and 9 × 10−9 s−1 at pH 9, 10, and 11, respectively.
These rates are slower than the oxidation rates at pH 9 and 10 in the initial 10−20 weeks, and also
slower than that at pH 11 at all times, according to Kohl and Bao (2011). Therefore, sulfate–water
oxygen isotope fractionation at acidic conditions does not apply to cases at alkaline conditions.
In addition, the pH buffered solution may affect the pyrite oxidation process. For example, the
fraction of O2 signal in sulfate decreased at pH 10 and 11 when compared to the acidic conditions
even without the addition of Fe3 + ion. A possible explanation is that carbonate ion in the buffered
solution accelerates the oxidation of Fe2 + to Fe3 + when pyrite is oxidized in aerated solution
(Caldeira et al. 2010) which enhances iron’s reactivity. Likewise, the fraction of the O2 signal
in sulfate is higher at pH 7 even with Fe3 + addition than at other pH conditions. A possible
explanation is that phosphate in the buffered solution was precipitated with Fe2 + and Fe3 + as
solids (Nriagu 1972; Singer 1972), resulting in a reduced iron’s reactivity in solution.
The concentration of sulfite, the most important intermediate during sulfur oxidation,
as well as sulfite/oxidant ratio, can affect the rate of oxidation (Lente and Fábián 2002),
and therefore the Δδ18OSO4–H2O. This can explain why different oxygen isotope effects were
observed in the similar sulfite oxidation experiments by O2 (Oba and Poulson 2009; Müller et
al. 2013a) or among experiments of sulfite, ZnS, and S oxidation by Fe3 + (see section Sulfite
oxidation by Fe3 + in solution).
476 Cao & Bao
It was observed that ~40% of the oxygen in sulfate derived from copper-catalyzed
aerated oxidation of sulfite is sourced from O2 (Holt et al. 1981). We think this explains the
43% O2-sourced oxygen in sulfate derived from the chalcopyrite abiotic oxidation experiments
conducted by Thurston et al. (2010). It also explains the 8% O2-sourced oxygen in sulfate
derived from pyrite and pure ZnS microbial oxidation experiments conducted by Balci et
al. (2007, 2012). In Balci et al.’s experiments, CuCl2 was added in their microbial culture
medium. Further examination of the role of copper ion is warranted.
Constraining intrinsic equilibrium and kinetic oxygen isotope effects during sulfide
oxidation
We need the EIEs and KIEs to predict the Δ′17O–δ18O space for endmember sulfates
quantitatively. These intrinsic isotope effects can be obtained by examining the published
experimental data. The knowledge synthesized in the previous sections allows us to avoid
potentially messy and misleading data while focusing on the Δδ18OSO4–H2O values obtained from
abiotic experiments that were conducted in acidic conditions with a sufficiently long duration.
Let us begin by considering the KIE and EIE for sulfite oxidation by O2 in solution.
Here we use Monte Carlo technique, an approach we have applied to analyze methane
KIEs (Cao et al. 2019), to constrain the values of KIESO3/O2SO4, SO EQ
2 , and KIEO2SO4 in
3 H2 O
Equation (16). Results from laboratory-controlled aerobic pyrite oxidation experiments at

pH 2 (Tichomirowa and Junghans 2009; Kohl and Bao 2011), in which about 25% oxygen
in sulfate was found to come from O2, were used as input constraints. We used 24.2‰ as
the δ18O for dissolved O2 (Kroopnick and Craig 1972; Reuer et al. 2007; Li et al. 2019). We
obtained values of KIESO3/O2SO4, SO
EQ
2 , and KIEO2SO4 in Equation (16) at 0.9916 ± 0.0003,
3 H2 O
1.0129 ± 0.0001, and 0.9850 ± 0.0010, respectively (Table 1). The determined KIE for sulfite is
close to the experimentally determined 0.9903 by Muller et al (2013a). The SO EQ
2
3 H2 O
of 1.0129
is within the range of 7.9‰ and 15.2‰ fractionation determined experimentally (Brunner et
al. 2006; Müller et al. 2013b; Wankel et al. 2014). The determined KIEO2SO4 of 0.9850 is,
however, rather different from 0.997 (Oba and Poulson 2009) and 1.007 (Müller et al. 2013a)
determined experimentally through sulfite oxidation by O2. It should be noted that the Monte
Carlo method determines possible EIE and KIE solutions and their probability. The isotope
effects presented above are the most likely solutions under given constraints. The results can
be changed when some of the parameters are constrained to be different in the future.
Experimental results from the abiotic ZnS oxidation by Fe3 + (Balci et al. 2012) are chosen to
represent an endmember case of sulfite oxidation by Fe3 + in solution. Instead of linear regression
used by Balci et al. (2012), the oxygen isotope composition in water and sulfate, as depicted by
Equation (17), is used here directly to calculate the apparent αSO4–H2O, which is determined to
be 1.0079 ± 0.0009. If SO
EQ
2
3 H2 O
is assumed to be 1.0129, KIESO3/Fe3+SO4 and KIEH2OSO4 are
determined to be 0.9984 ± 0.0008 and 0.9976 ± 0.0014, respectively by Monte Carlo method
(Table 1). Again, these values are the most probable solutions for the observed sulfate–water
oxygen isotope fractionation, and further constraints can revise and improve these values.
Abiotic anaerobic oxidation of pyrite conducted by Balci et al. (2007), Mazuzumdar et al.
(2008), and Heidel and Tichomirowa (2011) are chosen here to represent the endmember case
of thiosulfate oxidation by Fe3 + on pyrite surface. Sulfate–water oxygen isotope fractionation is
calculated to be 1.0027 ± 0.0007 using Equation (15). If SSOEQ
2
3 H2 O
and KIEH2OSO4 are set to be
1.0129 and 0.9976, respectively, KIESSO3SO4 is estimated to be 0.9916 ± 0.0005 (Table 1).
The θ values for the many important KIEs during sulfide oxidation are underconstrained
for now. If reduced masses, i.e., imaginary frequency term in transition state (Young et al.
2002; Bao et al. 2015), are used to calculate the values of θ for these KIEs, the range is 0.5052–
0.5168 (Table 1). Lower θKIE values are associated with thiosulfate or sulfite, while higher
values are obtained for H2O and O2. However, the exact θKIE values cannot be determined at
this time. An arbitrary value of 0.5110, i.e., the mean of the maximum (0.5168) and minimum
(0.5052) θKIE in Table 1, will be used for discussion hereafter.
Table 1. Constrained intrinsic or diagnostic KIE, EIE, and θ values for sulfide oxidation processes.
The 18SO
EQ
2
4 H2 O
is calculated at 25 °C based on Equation (6); 18SO
EQ
2 and KIEs are determined
3 H2 O
by Monte Carlo method. 18SSO32 H2 O is assumed to equal to 18SO

EQ EQ
2 ; θRM refers to θ value
3 H2 O
determined using reduced masses of associated reactants; The θ values for 18 SO EQ
2 , 18 SO
EQ
2 ,
4 H2 O 3 H2 O
and 18 SSO32 H2 O are set at 0.524 based on the value of SM H2 O (see section Sulfoxyanions–water
EQ EQ
oxygen isotope exchange); The θ value for all KIEs is set to 0.511 which is equal to the mean value
of maximum and minimum θRM.
Values of EIE/KIE θRM valuese θ values
18
SO
EQ
2
1.023 -- 0.524g
4 H2 O
18 a 1.0129 ± 0.0001 -- 0.524

SO
EQ
2
3 H2 O
18
 EQ
SSO32   H 2 O
1.0129 ± 0.0001 -- 0.524
KIEO2SO4b 0.9850 ± 0.0010 0.5109 0.511

KIESO3/O2SO4c 0.9916 ± 0.0003 0.5064 0.511
d f
KIEH2OSO4 0.9976 ± 0.0014 0.5168/0.5161 0.511
KIESO3/Fe3 + SO4d 0.9984 ± 0.0008 0.5067 0.511
KIESSO3SO4 0.9916 ± 0.0005 0.5052 0.511
a: SO H O was determined to be about 1.008 (Brunner et al. 2006), 1.009 (Wankel et al. 2014), and 1.015 (Müller et al. 2013b), respectively;
18 EQ
2
3 2
b: KIEO2SO4 was determined to be ~ 0.997 (Oba and Poulson 2009) and 1.007~1.023 (Müller et al. 2013a), respectively; c: KIESO3/O2SO4 was
determined to be ~ 0.990~0.995 (Müller et al. 2013a); d: 3/4 × KIESO3/Fe3 + SO4 + 1/4 × KIEH2OSO4 was determined to be ~ 0.994 (Müller et al. 2013a);
e: θRM = ln(17μ/16μ)/ln(18μ/16μ), where μ is the reduced mass of associated reactants, e.g. O2 and SO32−; f: 0.5168 and 0.5161 are for sulfite and
thiosulfate oxidation processes, respectively; g: the value of 0.524 was taken from the θ values for sulfate minerals-water determined by Schauble
and Young (2021, this volume).
APPLICATIONS
As shown in Equations (13) and (15–17), given the triple oxygen isotope compositions of
water and O2 and the constrained KIEs, EIEs, and θs, we can predict the values of δ18O and
small Δ′17O for endmember sulfate identified above. The predicted results can be compared
with field observations. Because the triple oxygen isotope compositions of water and O2
of the geological past are not well constrained, we will only explore the small Δ′17OSO4 in
modern sulfate. Specifically, modern sulfate from Ace lake in Antarctica and Mississippi and
Marsyangdi river basins will be analyzed.
Predicted sulfate δ18O and small Δ′17O
Let us first set the triple oxygen isotope compositions of water and dissolved O2 that are
required to make the prediction. The meteoric water is assumed to be on the Global Meteoric
Water Line (GMWL) (Luz and Barkan 2010),
 ’17 O  0.528   ’18 O  0.000033 (18)

and we will use the range of −20‰ to 0‰ for δ OH2O with a corresponding δ OH2O calculated
18 17
using Equation (18). For dissolved O2, we use fixed δ18O and Δ′17O at 24.2‰ and −0.554‰,
respectively (Kroopnick and Craig 1972; Reuer et al. 2007; Barkan and Luz 2011; Li et al. 2019).
Using the constrained endmember triple oxygen isotope parameters in Table 1 and Equations
(13, 15−17), we have constructed sulfate Δ′17O–δ18O space for four different endmember cases
given the range of δ18OH2O (Fig. 5). The results show that sulfate Δ′17O is apparently negatively
correlated with its δ18O. Sulfate derived from surface oxidation has the most positive Δ′17O
values. Its Δ′17O can be even more positive than that of its ambient water. As graphically
illustrated in Figure 6a, the combination of EIE and KIEs during sulfide oxidation can produce
478 Cao & Bao
a diagnostic θSO4–H2O value higher than 0.5305. Comparing with Figure 6c, Figure 6a shows
that a smaller KIESSO3SO4 (i.e., ln (KIESSO3SO4) being more negative than ln (KIESO3/Fe3 + SO4),
see Table 1) associated with thiosulfate oxidation to sulfate on pyrite surface is the key to
achieve a diagnostic θSO4–H2O value larger than 0.5305. Interestingly, the Δ′17O of sulfate derived
from solution oxidation by dissolved O2 is only slightly more negative than that of the water,
even though 25% oxygen of this sulfate is derived from O2, which carries a Δ′17O of −0.554‰
to begin with. Figure 6b shows that the negative Δ′17O signal in O2 is largely erased by its
KIE when O2 is incorporated into the sulfate. The Δ′17O value of sulfate derived from solution
oxidation by Fe3 + (Fig. 6c) is even more negative than that derived from the O2 oxidation
pathway (Fig. 6b). The most negative Δ′17O value is possessed by the equilibrated sulfate (Figs.
5 and 6d), attributed largely by its more positive δ18O value riding on a θEQ value smaller than
the reference 0.5305. All these results highlight the importance of knowing the intrinsic isotope
effects during sulfate formation when a small sulfate Δ′17O is of interest, and isotopologue
specific kinetic analysis is critical to correctly interpreting small Δ′17O.
The distinct curves in Figure 5 demonstrate that sulfate δ18O and Δ′17O can be used to
differentiate sulfate of different oxidation origins. The distribution of curves also reveals that
sulfates derived from the two solution oxidation pathways, by O2 and by Fe3 + , are difficult to
tell apart. Since the Fe3 + path is only dominant in very acidic conditions such as acid mine
drainages because Fe3 + precipitates at higher pH (Stefánsson 2007), its geological significance
is, therefore, limited. However, we should note that the results presented in Figure 5 are obtained
according to our best estimations for the intrinsic or diagnostic EIEs and KIEs presented in
Table 1. When the values of these EIEs and KIEs are further constrained to be different sulfate
δ18O and Δ′17O should be changed accordingly.
Figure 5. Triple oxygen isotope compositions of sulfate derived from different mechanisms in a given water
body. The value of δ18OH2O is set to range from −20‰ to 0‰ with the corresponding δ17OH2O being estimated
by Equation (18); “EQ. sulfate” refers to sulfate at oxygen isotope equilibrium with water; “sulfate derived
from surface oxidation” refers to the endmember from thiosulfate oxidation on pyrite surface; “sulfate derived
from solution oxidation by O2” refers to the endmember from sulfite oxidation by O2 in solution; “sulfate
derived from solution oxidation by Fe3 + ” refers to the endmember from sulfite oxidation by Fe3 + in solution;
EIEs, KIEs, and their corresponding θs presented in Table 1 are used to determine the triple oxygen isotope
compositions in these endmember sulfates. “Water line” indicates global meteoric water line.
Figure 6. Schematic illustration of the generation of small Δ′17O in sulfate derived from four different end-
member cases. Water is at the origin, and the black dashed line is the reference line going through the origin
with a slope of 0.5305. “EQ. sulfite” and “EQ. sulfate” refer to the sulfite and sulfate when at oxygen isotope
equilibrium with water, respectively; “KIE + EQ. sulfite”, “KIE O2”, and “KIE water” mark the triple oxygen
isotope compositions of sulfite, O2, and water when they are incorporated into sulfate, respectively; The
number “0.25” means that 25% oxygen of sulfate sourced from “KIE water” or “KIE O2”, respectively, and
“0.75” means that 75% oxygen of sulfate sourced from “KIE + EQ. sulfite”. The black dot “sulfate” located
at the middle of a grey dashed line connecting “0.25” and “0.75”is where the final product sulfate sits in
Δ′17O–δ18O space. The slope of a solid grey line marks specific equilibrium, kinetic, or apparent θs.
Lake sulfate
Ace lake, Antarctica, is a meromictic lake with an active microbial sulfur cycling
(Burton and Barker 1979; Lauro et al. 2011). Two sulfate samples, from a depth of 5.5 m
and 15.5 m, respectively, have been measured for their δ34S, δ18O and Δ′17O (Fig. 7) (Sun
et al. 2015). The water temperatures were ~ 1 °C and ~ 3 °C, respectively, for these two
samples (Lauro et al. 2011), and δ18OH2O value was ~ −16.7‰ (Sun et al. 2015). The observed
apparent Δδ18OSO4–H2O is, therefore, 25.2 ± 0.7‰ at 15.5 m (Fig. 7), which is very close to
the equilibrium Δδ18OSO4–H2O value of 27.2‰ calculated at 3 °C based on Equation (6). This
similarity suggests that sulfate and water have probably reached oxygen isotope equilibrium
at 15.5 m in this permanently stratified lake. Given the measured δ18O and δ17O of sulfate and
assuming a θSO4-H2OEQ of 0.524, we determined that the Δ′17O in water is −0.146 ± 0.043‰. A
negative Δ′17OH2O suggests a highly evaporated water body (Surma et al. 2015; Gázquez et
al. 2018; Passey and Ji 2019), as is likely the case based on Ace lake water history (Roberts
et al. 1999). At 5.5 m, the apparent Δδ18OSO4–H2O is 20.0 ± 0.7‰ which is much smaller than
Δδ18OSO4–H2Oeq, i.e., 27.7‰ at 1 °C, suggesting that sulfate and water are not in oxygen isotope
equilibrium at this shallower depth (Fig. 7). Given the lake water oxygen isotope composition,
we calculated the apparent θ for sulfate of the MSR pathway, app MSR, to be at 0.5288 ± 0.0028,
which is higher than the equilibrium θ value of 0.524 we concluded earlier and close to the
θMSR value of 0.5285 ± 0.0026 estimated by Waldeck et al. (2019). These data suggest that
the non-equilibrium MSR processes possess higher θ values than a sulfate–water equilibrium
system. The case of Ace Lake sulfate illustrates the potential of using small Δ′17O in sulfate
to constrain sulfur cycling in lake systems and lake water triple oxygen isotope compositions.
480 Cao & Bao
Figure 7. Oxygen and sulfur isotope compositions in sulfate from two different depths in Ace Lake, Ant-
arctica. The large δ18O difference between sulfate and water at 15.5 m, i.e., 25.2 ± 0.7‰, suggests sulfate
and water may have achieved oxygen isotope equilibrium.
Riverine sulfate
Mississippi river sulfate. Killingsworth et al. (2018) presented a four-year monthly to
biweekly δ34S, δ18O, and Δ′17O dataset for sulfate from Mississippi River Basin (MiRB). The
four-year average values of δ18OSO4 and Δ′17OSO4 are 3.4‰ and −0.09‰, respectively. Given
an average δ18OH2O value of −6.6‰ for MiRB (Killingsworth et al. 2018), if the sulfate were
all derived from thiosulfate oxidation on pyrite surface, we would predict a δ18OSO4 of −3.9‰.
If the sulfate were all derived from sulfite oxidation by O2 in solution, we would anticipate
a δ18OSO4 of 0.5‰. Both predicted endmember δ18OSO4 values are lower than the observed
average δ18OSO4 of 3.4‰. From the δ34SSO4 data and considering the contribution of sulfate
from evaporite source, Killingsworth et al. (2018) estimated the fraction of pyrite derived
sulfate in MiRB to be ~72%. Given this percentage and evaporite δ18OSO4 of 10‰ to 20‰
(Calmels et al. 2007), we estimate sulfate derived from sulfide oxidation in MiRB would have a
δ18O value of 0‰ to 2.8‰ or 3.2‰ to 6.0‰ if pyrite oxidation is through thiosulfate oxidation
on pyrite surface or sulfite oxidation by O2 in solution, respectively. These estimates offer no
information on the relative dominance of either of the endmember pyrite oxidation pathways
in MiRB due to their close δ18O values. However, the small Δ′17OSO4offers a better resolution
to this distinction. We estimate that the Δ′17OSO4 for surface and solution oxidation are 0.121‰
and 0.030‰, respectively (Fig. 5). Using a Δ′17O of −0.057‰ for evaporite sulfate (Cowie and
Johnston 2016) and the 72% fraction of pyrite derived sulfate, the Δ′17O of MiRB sulfate should
be 0.071‰ and 0.006‰, respectively, for the surface and solution oxidation endmembers.
Obviously, both values are higher than the observed value of −0.09‰. The observed more
negative Δ′17O in MiRB sulfate is not likely caused by MSR process (Hemingway et al. 2020)
because MSR can increase δ18OSO4 at the same time, which is not observed. One reason for
this apparent discrepancy is that the measured Δ′17OSO4 in the Mississippi river sulfate is more
negative than its real value due to the partial yield of O2 during laser fluorination. The data of
Δ′17OSO4 in Killingsworth et al. (2018) were measured at Bao’s laboratory at Louisiana State
University. Comparing Bao et al.’s NBS 127 Δ′17OSO4value to the one determined by Cowie
et al. in Johnston’s laboratory (Bao and Thiemens 2000; Cowie and Johnston 2016), Cowie et
al.’s Δ′17OSO4 is ~0.088‰ (or ~0.07‰, personal communication with David Johnston) higher
than the one obtained by Bao et al. If this 0.088‰ (or 0.07‰) is added to Killingsworth et al.’s
measured Δ′17OSO4 data, the averaged Δ′17OSO4 will be −0.002‰ (or −0.02‰), which matches
well with the pathway of sulfite oxidation by O2 in solution. However, this agreement might
be fortuitous because Δ′17OSO4 for seawater sulfate measured in Bao’s lab and Johnston’s lab
is almost identical, i.e., ~ −0.01‰ (Bao and Thiemens 2000; Cowie and Johnston 2016).
In addition, the average MiRB Δ′17OH2O may be off the GMWL given by Equation (18) (Li et
al. 2015; Sharp et al. 2018; Bindeman et al. 2019), and there are uncertainties in our estimated
EIE and KIE values and their corresponding θs. Nevertheless, this specific case reveals that
small Δ′17OSO4 can add additional constraints on the origin of riverine sulfate, although further
calibration work is required to substantiate and expand this utility.
Marsyangdi river sulfate. Hemingway et al. (2020) presented a δ18O and Δ′17O data
set for sulfate from Marsyangdi River Basin (MaRB), Nepal. Remarkably, some of the
upper-valley sulfates have Δ′17O values as positive as 0.18‰, even more positive than the
one estimated for the ambient water. Pyrite oxidation by atmospheric H2O2 was proposed
to interpret the rather positive Δ′17OSO4 values (Hemingway et al. 2020). However, it is not
clear how H2O2 is delivered to the pyrite oxidation site (Hemingway et al. 2020). From our
analysis (Fig. 8), a positive Δ′17OSO4 can be achieved if sulfate is derived from thiosulfate
oxidation on the pyrite surface. Using the measured water isotope data, we estimate this
endmember sulfate’s δ18O and Δ′17O at −12.1‰ and 0.14‰, respectively. Both of them are
in close agreement with the observed ones (Fig. 8). The discrepancy between our predicted
0.14‰ and the highest observed 0.18‰ may result from uncertainties associated with the
Δ′17O of ambient water, EIE and KIE, and their respective θ values.
MaRB Δ′17OSO4 decreases toward downstream and microbial sulfate reduction and
reoxidation are proposed as the cause (Hemingway et al. 2020). This interpretation is
comparable to the Ace lake case where MSR processes are shifting sulfate oxygen isotope
composition toward equilibrium (Fig. 8).
ANALYTICAL METHODS
O2 is the gaseous analyte for accurate triple oxygen isotope analysis due to its minimal
isobaric interference. Other gases, e.g., CO2, CO, or SO2, certainly cannot deliver a Δ′17O
uncertainty down to ± 0.01‰ required in studying small Δ′17O variations, e.g., from −0.3
to + 0.3‰. O2 generation from sulfate is done by converting SO42− ion into a solid form, either
BaSO4 or Ag2SO4. The powdery sulfate solids can be fluorinated (Bao and Thiemens 2000; Bao
2006; Cowie and Johnston 2016) or in the case of Ag2SO4 thermally decomposed (Savarino
et al. 2001; Schauer et al. 2012; Geng et al. 2013) to generate O2. Sample purification is
important during the precipitation of solid sulfate from the solution because other oxygen-
bearing compounds can be incorporated into BaSO4 or Ag2SO4. When the sample size is not
an issue, BaSO4 is the recommended solid to work with. BaSO4 is easy to handle and to be
purified using a chelating method (Bao 2006).
One important caveat in generating O2 from sulfate solids is that O2 yield is not quantitative,
ranging from 20−35% (Bao and Thiemens 2000) to 50% (Cowie and Johnston 2016) when
BrF5 and F2 vapor are used in laser fluorination, respectively. The partial O2 yield results in
the raw δ18O value being 8‰ to 20‰ lower than the true δ18O value we separately obtain by
analyzing CO gas generated by an online Temperature-conducive elemental analyzer connected
to an isotope-ratio mass spectrometer (IRMS). We calculate the Δ′17O, however, using the raw
δ18O and δ17O measured simultaneously in dual-inlet model on an IRMS. This is hardly an
issue when a large sulfate Δ′17O is of interest. However, a small sulfate Δ′17O value is sensitive
to both the δ18O value and the triple oxygen isotope exponent of the reaction that generates the
partial O2 yield. Laboratory tests show that the Δ′17O from three ~10% aliquots of O2 generated
482 Cao & Bao
Figure 8. Triple oxygen isotope compositions of Marsyangdi River Basin sulfate and sulfate of different
endmember pathways. The value of Δ′17O in “Marsyangdi river water” is estimated by Equation (18); triple
oxygen isotope compositions in “Marsyangdi river sulfate” is from Hemingway et al. (2020); “Sulfate
derived from surface oxidation” refers to the endmember from thiosulfate oxidation on pyrite surface, and
EIE, KIEs, and their corresponding θs presented in Table 1 are used to determine its triple oxygen isotope
compositions; triple oxygen isotope composition of the “Shifted Ace lake sulfate” represents sulfate of
non-equilibrium MSR at 5.5 m depth in Ace lake but the values are shifted by replacing Ace lake water with
the average MaRB water; “EQ. sulfate” represents sulfate calculated by Equation (13) using the average
water isotope data and at temperature 8.3 °C, i.e., the mean of Mean Annual Temperature (MAT) within
Marsyangdi River Basin.
sequentially from the same BaSO4 sample are nearly the same (Bao and Thiemens 2000), which
suggests that O2 yields between 10−30% will produce the same Δ′17O but does not exclude the
possibility that these O2 with lower than 30% yield are systematically different from the 100%
O2 in the Δ′17O. Cowie and Johnston (2016) concluded from a regression line generated by 38
analyses of an inhouse BaSO4 standard that the ~50%-yield processes have followed a slope
of 0.5301 which is analytically unresolvable from the reference slope 0.5305 when calculating
the Δ′17O. If 0.5301 were indeed the diagnostic θ value for the partial O2 generation reaction,
the Δ′17O calculated using the raw δ18O and δ17O of the ~50%-yielded O2 would represent the
Δ′17O of quantitative O2 from BaSO4. Unfortunately, the underlying reaction mechanism for the
spread of the inhouse raw δ18O and δ17O data is not known a priori. Therefore the diagnostic
θ value for the partial O2 generation from BaSO4 may not be close to the reference slope 0.5305.
As mentioned in the riverine sulfate cases, this is one of the uncertainties that have prevented us
from quantitatively interpreting the small sulfate Δ′17O variations.
FUTURE OPPORTUNITIES
Small triple oxygen isotope variation in sulfate, measured by Δ′17OSO4, is a high-dimensional
parameter that reveals the dynamics and pathway of sulfur cycling at present and in the past.
The four sulfate endmembers we identified in this chapter are going to help us to interpret the
measured small Δ′17OSO4. As discussed above, uncertainties exist even for these endmembers.
There are ample research opportunities along this line, and here we outline the most pressing ones.
Calibrating small Δ′17OSO4 measurement. Accurate measurement of Δ′17OSO4 is necessary

to power the small Δ′17OSO4 research. Although uncertainty for Δ′17OSO4 measurement can reach
0.01‰ or less (Cowie and Johnston 2016), the O2 generation reactions all have only partial O2
yields. Such obtained Δ′17O values are inherently less robust or stable than those of quantitative
yield. Linking Δ′17OSO4 with that of water or O2 at the intrinsic level demands a quantitative
yield or well-calibrated θ values for partial-yield reactions. At this time, there is an urgent need
to revisit the partial yield effect on triple oxygen isotope compositions of sulfate. Quantitative
conversion of sulfate to carbon dioxide and equilibrating CO2 and O2 at high temperature
assisted by a Pt catalyst (Mahata et al. 2013; Barkan et al. 2015) might be a way to resolve this
problem, although this issue is not important when dealing with large 17O anomalies.
Calibrating intrinsic triple oxygen isotope effects. Several EIEs, KIEs, and their
associated θs determine sulfate triple oxygen isotope compositions. For now, only the
equilibrium 18αSO4–H2O is relatively converged in literature, while some of the other key
parameters, e.g., αsulfite–H2O and αthiosulfate–H2O, are exhibiting a range of values or entirely absent.
Calibrating reaction rate constants. Multiple electron transfers and thus multiple
chemical reactions are involved in sulfur oxidation or sulfate reduction. Meanwhile, some
intermediates exchange oxygen isotopes with water with different rate constants. These rates
are competing with each other and determine the reversibility of a sulfur redox reaction and
isotope equilibrium or non-equilibrium state of relevant intermediates. For some endmember
cases, e.g. sulfite, its oxidation and oxygen exchange kinetics are well understood while
others, e.g. oxidation and oxygen exchange kinetics for thiosulfate and tetrathionate, are
poorly constrained or entirely unknown.
Sorting out reaction pathways. Sulfate can have different oxygen sources, and reaction
pathways control the relative fractions of these different oxygen sources. Pyrite oxidation has
the most geological implication. Molecular O2 and water can both be incorporated into sulfate
during pyrite oxidation. Several laboratory experiments have been conducted to constrain the
relative proportions of O2 and water in pyrite derived sulfate (Taylor et al. 1984a, Balci et al.
2007; Heidel et al. 2009; Tichomirowa and Junghans 2009; Heidel and Tichomirowa 2010;
Kohl and Bao 2011). Nevertheless, factors controlling the reaction pathways and thus the
relative fraction of incorporated O2 in sulfate are still nebulous, and the observed results are
not converging among different laboratories. Experiments using novel designs, e.g., 17O label
technique, are needed to understand reaction pathways, especially in pH-buffered solutions.
Knowing the role of microbes. Although biotic redox reaction is similar to the abiotic
one to a great extent, it has unique characteristics. For example, the O2 reduction and sulfur
oxidation sites are often physically separated in enzymatic reactions (Kelly 1982; Balci et al.
2017); multiple sulfur oxidation pathways can occur concurrently in one microbe community
(Bobadilla Fazzini et al. 2013); the reversibility of redox reactions is highly dependent on the
energy status or thermodynamic potential (Wing and Halevy 2014). Well-controlled experiments
and numerical models for abiotic redox reactions are essential to unsealing of the black boxes.
ACKNOWLEDGEMENTS
David Johnston and James Lyons are acknowledged for their constructive comments and
suggestions. Xiao Tan provided editorial assistance. This study is financially supported by
CAS “Light of West China” Program to XC and by the Strategic Priority Research Program
(B) of Chinese Academy of Sciences (XDB18010104), Chinese Natural Science Foundation
(41490635), and Jones Professorship to HB.
484 Cao & Bao
REFERENCES
alternative pathway on the triple isotopic composition of atmospheric O2. Global Biogeochem Cycles 17:1030
Antler G, Turchyn AV, Rennie V, Herut B, Sivan O (2013) Coupled sulfur and oxygen isotope insight into bacterial
sulfate reduction in the natural environment. Geochim Cosmochim Acta 118:98–117
Antler G, Turchyn AV, Ono S, Sivan O, Bosak T (2017) Combined 34S, 33S and 18O isotope fractionations record
different intracellular steps of microbial sulfate reduction. Geochim Cosmochim Acta 203:364–380
Balci N, Shanks WC, Mayer B, Mandernack KW (2007) Oxygen and sulfur isotope systematics of sulfate produced
by bacterial and abiotic oxidation of pyrite. Geochim Cosmochim Acta 71:3796–3811
Balci N, Mayer B, Shanks WC, III, Mandernack KW (2012) Oxygen and sulfur isotope systematics of sulfate produced
during abiotic and bacterial oxidation of sphalerite and elemental sulfur. Geochim Cosmochim Acta 77:335–351
Balci N, Brunner B, Turchyn AV (2017) Tetrathionate and elemental sulfur shape the isotope composition of sulfate
in acid mine drainage. Front Microbiol 8:1564
Bao HM (2006) Purifying barite for oxygen isotope measurement by dissolution and reprecipitation in a chelating
solution. Anal Chem 78:304–309
Bao H (2019) Triple Oxygen Isotopes. Cambridge University Press, Cambridge
Bao HM, Thiemens MH (2000) Generation of O2 from BaSO4 using a CO2-laser fluorination system for simultaneous
analysis of δ18O and δ17O. Anal Chem 72:4029–4032
Bao H, Campbell DA, Bockheim JG, Thiemens MH (2000a) Origins of sulphate in Antarctic dry-valley soils as
Bao HM, Thiemens MH, Farquhar J, Campbell DA, Lee CCW, Heine K, Loope DB (2000b) Anomalous 17O
compositions in massive sulphate deposits on the Earth. Nature 406:176–178
Bao H, Fairchild IJ, Wynn PM, Spoetl C (2009) Stretching the envelope of past surface environments: Neoproterozoic
glacial lakes from Svalbard. Science 323:119–122
Bao H, Yu S, Tong DQ (2010) Massive volcanic SO2 oxidation and sulphate aerosol deposition in Cenozoic North
America. Nature 465:909–912
Bao H, Cao X, Hayles JA (2016) Triple oxygen isotopes: Fundamental relationships and applications. Annu Rev Earth
Spectrom 19:3737–3742
Barkan E, Musan I, Luz B (2015) High-precision measurements of delta O−17 and O−17(excess) of NBS19 and
NBS18. Rapid Commun Mass Spectrom 29:2219–2224
Berner RA, Canfield DE (1989) A new model for atmospheric oxygen over Phanerozoic time. Am J Sci 289:333–361
Bertran E, Waldeck A, Wing BA, Halevy I, Leavitt WD, Bradley AS, Johnston DT (2020) Oxygen isotope effects
during microbial sulfate reduction: Applications to sediment cell abundances. ISME J
Betts RH, Voss RH (1970) The kinetics of oxygen exchange between sulfite ion and water. Can J Chem 48:2035–2041
Betts RH, Libich S (1971) 18O transfer in system thiosulfate-sulfite–water: An example of a set of consecutive
reversible first order rate processes. Can J Chem 49:180–186
Bigeleisen J, Wolfsberg M (1958) Theoretical and experimental aspects of isotope effects in chemical kinetics. In:
Adv Chem Phys. John Wiley & Sons, Inc., p 15–76
world’s rivers: Insights into weathering processes and global fluxes into the hydrosphere. Earth Planet Sci Lett 528
Bobadilla Fazzini RA, Cortes MP, Padilla L, Maturana D, Budinich M, Maass A, Parada P (2013) Stoichiometric
modeling of oxidation of reduced inorganic sulfur compounds (Riscs) in Acidithiobacillus thiooxidans.
Biotechnol Bioeng 110:2242–2251
Borilova S, Mandl M, Zeman J, Kucera J, Pakostova E, Janiczek O, Tuovinen OH (2018) Can sulfate be the first
dominant aqueous sulfur species formed in the oxidation of pyrite by Acidithiobacillus ferrooxidans? Front
Microbiol 9:3134
Bottrell SH, Newton RJ (2006) Reconstruction of changes in global sulfur cycling from marine sulfate isotopes. Earth
Sci Rev 75:59–83
Brunner B, Bernasconi SM, Kleikemper J, Schroth MH (2005) A model for oxygen and sulfur isotope fractionation in
sulfate during bacterial sulfate reduction processes. Geochim Cosmochim Acta 69:4773–4785
Brunner B, Mielke RE, Coleman M (2006) Abiotic oxygen isotope equilibrium fractionation between sulfite and
water, AGU Fall Meeting Abstracts, pp. V11C−0601
Brunner B, Yu J-Y, Mielke RE, MacAskill JA, Madzunkov S, McGenity TJ, Coleman M (2008) Different isotope
and chemical patterns of pyrite oxidation related to lag and exponential growth phases of Acidithiobacillus
ferrooxidans reveal a microbial growth strategy. Earth Planet Sci Lett 270:63–72
Brunner B, Einsiedl F, Arnold GL, Müller I, Templer S, Bernasconi SM (2012) The reversibility of dissimilatory
sulphate reduction and the cell-internal multi-step reduction of sulphite to sulphide: Insights from the oxygen
isotope composition of sulphate. Isotopes Environ Health Stud 48:33–54
Burton HR, Barker RJ (1979) Sulfur chemistry and microbiological fractionation of sulfur isotopes in a saline
Antarctic lake. Geomicrobiol J 1:329–340
Caldeira CL, Ciminelli VST, Osseo-Asare K (2010) The role of carbonate ions in pyrite oxidation in aqueous systems.
Calmels D, Gaillardet J, Brenot A, France-Lanord C (2007) Sustained sulfide oxidation by physical erosion processes
in the Mackenzie River basin: Climatic perspectives. Geology 35:1003–1006
Cao X, Bao H (2017) Redefining the utility of the three-isotope method. Geochim Cosmochim Acta 212:16–32
Cao X, Bao H, Peng Y (2019) A kinetic model for isotopologue signatures of methane generated by biotic and abiotic
CO2 methanation. Geochim Cosmochim Acta 249:59–75
Chandra AP, Gerson AR (2010) The mechanisms of pyrite oxidation and leaching: A fundamental perspective. Surf
Sci Rep 65:293–315
Chiba H, Sakai H (1985) Oxygen isotope exchange rate between dissolved sulfate and water at hydrothermal
temperatures. Geochim Cosmochim Acta 49:993–1000
Claypool GE, Holser WT, Kaplan IR, Sakai H, Zak I (1980) The age curves of sulfur and oxygen isotopes in marine
sulfate and their mutual interpretation. Chem Geol 28:199–260
Cowie BR, Johnston DT (2016) High-precision measurement and standard calibration of triple oxygen isotopic
compositions (δ18O, Δ′17O) of sulfate by F2 laser fluorination. Chem Geol 440:50–59
Craig H (1961) Standard for reporting concentrations of deuterium and oxygen−18 in natural waters. Science
133:1833–1834
Crockford PW, Hayles JA, Bao H, Planavsky NJ, Bekker A, Fralick PW, Halverson GP, Thi Hao B, Peng Y, Wing BA
Crockford PW, Cowie BR, Johnston DT, Hoffman PF, Sugiyama I, Pellerin A, Bui TH, Hayles J, Halverson GP,
Macdonald FA, Wing BA (2016) Triple oxygen and multiple sulfur isotope constraints on the evolution of the
post-Marinoan sulfur cycle. Earth Planet Sci Lett 435:74–83
Crockford PW, Kunzmann M, Bekker A, Hayles J, Bao H, Halverson GP, Peng Y, Bui TH, Cox GM, Gibson TM, Wörndle S
(2019) Claypool continued: Extending the isotopic record of sedimentary sulfate. Chem Geol 513:200–225
Druschel GK, Hamers RJ, Banfield JF (2003) Kinetics and mechanism of polythionate oxidation to sulfate at low pH
by O2 and Fe3 + . Geochim Cosmochim Acta 67:4457–4469
Eldridge DL, Mysen BO, Cody GD (2018) Experimental estimation of the bisulfite isomer quotient as a function of
temperature: Implications for sulfur isotope fractionations in aqueous sulfite solutions. Geochim Cosmochim
Acta 220:309–328
Feng C, Tollin G, Enernark JH (2007) Sulfite oxidizing enzymes. Biochim Biophys Acta Proteins Proteom 1774:527–539
Friedrich CG, Rother D, Bardischewsky F, Quentmeier A, Fischer J (2001) Oxidation of reduced inorganic sulfur
compounds by bacteria: Emergence of a common mechanism? Appl Environ Microbiol 67:2873–2882
Fritz P, Basharmal GM, Drimmie RJ, Ibsen J, Qureshi RM (1989) Oxygen isotope exchange between sulfate and
water during bacterial reduction of sulfate. Chem Geol 79:99–105
Geng L, Schauer AJ, Kunasek SA, Sofen ED, Erbland J, Savarino J, Allman DJ, Sletten RS, Alexander B (2013)
Analysis of oxygen−17 excess of nitrate and sulfate at sub-micromole levels using the pyrolysis method. Rapid
Ghosh W, Dam B (2009) Biochemistry and molecular biology of lithotrophic sulfur oxidation by taxonomically and
ecologically diverse bacteria and archaea. FEMS Microbiol Rev 33:999–1043
Goldhaber MB (1983) Experimental study of metastable sulfur oxyanion formation during pyrite oxidation at pH 6–9
and 30 °C. Am J Sci 283:193–217
486 Cao & Bao
Griffith EM, Paytan A (2012) Barite in the ocean – occurrence, geochemistry and palaeoceanographic applications.
Sedimentology 59:1817–1835
Guo W, Zhou C (2019) Triple oxygen isotope fractionation in the DIC-H2O-CO2 system: A numerical framework and
its implications. Geochim Cosmochim Acta 246:541–564
Hanor JS (2000) Barite–celestine geochemistry and environments of formation. Rev Mineral Geochem 40:193–275
Harris E, Sinha B, Van Pinxteren D, Tilgner A, Fomba KW, Schneider J, Roth A, Gnauk T, Fahlbusch B, Mertes S, Lee
T (2013) Enhanced role of transition metal ion catalysis during in-cloud oxidation of SO2. Science 340:727–730
Heidel C, Tichomirowa M (2010) The role of dissolved molecular oxygen in abiotic pyrite oxidation under acid pH
conditions—Experiments with 18O-enriched molecular oxygen. Appl Geochem 25:1664–1675
Heidel C, Tichomirowa M (2011) The isotopic composition of sulfate from anaerobic and low oxygen pyrite oxidation
experiments with ferric iron—New insights into oxidation mechanisms. Chem Geol 281:305–316
Heidel C, Tichomirowa M, Junghans M (2009) The influence of pyrite grain size on the final oxygen isotope difference
between sulphate and water in aerobic pyrite oxidation experiments. Isotopes Environ Health Stud 45:321–342
Hemingway JD, Olson H, Turchyn AV, Tipper ET, Bickle MJ, Johnston DT (2020) Triple oxygen isotope insight into
terrestrial pyrite oxidation. Proc Natl Acad Sci USA 117:7650–7657
Hodgskiss MSW, Crockford PW, Peng Y, Wing BA, Horner TJ (2019) A productivity collapse to end Earth’s Great
Oxidation. Proc Natl Acad Sci USA 116:17207–17212
Holt BD, Kumar R, Cunningham PT (1981) Oxygen−18 study of the aqueous-phase oxidation of sulfur dioxide.
Atmos Environ 15:557–566
Horner DA, Connick RE (2003) Kinetics of oxygen exchange between the two isomers of bisulfite ion, disulfite ion
(S2O52−), and water as studied by oxygen−17 nuclear magnetic resonance spectroscopy. Inorg Chem 42:1884–1894
Hulston JR, Thode HG (1965) Variations in S33, S34, and S36 contents of meteorites and their relation to chemical
Jørgensen BB (1977) Sulfur cycle of a coastal marine sediment (Limfjorden, Denmark). Limnol Oceanogr 22:814–832
Jørgensen BB (1982) Mineralization of organic matter in the sea bed—the role of sulphate reduction. Nature 296:643–645
Kasten S, Jørgensen BB (2000) Sulfate reduction in marine sediments. In: Marine Geochemistry. Schulz HD, Zabel
M, (eds). Springer Berlin Heidelberg, Berlin, Heidelberg, p 263–281
Kelly DP (1982) Biochemistry of the chemolithotrophic oxidation of inorganic sulphur. Philos Trans R Soc Lond B
Biol Sci 298:499–528
Kelly DP (2003) Microbial inorganic sulfur oxidation: The APS pathway. In: Biochemistry and Physiology of
Anaerobic Bacteria. Ljungdahl LG, Adams MW, Barton LL, Ferry JG, Johnson MK, (eds). Springer New York,
New York, NY, p 205–219
Killingsworth BA, Bao H, Kohl IE (2018) Assessing pyrite-derived sulfate in the Mississippi river with four years of
sulfur and triple-oxygen isotope data. Environ Sci Technol 52:6126–6136
Klatt JM, Polerecky L (2015) Assessment of the stoichiometry and efficiency of CO2 fixation coupled to reduced
sulfur oxidation. Front Microbiol 6
Kohl I, Bao H (2011) Triple-oxygen-isotope determination of molecular oxygen incorporation in sulfate produced
during abiotic pyrite oxidation (pH=2–11). Geochim Cosmochim Acta 75:1785–1798
Kroopnick P, Craig H (1972) Atmospheric oxygen: Isotopic composition and solubility fractionation. Science
175:54–55
Krouse HR, Gould WD, McCready RGL, Rajan S (1991) 18O incorporation into sulphate during the bacterial
oxidation of sulphide minerals and the potential for oxygen isotope exchange between O2, H2O and oxidized
sulphur intermediates. Earth Planet Sci Lett 107:90–94
Kuo DTF, Kirk DW, Jia CQ (2006) The chemistry of aqueous S(IV)–Fe–O2 system: state of the art. J Sulphur Chem
27:461–530
Lauro FM, DeMaere MZ, Yau S, Brown MV, Ng C, Wilkins D, Raftery MJ, Gibson JA, Andrews-Pfannkoch C, Lewis
M, Hoffman JM (2011) An integrative study of a meromictic lake ecosystem in Antarctica. ISME J 5:879–895
Lente G, Fábián I (2002) Kinetics and mechanism of the oxidation of sulfur(IV) by iron(III) at metal ion excess. J
Chem Soc, Dalton Trans:778–784
transfer and implications for tracing oxygen cycling in the ocean. Mar Chem 210:61–71
Lloyd RM (1968) Oxygen isotope behavior in the Sulfate–water System. J Geophys Res 73:6099–6110
Lloyd RM (1967) Oxygen−18 composition of oceanic sulfate. Science 156:1228–1231
Luther GW (1987) Pyrite oxidation and reduction: Molecular orbital theory considerations. Geochim Cosmochim
Acta 51:3193–3199
platinum: an innovative technique for measuring Δ17O in CO2. Anal Chem 85:6894–6901
Mazumdar A, Goldberg T, Strauss H (2008) Abiotic oxidation of pyrite by Fe(III) in acidic media and its implications
for sulfur isotope measurements of lattice-bound sulfate in sediments. Chem Geol 253:30–37
McCready RGL, Krouse HR (1982) Sulfur isotope fractionation during the oxidation of elemental sulfur by Thiobacilli
in a Solonetzic soil. Can J Soil Sci 62:105–110
Mizutani Y, Rafter TA (1973) Isotopic behaviour of sulphate oxygen in the bacterial reduction of sulphate. Geochem
J 6:183–191
Mook WG (2000) Environmental isotopes in the hydrological cycle: Principles and applications, Vol. I: Introduction
—Theory, Methods, Review. In: Isotope Effects. UNESCO/IAEA, Paris,
Moses CO, Herman JS (1991) Pyrite oxidation at circumneutral pH. Geochim Cosmochim Acta 55:471–482
Moses CO, Kirk Nordstrom D, Herman JS, Mills AL (1987) Aqueous pyrite oxidation by dissolved oxygen and by
ferric iron. Geochim Cosmochim Acta 51:1561–1571
Müller IA, Brunner B, Coleman M (2013a) Isotopic evidence of the pivotal role of sulfite oxidation in shaping the
oxygen isotope signature of sulfate. Chem Geol 354:186–202
Müller IA, Brunner B, Breuer C, Coleman M, Bach W (2013b) The oxygen isotope equilibrium fractionation between
sulfite species and water. Geochim Cosmochim Acta 120:562–581
Nriagu JO (1972) Stability of vivianite and ion-pair formation in the system Fe3(PO4)2–H3PO4–H2O. Geochim
Oba Y, Poulson SR (2009) Oxygen isotope fractionation of dissolved oxygen during abiological reduction by aqueous
sulfide. Chem Geol 268:226–232
Pack A, Herwartz D (2014) The triple oxygen isotope composition of the Earth mantle and understanding variations
in terrestrial rocks and minerals. Earth Planet Sci Lett 390:138–145
Peng Y, Bao H, Zhou C, Yuan X (2011) 17O-depleted barite from two Marinoan cap dolostone sections, South China.
Pryor WA, Tonellat U (1967) Nucleophilic displacements at sulfur. III. The exchange of oxygen−18 between sodium
thiosulfate−18O and water. J Am Chem Soc 89:3379–3386
Rennie VCF, Turchyn AV (2014) Controls on the abiotic exchange between aqueous sulfate and water under laboratory
conditions. Limnol Oceanogr Methods 12:166–173
Reuer MK, Barnett BA, Bender ML, Falkowski PG, Hendricks MB (2007) New estimates of Southern Ocean
biological production rates from O2/Ar ratios and the triple isotope composition of O2. Deep Sea Res Part I
Oceanogr Res Pap 54:951–974
Rimstidt JD, Vaughan DJ (2003) Pyrite oxidation: A state-of-the-art assessment of the reaction mechanism. Geochim
Roberts D, Roberts JL, Gibson JAE, McMinn A, Heijnis H (1999) Palaeohydrological modelling of Ace Lake,
Vestfold Hills, Antarctica. Holocene 9:515–520
Rosso KM, Vaughan DJ (2006) Reactivity of sulfide mineral surfaces. Rev Mineral Geochem 61:557–607
Sand W, Gerke T, Hallmann R, Schippers A (1995) Sulfur chemistry, biofilm, and the (in)direct attack mechanism - a
critical-evaluation of bacterial leaching. Appl Microbiol Biotechnol 43:961–966
Savarino J, Lee CCW, Thiemens MH (2000) Laboratory oxygen isotopic study of sulfur (IV) oxidation: Origin of
the mass-independent oxygen isotopic anomaly in atmospheric sulfates and sulfate mineral deposits on Earth. J
Geophys Res Atmos 105:29079–29088
Savarino J, Alexander B, Darmohusodo V, Thiemens MH (2001) Sulfur and oxygen isotope analysis of sulfate at
micromole levels using a pyrolysis technique in a continuous flow system. Anal Chem 73:4457–4462
Schauer AJ, Kunasek SA, Sofen ED, Erbland J, Savarino J, Johnson BW, Amos HM, Shaheen R, Abaunza M, Jackson
TL, Thiemens MH (2012) Oxygen isotope exchange with quartz during pyrolysis of silver sulfate and silver
nitrate. Rapid Commun Mass Spectrom 26:2151–2157
Schippers A, Jozsa PG, Sand W (1996) Sulfur chemistry in bacterial leaching of pyrite. Appl Environ Microbiol
62:3424–3431
Singer PC (1972) Anaerobic control of phosphate by ferrous iron. J Water Pollut Control Fed 44:663–669
488 Cao & Bao
Singer PC, Stumm W (1970) Acidic mine drainage: The rate-determining step. Science 167:1121–1123
Smith LA, Hendry MJ, Wassenaar LI, Lawrence J (2012) Rates of microbial elemental sulfur oxidation and 18O and
34
S isotopic fractionation under varied nutrient and temperature regimes. Appl Geochem 27:186–196
Spencer RJ (2000) Sulfate minerals in evaporite deposits. Rev Mineral Geochem 40:173–192
Stefánsson A (2007) Iron(III) hydrolysis and solubility at 25 °C. Environ Sci Technol 41:6117–6123
Sun T, Socki RA, Bish DL, Harvey RP, Bao H, Niles PB, Cavicchioli R, Tonui E (2015) Lost cold Antarctic deserts
inferred from unusual sulfate formation and isotope signatures. Nat Commun 6:7579
Taylor BE, Wheeler MC, Nordstrom DK (1984a) Stable isotope geochemistry of acid mine drainage: Experimental
oxidation of pyrite. Geochim Cosmochim Acta 48:2669–2678
Taylor BE, Wheeler MC, Nordstrom DK (1984b) Isotope composition of sulphate in acid mine drainage as measure
of bacterial oxidation. Nature 308:538–541
Thiemens MH (2006) History and applications of mass-independent isotope effects. Annu Rev Earth Planet Sci
34:217–262
Thurston RS, Mandernack KW, Shanks WC, III (2010) Laboratory chalcopyrite oxidation by Acidithiobacillus
ferrooxidans: Oxygen and sulfur isotope fractionation. Chem Geol 269:252–261
Tichomirowa M, Junghans M (2009) Oxygen isotope evidence for sorption of molecular oxygen to pyrite surface sites
and incorporation into sulfate in oxidation experiments. Appl Geochem 24:2072–2092
Turchyn AV, Brüchert V, Lyons TW, Engel GS, Balci N, Schrag DP, Brunner B (2010) Kinetic oxygen isotope
effects during dissimilatory sulfate reduction: A combined theoretical and experimental approach. Geochim
Van Stempvoort DR, Krouse HR (1993) Controls of δ18O in sulfate: Review of experimental data and application
to specific environments. In: Environmental Geochemistry of Sulfide Oxidation. Vol 550. American Chemical
Society, p 446–480
Vera M, Schippers A, Sand W (2013) Progress in bioleaching: Fundamentals and mechanisms of bacterial metal
sulfide oxidation-part A. Appl Microbiol Biotechnol 97:7529–7541
Waldeck AR, Cowie BR, Bertran E, Wing BA, Halevy I, Johnston DT (2019) Deciphering the atmospheric signal in
marine sulfate oxygen isotope composition. Earth Planet Sci Lett 522:12–19
Wankel SD, Bradley AS, Eldridge DL, Johnston DT (2014) Determination and application of the equilibrium oxygen
isotope effect between water and sulfite. Geochim Cosmochim Acta 125:694–711
Wing BA, Halevy I (2014) Intracellular metabolite levels shape sulfur isotope fractionation during microbial sulfate
respiration. PNAS 111:18116–18125
Wortmann UG, Chernyavsky B, Bernasconi SM, Brunner B, Böttcher ME, Swart PK (2007) Oxygen isotope
biogeochemistry of pore water sulfate in the deep biosphere: Dominance of isotope exchange reactions with
ambient water during microbial sulfate reduction (ODP Site 1130). Geochim Cosmochim Acta 71:4221–4232
Xu Y, Schoonen MAA (1995) The stability of thiosulfate in the presence of pyrite in low-temperature aqueous
solutions. Geochim Cosmochim Acta 59:4605–4622
Zak I, Sakai H, Kaplan IR (1980) Factors controlling 18O/16O and 34S/32S isotope rations of ocean sulfates, evaporites,
and interstitial sulfates from modern deep sea sediments. In: Isotope marine chemistry. Goldberg ED, Horibe Y,
Sakurashi K, (eds). Rokakuho, Tokyo, p 339–373
Zeebe RE (2010) A new value for the stable oxygen isotope fractionation between dissolved sulfate ion and water.
Zhang JZ, Millero FJ (1991) The rate of sulfite oxidation in seawater. Geochim Cosmochim Acta 55:677–685
Ziegler K, Coleman ML, Mielke RE, Young ED (2010) Sources and contributions of oxygen during microbial pyrite
oxidation: The triple oxygen isotopes of sulfate as a biosignature. Lunar and Planetary Science Conference, p. 2245
RiMG Series
HISTORY OF RiMG
Volumes 1–38 were published as “Reviews in Mineralogy” (ISSN 0275-0279). Volumes 1-6
originally appared as “Short Course Notes” (no ISSN). The name was changed to “Reviews in
Mineralogy & Geochemistry” (RiMG) (ISSN 1529-6466) starting with Volume 39. Paul Ribbe
was sole editor for volumes 1–41. He was joined by Jodi Rosso as series editor for volumes 42–53
in the RiMG series submitted through the Geochemistry Society. With his retirement, Jodi Rosso
became sole editor for volumes 54–79 in the RiMG series. With Jodi Rosso’s move to Executive
Editor of Elements magazine, Ian Swainson became Series Editor starting with volume 80.
HOW TO PUBLISH IN RiMG

RiMG volumes are based on topics that have been proposed and appoved by the MSA
Council or Geochemical Society Board of Directors. If you have an idea for a future RiMG
volume, or a short course accompanied by a RiMG volume, you should read the Short Course
Guide which describes how to develop and propose a topic for consideration for either case.
Proposals should be submitted to the Short Course Committee (http://www.minsocam.org/
msa/SC/SCCommittee.html). Contributions to an appoved volume are by invitation only.
A listing of the previous volume numbers and their volume editors is below. Selecting
the titles at http://www.minsocam.org/msa/RIM/index2.html gives you to a detailed
description of each volume, the table of contents, and any errata or supplementary material
Previous volumes can be ordered from https://msa.minsocam.org/orders.html.
Volume 85: Reactive Transport in Natural and Engineered Systems
2019 ISBN 978-1-946850-01-0 J Druhan C Tournassat i-xv + 534 pp
Volume 84: High Temperature Gas–Solid Reactions in Earth and Planetary Processes
2018 ISBN 978-1-946850-00-3 PL King, BJ Fegley, Jr., T Seward i-xiv + 520 pp
Volume 83: Petrochronology: Methods and Applications
2017 ISBN 978-0-939950-05-8 MJ Kohn, M Engi, P Lanari i-xiv + 575 pp
Volume 82: Non-Traditional Stable Isotopes
2017 ISBN 978-0-939950-98-0 F-X Teng, J Watkins, N Dauphas i-xvi + 892 pp
Volume 81: Highly Siderophile and Strongly Chalcophile Elements in High-Temperature Geochemistry and Cosmochemistry
2015 ISBN 978-0-939950-97-3 J Harvey, JMD Day i-xxiii + 774 pp
Volume 80: Pore-Scale Geochemical Processes
2015 ISBN 978-0-939950-96-6 CI Steefel, S Emmanuel, LM Anovitz i-xiv + 491 pp
Volume 79: Arsenic, Environmental Geochemistry, Mineralogy, and Microbiology
RJ Bowell, CN Alpers, HE Jamieson, DK
2014 ISBN 978-0-939950-94-2 i-xvi + 635 pp
Nordstrom, J Majzlan
Volume 78: Spectroscopic Methods in Mineralogy and Materials Sciences
2014 ISBN13 978-0-939950-93-5 GS Henderson, DR Neuville, RT Downs i-xviii + 800 pp
Volume 77: Geochemistry of Geologic CO2 Sequestration
2013 ISBN 978-0-939950-92-8 DJ DePaolo, DR Cole, A Navrotsky, IC Bourg i-xiv + 539 pp
Volume 76: Thermodynamics of Geothermal Fluids
2013 ISBN 978-0-939950-91-1 A Stefánsson, T Driesner, P Bénézeth i-x + 350 pp
Volume 75: Carbon in Earth
2013 ISBN 978-0-939950-90-4 RM Hazen, AP Jones, JA Baross i-xv + 698 pp
Volume 74: Appied Mineralogy of Cement & Concrete
2012 ISBN 978-0-939950-88-1 MATM Broekmans, Pöllmann, i-x + 364 pp
Volume 73: Sulfur in Magmas and Melts: Its Importance for Natural and Technical Processes
2011 ISBN 978-0-939950-87-4 H Behrens, JD Webster i-xiv + 578 pp
Volume 72: Diffusion in Minerals and Melts
2010 ISBN 978-0-939950-86-7 Y Zhang, DJ Cherniak i-xviii + 1036 pp
Volume 71: Theoretical and Computational Methods in Mineral Physics: Geophysical Appications
2010 ISBN 978-0-939950-85-0 R Wentzcovitch, L Stixrude i-xviii + 484 pp
Volume 70: Thermodynamics and Kinetics of Water-Rock Interaction
ISBN 0-939950-84-7;
2009 ISBN13 978-0-939950-84-3 EH Oelkers J Schott i-xvii + 569 pp
Volume 69: Minerals, Inclusions and Volcanic Processes

ISBN 0-939950-83-9;
2008 ISBN13 978-0-939950-83-6 KD Putirka, FJ Tepley III i-xiv + 674 pages.
Volume 68: Oxygen In the Solar System

ISBN 0-939950-80-4; GJ MacPherson, DW Mittlefehldt, JH Jones,
2008 ISBN13 978-0-939950-80-5 i-xx + 598 pages
SB Simon, JJ Papike, S Mackwell
Volume 67: Amphiboles: Crystal Chemistry, occurrences, and Health Issues
ISBN 0-939950-79-0; FC Hawthorne, R Oberti, G Della Ventura,
2007 ISBN13 978-0-939950-79-9 i-xxv + 545 pages.
A Mottana
Volume 66: Paleoaltimetry: Geochemical and Thermodynamic Appoaches
ISBN 0-939950-78-2;
2007 ISBN13 978-0-939950-78-2 MJ Kohn i-x + 278 pages
Volume 65: Fluid-Fluid Interactions

ISBN 0-939950-77-4;
2007 ISBN13 978-0-939950-77-5 A Liebscher CA Heinrich i-xii + 430 pages
Volume 64: Medical Mineralogy and Geochemistry

ISBN 0-939950-75-8;
2006 ISBN13 978-0939950-75-1 N Sahai, MAA Schoonen i-xi + 332 pp
Volume 63: Neutron Scattering in Earth Sciences

ISBN 0-939950-75-8;
2006 ISBN13 978-0939950-75-1 HR Wenk i-xx + 471 pp
Volume 62: Water in Nominally Anhydrous Minerals

ISBN 0-939950-74-X;
2006 ISBN13 978-0-939950-74-4 H Kepper, JR. Smyth i-viii + 478 pp
Volume 61: Sulfide Mineralogy and Geochemistry

ISBN 0-939950-73-1;
2006 ISBN13 978-0-939950-73-7 DJ Vaughan i-xiii + 714 pp
Volume 60: New Views of the Moon

ISBN 0-939950-72-3;
2006 ISBN13 978-0-939950-72-0 BL Jolliff, MA Wieczorek, CK Shearer, CR Neal i-xxii + 772 pp
Volume 59: Molecular Geomicrobiology

ISBN 0-939950-71-5;
2005 ISBN13 978-0-939950-71-3 JF Banfield, J Cervini-Silva, KH Nealson i-xiv + 294 pp
Volume 58: Low-Temperature Thermochronology: Techniques, Interpretations, and Appications

ISBN 0-939950-70-7;
2005 ISBN13 978-0-939950-70-6 PW Reiners, TA Ehlers i-xxii + 620 pp
Volume 57: Micro- and Mesoporous Mineral Phases
ISBN 0-939950-69-3;
2005 ISBN13 978-0-939950-69-0 G Ferraris, S Merlino i-xiii + 448 pp
Volume 56: Epidotes

ISBN 0-939950-68-5;
2004 ISBN13 978-0-939950-68-3 A Liebscher, G Franz i-xviii + 628 pp
Volume 55: Geochemistry of Non-Traditional Stable Isotopes

ISBN 0-939950-67-7;
2004 ISBN13 978-0-939950-67-6 CM Johnson, BL Beard, F Albarede i-xvi + 454 pp
Volume 54: Biomineralization

ISBN 0-939950-66-9;
2003 ISBN13 978-0-939950-66-9 PM Dove, JJ De Yoreo, S Weiner i-xiv + 381 pp
Volume 53: Zircon

ISBN 0-939950-65-0;
2003 ISBN13 978-0-939950-65-2 JM Hanchar, PWO Hoskin i-xviii + 500 pp
Volume 52: Uranium-Series Geochemistry

ISBN 0-939950-64-2; B Bourdon, GM Henderson, CC Lundstrom, SP
2003 ISBN13 978-0-939950-64-5 i-xx + 656 pp
Turner,
Volume 51: Plastic Deformation of Minerals and Rocks
ISBN 0-939950-63-4;
2002 ISBN13 978-0-939950-63-8 S Karato, H-R Wenk i-xiv + 420 pp
Volume 50: Beryllium Mineralogy, Petrology, and Geochemistry

ISBN 0-939950-62-6;
2002 ISBN13 978-0-939950-62-1 E Grew i-xii + 691 pp
Volume 49: Appications of Synchrotron Radiation in Low-Temperature Geochemistry and Environmental Science
ISBN 0-939950-61-8;
2002 ISBN13 978-0-939950-61-4 PA Fenter, ML Rivers, NC Sturchio, SR Sutton, i-xxii + 579 pp
Volume 48: Phosphates Geochemical, Geobiological, and Materials Importance

ISBN 0-939950-60-X;
2002 ISBN13 978-0-939950-60-7 ML Kohn, J Rakovan, JM Hughes i-xvi + 742 pp
Volume 47: Nobel Gases in Geochemistry and Cosmochemistry

ISBN 0-939950-59-6;
2002 ISBN13 978-0-939950-59-1 DP Porcelli, CJ Ballentine, R Wieler i-xviii + 844 pp
Volume 46: Micas: Crystal Chemistry & Metamorphic Petrology

ISBN 0-939950-58-8; A Mottana, FP Sassi, JB Thompson, Jr., S
2002 ISBN13 978-0-939950-58-4 i-xiv + 499 pp
Guggenheim
Volume 45: Naturnal Zeolites: Occurrence, Properties, Appications
ISBN 0-939950-57-X;
2001 ISBN13 978-0-939950-57-7 DL Bish, DW Ming i-xiv + 654 pp
Volume 44: Nanoparticles and the Environment

ISBN 0-939950-56-1;
2001 ISBN13 978-0-939950-56-0 JF Banfield ,A Navrotsky i-xiv + 349 pp
Volume 43: Stable Isotope Geochemistry

ISBN 0-939950-55-3;
2001 ISBN13 978-0-939950-55-3 JW Valley, D Cole i-x11 + 531 pp
Volume 42: Molecular Modeling Theory: Applications in the Geosciences

2001 ISBN 0-939950-54-5; RT Cygan, JB Kubicki i-v +662 pp
ISBN13 978-0-939950-54-6
Volume 41: High-Temperature and High-Pressure Crystal Chemistry
ISBN 0-939950-53-7;
2001 ISBN13 978-0-939950-53-9 RM Hazen, RT Downs i-viii + 596 pp
Volume 40: Sulfate Minerals - Crystallography, Geochemistry, and Environmental Significance

ISBN 0-939950-52-9;
2000 ISBN13 978-0-939950-52-2 CN Alpers, JL Jambor, DK Nordstrom i-xii + 608 pp
Volume 39: Transformation Processes in Minerals

ISBN 0-939950-51-0;
2000 ISBN13 978-0-939950-51-5 SAT Redfern, MA Carpenter, i-x + 361 pp
Volume 38: Uranium: Mineralogy, Geochemistry and the Environment

ISBN 0-939950-50-2;
1999 ISBN13 78-0-939950-50-8 PC Burns, R Finch i-xvi + 679 pp
Volume 37: Ultrahigh-Pressure Mineralogy: Physics and Chemistry of the Earth’s Deep Interior
ISBN 0-939950-48-0;
1998 ISBN13 978-0-939950-48-5 R Hemley i-xx + 671 pp
Volume 36: Planetary Materials

ISBN 0-939950-46-4;
1998 ISBN13 978-0-939950-46-1 JJ Papike i-xx + 864 pp
Volume 35: Geomicrobiology: Interaction Between Microbes and Minerals

ISBN 0-939950-45-6;
1997 ISBN13 978-0-939950-45-4 JF Banfield ,KH Nealson i-xvi + 448 pp
Volume 34: Reactive Transport in Porous Media

ISBN 0-939950-45-6;
1997 ISBN13 978-0-939950-45-4 PC Lichtner, CI Steefel, EH Oelkers i-xiv + 438 pp
Volume 33: Boron Mineralogy, Petrology and Geochemistry

ISBN 0-939950-41-3;
1996 ISBN13 978-0-939950-41-6 LM Anovitz, ES Grew i-xx + 864 pp
Volume 32: Structure, Dynamics and Properties of Silicate Melts

ISBN 0-939950-39-1;
1995 ISBN13 978-0-939950-39-3 JF Stebbins, PF McMillan, DB Dingwell i-xvi + 616 pp
Volume 31: Chemical Weathering Rates of Silicate Minerals

SBN 0-939950-38-3;
1995 ISBN13 978-0-939950-38-6 AF White, SL Brantley i-xvi + 583 pp
Volume 30: Volatiles in Magmas

ISBN 0-939950-36-7;
1994 ISBN13 978-0-939950-36-2 MR Carroll, JR Holloway i-xviii + 517 pp
Volume 29: Silica: Physical Behavior, Geochemistry and Materials Appcations

ISBN 0-939950-35-9;
1994 ISBN13 978-0-939950-35-5 PJ Heaney, CT Prewitt, GV Gibbs i-xviii + 606 pp
Volume 28: Health Effects of Mineral Dusts

ISBN 0-939950-33-2;
1993 ISBN13 978-0-939950-33-1 GD Guthrie, Jr., BT Mossman i-xvi + 584 pp
Volume 27: Minerals and Reactions at the Atomic Scale: Transmission Electron Microscopy
ISBN 0-939950-32-4;
1992 ISBN13 978-0-939950-32-4 PR Buseck i-xvi + 516 pp
Volume 26: Contact Metamorphism

ISBN 0-939950-31-6;
1991 ISBN13 978-0-939950-31-7 DM Kerrick i-xvi + 672 pp
Volume 25: Oxide Minerals:Petrologic and Magnetic Significance
1991 ISBN 0-939950-30-8; DH Lindsley i-xiv + 509 pp
ISBN13 978-0-939950-30-0
Volume 24: Modern Methods of Igneous Petrology: Understanding Magmatic Processes

1990 ISBN 0-939950-29-4; J Nicholls, JK Russell i-viii + 314 pp
ISBN13 978-0-939950-29-4
Volume 23: Mineral–Water Interface Geochemistry

1990 ISBN 0-939950-28-6; MF Hochella, Jr., AF White i-xvi + 603 pp
ISBN13 978-0-939950-28-7
Volume 22: The Al2SiO5 Polymorphs

1990 ISBN 0-939950-27-8; DM Kerrick i-xii + 406 pp
ISBN13 978-0-939950-27-0
Volume 21: Geochemistry and Mineralogy of Rare Earth Elements

1989 ISBN 0-939950-25-1; BR Lipin, GA McKay i-x + 348 pp
ISBN13 978-0-939950-25-6
Volume 20: Modern Powder Diffraction

1989 ISBN 0-939950-24-3; DL Bish, JE Post i-xii + 369 pp
ISBN13 978-0-939950-24-9
Volume 19: Hydrous Phyllosilicates (exclusive of micas)

1988 ISBN 0-939950-23-5; SW Bailey, i-xiii + 725 pp
ISBN13 978-0-939950-23-2
Volume 18: Spectroscopic Methods in Mineralogy and Geology

1988 ISBN 0-939950-22-7; FC Hawthorne i-xvi + 512 pp
ISBN13 978-0-939950-22-5
Volume 17: Thermodynamic Modeling of Geological Materials: Minerals, Fluids and Melts
1987 ISBN 0-939950-21-9; ISE Carmichael, HP Eugster i-xiv + 499 pp
ISBN13 978-0-939950-21-8
Volume 16: Stable Isotopes in High Temperature Geological Processes

1986 ISBN 0-939950-20-0; JW Valley, HP Taylor, Jr., JR O’Neil i-xvi + 570 pp
ISBN13 978-0-939950-20-1
Volume 15: Mathematical Crystallography

1985 ISBN 0-939950-19-7; MB Boisen, Jr., GV Gibbs i-xii + 460 pp
ISBN13 978-0-939950-19-5
Volume 14: Microscopic to Macroscopic

1985 ISBN 0-939950-18-9; SW Kieffer, A Navrotsky i-x + 428 pp
ISBN13 978-0-939950-18-8
Volume 13: Micas

1984 ISBN 0-939950-17-0; SW Bailey i-xii + 584 pp
ISBN13 978-0-939950-17-1
Volume 12: Fluid Inclusions

1984 ISBN 0-939950-16-2; Edwin Roedder i-vi + 646 pp
ISBN13 978-0-939950-16-4
Volume 11: Carbonates: Mineralogy and Chemistry

1983, ISBN 0-939950-15-4; RJ Reeder i-xii + 399 pp
ISBN13 978-0-939950-15-7
1990
Volume 10: Characterization of Metamorphism through Mineral Equilibria
1982 ISBN 0-939950-12-X; JM Ferry i-xiv + 397 pp
ISBN13 978-0-939950-12-6
Volume 9B: Amphiboles and Other Hydrous Pyriboles—Mineralogy
1982 ISBN 0-939950-11-1; DR Veblen, PH Ribbe i-x + 390 pp
ISBN13 978-0-939950-11-9
Volume 9A: Amphiboles: Petrology and Experimental Phase Relations

1981 ISBN 0-939950-10-3; DR Veblen, PH Ribbe i-xii + 372 pp
ISBN13 978-0-939950-10-2
Volume 8: Kinetics of Geochemical Processes

1981 ISBN 0-939950-08-1; AC Lasaga, RJ Kirkpatrick i-x + 398 pp
ISBN13 978-0-939950-08-9
Volume 7: Pyroxenes
1980 ISBN 0-939950-07-3; CT Prewitt i-x + 525 pp
ISBN13 978-0-939950-07-2
Volume 6: Marine Minerals

1979 ISBN 0-939950-06-5; RG Burns i-x + 380 pp
ISBN13 978-0-939950-06-5
Volume 5: Orthosilicates
1980 ISBN 0-939950-13-8; RG Burns i-xii + 450 pp
ISBN13 978-0-939950-13-3
Volume 4: Mineralogy and Geology of Natural Zeolites

1977 ISBN 0-939950-04-9; FA Mumpton i-xii + 233 pp
ISBN13 978-0-939950-04-1
Volume 3: Oxide Minerals

1976 ISBN 0-939950-03-0; D Rumble, III i-3 + 706 pp
ISBN13 978-0-939950-03-4
Volume 2: Feldspar Mineralogy

1975 ISBN 0-939950-14-6; PH Ribbe i-vii + 362 pp
ISBN13 978-0-939950-14-0
1983
Volume 1: Sufide Mineralogy
1974 ISBN 0-939950-01-4; PH Ribbe i-v + 301 pp
ISBN13 978-0-939950-01-0

Ebin - Pub Triple Oxygen Isotope Geochemistry 9781501524677 9781946850065

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ebin - Pub Triple Oxygen Isotope Geochemistry 9781501524677 9781946850065

Uploaded by

Copyright:

Available Formats

REVIEWS in MINERALOGY

Triple Oxygen Isotope

Series Editor: Ian Swainson

ISSN 1529-6466 (print)

86 Reviews in Mineralogy and Geochemistry 86

1 Why Measure 17O? Historical Perspective,

2 Discoveries of Mass Independent Isotope Effects in the

THE BEGINNING OF ISOTOPES.........................................................................................35

3 Climbing to the Top of Mount Fuji: Uniting Theory and

4 Mass Dependence of Equilibrium Oxygen Isotope Fractionation in

5 Standardization for the Triple Oxygen Isotope System:

6 Mass-Independent Fractionation of Oxygen Isotopes

7 Isotopic Traces of Atmospheric O2 in Rocks, Minerals, and Melts

8 Triple Oxygen Isotopes in Evolving Continental Crust, Granites,

PART III: THE TRIPLE OXYGEN ISOTOPE ANALYSIS OF GRANITES IN

9 Triple Oxygen Isotope Variations in Earth’s Crust

10 Triple Oxygen Isotope Trend Recorded by Precambrian Cherts:

11 Triple Oxygen Isotopes in Silica–Water and Carbonate–Water

TRIPLE OXYGEN ISOTOPES OF COEXISTING QUARTZ AND CALCITE..................393

12 Triple Oxygen Isotope Systematics in the Hydrologic Cycle

13 Triple Oxygen Isotopes in Meteoric Waters, Carbonates, and

14 Small Triple Oxygen Isotope Variations in Sulfate:

Why Measure 17O? Historical Perspective,

1529-6466/21/0086-0001$05.00 (print) http://dx.doi.org/10.2138/rmg.2021.86.01

OXYGEN TRIPLE-ISOTOPE SYSTEMATICS

and as expressed in delta values:

m16  m17 (11)

Experimental measurements of ln(17/16α) /ln(18/16α) = θ

which is of a similar form to the definition of the δ value. In linear format,

5 Data reported relative to VSMOW:

REFERENCE MATERIALS AND STANDARDS

Table 1. Recent high-precision oxygen triple-isotope measurements of widely-used silicate

(1   18 OSLAP/VSMOW, measured )0.528  1 (24)

(1   18 OSLAP/VSMOW, assigned )0.528  1

STANDARDIZING Δ′17O DATA FROM ROCKS AND MINERALS

SOME EXAMPLES OF THE APPLICATION OF

Corrections to δ13C measurements of CO2

Zakharov and Bindeman (2019) reported that oxygen triple-isotope analyses of

CONCLUSIONS AND OUTLOOK

Discoveries of Mass Independent Isotope Effects

University of Chinese Academy of Sciences

THE BEGINNING OF ISOTOPES

AN OVERVIEW OF STABLE ISOTOPE GEOCHEMISTRY

FUNDAMENTAL CHEMICAL PHYSICS OF

OH + SiO → SiO2 + H (R3)

Though isotopic self-shielding of O2 does not allow the sequestration of an isotopic

isotopic composition. Though the early balloon-borne mass spectrometer measurement of a

O(1D) + N2O ↔ O2 + N2 (R11)

In the Martian atmosphere, CO2 is exposed to UV light and subject to photodissociation.

As discussed by Bhattacharya et al. (2000) the CO2 photodissociation experiments and

PHOTODISSOCIATION IN THE EARLY SOLAR SYSTEM AND

Mass independent isotope effects in CO photodissociation

electronic structure. A historically interesting review of Rydberg states is given by Mulliken

Self-shielding final test and errors

Nebular fate of water and water ice from photodissociation

Figure 25. Schematic diagram defining the reaction net-

either an oxidant of reduced species, or formation of O2 and OH. The OH is spectroscopically

Figure 27. Schematic diagram showing

A requirement in the transport phenomena and concomitant photochemical, nebular

SOLID FORMATION IN THE EARLY SOLAR SYSTEM

3FeS + 4H2O ↔ Fe3O4 + 3H2S + H2 (R12)

3Fe + 4H2O ↔ Fe3O4 + 4H2 (R13)

Fe + OH ↔ FeOH ↔ FeO + 1/2 H2 (R14)